Article 2730 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!rill.news.pipex.net!pipex!uunet!in2.uu.net!204.171.44.51!scramble.lm.com!news.psc.edu!nntp.sei.cmu.edu!news.sei.cmu.edu!gjp From: gjp@sei.cmu.edu (George Pandelios) Newsgroups: alt.sys.pdp10 Subject: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 3 Apr 1997 22:21:41 GMT Organization: The Software Engineering Institute Lines: 36 Distribution: world Message-ID: <5i1ahl$c6i@news.sei.cmu.edu> NNTP-Posting-Host: ts5d.sei.cmu.edu Xref: nntp1.ba.best.com alt.sys.pdp10:2730 Folks, I would be interested in any accounts or insights into what caused Digital to cancel the 36-bit successor to the KL10 processor. This was announced in May 1983 at Spring DECUS. At the time, I was a senior software specialist with Digital and the primary -10 guy for the Pittsburgh office. I also supported a lot of VAX/VMS systems as well. I worked for Digital Software Services from 1980-85. I am specifically interested in: - the primary cause of the cancellation (technical or business basis, or both?) - any discussions or debates over the decision within LCG - estimates of impact to existing customers made before the decision - actual fallout from the decision (what customers were lost as a result) If you would feel more comfortable sending e-mail, please do so. Thanks, George =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= George J. Pandelios Senior Member of Technical Staff Software Engineering Institute Internet: gjp@sei.cmu.edu 4500 Fifth Avenue Voice: (412) 268-7186 Pittsburgh, PA 15213 FAX: (412) 268-5758 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Disclaimer: These opinions are my own and do not reflect those of the Software Engineering Institute, its sponsors, customers, clients, affiliates, or Carnegie Mellon University. In fact, any resemblence of these opinions to any individual, living or dead, fictional or real, is purely coincidental. So there. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Article 2735 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!ais.net!uunet!in2.uu.net!199.232.56.18!news.ultranet.com!not-for-mail From: "Carl R. Friend" Newsgroups: alt.sys.pdp10,alt.folklore.computers Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: Mon, 07 Apr 1997 22:20:02 -0400 Organization: as little as possible! Lines: 41 Message-ID: <3349AB52.1A1FAD84@swec.com> References: <5i1ahl$c6i@news.sei.cmu.edu> <5i9k80$jh5@shell3.ba.best.com> <5ib4lg$ap4@tor-nn1-hb0.netcom.ca> NNTP-Posting-Host: zephyr.ultranet.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 2.0 (X11; I; Linux 1.2.1 i586) Xref: nntp1.ba.best.com alt.sys.pdp10:2735 alt.folklore.computers:83378 John Savard wrote in article Nr. <5ib4lg$ap4@tor-nn1-hb0.netcom.ca>: > > inwap@shell3.ba.best.com (Smith and O'Halloran) wrote, in part: > > >In article <5i1ahl$c6i@news.sei.cmu.edu>, > >George Pandelios wrote: > > >>I would be interested in any accounts or insights into what caused > >>Digital to cancel the 36-bit successor to the KL10 processor. > > >I have no details, but I can tell you the about the unsubstantiated > >and vicious rumors that LCG customers were hearing. > > I remember news stories at the time had a Digital spokesperson saying > that they discontinued that architecture because customers were > reluctant to migrate to the VAX, and Digital believed that the VAX was > its architecture for the future. This, of course, begs the question: "Where's the VAX now?" Given the architecture of the VAX, with its' rather bloated instruction set, there's no way that the raw speed could have approached that of a cleanly-designed -10-based implementation. The PDP-10 (and its' progenitor, the PDP-6) had one of the finest instruction sets known to man; it combined simplicity of implementation with elegance that remains unmatched to this day. While I might not be the biggest fan of the KL- implementation (it was a decent, workable, mid-life design), I think that a modern -10 design would be a machine to reckon with. I suspect it would lend itself to ASIC silicon rather well, and might achieve a good bit of parallelism in the process. (How many SMP -10s can we pack on a single die?) -- ______________________________________________________________________ | | | | Carl Richard Friend (UNIX Sysadmin) | West Boylston | | Minicomputer Collector / Enthusiast | Massachusetts, USA | | mailto:carl.friend@stoneweb.com | | | http://www.ultranet.com/~engelbrt/carl/museum | ICBM: N42:22 W71:47 | |________________________________________________|_____________________| Article 2738 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news-dc-10.sprintlink.net!news-dc-26.sprintlink.net!news-peer.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!sprint!newsfeed.internetmci.com!news.wco.com!not-for-mail From: "Henry W. Miller" Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: Wed, 09 Apr 1997 02:19:41 -0700 Organization: USBR Lines: 58 Message-ID: <334B5F2C.61C7@sacto.mp.usbr.gov> References: <5i1ahl$c6i@news.sei.cmu.edu> NNTP-Posting-Host: hmiller.mistic.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (Win95; I) Xref: nntp1.ba.best.com alt.sys.pdp10:2738 George, George Pandelios wrote: > > Folks, > > I would be interested in any accounts or insights into what caused Digital to > cancel the 36-bit successor to the KL10 processor. This was announced in May > 1983 at Spring DECUS. At the time, I was a senior software specialist with > Digital and the primary -10 guy for the Pittsburgh office. I also supported a > lot of VAX/VMS systems as well. I worked for Digital Software Services from > 1980-85. > > I am specifically interested in: > > - the primary cause of the cancellation (technical or business basis, or both?) > - any discussions or debates over the decision within LCG > - estimates of impact to existing customers made before the decision > - actual fallout from the decision (what customers were lost as a result) > > If you would feel more comfortable sending e-mail, please do so. > > Thanks, > > George > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > George J. Pandelios Senior Member of Technical Staff > Software Engineering Institute Internet: gjp@sei.cmu.edu > 4500 Fifth Avenue Voice: (412) 268-7186 > Pittsburgh, PA 15213 FAX: (412) 268-5758 > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > Disclaimer: These opinions are my own and do not reflect those of the > Software Engineering Institute, its sponsors, customers, > clients, affiliates, or Carnegie Mellon University. In fact, > any resemblence of these opinions to any individual, living > or dead, fictional or real, is purely coincidental. So there. > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= I can't offer too much in the way of internal knowledge of what was going on in DEC's collective head at the time of the cancellation, but I can offer this interesting sidebar: the day that the cancellation was announced was also the day that the Russians shot down flight KAL-007. I was working at SRI at the time in the Network Information Center. Up to this point in time, we had NEVER seen the network so congested. Of course, at that time, the backbone of the ARPANET was 56KB lines, and the military traffic used the same lines as did the limited commercial traffic and the university traffic. We could not quite figure out if the traffic was due to the cancellation of Jupiter, of the downing of KAL-007. From what I heard from a friend at DEC, the prototype for the Jupiter was very erratic. There could be as much as a 3:1 difference in it's performance during test benchmarks between runs. The prevailing theory as I recall was a design flaw in the pipeline queue that caused the pipeline to stall. -HWM Article 2744 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news-dc-10.sprintlink.net!news-dc-26.sprintlink.net!news.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!sprint!uunet!in1.uu.net!207.112.248.89!news.nap.net!news2.nap.net!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 12 Apr 1997 15:58:11 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 42 Message-ID: <5iobej$f20$1@kirin.wwa.com> References: <5i1ahl$c6i@news.sei.cmu.edu> <334B5F2C.61C7@sacto.mp.usbr.gov> <5ill76$cne$1@synthemesc.insync.net> NNTP-Posting-Host: pool3-007.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:2744 In article <5ill76$cne$1@synthemesc.insync.net>, spring1@insync.net says... > >I was with DEC at the time, and was involved in the decision process on >Jupiter. As I recall, there were technical problems with the design and >prototypes. Something to do with pipelined instruction performance (I'm no >expert in the design...). It would be interesting to know who "spring1" is. At the time I was Director of Systems Development for ADP's Network Services Division. We were one of DEC's National Accounts, and I sat in on many "non-disclosure" presentations on the Jupiter. I took copious notes and still have my notebooks from that era. ADPNS was LCG's largest customer. We had over 100 PDP-10's of one sort or another, all running TOPS-10. During 1982 I was involved in negotiations with LCG on the Jupiter implementation of TOPS-10. We (ADPNS) had done the portation to the KS-10 (2020), and as LCG was firmly committed to TOPS-20 they were once again looking to us to do the portation. The last notes I have on the subject are dated 9/21/82, a meeting with John Montesian, Don Turner, and Chuck Pickel. At that time Jupiter product announcement was scheduled for 12/82, first TOPS-20 customer ship for 6/84, and first TOPS-10 customer ship 6/85. BTW, in my notes the Jupiter is referred to as a 4050, although in earlier notes it's called the 2080. The surprising thing is that I've scanned my notes from there to the Las Vegas DECUS, which began 10/24/83. The term Jupiter never appears, which I find somewhat surprising. I guess the death announcement was such a big deal I didn't think I needed to take notes about it. Today I wish I had. I do recall that there were significant technical problems with the prototype(s), mostly imvolved with the EBOXes pipelining. DEC was looking at a complete redesign which would have pushed the product schedule out at least a year. They were concerned that by that time they would probably have been introducing an obsolete product into a market increasingly dominated by "power of two" architectures. That and the apparent success of the VAX line led to the decision to put the Jupiter project out of its misery. While it's easy today to take shots at the VAX, it was a huge success in the late '70s, early '80s. IBM was scared to death of it, so much so that the 4341 was being referred to internally as "the VAX killer". -- jeverett@wwa.com (John V. Everett) http://www.wwa.com/~jeverett Article 2747 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!news.nacamar.de!news.apfel.de!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!europa.clark.net!newsfeeds.sol.net!ix.netcom.com!pfarrell From: pfarrell@netcom.com (Pat Farrell) Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Message-ID: Sender: pfarrell@netcom17.netcom.com Organization: Netcom On-Line Services X-Newsreader: Forte Free Agent 1.0.82 References: <5i1ahl$c6i@news.sei.cmu.edu> <334B5F2C.61C7@sacto.mp.usbr.gov> <5ill76$cne$1@synthemesc.insync.net> <5iobej$f20$1@kirin.wwa.com> Date: Sun, 13 Apr 1997 21:31:33 GMT Lines: 71 Xref: nntp1.ba.best.com alt.sys.pdp10:2747 jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: > . At the time I was Director >of Systems Development for ADP's Network Services Division. We were one of >DEC's National Accounts, and I sat in on many "non-disclosure" presentations >on the Jupiter. At that time, I was manager of DEC operating systems development and Technical Services at American Management Systems. We were smaller than ADP or Compuserv, but not tiny. We had six KLs running Tops20, monitor sources and more than a thousand incomming modems.... >I do recall that there were significant technical problems with the >prototype(s), mostly imvolved with the EBOXes pipelining. I too heard early discussions of serious pipelinging performance problems. We were nearly desparate for a four times performance increase, we were turning away customers who wanted one huge machine supporting large commercial database work. And the HSC70 wasn't the equivalent to a mainframe channel as advertised. Given today's knowledge of the sensitivity of heavily pipelined processors to optimizing compilers, combined with the LCG community's love of Macro-10/20 programs (All of the OS, nearly all the CUSPs, languages except for Fortran, and the ever popular System 1022). it isn't even clear that a multi-stage heavily pipelined processor running typical 10/20 program mixes could have delivered much without all sorts of pipeline stalls. [no modern load/store RISC machine attempts to do KL effective address calculations, etc.]. > That and the apparent success of the VAX line >led to the decision to put the Jupiter project out of its misery. I think a major problem with LCG was the success of the larger Vaxen. The Vax community was not nearly as hard to support as those of us in LCG. They would put up with whatever DEC sold. There was a lot of internal politics inside DEC at the time, so much so that it was obvious to outsiders. There were also a number of secondary issues, which may have combined to kill LCG. These included: 1) KO's reported death fear of competing too closely with IBM. As LCG machines got bigger and 438x systems (with VM/CMS) got cheaper, the difference were no longer quite so clear. [Not that I think a 438x with VM/CMS is even in the neighborhood of TOPS-20] 2) limited heavy duty resource availability to do design work. Not long after the Jupiter was killed, while in MR01, I was told that the circuit simulations for both the Jupiter and Vax8600 were taking much more resources (KL time?) than earlier boxes, so part of the rationale for killing the Jupiter was that it would free modeling cycles for the 8600 and followon machines. 3) inability to develop a TP monitor for heavy duty transaction processing. I know there were several efforts within DEC to build something to remove the converstational nature of Tops-10 and Tops-20 jobs, so that hundreds and thousands of terminals could execute the same job streams. At least one product was fairly far through engineering before it was killed, at least in part to avoid #1 above. At one point, we were offered that product, complete with sources, if we'd just run it commercially [I know, why ruin a nice OS by adding CICS to it, but that is what at least a portion of the large commercial shops wanted....] . Pat Article 2748 of alt.sys.pdp10: Path: nntp1.ba.best.com!shell3.ba.best.com!not-for-mail From: inwap@shell3.ba.best.com (Smith and O'Halloran) Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 13 Apr 1997 18:30:28 -0700 Organization: Chez INWAP (people, computers, cats) Lines: 26 Message-ID: <5is1bk$63u@shell3.ba.best.com> References: <5i1ahl$c6i@news.sei.cmu.edu> <334B5F2C.61C7@sacto.mp.usbr.gov> <5ill76$cne$1@synthemesc.insync.net> NNTP-Posting-Host: shell3.ba.best.com Xref: nntp1.ba.best.com alt.sys.pdp10:2748 >I remember the announcement to customers at DECUS in St Louis, and the >subsequent migration strategy presentations and discussions ath the following >DECUS in Las Vegas. Somewhere in my boxes of stuff, I have micro-cassettes (from a dictation recorder) of the Las Vegas sessions. I really must find them, and get the exact wording of Ralph Gorin's famous remark. It was in a room on the top floor of a hotel, when DEC was telling us that two VAX-8600's and a VAX-780 had the combined computing power of a single KL. And that the problems they had in clustering that much CPU power together would be fixed by the time VAX-clusters shipped. Ralph said something like: What your asking is for us to jump out this window, on the assurances that the holes in the net below will be mended by the time we get there. This would have been the perfect opportunity for Systems Concepts to take orders for their "Mars" 36-bit processor, but the head of the company was a perfectionist, and would not announce the product until the hardware was prefect. He missed the window of opportunity. -Joe -- INWAP.COM is Joe and Sally Smith, John and Chris O'Halloran and our cats See http://www.inwap.com/ for "ReBoot", PDP-10, and Clan MacLeod. Article 2739 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!news-dc-10.sprintlink.net!news-dc-26.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!sprint!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!ais.net!uunet!in1.uu.net!192.67.173.4!news.decus.org!eisner!walsh_b From: walsh_b@eisner.decus.org (Brian Walsh) Subject: BOCES NCODE Etc. X-Nntp-Posting-User: WALSH_B Lines: 39 Organization: DECUServe Message-ID: <1997Apr10.141753.1@eisner> X-Trace: 860696301/13015 X-Nntp-Posting-Host: eisner.decus.org Date: Thu, 10 Apr 1997 18:17:53 GMT Xref: nntp1.ba.best.com alt.sys.pdp10:2739 In an archive of this newsgroup from last year I came across a thread about DECsystem-10 timesharing systems on Long Island. The thread referred to LIRICS, NCODE, and NAG. Since I was one of the original staff of NCODE, I thought I might share a little history and also maybe find someone from my past at the same time. In 1975, the Nassau County Board of Cooperative Educational Services obtained their own KA-10 (R109 I think) to offer the same sort of service that LIRICS was offering in Suffolk county. At the time I was a night-shift computer operator for an IBM 360/30 in the same room as the DEC-10 was being installed into. This was located in Westbury at the BOCES County Center. Being bored with loading cards into the 360, I started to mess around with the DEC-10 which was running overnight disk diagnostics. I really messed things up and was almost fired for it. But instead they offered me the job of system programmer to learn from and take over from the DEC resident, Pat Sweeney. At the same time, BOCES hired Bob Liquori and he and I became NCODE. We worked out of two desks in the telephone equipment room and supported users at many Nassau County High Schools. After a few years NCODE obtained 2 PDP-11/70s running RSTS/E and these replaced the -10 for educational computing. The -10 then became abusiness applications platform (accounting, etc). NCODE grew and I stayed with it until 1979 when I went to work for DEC in Colorado Springs as one of the first members of the Telephone Support Center on the RSTS/E support team. I could say lots more but I'm not even sure anyone cares. Brian Article 2741 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!news-dc-10.sprintlink.net!news-dc-26.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!sprint!uunet!in2.uu.net!192.67.173.4!news.decus.org!eisner!walsh_b From: walsh_b@eisner.decus.org (Brian Walsh) Subject: Re: BOCES NCODE Etc. X-Nntp-Posting-User: WALSH_B Lines: 46 Organization: DECUServe Message-ID: <1997Apr11.095523.1@eisner> References: <1997Apr10.141753.1@eisner> <860749091.10200.2@balti.tnx.djmarkets.co.uk> X-Trace: 860766937/10713 X-Nntp-Posting-Host: eisner.decus.org Date: Fri, 11 Apr 1997 13:55:23 GMT Xref: nntp1.ba.best.com alt.sys.pdp10:2741 In article <860749091.10200.2@balti.tnx.djmarkets.co.uk>, cbh@bankersnet.co.uk (Chris Hedley) writes: > > In article <1997Apr10.141753.1@eisner>, > walsh_b@eisner.decus.org (Brian Walsh) writes: >> I could say lots more but I'm not even sure anyone cares. > > Oh, go on. I'm always interested in stories related to old DEC kit. > > Chris. I remember learning MACRO-10 and a lot of Monitor internals because the system crashed so often due to hacking by our wonderful students! The story was that the LICUS meetings (I never went to one) were used to exchange tips on how to crash the NCODE DEC-10. This is ultimately why the instructional service (mostly BASIC programming) was moved to RSTS. We had to get the students off of the -10. Once I started working on the -10 I got a crash course by attending Assembly Language Programming, Advanced ALP, Monitor Structure, and Monitor internals in Marlboro. I also learned a tremendous amount from Pat Sweeny whose initial advice to me was to get listings of the entire monitor, "You'll need them", he said and he was right. A couple of the courses I tookk in MR were taught be Dave Cressey (Hi, Dave) who I recently ran into when he did some consulting for my present employer. I still have my "diplomas" from those courses. The NCODE -10 was a refurbished KA-10 with 4 MA-10s, the PDP-11/40-based front end (I can't think of the name), TU10, and a couple of RP02s. If I remember correctly, we couldn't boot directly from disk at first and had to use paper tape. Eventually the system was upgraded to boot directly from disk. We had something like 20 dial-in lines, telco DAAs mounted on the wall, that over time were replaced by dedicated circuits to each High School. One of the biggest advances in those years was the switch from teletypes to LA-36s. Video terminals were unheard of. Everything was hardcopy. I was very good at Teco on a hardcopy terminal. It wasn't until I went to work for DEC that I worked on my first VT52. Well, I need to get to work. Regards, Brian Article 2752 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news.sprintlink.net!sprint!newsfeed.internetmci.com!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 14 Apr 1997 18:03:06 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 25 Message-ID: <5itrgq$or0$1@kirin.wwa.com> References: <5i1ahl$c6i@news.sei.cmu.edu> <334B5F2C.61C7@sacto.mp.usbr.gov> <5ill76$cne$1@synthemesc.insync.net> <5iobej$f20$1@kirin.wwa.com> <5is1ln$6o4@shell3.ba.best.com> NNTP-Posting-Host: pool2-049.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:2752 In article <5is1ln$6o4@shell3.ba.best.com>, inwap@shell3.ba.best.com says... > >In article <5iobej$f20$1@kirin.wwa.com>, >John Everett wrote: >>In article <5ill76$cne$1@synthemesc.insync.net>, spring1@insync.net says... >>>I was with DEC at the time, and was involved in the decision process on >>>Jupiter. As I recall, there were technical problems with the design and >>>prototypes. Something to do with pipelined instruction performance (I'm no >>>expert in the design...). >> >>It would be interesting to know who "spring1" is. > >finger spring1@insync.net >Login Name TTY Idle When Where >spring1 Dave Doxey < . . . . > > Ah yes, the same Dave Doxey who on October 27, 1983 gave a major part of DEC's presentation at the 10/20/VAX Migration User Panel session during the Las Vegas DECUS. My notes have him as "DEC Internal MIS". -- jeverett@wwa.com (John V. Everett) http://www.wwa.com/~jeverett Article 2755 of alt.sys.pdp10: Path: nntp1.ba.best.com!shell3.ba.best.com!not-for-mail From: inwap@shell3.ba.best.com (Smith and O'Halloran) Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 14 Apr 1997 18:52:28 -0700 Organization: Chez INWAP (people, computers, cats) Lines: 33 Message-ID: <5iun0s$oh4@shell3.ba.best.com> References: <5i1ahl$c6i@news.sei.cmu.edu> <334B5F2C.61C7@sacto.mp.usbr.gov> <5ill76$cne$1@synthemesc.insync.net> <5is1bk$63u@shell3.ba.best.com> NNTP-Posting-Host: shell3.ba.best.com Xref: nntp1.ba.best.com alt.sys.pdp10:2755 In article <5is1bk$63u@shell3.ba.best.com>, Smith and O'Halloran wrote: >Somewhere in my boxes of stuff, I have micro-cassettes (from a dictation >recorder) of the Las Vegas sessions. I really must find them, and get the >exact wording of Ralph Gorin's famous remark. Found it. From audio tape 83267-3300 "Technical Comparison of TOPS-10/20 to VAX/VMS". DECUS in Las Vegas, October 1983. [This excerpt taken from approximately halfway into side 1 of the tape.] Ralph Gorin, Stanford University. Well, I've listened carefully to the "Integration Plan", as such, and alas, I'm unimpressed. The Integeration Plan, so far as I understand it, is: I'm supposed to walk to that window and jump out, on the assurance that the holes in that are in the net below will be fixed by the time I get there. ... I am looking for an assurance that Digital will NOT do something And the thing that I want you to not do is: I want you to promise to not obstruct our efforts to find third party manufacturers who will build us hardware so that we can run the software. ... want you to unbundle everything. Which is to say, license to all those who want it, all the software packages that you now posses. [Both comments received wild applause from the audience.] -- INWAP.COM is Joe and Sally Smith, John and Chris O'Halloran and our cats See http://www.inwap.com/ for "ReBoot", PDP-10, and Clan MacLeod. Article 2760 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!sprint!news-pull.sprintlink.net!news.sprintlink.net!gatech!news.niehs.nih.gov!not-for-mail From: ed@pigdog.niehs.nih.gov (Edward C. Bailey) Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: 17 Apr 1997 16:06:26 -0400 Organization: NIEHS Lines: 46 Message-ID: References: <5i1ahl$c6i@news.sei.cmu.edu> <5i9k80$jh5@shell3.ba.best.com> NNTP-Posting-Host: pigdog.niehs.nih.gov X-Newsreader: Gnus v5.3/Emacs 19.34 Xref: nntp1.ba.best.com alt.sys.pdp10:2760 > In article <5i1ahl$c6i@news.sei.cmu.edu>, > George Pandelios wrote: >I would be interested in any accounts or insights into what caused Digital to >cancel the 36-bit successor to the KL10 processor. Well, back in '81-82, I answered SPRs for galaxy on the '10s and '20s. Our group's offices were in the same building as the Jupiter and Venus (aka 8600) groups; I had to walk by them to go to the cafeteria. I didn't have any close ties with either group, but I was in a position to hear things. What did I hear? Lots of the crazy rumors that we've all heard over the years. But the truth as near as I could tell was that Jupiter would have been a fractional improvement in speed over a KL at a cost way out of proportion to the performance improvement. Contrast this with Venus (which was being developed at roughly the same time) -- a good solid performance bump for a reasonable amount of money. It was simply a more cost-effective solution, and the powers-that-be knew customers would compare the two systems, and that Jupiter would've only been a viable product for a very small portion of DEC's 36-bit customer base. The fact that VAX was DEC's most recent architecture certainly didn't hurt Venus, either. When they gathered the Technical Support Group to tell us that Jupiter was being canceled, people started jumping ship -- some to the SuVAX (VAXstation prototype) project across the hall, others left the company entirely, and one or two, I seem to recall, actually ended up getting into VMS Engineering. However, there was still a solid core of individuals in the 36-bit engineering and support organizations that remained for quite some time. Probably the biggest tragedy of that whole era was that a lot of good people and a lot of good technology went down the drain with the 36-bit systems. As I recall, the attitudes between the VMS and LCG organizations were fairly antagonistic at the time. I knew it was time to move on when I was eating lunch in the BU cafeteria and a customer from the U.S. Navy noticed the sailboat on my badge. Almost in tears, he begged me to tell him what I knew about Jupiter, and that the project he'd been working on for the past five years would be down the tubes unless he could get a faster 36-bit processor soon. I wanted to tell him to get his resume current, but I couldn't. I hated that... Ed -- Ed Bailey, bailey@niehs.nih.gov Article 2766 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news.sprintlink.net!sprint!uunet!in2.uu.net!uucp2.uu.net!world!bzs From: bzs@world.std.com (Barry Shein) Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? In-Reply-To: ed@pigdog.niehs.nih.gov's message of 17 Apr 1997 16:06:26 -0400 Message-ID: Sender: bzs@world.std.com Organization: The World @ Software Tool & Die X-Newsreader: Gnus v5.1 References: <5i1ahl$c6i@news.sei.cmu.edu> <5i9k80$jh5@shell3.ba.best.com> Date: Fri, 25 Apr 1997 04:12:00 GMT Lines: 24 Xref: nntp1.ba.best.com alt.sys.pdp10:2766 I was on the fast path for a Jupiter (Boston U), having taken delivery of a brand new 2060 in June 1983, DEC came over that summer to pitch the Jupiter to us and how great it was going to be. I think I remember speaking to Roseanne Giordano the day after announcement, on the phone, it was more of a lost opportunity for DEC than us since BU spent most of its money on IBM mainframes, the academic side went on in other directions, Sun etc. Wasn't there some story that the senior CPU architecture designer for Jupiter got deathly ill a coupla months before deadline, hepatitis or something like that, and a more junior fellow took his place. The latter had all these different ideas he felt had not been given sufficient shrift by his now out of the way superior, made a lot of changes to the design, and it just didn't work so there was your "performance problem". I dunno, maybe it was an apocryphal story. -- -Barry Shein Software Tool & Die | bzs@world.std.com | http://www.std.com Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD Article 2770 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-out.communique.net!communique!news.ultranet.com!d14 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: Mon, 28 Apr 97 13:26:39 GMT Organization: UltraNet Communications, Inc. Lines: 42 Message-ID: <5k27ub$97k$1@decius.ultra.net> References: <5i1ahl$c6i@news.sei.cmu.edu> <5i9k80$jh5@shell3.ba.best.com> <33647F2E.10EE@sacto.mp.usbr.gov> NNTP-Posting-Host: d14.dial-16.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:2770 In article <33647F2E.10EE@sacto.mp.usbr.gov>, "Henry W. Miller" wrote: > >Barry Shein wrote: >> >> I was on the fast path for a Jupiter (Boston U), having taken delivery >> of a brand new 2060 in June 1983, DEC came over that summer to pitch >> the Jupiter to us and how great it was going to be. I think I remember >> speaking to Roseanne Giordano the day after announcement, on the >> phone, it was more of a lost opportunity for DEC than us since BU >> spent most of its money on IBM mainframes, the academic side went on >> in other directions, Sun etc. >> > >Barry, > >> Wasn't there some story that the senior CPU architecture designer for >> Jupiter got deathly ill a coupla months before deadline, hepatitis or >> something like that, and a more junior fellow took his place. The >> latter had all these different ideas he felt had not been given >> sufficient shrift by his now out of the way superior, made a lot of >> changes to the design, and it just didn't work so there was your >> "performance problem". >> >> I dunno, maybe it was an apocryphal story. >> > > FWIW, now that you reminded me of this, I do recall hearing something >similar from a friend at DEC during that peroid. > All right! THAT'S ENOUGH! AFAIK there exists only one person left alive that can tell the real story and s/he doesn't seem to be willing to tell it I had not realized that this was the flavor of the speculations back then (I was not allowed to attend DECUS after the announcement that 6.04 was going to be the last TOPS10 monitor). No wonder there was such angst when the Jupiter cancellation was done! I, for one, was glad that it happened because it gave new life to the software (TOPS10 and TOPS20) side of the PDP-10 product line. /BAH - subtract twenty for e-mail Article 2783 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!noos.hooked.net!news.hooked.net!decwrl!pa.dec.com!nntpd.lkg.dec.com!leggy.zk3.dec.com!orb!news.ultranet.com!d17 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Accounts of the death of the KL-10 (aka Jupiter, 2080)? Date: Thu, 01 May 97 13:09:24 GMT Organization: UltraNet Communications, Inc. Lines: 31 Message-ID: <5ka42r$j3o$1@decius.ultra.net> References: <5i1ahl$c6i@news.sei.cmu.edu> <5k1frl$2ej$1@news.ece.nwu.edu> NNTP-Posting-Host: d17.dial-19.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:2783 This is an apology for yelling in a post I made a few days ago about the above subject. [Note to HWM: I was _not_ yelling at you; it just happened that it was your post that was on my screen when I lost my temper. So I owe you personal apology if you thought my post was directed at you. I assure you, it was not.] I have my own speculations about why Jupiter was cancelled. However, I do not plan to voice them here because they are speculations. JMF and TW did not tell me what was going on at that time. If they (especially Jim) wanted to keep the facts of that situation private, I am going to respect their decision. Now, the reason that I was happy the day the announcement happened was because I knew that the money in the PDP-10 budgets would be directed into Software Engineering's pot rather than in [what seemed at the time] the black hole of the Jupiter project. TOPS10 was getting very little funding and we were hurting. I am not speaking for TOPS20 because that wasn't where I worked. Because of the new funding that Software Engineering got after the Jupiter cancellation, we were able to hire _more_ people. A few of the really good ones that come to mind are Carl, Dawn and Bill. We got funding for implementing such things as DECnet, new peripherals and other things that I can't remember. If the Jupiter project had been allowed to continue the way it was going, none of this stuff would have been done. Again, my apologies for yelling; for those of you who know me, I did edit out a lot of blue material before I clicked the send icon [smiling emoticon here]. /BAH - subtract twenty for e-mail Article 2864 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!newsfeed.internetmci.com!news.wco.com!not-for-mail From: "Henry W. Miller" Newsgroups: alt.sys.pdp10 Subject: Happy Tony Wachs Day!!! Date: Fri, 06 Jun 1997 02:50:45 -0700 Organization: USBR Lines: 7 Message-ID: <3397DD75.1DBE@mp.usbr.gov> NNTP-Posting-Host: hmiller.mistic.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (Win95; I) Xref: nntp1.ba.best.com alt.sys.pdp10:2864 Hi, all, I remember reading in some ancient PDP-10 trivia-lore, that Tony Wachs started work at DEC on 6/6/66, and that his last day at DEC was also June 6th, some year. -HWM Article 2867 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newsfeed.internetmci.com!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: Happy Tony Wachs Day!!! Date: 7 Jun 1997 00:16:28 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 20 Message-ID: <5na98s$b2r$1@kirin.wwa.com> References: <3397DD75.1DBE@mp.usbr.gov> NNTP-Posting-Host: pool1-043.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.9 (Released Version) (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:2867 In article <3397DD75.1DBE@mp.usbr.gov>, henrym@mp.usbr.gov says... > >Hi, all, > > I remember reading in some ancient PDP-10 trivia-lore, that >Tony Wachs started work at DEC on 6/6/66, and that his last day at >DEC was also June 6th, some year. I'm not sure of the exact date, but you're in the ballpark. Hell, I can't even remember the date I started at DEC, but it was in the summer of '66. Tony had only been there a month or two. He was my office mate for the first year or so we worked at DEC. He was my close friend much longer. We both lived in Holliston at that time and regularly car-pooled the 20 or so miles to Maynard. We kind of lost track of each other after Tony left DEC (I had left years earlier). I could fill many screens with TW stories, but I'm not sure anyone would be interested. I was really saddened when I learned of his death. I'm glad he's still remembered in this forum. -- jeverett@wwa.com (John V. Everett) http://www.wwa.com/~jeverett Article 2915 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!newshub2.home.com!news.home.com!howland.erols.net!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d8 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Fri, 04 Jul 97 11:32:15 GMT Organization: UltraNet Communications, Inc. Lines: 47 Message-ID: <5pimve$es$2@decius.ultra.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <33b88c35.14236771@news.msy.bellsouth.net> <33BA6787.167E@ncsa.uiuc.edu> <6vtep5.b5.ln@dass.Prestel.Co.UK> <5pfvrg$goj$1@decius.ultra.net> <33bbcafd.3174005@news.vip.net> NNTP-Posting-Host: d8.dial-14.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.folklore.computers:88924 alt.sys.pdp10:2915 In article <33bbcafd.3174005@news.vip.net>, genew@vip.net (Gene Wirchenko) wrote: >jmfbahxx@ma.ultranet.com wrote: > >>In article <6vtep5.b5.ln@dass.Prestel.Co.UK>, >> brenc@dass.Prestel.Co.UK (Ben Clifford) wrote: >>>Alan Shutko wrote: >>> >>>: Bill> Any properly configured machine with an unlimited power supply >>>: Bill> will run for years (or should). >>> >>>: How many PCs have hot-swappable CPUs? >>> >>>It might be possible with the dual/quad processor machines... >>>unload all the tasks from one processor onto the others, take out >>>the CPU, put in a new one and load it back up. I expect if it can >>>be done, the linux people will do it :-) >>> >>It was done in 1980'ish [can't remember the ship date) by TOPS10 V7.01 >>which was an operating system that ran on PDP-10s manufactured by DEC. Not >>only could the OS allow one to "hot-swap" hardware (we never used that >>term), we often split our systems into two separate systems for >>stand-alone, F/S maintenance, etc. Our systems were the development >>systems for the TOPS10 Monitor Group (where monitor is defined as that >>piece of software that monitored timesharing. On our site we had 3 CPUs >>that could be used. We had one customer who ran 5 CPUs. > > This sounds interesting. Could you please post some more details >or some pointers? > >>/BAH > >Sincerely, > >Gene Wirchenko > >C Pronunciation Guide: > y=x++; "wye equals ex plus plus semicolon" > x=x++; "ex equals ex doublecross semicolon" JMF was one of the architects of the implementation and wrote a paper about the techniques he used. He also did SMP on a Unix OS but I don't know any of those details. I don't know where this paper resides so I'm posting to a group that is read by the keepers of such lore. /BAH Article 2919 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!howland.erols.net!news2.digex.net!digex.net!not-for-mail From: doug@ss1.digex.net (Doug Humphrey) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: 5 Jul 1997 00:36:24 -0400 Organization: Express Access Online Communications USA: 800-969-9090 Lines: 32 Message-ID: <5pkj08$b2c@ss1.digex.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pfvrg$goj$1@decius.ultra.net> <33bbcafd.3174005@news.vip.net> <5pimve$es$2@decius.ultra.net> NNTP-Posting-Host: ss1.digex.net Xref: nntp1.ba.best.com alt.folklore.computers:88949 alt.sys.pdp10:2919 Wow... I haven't touched on this subject in a while... "mainframes" and "pcs" in general have a very large thing seperating them. PCs tend to focus on processor acrhitecture, while mainframes focused more on "systems architecture" of which processor architecture is an important part, but just a part. PCs are starting to get into system architecture a little bit, but since they generally are starting from scratch, and many of their designers are not familiar with the lessons provided by the history of mainframes, they are in many cases reinventing the wheel (note that most of the "good" things in windows/NT are from DEC mainframe land, and most of the good things that were taken OUT of NT by the microserfs were taken out because they didn't really understand them...) Because of this, people are always astounded that a 1 MIP KL10 was able to do the work it was (it did it because it had a lot of help from the rest of the system architectual elements) and are quite often disapointed at the bottlenecks that they run into in a 100 or 200 MIP "pc" where the CPU is like lightning, but it can't overcome the shortcommings of the poor systems architecture that it has been inserted in (or in many cases the poor ways that the OS it is running handles some situations). I think that a lot of good knowledge has been lost in terms of systems design. The 10 series left things on the floor when the VAX people refused, and then only grudgingly, took some of the righteous things from the 10, and in general I think that the PC world would benefit form looking at some of the ways that it was done. Faster CPUs are not everything... meta X Soap-box-mode-off Article 2921 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: 6 Jul 1997 12:48:33 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 21 Message-ID: <5po471$lsa$1@kirin.wwa.com> References: <01bc8523$84da9b80$e12185c2@rashid1> <33b88c35.14236771@news.msy.bellsouth.net> <33BA6787.167E@ncsa.uiuc.edu> <6vtep5.b5.ln@dass.Prestel.Co.UK> <5pfvrg$goj$1@decius.ultra.net> <33bbcafd.3174005@news.vip.net> <5pimve$es$2@decius.ultra.net> <33BCE3A2.21EE86C@stoneweb.com> NNTP-Posting-Host: poolf1-055.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.folklore.computers:88996 alt.sys.pdp10:2921 In article <33BCE3A2.21EE86C@stoneweb.com>, carl.friend@stoneweb.com says... > > Joe Smith keeps a copy of "TOPS-10 Symmetric Multiprocessing" by >Allan B. Wilson and James M. Flemming Copyright (c) 1989, 1991, on his >excellent PDP-10 page, "http://www.inwap.com/pdp10/", at URL: >"http://www.inwap.com/pdp10/paper-smp.txt" I was an operating systems programmer at DEC from 1966 to 1974, the last five years in the PDP-10 Monitor (TOPS-10) group. After leaving DEC I remained close to the -10 as a customer until 1988, thus thought I knew all the key people involved in TOPS-10. Since I started monitoring this group a few years ago the name Allan B. Wilson has appeared a couple of times, and each time I thought, "Who the hell is Allan B. Wilson?" Last night it finally hit me, Allan B. is Pete Wilson's real name! Can anyone confirm my epiphany? -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 2923 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d10 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Sun, 06 Jul 97 13:58:08 GMT Organization: UltraNet Communications, Inc. Lines: 31 Message-ID: <5po89h$uts$1@decius.ultra.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <33b88c35.14236771@news.msy.bellsouth.net> <33BA6787.167E@ncsa.uiuc.edu> <6vtep5.b5.ln@dass.Prestel.Co.UK> <5pfvrg$goj$1@decius.ultra.net> <33bbcafd.3174005@news.vip.net> <5pimve$es$2@decius.ultra.net> <33BCE3A2.21EE86C@stoneweb.com> <5po471$lsa$1@kirin.wwa.com> NNTP-Posting-Host: d10.dial-18.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.folklore.computers:88998 alt.sys.pdp10:2923 In article <5po471$lsa$1@kirin.wwa.com>, jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: >In article <33BCE3A2.21EE86C@stoneweb.com>, carl.friend@stoneweb.com says... >> >> Joe Smith keeps a copy of "TOPS-10 Symmetric Multiprocessing" by >>Allan B. Wilson and James M. Flemming Copyright (c) 1989, 1991, on his >>excellent PDP-10 page, "http://www.inwap.com/pdp10/", at URL: >>"http://www.inwap.com/pdp10/paper-smp.txt" > >I was an operating systems programmer at DEC from 1966 to 1974, the last five >years in the PDP-10 Monitor (TOPS-10) group. After leaving DEC I remained >close to the -10 as a customer until 1988, thus thought I knew all the key >people involved in TOPS-10. Since I started monitoring this group a few years >ago the name Allan B. Wilson has appeared a couple of times, and each time I >thought, "Who the hell is Allan B. Wilson?" Last night it finally hit me, >Allan B. is Pete Wilson's real name! Can anyone confirm my epiphany? > I can. They're two different people. Alan worked for Alan Titcomb (? I think that's the name) in marketing. I don't remember when he first appeared but I seem to recall that he did some special development work on the 1091. When he left DEC I think he went to Atlanta but I cannot recall the company. I don't remember what happened to Pete Wilson. /BAH - subtract twenty for e-mail Article 2920 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsfeed.dacom.co.kr!newsfeed.direct.ca!newsfeed.internetmci.com!howland.erols.net!news-peer.sprintlink.net!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d9 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Sat, 05 Jul 97 11:24:04 GMT Organization: UltraNet Communications, Inc. Lines: 26 Message-ID: <5plase$eek$1@decius.ultra.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <33b88c35.14236771@news.msy.bellsouth.net> <33BA6787.167E@ncsa.uiuc.edu> <6vtep5.b5.ln@dass.Prestel.Co.UK> <5pfvrg$goj$1@decius.ultra.net> <33BCE63F.59BE779A@stoneweb.com> NNTP-Posting-Host: d9.dial-11.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.folklore.computers:88956 alt.sys.pdp10:2920 In article <33BCE63F.59BE779A@stoneweb.com>, "Carl R. Friend" wrote: >jmfbahxx@ma.ultranet.com wrote in article nr. ><5pfvrg$goj$1@decius.ultra.net>: > >> It [PDP-10 SMP] was done in 1980'ish [can't remember the ship date) >> by TOPS10 V7.01 which was an operating system that ran on PDP-10s >> manufactured by DEC. Not only could the OS allow one to "hot-swap" >> hardware (we never used that term), we often split our systems into >> two separate systems for stand-alone, F/S maintenance, etc. [...] On >> our site we had 3 CPUs that could be used. We had one customer who >> ran 5 CPUs. > > Wow! I had the good fortune to configure a quad-KI at my first >employer which was a wonderful system. Who had the quint and what >machine types did they use? > The customer site was at Oak Ridge National Labs; they had KLs but had to use AMPEX memory. IIRC, JOBMAX was set to 500 and was filled daily. One of the pictures that we were going to scan is a system configuration of of that site. Since Thor stomped his way through this area the other night without bringing a tornado, could we think again about getting this stuff in bits before it's too late? /BAH - subtract twenty for e-mail or ship it to AOL Article 2929 of alt.sys.pdp10: Newsgroups: alt.folklore.computers,alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!newshub2.home.com!news.home.com!howland.erols.net!newsfeed.internetmci.com!in2.uu.net!uucp1.uu.net!world!bzs From: bzs@world.std.com (Barry Shein) Subject: Re: Why Mainframes? In-Reply-To: peter@taronga.com's message of 8 Jul 1997 07:48:35 -0500 Message-ID: Sender: bzs@world.std.com Organization: The World @ Software Tool & Die X-Newsreader: Gnus v5.1 References: <01bc8523$84da9b80$e12185c2@rashid1> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5ptcv3$6tf@bonkers.taronga.com> Date: Tue, 8 Jul 1997 16:58:43 GMT Lines: 43 Xref: nntp1.ba.best.com alt.folklore.computers:89089 alt.sys.pdp10:2929 A mainframe was typically a machine with separate I/O processors (in IBM parlance, "channels") which normally had independant paths to memory (no single bus.) For example, to perform disk I/O on a 370 you created a channel program in memory which the IOP executed directly (think of it as another CPU living off the main system's memory.) Besides all the obvious stuff ("get me this block") the IOP could do things such as search a range of tracks for a particular data key (match, high, low, etc.) Large 370 mainframes had 16 then later 256 separate channels (well, you bought as many as you could afford, but "up to...") These allowed true I/O parallelism, you could have a disk farm all lit up doing operations simultaneously, including transferring data to/from memory without contention (i.e., no serialization over a single bus.) In some sense all the other terms, mini/micro/pc, refer to just a progression of making a simple computer (single I/O bus) smaller and cheaper. I suppose the important distinction is that PCs center around a motherboard, whereas minis were typically a rack of boards (CPU board, memory board, serial I/O board) on a common backplane such as a Unibus. Anyhow, in this world of fascination with prices and CPU clock speeds it's not surprising that people have trouble understanding that mainframes were defined by their I/O speed and generally only required to have adequate ("balanced") CPU speed so that wasn't a choke point for the I/O. For example, I remember being awed at watching a 3090 (big IBM 370 architecture mainframe) simultaneously reading and/or writing 8 tape drives at what looked like fast rewind speed, 250ips, when my supposedly high-end minis would stop/start and stutter, maybe on a good day stream for short bursts, trying to turn one tape drive. I remember that mainframe could read or write a tape in about 2 minutes, several of them simultaneously, where the minis could easily take an hour, even with high-end tape drives (eg, TU78s.) -- -Barry Shein Software Tool & Die | bzs@world.std.com | http://www.std.com Purveyors to the Trade | Voice: 617-739-0202 | Login: 617-739-WRLD Article 2932 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsfeed.dacom.co.kr!newsfeed.internetmci.com!europa.clark.net!dispatch.news.demon.net!demon!mail2news.demon.co.uk!tnglwood.demon.co.uk!unclebob From: Robert Billing Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Tue, 08 Jul 97 16:55:53 GMT Message-ID: <868380953snz@tnglwood.demon.co.uk> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pimve$es$2@decius.ultra.net> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5psm6s$sbh$1@skat.usc.edu> Reply-To: unclebob@tnglwood.demon.co.uk X-Mail2News-User: unclebob@tnglwood.demon.co.uk X-Mail2News-Path: punt-2.mail.demon.net!tnglwood.demon.co.uk X-Newsreader: Demon Internet Simple News v1.29 Lines: 19 Xref: nntp1.ba.best.com alt.folklore.computers:89096 alt.sys.pdp10:2932 In article <5psm6s$sbh$1@skat.usc.edu> tli@skat.usc.edu "tli" writes: > I believe that they did some stuff in the I/O processor (IOP), which was a > programmable front end, not wholly unlike an IBM channel. In fact the terminal channel on the 370/165 in Cambridge was replaced with a PDP11 (an 11/10 IIRC) which did ascii-ebcidic, line editing and a few other odds & ends and allowed TCAM to be replaced with the much smaller & simpler parrot. CDC may well have done the same thing. -- I am Robert Billing, Christian, inventor, traveller, cook and animal lover, I live near 0:46W 51:22N. http://www.tnglwood.demon.co.uk/ "If ladies wish to change compartments during the journey, the staff must enable them to do so." LNER rule book, rule 161, 1933 edition. Article 2933 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsfeed.dacom.co.kr!newsfeed.internetmci.com!cpk-news-hub1.bbnplanet.com!cam-news-feed1.bbnplanet.com!news.bbnplanet.com!nntp.neu.edu!grapevine.lcs.mit.edu!bone.think.com!blanket.mitre.org!news.mitre.org!mwunix!jcmorris From: jcmorris@mwunix.mitre.org (Joe Morris) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: 8 Jul 1997 20:25:36 GMT Organization: The MITRE Corporation Lines: 47 Message-ID: <5pu7o0$ndc@top.mitre.org> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5ptcv3$6tf@bonkers.taronga.com> NNTP-Posting-Host: mwunix.mitre.org Xref: nntp1.ba.best.com alt.folklore.computers:89102 alt.sys.pdp10:2933 bzs@world.std.com (Barry Shein) writes: >For example, to perform disk I/O on a 370 you created a channel >program in memory which the IOP executed directly (think of it as >another CPU living off the main system's memory.) Besides all the >obvious stuff ("get me this block") the IOP could do things such as >search a range of tracks for a particular data key (match, high, low, >etc.) Large 370 mainframes had 16 then later 256 separate channels >(well, you bought as many as you could afford, but "up to...") These >allowed true I/O parallelism, you could have a disk farm all lit up >doing operations simultaneously, including transferring data to/from >memory without contention (i.e., no serialization over a single bus.) That was a function of the complexity (and therefore the price) of the systems. The low-end IBM S/360 machines, for example, had the channels built into the main box, and these channels were serviced by the CPU circuitry. This meant that whenever the channel had an event that required more than minimial processing, the active task in the computer was effectively suspended while the CPU cycles were used to process the I/O event. Such events included starting or ending any channel control word (a CCW, the opcode that directed the channel and device), as well as periodic loading and unloading of buffers as bytes moved between the I/O device and main memory. Such configurations were called "cycle-stealing channels". Except for performance, this was totally transparent to the operating system; it generally needed only to tell the channel logic to start an operation, and later responded to an interrupt telling it that the channel had finished. CCW handling and channel buffer management needed no assistance from the OS. The same design was used for the integrated control units in the low-end 4300-series IBM boxes, where you could have not only data channels, but device controllers built into the main computer box. Having a lot of high-speed (for that day) communications lines could place a quite heavy load on the CPU and cause significant reductions in throughput. Each system with these cycle-stealing control units came with a manual that stepped you through the calculations to see how bad it could be. The higher-end models of all of the IBM mainframe lines used channels and control units that had their own dedicated logic to control the operation, and required only an arbitration mechanism to mediate access to memory once the I/O operation was begun. Joe Morris Article 2934 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.224.117.11!news1.epix.net!not-for-mail From: "Julian Thomas" Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Tue, 08 Jul 97 19:34:59 -0500 Organization: epix Internet Services Lines: 22 Message-ID: <5puiv7$a5b$1@news1.epix.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5ptcv3$6tf@bonkers.taronga.com> NNTP-Posting-Host: tmbg-115ppp41.epix.net X-Newsreader: MR/2 Internet Cruiser Edition for OS/2 v1.29f b24 Xref: nntp1.ba.best.com alt.folklore.computers:89105 alt.sys.pdp10:2934 In , on 07/08/97 at 04:58 PM, bzs@world.std.com (Barry Shein) said: ::For example, to perform disk I/O on a 370 you created a channel program ::in memory which the IOP executed directly (think of it as another CPU ::living off the main system's memory.) Besides all the obvious stuff ::("get me this block") the IOP could do things such as search a range of ::tracks for a particular data key (match, high, low, etc.) One minor quibble with an otherwise excellent and accurate description of the IBM IO subsystems. The IOP was limited to the basic input and output transfers; searching tracks for key match was the function of the outboard disk control unit (yet another processor living on the other side of the io interface!). -- Julian Thomas (and/or if appropriate, Mary Jane Thomas) jt@epix.net In the beautiful Finger Lakes Wine Country of New York State! -------------------------------------------------- You said Windows was a Power Tool??? Article 2936 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!dispatch.news.demon.net!demon!mail2news.demon.co.uk!p850ug1.demon.co.uk!ard From: ard@p850ug1.demon.co.uk (Tony Duell) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Thu, 10 Jul 97 00:06:16 GMT Organization: P850 User Group Distribution: world Message-ID: <868493176snz@p850ug1.demon.co.uk> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pimve$es$2@decius.ultra.net> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5psm6s$sbh$1@skat.usc.edu> <868380953snz@tnglwood.demon.co.uk> Reply-To: ard@p850ug1.demon.co.uk X-Mail2News-User: ard@p850ug1.demon.co.uk X-Mail2News-Path: p850ug1.demon.co.uk X-Newsreader: Demon Internet Simple News v1.29 Lines: 37 Xref: nntp1.ba.best.com alt.folklore.computers:89163 alt.sys.pdp10:2936 In article <868380953snz@tnglwood.demon.co.uk> unclebob@tnglwood.demon.co.uk "Robert Billing" writes: > In article <5psm6s$sbh$1@skat.usc.edu> tli@skat.usc.edu "tli" writes: > > > I believe that they did some stuff in the I/O processor (IOP), which was a > > programmable front end, not wholly unlike an IBM channel. > > In fact the terminal channel on the 370/165 in Cambridge was replaced > with a PDP11 (an 11/10 IIRC) which did ascii-ebcidic, line editing and > a few other odds & ends and allowed TCAM to be replaced with the much > smaller & simpler parrot. I believe that the front-end to the later IBM 3084Q in Cambridge was also PDP11-based. I think it consisted of racks of PDP11/34's (which replaced 11/45's some years ago, but the system was essentiall unaltered apart from the change in CPU). Some of the 11/34's were stuffed with DZ11's and DJ11's that provided the async lines. They communicated with 'master' PDP11/34's using DMR11's and DMC11's. There were also DUP11's in there somewhere - maybe as interprocessor links, maybe as sync line inputs. The 2 master PDP11/34's talked to the IBM channel using a couple of DX11's (8 rows of flip-chip cards that linked the IBM channel to Unibus. Somewhat impressive to look at!). I am fairly sure about the model numbers of the cards used, but exactly how they were all interconnected is a mystery to me, alas, so the above is somewhat guesswork. > I am Robert Billing, Christian, inventor, traveller, cook and animal -tony Article 2941 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!su-news-feed4.bbnplanet.com!news.bbnplanet.com!enews.sgi.com!news.sgi.com!news.spies.com!not-for-mail From: eric@brouhaha.com (Eric Smith) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: 10 Jul 1997 14:25:57 -0400 Organization: Brouhaha Computer Mercenary Services Lines: 31 Message-ID: References: <01bc8523$84da9b80$e12185c2@rashid1> <5pfvrg$goj$1@decius.ultra.net> <33bbcafd.3174005@news.vip.net> <5pimve$es$2@decius.ultra.net> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5pu79l$vha$1@ausnews.austin.ibm.com> NNTP-Posting-Host: houhah.brouhaha.com In-reply-to: glass2@glass2.cv.lexington.ibm.com's message of 8 Jul 1997 20:17:57 GMT To: wa4qal@vnet.ibm.nospam.net X-Newsreader: Gnus v5.1 Xref: nntp1.ba.best.com alt.folklore.computers:89195 alt.sys.pdp10:2941 In article <5pu79l$vha$1@ausnews.austin.ibm.com> glass2@glass2.cv.lexington.ibm.com writes: > But, what about something like > the P/390 card, which implements the same instruction set > and architecture, but which plugs into a PC (and weighs less > than a pound)? Do we have a new category? The Micro-mainframe? Micromainframe(tm) is (or was) a trademark of Intel(tm), for the iAPX-432(tm). But they haven't used the trademark in years, and perhaps they aren't claiming it any more. All of the iAPX-432 systems I used weighed more than 50 kg, so by Andrew K Bressen's criteria they would have been minis. They were so slow and had such low I/O bandwidth that it's not clear why Intel thought that Micromainframe was a good name for them. On the other hand, they had a very intriguing object-oriented capability-based architecture. It was a fun machine to play with, even if it wasn't practical for anything. I would have removed alt.sys.pdp10 from the Newsgroups: header, except that I found out (just recently) that all of Intel's early iAPX-432 development tools ran on the DECsystem-10. The tools that they released or sold to customers ran on VAXen under BSD 4.1 and were written in Berkeley Pascal. Cheers, Eric -- ----------------------------------------------------------------------------- Don't blame me, I voted Libertarian. http://www.lp.org/ (800) 682-1776 ----------------------------------------------------------------------------- Article 2954 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!newshub2.home.com!news.home.com!howland.erols.net!infeed1.internetmci.com!newsfeed.internetmci.com!in2.uu.net!204.177.208.6!news.iwl.net!usenet From: Randall Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Fri, 11 Jul 1997 23:16:18 -0500 Organization: A poorly-installed InterNetNews site Lines: 23 Message-ID: <33C70512.3EB0@iwl.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <5pkj08$b2c@ss1.digex.net> <0q4ta6r06p.fsf@rorschach.hks.net> <5pto0i$1md@polo.demon.co.uk> NNTP-Posting-Host: a60.iwl.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01 (Win95; I) Xref: nntp1.ba.best.com alt.folklore.computers:89274 alt.sys.pdp10:2954 John Winters wrote: > > In article , > Mark van Walraven wrote: > >Andrew K Bressen wrote: {snip weights and dimensions of 1979 micro, mini and mainframe computers] > > From 'computing with mini computers' 1973; by Gruenberger & Babcock: CPU costs between 3K - 20K word length 12 or 16 bits usually 16 19X11X21 inches Internal operating speed is from 1 to 4 microseconds per full word addition Some offer microprogramming capability The current leaders, in terms of machines installed, were DEC Varian Data machines, HP, Microdata and General Automation. The book then goes on to describe programming and operation of the Varian 620L computer. Article 2973 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!not-for-mail From: "Carl R. Friend" Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Sun, 13 Jul 1997 19:07:40 -0400 Organization: as little as possible! Lines: 27 Message-ID: <33C95FBC.6E6C1B44@stoneweb.com> References: <01bc8523$84da9b80$e12185c2@rashid1> <19970711234045457209@n243-157.berlin.snafu.de> <5q9cg3$8nd@bonkers.taronga.com> <33C8D181.7125F907@stoneweb.com> NNTP-Posting-Host: zephyr.ultranet.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 2.0 (X11; I; Linux 2.0.29 i586) Xref: nntp1.ba.best.com alt.folklore.computers:89362 alt.sys.pdp10:2973 Eric S. Smith, in article nr. wrote: > > In article <33C8D181.7125F907@stoneweb.com>, > Carl R. Friend wrote: > > >definition far enough, the PDP-10 is a minicomputer (it, in native > > This point of view is supported by a book in our library which talks > about doing sciency stuff with "the PDP-11 minicomputer" if the 11 is > substantially similar to the 10. Other than a common manufacturer, the two share very, very, few similarities. The two have widely divergent bus structures, instruction sets, word sizes, and memory capacities. They are quite un-alike. The KL-10, though, did use a PDP-11/40 to spoon-feed its brains to it at power-on. (Sorry, KL-10 fans.) -- ______________________________________________________________________ | | | | Carl Richard Friend (UNIX Sysadmin) | West Boylston | | Minicomputer Collector / Enthusiast | Massachusetts, USA | | mailto:carl.friend@stoneweb.com | | | http://www.ultranet.com/~engelbrt/carl/museum | ICBM: N42:22 W71:47 | |________________________________________________|_____________________| Article 3032 of alt.sys.pdp10: Newsgroups: alt.folklore.computers,alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!ix.netcom.com!netcom13!alderson From: alderson@netcom13.netcom.com (Richard M. Alderson III) Subject: Re: TU78 In-Reply-To: fetrow@biostat.washington.edu's message of 21 Jul 1997 22:43:39 GMT Message-ID: Sender: alderson@netcom13.netcom.com Reply-To: alderson@netcom.com Organization: NETCOM On-line services References: <01bc8523$84da9b80$e12185c2@rashid1> <5qac8j$ub2$1@decius.ultra.net> <33D32FE1.52AC@mp.usbr.gov> <5r0omr$nah@nntp4.u.washington.edu> Date: Mon, 21 Jul 1997 23:19:13 GMT Lines: 23 Xref: nntp1.ba.best.com alt.folklore.computers:89725 alt.sys.pdp10:3032 In article <5r0omr$nah@nntp4.u.washington.edu> fetrow@biostat.washington.edu (David Fetrow) writes: >I always wondered why the local PDP-10 computer once upon a time was filled >with DEC this, DEC that plus 1 adaptor from Systems Concepts connected to a >long line of IBM tape drives. Ah, yes, the SA-10 interface adapter. Systems Concepts' first product. When they were still in the 36-bit computer business, they included an SA channel in the SC30M (and other members of their line), so the one I ran at Stanford had the StorageTek drives talked about in another thread, along with RP07 and RA81 disks. As a matter of fact, when they first delivered it, the MI (Massbuss Interface) channel was no yet available, so we had to use the SA channel to connect to some IBM 3380 single-density drives we happened to have in the room, so we could bet Tops-20 running on the thing. (Mike and Stew were much more at home with Tops-10.) -- Rich Alderson You know the sort of thing that you can find in any dictionary of a strange language, and which so excites the amateur philo- logists, itching to derive one tongue from another that they know better: a word that is nearly the same in form and meaning as the corresponding word in English, or Latin, or Hebrew, or what not. --J. R. R. Tolkien, alderson@netcom.com _The Notion Club Papers_ Article 3035 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d12 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: TU78 Date: Tue, 22 Jul 97 10:31:41 GMT Organization: UltraNet Communications, Inc. Lines: 29 Message-ID: <5r22b7$ipb$1@decius.ultra.net> References: <01bc8523$84da9b80$e12185c2@rashid1> <5qac8j$ub2$1@decius.ultra.net> <33D32FE1.52AC@mp.usbr.gov> <5r0omr$nah@nntp4.u.washington.edu> NNTP-Posting-Host: d12.dial-16.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.folklore.computers:89751 alt.sys.pdp10:3035 In article <5r0omr$nah@nntp4.u.washington.edu>, fetrow@biostat.washington.edu (David Fetrow) wrote: >>> >tape forward so that the unsuspecting software wouldn't know its place on the >>> >tape, the TU45 wouldn't even allow a tape to be loaded and generate that >>> >on-line interrupt. > > I always wondered why the local PDP-10 computer once upon a time >was filled with DEC this, DEC that plus 1 adaptor from Systems Concepts >connected to a long line of IBM tape drives. > And that's only from the hardware side. I remember one morning, after a particularly difficult session debugging his new magtape driver, TW glared at an offending magtape drive and declared that "God never meant for there to be magtapes." [chuckling emoticon here] We waited until we got out of TW's range before we laughed. Oh, yea...After developing/maintaining BACKUP, I agree. /BAH Article 3042 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!francini From: francini@ultranet.com (John Francini) Newsgroups: alt.folklore.computers,alt.sys.pdp10 Subject: Re: Why Mainframes? Date: Tue, 22 Jul 1997 23:44:12 -0400 Organization: Appletree Systems Lines: 67 Message-ID: References: <01bc8523$84da9b80$e12185c2@rashid1> <5pkj08$b2c@ss1.digex.net> <1997Jul10.122310.28825@indyvax.iupui.edu> <5qaed0$f0n$1@nntp2.ba.best.com> <5qn82p$h0k$1@nntp2.ba.best.com> <33D25DBA.A38@acm.org> <01bc9636$64ea0530$72a036cf@alpha> <869565631.25814.10.nnrp-2.c29fa409@news.demon.co.uk> NNTP-Posting-Host: d2.dial-1.nsh.nh.ultra.net Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Newsreader: Yet Another NewsWatcher 2.4.0 Xref: nntp1.ba.best.com alt.folklore.computers:89774 alt.sys.pdp10:3042 In article <869565631.25814.10.nnrp-2.c29fa409@news.demon.co.uk>, cbh@vodka.tnx.djmarkets.co.uk (Chris Hedley) wrote: >In article <01bc9636$64ea0530$72a036cf@alpha>, > "Eugene A. Pallat" writes: >> Performance is "in the eye of the beholder." If you just mean raw CPU >> speed, OK. Just remember that the big iron uses other computers for the >> I/O channels. > >The larger Vax machines used something called HSC-11s for its I/O >work, which, IIRC, were basically just renamed PDP-11s! > >Chris. Just to enlarge a bit on what Chris said. You mean the HSC-50, followed by the higher-performing HSC-70 and HSC-90, and the cost-reduced HSC-40. Yes, they indeed had/have PDP-11s in them, but the PDP didn't do the actual work of flinging the data around; dedicated microprocessors on controller cards that spoke directly to the disks and tapes did the work; the data was then moved to/from the CI bus using a different controller card. The controller cards plugged into a high-speed bus. Their overall activities were controlled by the RSX-11 derived code running in the 11, but it did NOT actually do the bulk of the work. It would never be up to it. The disks were Digital's own SDI-speaking disks: RA60/70/71/72/73/80/81/82/90/91/92 etc. It also supported TA-variants of the common Digital-repackaged tape drives, like the TA78, TA79, TA90, TA92, etc. The very latest descendants, the HSD-xx/HSJ-xx/HSZ-xx use a third-party controller card set (I forget the vendor), with a common version of firmware. The disks are now fast SCSI-2 and/or fast/wide SCSI-2. SCSI tape drives (DAT, DLT, 9-track, and (i think) 8MM) are supported. All three controller families support hardware striping, mirroring, and RAID 3/5. HSD-xx controllers use DSSI to speak to the host(s). Each "disk unit" presented to the hosts looks like a DSSI disk with dedicated controller. HSJ-xx controllers use MSCP to speak to the hosts over a CI bus. It looks to the hosts much like an HSC-xx. HSZ-xx controllers use fast/wide/differential SCSI to speak to the hosts. Each "disk unit" presented to the host looks like a SCSI device on that bus. There are also variants of the HSC series that use a fiber-optic interconnect between the controller and the hosts. I may have a couple of minor details wrong in the above descriptions, but this is generally what the lay of the Digital land is for "independent" disk controllers (i.e., ones not on a system's backplane bus). If I've got it wrong, please let me know! -- ---- +------------------------------------------------------------------------------+ | "I have come to the conclusion that one useless man is called a disgrace; | | that two or more are called a law firm; and that three or more become | | a Congress. And by God I have had _this_ Congress!" | | -- John Adams | +------------------------------------------------------------------------------+ Article 3101 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!gatech!huron.eel.ufl.edu!usenet.eel.ufl.edu!news.ultranet.com!d17 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: SCAN/WILD questions Date: Thu, 31 Jul 97 12:50:22 GMT Organization: UltraNet Communications, Inc. Lines: 74 Message-ID: <5rq1tt$gdr$1@decius.ultra.net> Reply-To: Elrond74@worldnet.att.net NNTP-Posting-Host: d17.dial-16.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3101 I received this mail message and I thought it was quite appropriate to put here. At 12:38 AM 7/30/97 -0700, you wrote: >jmfbahxx@ma.ultranet.com wrote: >>One of our roaring debates at that time was whether a CUSP update tape >>should contain only the new software, or a complete set of all CUSPs even >>though they hadn't changed. By that time, I believe that a new customer >>got to "install" around 20 CUSP Update Tapes in order to get all the latest >>and greatest of ships. This organization of software also (unfortunately) >>produced many flavors of versions of each CUSP (including SCAN and WILD >>which was very, very bad--caused all sorts of support problems).^^^^^^^ > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > >I guess that's why I noticed some odd SCAN/WILD behaviour from some of >the SYS:*.SAV (and OLD:*.SAV) programs... (we were on 7.01A at the time). I didn't know we even supported those with the advent of SMP; I am amazed. For oldtimers, remember SV2EXE (guess I may not...that doesn't look right). > >[and perhaps why even older .SHR/.LOW programs had no SCAN/WILD at all; > I take it that was a later invention, so the repercussions took a few > years to shake out....] Possible. > > >>As a side effect of remastering the CUSPs to only one tape, we were able >>to get the bricks laid for shipping/supporting only one version of >>SCAN/WILD. > > >Ah, yes, my real question is: > Where in the DECSystem-10 Manual Set did one find the documentation >on exactly how to make use of SCAN/WILD in a program? Chortle. Nowhere. Here's how I did my first one. I read CREFed listings of SCAN and WILD, and any other program that used them _properly_. DIRECT was an example. Then I tried to write the program. (Now JMF taught me to do this.) When I became so confused that I didn't know my ass from a hole in the ground, I asked an "expert", where expert is defined as someone who wrote a SCAN/WILD interface before me. Of course, we were fortunate because we could always ask the developer/maintainer (Hey, Bob!)[grinning emoticon here]. Just a note: Even though Jim managed to write an interface in user mode (FILDAE.MAC), he never, ever admitted it to anyone with the intention of avoiding the label of SCAN expert. > >[ As a lowly undergrad student in 1982 (and aspiring DEC-10 systems' >programmer), I could not get any help with this from our Computer Center >staff ...] Not surprising. Hmmm....Was the ability to use SCAN/WILD a rite of passage in the TOPS10 world? > >>Ahem....Sorry I didn't mean to get so windy, but once one starts >>remembering the good ol' days, one could wax on and on and on.... > >That's OK by me, I feel the same way about my TOPS-10 years (short 3 or 4 >that they were) [and I don't think it was simply due to "college daze", >but really more to _sane_software_design_ (combined with clever H/W >design)]! Amen. /BAH - subtract twenty for e-mail Article 3113 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!not-for-mail From: "Carl R. Friend" Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: Sat, 02 Aug 1997 10:21:41 -0400 Organization: as little as possible! Lines: 32 Message-ID: <33E34275.2AF7A3C8@stoneweb.com> References: <5rq1tt$gdr$1@decius.ultra.net> <5rv4mn$7bi$1@decius.ultra.net> NNTP-Posting-Host: zephyr.ultranet.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 2.0 (X11; I; Linux 2.0.29 i586) Xref: nntp1.ba.best.com alt.sys.pdp10:3113 jmfbahxx@ma.ultranet.com, in article nr.<5rv4mn$7bi$1@decius.ultra.net> wrote: > > This person's question about learning how to use TOPS10's software > started me thinking. We never really provided good instructions > [hmmm...I just did a pun :-)], I mean, documentation, describing how > to use the software. Heaven knows that we had enough work to do just > to try to get "approval" to work on our next version; but could this > failure to document at a system level be where we fucked up? For some odd reason, I doubt it. Perhaps that was one of the very reasons why the -10 was a successful as it was - people who were new to the machines _had_ to learn from experienced folks. In learning the ropes from more experienced individuals, newcomers got a hefty dose of understanding about the machines based on the teacher's level of knowledge and exposure. What few books existed were pretty paltry, but the best instruction most always comes from someone who already knows the subject, and has screwed up in the past (if they're any good, they'll tell you about it, too, so you don't make the same blunder). Does this mean that the world of the -10 is an "oral tradition"? -- ______________________________________________________________________ | | | | Carl Richard Friend (UNIX Sysadmin) | West Boylston | | Minicomputer Collector / Enthusiast | Massachusetts, USA | | mailto:carl.friend@stoneweb.com | | | http://www.ultranet.com/~engelbrt/carl/museum | ICBM: N42:22 W71:47 | |________________________________________________|_____________________| Article 3115 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!infeed2.internetmci.com!newsfeed.internetmci.com!news.wco.com!not-for-mail From: "Henry W. Miller" Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: Sat, 02 Aug 1997 09:11:10 -0700 Organization: USBR Lines: 55 Message-ID: <33E35C1E.5DC6@sacto.mp.usbr.gov> References: <5rq1tt$gdr$1@decius.ultra.net> <5rv4mn$7bi$1@decius.ultra.net> <33E34275.2AF7A3C8@stoneweb.com> NNTP-Posting-Host: hmiller.mistic.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (Win95; I) To: "Carl R. Friend" Xref: nntp1.ba.best.com alt.sys.pdp10:3115 Carl R. Friend wrote: > > jmfbahxx@ma.ultranet.com, in article nr.<5rv4mn$7bi$1@decius.ultra.net> > wrote: > > > > This person's question about learning how to use TOPS10's software > > started me thinking. We never really provided good instructions > > [hmmm...I just did a pun :-)], I mean, documentation, describing how > > to use the software. Heaven knows that we had enough work to do just > > to try to get "approval" to work on our next version; but could this > > failure to document at a system level be where we fucked up? > > For some odd reason, I doubt it. Perhaps that was one of the very > reasons why the -10 was a successful as it was - people who were new > to the machines _had_ to learn from experienced folks. > > In learning the ropes from more experienced individuals, newcomers > got a hefty dose of understanding about the machines based on the > teacher's level of knowledge and exposure. What few books existed > were pretty paltry, but the best instruction most always comes from > someone who already knows the subject, and has screwed up in the > past (if they're any good, they'll tell you about it, too, so you > don't make the same blunder). > > Does this mean that the world of the -10 is an "oral tradition"? > I think so, or something similar. My first week at DEC, I had to sit by the elbows of two senior engineers and watch them debug a system crash, and otherwise be their gofer. Then, after I had this impromtu "crash" course, they told me to go figure out why a customer's card punch was throwing out a duplicate card after they refilled the hopper. Damned if I didn't find the "JRST .-n" off-by-one bug within a couple of hours, and fix it. They were shocked - they had never had a "newhire" catch on and fix a monitor bug that quickly. (I've been wanting to tell this CDPSER story for a while - yes, folks, it was I who found and fixed that duplicate card problem. The driver would jump back to the DATAO instruction when it detected a hopper emptpy after punching the card then doing a CONI, instead of advancing to the next buffer.) Then they had me write some code to use the DAEMON FACT. function, and I found a bug in DAEMON. But that's a story for another time. > -- > ______________________________________________________________________ > | | | > | Carl Richard Friend (UNIX Sysadmin) | West Boylston | > | Minicomputer Collector / Enthusiast | Massachusetts, USA | > | mailto:carl.friend@stoneweb.com | | > | http://www.ultranet.com/~engelbrt/carl/museum | ICBM: N42:22 W71:47 | > |________________________________________________|_____________________| -HWM Article 3120 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d22 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Spring '78 CUSP Tape Date: Mon, 04 Aug 97 10:44:38 GMT Organization: UltraNet Communications, Inc. Lines: 70 Message-ID: <5s4c3b$kim$1@decius.ultra.net> References: <33D70A18.188@mp.usbr.gov> <5r7fvn$58j$1@decius.ultra.net> <33D8429E.60D@mp.usbr.gov> <5ra50e$8mv$1@decius.ultra.net> <5rmsvp$e6$1@shell3.ba.best.com> NNTP-Posting-Host: d22.dial-33.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3120 In article <5rmsvp$e6$1@shell3.ba.best.com>, inwap@best.com (Smith and O'Halloran) wrote: >In article <5ra50e$8mv$1@decius.ultra.net>, wrote: >>As a side effect of remastering the CUSPs to only one tape, we were able to >>get the bricks laid for shipping/supporting only one version of SCAN/WILD. >> >>And _that_ was a raging knock-down drag-out fight, too. >> >>How's that for airing some old dirty laundry? [grinning emoticon here]. > >Ah, yes. And I helped force the issue. > >One day I noticed that a particular bug in WILD (that had a patch published >in the Software Dispatch) was still present, and was preventing DIRECT >from listing certain subdirectories. I sent in an SPR, and it came back >as "the bug is not in DIRECT, it is in WILD, we do not support WILD". > >I sent back a nasty-gram, stating that the bug *WAS* in SYS:DIRECT.EXE as >supplied on the CUSP tape. It did not matter whether the bug was in >DIRECT.REL or WLD7A.REL, the fact remained that the resulting DIRECT.EXE >exhibited the bug. > >The fine print for SCAN and WILD stated that only problems involving supported >CUSPs would be fixed. Well, DIRECT was a supported CUSP. I demanded that >DEC provide me with a fix that would cause DIRECT.EXE to work correctly. >"It appears that Digital has three alternatives: > 1. Provide a patch for DIRECT.MAC that will provide a workaround > for the bug in WILD.REL, > 2. Fix WILD, or > 3. Declare that they have no intention of honoring the software > support contract for supported CUSPs." Ah, Joe. I'm deep in the golden mists of aging memories and up you pop reminding me how it really was in those days [sighing emoticon here]. You will never know how close we came to _not_ supporting CUSPs because of the mess we had with multiple copies of pieces of software that was stated in the SPD as Category B support. No matter what the legalese said, fixing bugs in that software was at the behest of the programmer and it really depended on how much time was available (which was almost non-existent). The suggestion that all CUSPs no longer be supported was brought up and seriously considered. Just as a side note, the CUSP tape started as a "oh by the way, customers, we found this software to be rather useful; maybe you will, too". It would not have been unprecedented to just ship the CUSPs on a Customer Supported Tape (or not at all). I seem to remember a maintainer in a meeting (and this could be a memory error) considering all the SPRs he could close and get off his backlog if we did drop support. At the end of the discussion, he exhorted that the developers never tell the other maintainers since most of them would have supported the notion. Also, as another side note, in those days the developers rarely saw SPRs. After SPRs were answered by the maintainers, packets of them were delivered down the hall to the developers who rarely had time to go through them. Sometimes, when taking a break, a Bulletin would be rumaged through. Only if an SPR presented a bad design flaw, or was a serious system breaker preventing the customer from doing work (this was 99% monitor software) would a maintainer walk down and talk to a developer. We "improved" our methodology as time went on. /BAH Article 3122 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d22 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: Mon, 04 Aug 97 10:58:41 GMT Organization: UltraNet Communications, Inc. Lines: 76 Message-ID: <5s4ctk$kim$3@decius.ultra.net> References: <5rq1tt$gdr$1@decius.ultra.net> <5rv4mn$7bi$1@decius.ultra.net> <33E34275.2AF7A3C8@stoneweb.com> <33E35C1E.5DC6@sacto.mp.usbr.gov> NNTP-Posting-Host: d22.dial-33.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3122 In article <33E35C1E.5DC6@sacto.mp.usbr.gov>, "Henry W. Miller" wrote: >Carl R. Friend wrote: >> >> jmfbahxx@ma.ultranet.com, in article nr.<5rv4mn$7bi$1@decius.ultra.net> >> wrote: >> > >> > This person's question about learning how to use TOPS10's software >> > started me thinking. We never really provided good instructions >> > [hmmm...I just did a pun :-)], I mean, documentation, describing how >> > to use the software. Heaven knows that we had enough work to do just >> > to try to get "approval" to work on our next version; but could this >> > failure to document at a system level be where we fucked up? >> >> For some odd reason, I doubt it. Perhaps that was one of the very >> reasons why the -10 was a successful as it was - people who were new >> to the machines _had_ to learn from experienced folks. >> >> In learning the ropes from more experienced individuals, newcomers >> got a hefty dose of understanding about the machines based on the >> teacher's level of knowledge and exposure. What few books existed >> were pretty paltry, but the best instruction most always comes from >> someone who already knows the subject, and has screwed up in the >> past (if they're any good, they'll tell you about it, too, so you >> don't make the same blunder). >> >> Does this mean that the world of the -10 is an "oral tradition"? >> > > I think so, or something similar. My first week at DEC, I had >to sit by the elbows of two senior engineers and watch them debug a >system crash, and otherwise be their gofer. Then, after I had this >impromtu "crash" course, they told me to go figure out why a customer's >card punch was throwing out a duplicate card after they refilled the >hopper. Damned if I didn't find the "JRST .-n" off-by-one bug within >a couple of hours, and fix it. They were shocked - they had never had a >"newhire" catch on and fix a monitor bug that quickly. Yup. We would have been shocked in the 80s, too. We usually figured 3-6 months for "breaking in" a programmer; (we gave 1 year to supervisors :-)). The tradition of sitting and watching the senior developers look at crash dumps never stopped until Jim and Tony went away. Part of most people's morning ritual was to get a cup of coffee, go across the hall to the machine room, and find Jim and Tony looking at dumps at the console. Their preference was always the KI. One of the things that the kiddies never understood was the role of being a gofer. It was an honor and part of the breaking in ritual; I remember a female being very offended that she was asked to get Jim coffee while he was fire-fighting at a customer site (she had no idea what he was doing, by the way). /BAH > > (I've been wanting to tell this CDPSER story for a while - yes, >folks, it was I who found and fixed that duplicate card problem. The >driver would jump back to the DATAO instruction when it detected a >hopper emptpy after punching the card then doing a CONI, instead of >advancing to the next buffer.) > > Then they had me write some code to use the DAEMON FACT. function, >and I found a bug in DAEMON. But that's a story for another time. > >> -- >> ______________________________________________________________________ >> | | | >> | Carl Richard Friend (UNIX Sysadmin) | West Boylston | >> | Minicomputer Collector / Enthusiast | Massachusetts, USA | >> | mailto:carl.friend@stoneweb.com | | >> | http://www.ultranet.com/~engelbrt/carl/museum | ICBM: N42:22 W71:47 | >> |________________________________________________|_____________________| > >-HWM Article 3125 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!europa.clark.net!199.60.229.3!newsfeed.direct.ca!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d8 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: Tue, 05 Aug 97 11:03:01 GMT Organization: UltraNet Communications, Inc. Lines: 143 Message-ID: <5s71i9$ofu$1@decius.ultra.net> References: <5rq1tt$gdr$1@decius.ultra.net> <33E31552.4636@sacto.mp.usbr.gov> NNTP-Posting-Host: d8.dial-17.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3125 In article <33E31552.4636@sacto.mp.usbr.gov>, "Henry W. Miller" wrote: >jmfbahxx@ma.ultranet.com wrote: >> >> I received this mail message and I thought it was quite appropriate to put >> here. >> >> At 12:38 AM 7/30/97 -0700, you wrote: >> >jmfbahxx@ma.ultranet.com wrote: >> >>One of our roaring debates at that time was whether a CUSP update tape >> >>should contain only the new software, or a complete set of all CUSPs even >> >>though they hadn't changed. By that time, I believe that a new customer >> >>got to "install" around 20 CUSP Update Tapes in order to get all the >> latest >> >>and greatest of ships. This organization of software also >> (unfortunately) >> >>produced many flavors of versions of each CUSP (including SCAN and WILD >> >>which was very, very bad--caused all sorts of support problems). >> > >> >I guess that's why I noticed some odd SCAN/WILD behaviour from some of >> >the SYS:*.SAV (and OLD:*.SAV) programs... (we were on 7.01A at the time). >> >> I didn't know we even supported those with the advent of SMP; I am amazed. >> For oldtimers, remember SV2EXE (guess I may not...that doesn't look right). >> > > No, that doesn't look right, but I recall the program that you >are speaking about. But I thought the conversion was due to the >migration >to KL paging, and/or the changed JOBDAT area??? No. The conversion was done before the KL was glint in anybody's eye, IIRC. Now, let's see if we can reconstruct what was happening then. The EXE file eliminated the high and low segs being in separate files. VM was being done around then (that was the significance of uping the monitor's version number from 5.xx to 6.xx); and this is KI days...not KL. EXEs came with LINK V5.0. > >> > >> >[and perhaps why even older .SHR/.LOW programs had no SCAN/WILD at all; >> > I take it that was a later invention, so the repercussions took a few >> > years to shake out....] >> >> Possible. >> > > As I recall the history, in the early days, there was no standard >parser - everyone wrote their own. As time went on, and more programs >were developed, people tended to take portions of their earlier code and >use in in the new programs, hacking it suitably along the way. Thus, >eventually, SCAN and WILD became sort of defacto standards within DEC, >even though they were never formally supported until late in their life >span. That's about how things evolved. It also helped that one of the midnight projects of Pete Conklin (who, I think, wrote SCAN and WILD) was to convert some of the major CUSPs to use the interface. There were times that this was not a popular idea. > >> > >> > >> >>As a side effect of remastering the CUSPs to only one tape, we were able >> >>to get the bricks laid for shipping/supporting only one version of >> >>SCAN/WILD. >> > >> > >> >Ah, yes, my real question is: >> > Where in the DECSystem-10 Manual Set did one find the documentation >> >on exactly how to make use of SCAN/WILD in a program? >> >> Chortle. Nowhere. Here's how I did my first one. I read CREFed listings >> of SCAN and WILD, and any other program that used them _properly_. DIRECT >> was an example. Then I tried to write the program. (Now JMF taught me to >> do this.) When I became so confused that I didn't know my ass from a hole >> in the ground, I asked an "expert", where expert is defined as someone who >> wrote a SCAN/WILD interface before me. Of course, we were fortunate >> because we could always ask the developer/maintainer (Hey, Bob!)[grinning >> emoticon here]. >> > > I recall that there was a really dinky section in the "white wall". >But I used a set of documents that was prepared by Reed Powell. It must have been something Reed provided. Maybe, a RUNOFFed SCAN.MAC? > >> Just a note: Even though Jim managed to write an interface in user mode >> (FILDAE.MAC), he never, ever admitted it to anyone with the intention of >> avoiding the label of SCAN expert. >> >> > >> >[ As a lowly undergrad student in 1982 (and aspiring DEC-10 systems' >> >programmer), I could not get any help with this from our Computer Center >> >staff ...] >> >> Not surprising. Hmmm....Was the ability to use SCAN/WILD a rite of passage >> in the TOPS10 world? >> > > I would accept that as a truism: a rite of passage to advanced >programming. I'd rate it as more difficult than doing a run of the mill >(no pun intended...) _I_ liked it [grinning emoticon here] Monitor or DAEMON patch, but not quite as difficult >as screwing around with the source code for the Cobol compile, LIBOL or >SORT... Well, maybe as a user mode programmer. TW and JMF were not user mode programmers; I don't recall TW ever touching SCAN and WILD. Here's a JMF story to illustrate his inability to do user mode programming. When he began writing the FILDAE, he ran into a slight difficulty; he couldn't lay out a programming structure to do I/O in user mode. Remember, he did everything in exec mode. Now, this was all done on his weekend time which was the time he reserved for doing fun things. Since, it was the weekend, there wasn't anybody to ask how to do the I/O. He raided my office for documentation (Monitor Calls Manual) [grumble...leaving a mess] and couldn't figure it out. So this man went into the machine room, set a breakpoint, and wrote his I/O routines based on what the internals of the monitor did when his user mode program executed the I/O monitor calls. And that, boys and girls, is how Jim wrote and debugged the FILDAE...with EDDT. /BAH Article 3137 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newsfeed.direct.ca!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: 8 Aug 1997 14:55:28 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 62 Message-ID: <5sfc10$qa2$1@kirin.wwa.com> References: <5rq1tt$gdr$1@decius.ultra.net> <5rv4mn$7bi$1@decius.ultra.net> <33E34275.2AF7A3C8@stoneweb.com> <5s4cf8$kim$2@decius.ultra.net> <33E7BD06.CDE1A1C@stoneweb.com> <5scbid$nvh$1@decius.ultra.net> <33EA5F91.63441604@stoneweb.com> NNTP-Posting-Host: poolf1-034.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3137 In article <33EA5F91.63441604@stoneweb.com>, carl.friend@stoneweb.com says... > >> I don't think people talk about styles when they're documenting >> history. For instance, there will never be a description of the >> [what I considered] (jmfbah's brackets) art of JMF and TW. > > This is a tragedy. Part of what the machine was, and is, is >inextricably linked to the personalities and talents of its designers >and programmers. This facet certainly deserves preservation, and >hopefully is beginning to attract some attention. I never met either >JMF or TW, but I've been able to garner quite a bit of respect for >both from what I've seen here; the "folklore" aspect is as important >as the absolute "historic" one. In the eight years I spent at DEC, I at one time or another shared an office with both TW and JMF, and counted each of them among my best friends. During what I consider the most groundbreaking period of TOPS-10 development we constituted 60% of the monitor group. That being said, I think there's been a tendency here to greatly romanticize our accomplishments. Remember that companies who relied upon the DEC-10 for there livelihood needed good monitor people to keep things running. The quality of the software was that bad. I'm a living example, I was lured away from DEC by a timesharing bureau waving more money than I could see earning at DEC in several years of raises. During TOPS-10's formative years DEC was a real meritocracy. In the Programming Department, the monitor group was the top of the pyramid. The PDP-10 monitor was a going concern before either TW, JMF or I joined the group. Over a period of years we supplanted the "founding fathers", who included Tom Hastings, Don Witcraft, Dave Plummer, Dick Gruen, Chris White, and others who's names I've forgotten. It should be noted that there was precious little attention given to formal quality control. Tom Hasting (to his credit) tried to create an automated test suite at one time, but after he left the group it quickly fell out of date and its running was abandoned. It wasn't until years later I heard the term "regression testing", but in retrospect that was exactly what Tom was trying to initiate. We were too arrogant and shortsighted to follow his lead. Keeping the test suite current with development was seen as just too much of a pain in the ass. To our credit, the peer pressure within the group was intense. Any quality built into the monitor was due more to the need to not be seen screwing up in front of the group than concern for customers. During the period I was responsible for magtape support, we had the "MTREQ goes to -2" problem. (BTW, another example of why customers needed monitor programmers on staff, patching MTREQ back to -1 in the running monitor freed up the hung magtapes.) I still recall how fanatically I undertook this bug hunt, particularly since so many others had failed to find it. It wasn't because of concern for customers (after all, they had a "get around"), it was because I wanted so badly to be the person who cracked the tough one. Having over the years managed a number of software development departments, I'm not so sure I'd like to manage a group like the late 60s, early 70s TOPS-10 group. We were way too elitist and arrogant, and practically unmanagable. It's a good thing we could back it up with performance. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3145 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!infeed2.internetmci.com!newsfeed.internetmci.com!news.wco.com!not-for-mail From: "Henry W. Miller" Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: Sat, 09 Aug 1997 07:47:16 -0700 Organization: USBR Lines: 148 Message-ID: <33EC82F3.66E9@sacto.mp.usbr.gov> References: <5rq1tt$gdr$1@decius.ultra.net> <33E31552.4636@sacto.mp.usbr.gov> <5s71i9$ofu$1@decius.ultra.net> <33EAE2DA.36D5@sacto.mp.usbr.gov> <5shn23$dcm$4@decius.ultra.net> NNTP-Posting-Host: hmiller.mistic.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (Win95; I) CC: henrym@sacto.mp.usbr.gov Xref: nntp1.ba.best.com alt.sys.pdp10:3145 jmfbahxx@ma.ultranet.com wrote: > > In article <33EAE2DA.36D5@sacto.mp.usbr.gov>, > "Henry W. Miller" wrote: > >jmfbahxx@ma.ultranet.com wrote: > >> > >> In article <33E31552.4636@sacto.mp.usbr.gov>, > >> "Henry W. Miller" wrote: > >> >jmfbahxx@ma.ultranet.com wrote: > >> >> > >[SNIP] > >> >> For oldtimers, remember SV2EXE (guess I may not...that doesn't look > >> right). > >> >> > >> > > >> > No, that doesn't look right, but I recall the program that you > >> >are speaking about. But I thought the conversion was due to the > >> >migration > >> >to KL paging, and/or the changed JOBDAT area??? > >> > >> No. The conversion was done before the KL was glint in anybody's eye, > >> IIRC. Now, let's see if we can reconstruct what was happening then. > >> The EXE file eliminated the high and low segs being in separate files. > >> VM was being done around then (that was the significance of uping the > >> monitor's version number from 5.xx to 6.xx); and this is KI days...not > >>KL. EXEs came with LINK V5.0. > >> > > > > > OK, that's right. I recall the elimination of the .SHR segs now, > >but didn't it also entail some reordering of the JOBDAT table? Or was > >it the other way around - was the JOBDAT table modified to get rid of > >the SHR segs? > > I don't know. I doubt it was a reordering. One of the ongoing > frustrations with TOPS10 development was our self-imposed restriction of > _not_ modifying data structures that could conceivably be used by some > customer's program. With the implementation of SMP, one of the meetings > that JMF and TW had was to form a wish list of what junk they would like to > see eliminated from the monitor. It was DEC's one and only chance to get > data structures into a manageable state. > Yes, as I recall now, weren't certain locations of JOBDAT "depreciated", meaning that they no longer had a purpose in later releases of the monitor? > > >> >> > >[SNIP] > >> > I would accept that as a truism: a rite of passage to advanced > >> >programming. I'd rate it as more difficult than doing a run of the > >> >mill > >> >(no pun intended...) > >> > >> _I_ liked it [grinning emoticon here] > >> > >> Monitor or DAEMON patch, but not quite as difficult > >> >as screwing around with the source code for the Cobol compile, LIBOL or > >> >SORT... > >> > >> Well, maybe as a user mode programmer. TW and JMF were not user mode > >> programmers; I don't recall TW ever touching SCAN and WILD. Here's a > >>JMF > >> story to illustrate his inability to do user mode programming. > >> > > > > That's quite true - it usually took a different breed of > >programmer to handle both. I remember in surprise when I first > >started looking at the guts of TOPS20, and here it is with JSYS > >calls inside of JSYS calls... Wait a minute, here... > > Right. I worked in the FORTRAN maintenance group after the monitor group > for a while. You wouldn't believe the difference language that had to be > used just to communicate with the compiler types. There definitely was a > completely different mindset. People who worked on the OTS sort of had an > idea about the relationship between compiler type work and operating system > work. > That was spooky in it's own right. I mean, COBOL was pretty straight forward - it basically generated threaded code with parameter lists and called LIBOL to do almost everything. The Fortran compiler was down right eerie. It was almost as if the compiler could read your mind and perform a DWIM. I spent a morning rewriting a customer program for a benchmark one time, to attempt to get a little more efficency out of it. The object code generated by the orignal version and my hand optimized version were almost identical. After a while, though, I got so that I could predict what sort of object code the compiler would spit out. > > > >> When he began writing the FILDAE, he ran into a slight difficulty; he > >> couldn't lay out a programming structure to do I/O in user mode. > >> Remember, > >> he did everything in exec mode. Now, this was all done on his weekend > >> time > >> which was the time he reserved for doing fun things. Since, it was the > >> weekend, there wasn't anybody to ask how to do the I/O. He raided my > >> office for documentation (Monitor Calls Manual) [grumble...leaving a > >> mess] > >> and couldn't figure it out. So this man went into the machine room, set > >> a breakpoint, and wrote his I/O routines based on what the internals of > >> the monitor did when his user mode program executed the I/O monitor > >> calls. And that, boys and girls, is how Jim wrote and debugged the > >> FILDAE...with EDDT. > >> > > > > {chortle} Sort of reminds me of the way I debugged some portions > >of TCP/IP for Tenex on a Foonly F2/F3... > > Ah! A telling sign of the type of programmer you were [grinning emoticon > here]. > Yep, a feature I sorely missed when attempting to debug a tape driver problem at a customer site, with their System's Comcepts SA-10 controller. > I wanted to ask you...what FACT. work did you do. As someone whose goal > was to eliminate that code, I'm curious. > A customer needed a way to charge back to individual departments for each use of certain programs. The best way, we determined was to use the FACT. function of DAEMON. Now, there was a problem, as only privileged code could call some DAEMON functions. So I wrote, first, an ugly little hack that used PEEK./POKE. to change the PPN to [1,2] to allow the DAEMON call to work, then back to it's original PPN. The second problem was that the FACT. function would fail silently, giving the impression that it had indeed worked, when in fact, it had appended nothing to the FACT file. It turned out to be an off-by-one bug - if the amount of data that you were attempting to log was exactly 16 words, the maximum allowed, which of course was the size that I was using, it would fail. Simple enough to fix, but a real bitch to track down. Even at that, I had to use byte stuffing to get all of the required data into one FACT entry. Now that I think of it, this may have been the only customer site using that function. This was long before, of course, TOPS10 supported the USAGE. UUO, which made the job much easier down the road. > /BAH -HWM Article 3155 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!infeed2.internetmci.com!newsfeed.internetmci.com!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: SCAN/WILD questions Date: 10 Aug 1997 15:13:47 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 29 Message-ID: <5sklrb$gcc$1@kirin.wwa.com> References: <5rq1tt$gdr$1@decius.ultra.net> <33E31552.4636@sacto.mp.usbr.gov> <5s71i9$ofu$1@decius.ultra.net> <33EAE2DA.36D5@sacto.mp.usbr.gov> <5shn23$dcm$4@decius.ultra.net> <33EC82F3.66E9@sacto.mp.usbr.gov> <5sk854$fs5$2@decius.ultra.net> NNTP-Posting-Host: poolf1-051.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3155 In article <5sk854$fs5$2@decius.ultra.net>, jmfbahxx@ma.ultranet.com says... > >> The Fortran compiler was down right eerie. It was almost as >>if the compiler could read your mind and perform a DWIM. I spent a >>morning rewriting a customer program for a benchmark one time, to >>attempt to get a little more efficency out of it. The object code >>generated by the orignal version and my hand optimized version were >>almost identical. After a while, though, I got so that I could >>predict what sort of object code the compiler would spit out. > >They had a really good team. John, didn't the original implementers come >from the monitor group? Actually, F10 was purchased (sometime around '67 or '68, I think) from an outside contractor who's name I've long forgotten. I don't recall there being any ex-monitor folk involved. While I'm here typing, and the thread is "Scan/Wild" after all, I seem to recall another parser caller SCNNER (or something like that). When I was working on DAEMON (making DUMP and DCORE work with VM, a non-trivial exercise) it seems to me it was built with a different parser than SCAN. Anyone have a better memory than I? -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3156 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!howland.erols.net!newsxfer.itd.umich.edu!newbabylon.rs.itd.umich.edu!sarr From: sarr@engin.umich.edu (Sarr J. Blumson) Newsgroups: alt.sys.pdp10 Subject: Re: Fortran (was SCAN/WILD questions) Date: 10 Aug 1997 20:46:00 GMT Organization: CITI, University of Michigan Lines: 27 Message-ID: <5sl9a8$cfr$1@newbabylon.rs.itd.umich.edu> References: <5rq1tt$gdr$1@decius.ultra.net> <33EC82F3.66E9@sacto.mp.usbr.gov> <5sk854$fs5$2@decius.ultra.net> <5sklrb$gcc$1@kirin.wwa.com> Reply-To: sarr@umich.edu NNTP-Posting-Host: sinshan.engin.umich.edu Xref: nntp1.ba.best.com alt.sys.pdp10:3156 In article <5sklrb$gcc$1@kirin.wwa.com>, John Everett wrote: >In article <5sk854$fs5$2@decius.ultra.net>, jmfbahxx@ma.ultranet.com says... >> >>They had a really good team. John, didn't the original implementers come >>from the monitor group? > >Actually, F10 was purchased (sometime around '67 or '68, I think) from an >outside contractor who's name I've long forgotten. I don't recall there being >any ex-monitor folk involved. CSC? I may have it mixed up with a really amazing Fortran for the Philco 2000 from around the same era. F40, bless it's twisted little heart, had been hacked on by some of the old hadrware/monitor crew, from the original Digitek code Tom Eggers, in particular, wrote the common sub expression hackes (accor4ding to the comments, anyway). I spent many happy hours trying to figure out how it worked (or what the instructions did, since it was writtne in machined code for pseudo hardware). -- -------- Sarr Blumson sarr@umich.edu voice: +1 313 764 0253 home: +1 313 665 9591 ITD, University of Michigan http://www-personal.umich.edu/~sarr/ 535 W William, Ann Arbor, MI 48103-4943 Article 3157 of alt.sys.pdp10: Path: nntp1.ba.best.com!not-for-mail From: inwap@shell3.ba.best.com (Smith and O'Halloran) Newsgroups: alt.sys.pdp10,alt.folklore.computers Subject: Re: Stupid Administration Tricks (Was Re: Identical connectors) Date: 11 Aug 1997 00:42:42 GMT Organization: Chez INWAP (people, computers, cats) Lines: 43 Message-ID: <5sln62$sq6$1@nntp1.ba.best.com> References: <5qln0i$qkb$1@shell3.ba.best.com> <5sioda$9sd$1@news.wizvax.net> <33EDC7B7.37EF6F71@stoneweb.com> NNTP-Posting-Host: shell3.ba.best.com Xref: nntp1.ba.best.com alt.sys.pdp10:3157 alt.folklore.computers:90422 In article <33EDC7B7.37EF6F71@stoneweb.com>, > Is there truth to the story that the KS-10's design criteria was: >to produce a machine with the computational capabilities of a PDP-6? No, the 2020 was supposed to be 20% of a KL. CSM (the Colorado School of Mines) upgraded from a 192K KA to a KL and a KS. A particular job that took an hour to run on the KA took 10 minutes on the KL and 50 minutes on the KS. KL-1091 = 1 * KL KS-2020 = KL / 5 KA-1050 = KL / 6 There was one test that the KS was really slow on: buffered disk I/O. It could only do 60 blocks per second (one per disk revolution). The RH11 disk controller ran in 18-bit mode, using the parity lines as extra data lines. It could transfer several consecutive disk blocks into contiguous UNIBUS virtual addresses. But consecutive disk blocks going into different buffers (each with a 3-word buffer header) required CPU intervention and the CPU was not quick enough. (The RH20 did not have this problem since it had an alternate set of registers, so the 2nd command could be loaded into the primary registers by the hardware when the 1st command was done.) A typical read of 3 blocks, using TOPS10 buffered I/O, was: Locations N through N+2 were the 3-word buffer header. As were locations N+203 to N+205 and N+406 to N+410. Read 128 words into location N+3 to N+202. Get interrupt from RH11 disk controller (running in 18-bit mode). Tell controller to read next block into locations N+206 to N+405. Oops, the disk sector is already partway past the read heads. Wait for sector to come around again. Transfer data. Get interrupt from RH11 disk controller when transfer is done. Tell controller to read next block into locations N+411 to N+610. Oops, the disk sector is already partway past the read heads. Wait for sector to come around again. Transfer data. Get interrupt from RH11. Done. -Joe -- INWAP.COM is Joe and Sally Smith, John and Chris O'Halloran and our cats See http://www.inwap.com/ for "ReBoot", PDP-10, and Clan MacLeod. Article 3159 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!news.home.com!newsfeed.direct.ca!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!francini From: francini@ultranet.com (John Francini) Newsgroups: alt.sys.pdp10 Subject: Re: SET HOSTess Date: Mon, 11 Aug 1997 02:27:36 -0400 Organization: Appletree Systems Lines: 52 Message-ID: References: <33EC61B8.158E@sacto.mp.usbr.gov> NNTP-Posting-Host: d11.dial-1.nsh.nh.ultra.net Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Newsreader: Yet Another NewsWatcher 2.4.0 Xref: nntp1.ba.best.com alt.sys.pdp10:3159 In article , mbg@world.std.com (Megan) wrote: >"Henry W. Miller" writes: > >>Someone once told me the story about why the TOPS10 COMCON >>command for SET HOST was, in full, actually SET HOSTESS, but I have >>forgotten it over the years. Did it have something to do with the >>node TWINKY, or was the naming of that node just a continuation of >>the joke ? > >Sounds like what we used to do at college (WPI) back in the 70's >when people said that the text went by on their screen too fast... > > "Just set your terminal in slow mode... Do this, type > 'SET TT SLOW'" > >which of course, actually slaved their terminals... > >And as for twinky, I remember using that system on occasion when >I was an operator (3rd shift) at PKO1. > And someone else mentioned that it was a small system. Actually, when I got on board at Digital (oops, I mean DEC) in 1983 in their Internal Software Services group, TWINKY had just been re-potted from the KS-10 it had lived in to a 1090. The KS was used to run TOPS-20, and was called TWONKY. (That system was an interesting story in itself, as the TOPS-20 support people in my group managed to get V5.x of TOPS-20 to run on that beast. Amazing what you can do when you have sources.) I ended up being in charge of the Monitor builds on TWINKY until I went to the DEC-10 development group in late 1986. After being made aware of the true name of the SET HOSTES command (remember, Monitor command tables only held 6 characters/entry), I changed the ID string in the Monitor (printed out when it booted and every time INITIA and LOGIN ran, and advertised on the ANF-10 network) to be "Hostess Twinky" followed by the version/build info. There's still a TWINKY system inside Digital, but, alas, it's no longer a -10... John Francini Former DEC-10 hacker -- ---- +------------------------------------------------------------------------------+ | "I have come to the conclusion that one useless man is called a disgrace; | | that two or more are called a law firm; and that three or more become | | a Congress. And by God I have had _this_ Congress!" | | -- John Adams | +------------------------------------------------------------------------------+ Article 3204 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!news.home.com!feta.direct.ca!newsfeed.direct.ca!news-feed1.tiac.net!decius.ultra.net!d14 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: What's with the pdp-10 emulator? Date: Tue, 26 Aug 97 11:47:45 GMT Organization: UltraNet Communications, Inc. Lines: 44 Message-ID: <5tug8h$g90$1@decius.ultra.net> References: <01bcafc2$e24acb40$29c906c6@hsnewman.ccsi.com> <5to3ud$ria$1@news.wizvax.net> <34016536.2FA8@cisco.com> <5trvjl$adu$1@news.wizvax.net> NNTP-Posting-Host: d14.dial-18.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3204 In article <5trvjl$adu$1@news.wizvax.net>, wilson@dbit.com (John Wilson) wrote: >In article <34016536.2FA8@cisco.com>, >Henry W. Miller wrote: >>any way you cut it, emulating a 36 bit machine on a 32 bit machine could >>be a little hairy. You could, of course, map it into a 64 bit word, but >>what a waste of storage, unless you do some pretty byte stuffing. > >It's pretty cumbersome no matter what you do with the extra 4 bits. It >would sure be nice to get some help from the host CPU with emulating CRY0 >and CRY1 using the carry and overflow flags, but that probably ain't gonna >happen. > >> How about I/O ? Do you really need to emulate the I/O completely, >>or do you make the I/O instructions a sort of emulator UUO, which would >>map to a virtual device, like a large file on a disk to emulate a >>virtual RP07? > >Yes and yes -- you use the image file in conjunction with a complete >emulation of the I/O instructions and device registers. Even given the >gigantic PDP-10 instruction set, the I/O is the *really* hard part of >writing the emulator. Especially because the documentation is so bad -- >for example, the TM03 manual *looks* pretty complete, until you get into >it and discover the key issues (drive attentions, spacing over tape marks, >command queueing) that it's either vague on or totally wrong about. And >every time you get a device emulation working with one OS, you find that >it doesn't work with the other OSes because the authors of those knew some >more details about the device (either discovered experimentally or maybe >by interviewing the hardware designer in person) that aren't in the docs. >The prints help tremendously but old-style DEC prints tend to be pretty >disorganized, with tangled combinatorial logic that wanders aimlessly from >sheet to sheet (and often from board to board) and signal names that don't >seem to mean anything. Or one could read the TOPS10 module that implemented the device. I remember the frustration that TW had when he was trying to write the RH20 driver. Out of that came a document (written by Tony whilst he was debugging the hardware) called "Tony in RH20 Land" or "I Should Have Listened When Mother Told Me There Was A Great Future In Encyclopedia Sales". Is that document in the archieves? /BAH Article 3212 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!infeed1.internetmci.com!newsfeed.internetmci.com!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: What's with the pdp-10 emulator? Date: 27 Aug 1997 14:49:50 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 31 Message-ID: <5u1eqe$o7l$3@kirin.wwa.com> References: <01bcafc2$e24acb40$29c906c6@hsnewman.ccsi.com> <5to3ud$ria$1@news.wizvax.net> <34016536.2FA8@cisco.com> <5trvjl$adu$1@news.wizvax.net> <5tug8h$g90$1@decius.ultra.net> NNTP-Posting-Host: poolf3-033.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3212 In article <5tug8h$g90$1@decius.ultra.net>, jmfbahxx@ma.ultranet.com says... > >Or one could read the TOPS10 module that implemented the device. I >remember the frustration that TW had when he was trying to write the RH20 >driver. Out of that came a document (written by Tony whilst he was >debugging the hardware) called "Tony in RH20 Land" or "I Should Have >Listened When Mother Told Me There Was A Great Future In Encyclopedia >Sales". One of the real frustrations in dealing with post-TOPS20 devices is that the engineers sort of forgot that not all I/O operations start and end at a page boundary. I recall the nastiness I had to dig through when I implemented the RH11/UBA driver for the KS10. Later when I was managing software development at ADP's Network Services division we implemented CI20/HSC50/RA?? support in TOPS10 for DEC. We had this strange data corruption problem which occured once in a blue moon. After months of bug chasing we found that the microcode in the CI20 couldn't handle "gather writes" properly, but only during an error recovery retry. By that time the guy who implemented the microcode at DEC was long gone and it took what seemed like six months or so to get DEC to correct the problem. Meanwhile we had thousands of dollars of "new technology" sitting around useless because we wouldn't subject our clients to a known data corruption potential. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3221 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!newshub2.home.com!news.home.com!howland.erols.net!ix.netcom.com!netcom16!alderson From: alderson@netcom16.netcom.com (Richard M. Alderson III) Subject: Re: Any celebrations planned for 36th anniversary of PDP-6? In-Reply-To: eric@brouhaha.com's message of 27 Aug 1997 17:19:16 -0400 Message-ID: Sender: alderson@netcom16.netcom.com Reply-To: alderson@netcom.com Organization: NETCOM On-line services References: Date: Thu, 28 Aug 1997 19:13:07 GMT Lines: 22 Xref: nntp1.ba.best.com alt.sys.pdp10:3221 In article eric@brouhaha.com (Eric Smith) writes: >With all the hoopla surrounding DEC's 40th anniversary, has anyone given >consideration to planning a celebration of the 36th anniversary of DEC's >36-bit computer family, in October of 2000? Well, at the 30th Anniversary get-together at the spring DECUS in DC, Clive Dawson and I ended the session with an invitation to all to attend the next planned event, the 36th Anniversary. That assumes that DECUS will still exist, and that *someone* will still be willing to schedule a session for us. We had a pretty good turnout, with attendees including the DECUS Board, and Roseann Giordano, who was forgiven her announcement of the Jupiter Cancellation by acclamation. I'm maintaining my DECUS membership just so that I can answer the Call for Sessions in 2000... -- Rich Alderson You know the sort of thing that you can find in any dictionary of a strange language, and which so excites the amateur philo- logists, itching to derive one tongue from another that they know better: a word that is nearly the same in form and meaning as the corresponding word in English, or Latin, or Hebrew, or what not. --J. R. R. Tolkien, alderson@netcom.com _The Notion Club Papers_ Article 3225 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsfeed.dacom.co.kr!infeed1.internetmci.com!newsfeed.internetmci.com!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10,alt.folklore.computers,comp.sys.dec.micro Subject: Re: Dec 40th anniversary web pages Date: 29 Aug 1997 12:12:34 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 26 Message-ID: <5u6ebi$8es$1@kirin.wwa.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> NNTP-Posting-Host: poolf1-017.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3225 alt.folklore.computers:91282 comp.sys.dec.micro:3248 In article <5u50u7$iiu$1@shell3.ba.best.com>, inwap@best.com says... > >The timeline is pretty good (if you ignore the few technical errors that >the marketdroids let slip through). I created a 36-bit specific index >to Digital's timeline web pages; it is in > > http://www.inwap.com/pdp10/timeline.html > > > * 1978 [17]DECsystem-2020 introduced as "the world's lowest cost > mainframe computer system". > Can anyone identify the KS-10 developers in the pictur? I checked out the picture, and although I recognize most of the faces find I can't put a single name to one. What's worse is that I worked closely with many of them while porting TOPS-10 to the KS. Sigh! Barb (jmfbah), hope you can do better than I. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3239 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!infeed1.internetmci.com!newsfeed.internetmci.com!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10,alt.folklore.computers,comp.sys.dec.micro Subject: Re: Dec 40th anniversary web pages Date: 30 Aug 1997 17:41:38 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 24 Message-ID: <5u9m0i$g59$1@kirin.wwa.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u6ebi$8es$1@kirin.wwa.com> <5u75h1$obf$1@gte2.gte.net> NNTP-Posting-Host: poolf4-034.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3239 alt.folklore.computers:91357 comp.sys.dec.micro:3255 In article <5u75h1$obf$1@gte2.gte.net>, allenb1@gte.net says... > >Top row , third for left - Possibly Alan Kotok ? > >> Can anyone identify the KS-10 developers in the pictur? >> >> I checked out the picture, and although I recognize most of the faces >find I >> can't put a single name to one. What's worse is that I worked closely >with >> many of them while porting TOPS-10 to the KS. Sigh! >> >> I took the .jpg, blew it up, and applied a smoothing algoithm, and I can't swear it isn't Alan; but I doubt it because he wasn't involved in the KS-10 project. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3238 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!decius.ultra.net!d17 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Sat, 30 Aug 97 11:43:03 GMT Organization: UltraNet Communications, Inc. Lines: 33 Message-ID: <5u91gt$iu4$2@decius.ultra.net> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u6ebi$8es$1@kirin.wwa.com> NNTP-Posting-Host: d17.dial-21.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3238 In article <5u6ebi$8es$1@kirin.wwa.com>, jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: >In article <5u50u7$iiu$1@shell3.ba.best.com>, inwap@best.com says... >> >>The timeline is pretty good (if you ignore the few technical errors that >>the marketdroids let slip through). I created a 36-bit specific index >>to Digital's timeline web pages; it is in >> >> http://www.inwap.com/pdp10/timeline.html >> > >> >> * 1978 [17]DECsystem-2020 introduced as "the world's lowest cost >> mainframe computer system". >> Can anyone identify the KS-10 developers in the pictur? > >I checked out the picture, and although I recognize most of the faces find I >can't put a single name to one. What's worse is that I worked closely with >many of them while porting TOPS-10 to the KS. Sigh! > >Barb (jmfbah), hope you can do better than I. > Well, I took a look (you realize that I sacrificed a lot since using AOL's web browser now means that it's top level software won't work for weeks now). The picture was a blur; I had a _lot_ of &nsbp characters sprinkled all over the place. So until someone can tell me what I'm doing wrong, I can't help you. One of the blobs looked a lot like Ted Hess; he had a distinctive facial configuration. /BAH -subtract twenty for e-mail - this is _NOT_ an invitation Article 3256 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10,alt.folklore.computers,comp.sys.dec.micro Subject: Re: Dec 40th anniversary web pages Date: 3 Sep 1997 17:04:46 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 36 Message-ID: <5uk5be$quq$1@kirin.wwa.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u8m2t$9o2$1@shell3.ba.best.com> NNTP-Posting-Host: poolf2-002.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3256 alt.folklore.computers:91465 comp.sys.dec.micro:3264 In article <5u8m2t$9o2$1@shell3.ba.best.com>, inwap@best.com says... > >Also, DEC's web page states that TOPS-10 was created for the PDP-6. >IIRC, it was "the Monitor"; the name was changes to TOPS-10 after the >KA10 was in use. Yes, quite true. I was in the Monitor Group when David Stone announced at one of our regular Monday morning Monitor Group meetings that he had had this epiphany over the weekend and had decided to call the Monitor TOPS-10 (for Timeshared OPerating System -10). We all thought the name sucked. As I recall we left the meeting shaking our heads and chuckling under our collective breaths. We thought it sounded too much like a Saturday morning AM radio show. I often wish I had a map of The Mill with the building numbers on it. I remember where that meeting was held which helps with chronology. The group started on the first floor of Building 12 (I think), migrated to the second, then third floor, then to the top floor of what I recall as Building 1 (it ran east/west and connected 12 with 5), where KA-10 System 2 was located. We were there when DS made his proclamation, which makes it perhaps 1970 or 1971. The Group later moved to the top floor of Building 5 before departing for Marlboro in 1973. Which reminds me (for some strange reason), how many others remember (accurately) the origin of the name Leprechaun Lounge for a conference room in the Marlboro facility? I'll bet the story has been through a few permutations over the years. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3261 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-feed1.tiac.net!decius.ultra.net!d7 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Thu, 04 Sep 97 10:30:54 GMT Organization: UltraNet Communications, Inc. Lines: 56 Message-ID: <5um370$k2u$1@decius.ultra.net> References: <5tfkn8$8uh@src-news.pa.dec.com> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u8m2t$9o2$1@shell3.ba.best.com> <5uk5be$quq$1@kirin.wwa.com> NNTP-Posting-Host: d7.dial-33.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3261 [unfamiliar newsgroups trimmed] In article <5uk5be$quq$1@kirin.wwa.com>, jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: > >In article <5u8m2t$9o2$1@shell3.ba.best.com>, inwap@best.com says... >> >>Also, DEC's web page states that TOPS-10 was created for the PDP-6. >>IIRC, it was "the Monitor"; the name was changes to TOPS-10 after the >>KA10 was in use. > >Yes, quite true. I was in the Monitor Group when David Stone announced at >one of our regular Monday morning Monitor Group meetings that he had had >this epiphany over the weekend and had decided to call the Monitor TOPS-10 >(for Timeshared OPerating System -10). We all thought the name sucked. As >I recall we left the meeting shaking our heads and chuckling under our >collective breaths. We thought it sounded too much like a Saturday morning >AM radio show. Now I hadn't heard that one! Thanks. > >I often wish I had a map of The Mill with the building numbers on it. I >remember where that meeting was held which helps with chronology. The >group started on the first floor of Building 12 (I think), migrated to the >second, then third floor, then to the top floor of what I recall as >Building 1 Nope. Building 3 and it was the 5th floor (there were 6 in building 3). You had to go up to the 6th floor in order to get to building 21 which housed field service at the time. >(it ran east/west and connected 12 with 5), where KA-10 System 2 was >located. We were there when DS made his proclamation, which makes it >perhaps 1970 or 1971. Right. You guys were on 3-5 when I started working at DEC which was July, 1971. I think it was later in the year (or maybe the beginning of the 1972) when you moved to 5-5. IIR, there wasn't enough space for the second processor to implement the master/slave configuration. Remember the slanting floors on 3-5? > >The Group later moved to the top floor of Building 5 before departing for >Marlboro in 1973. > >Which reminds me (for some strange reason), how many others remember >(accurately) the origin of the name Leprechaun Lounge for a conference >room in the Marlboro facility? I'll bet the story has been through a few >permutations over the years. > I remember how the aisles got their names but, for the life of me, I can't recall the naming of the lounge. /BAH Article 3262 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!uunet!in5.uu.net!world!world!mbg From: mbg@world.std.com (Megan) Subject: Re: Dec 40th anniversary web pages Message-ID: Organization: The World Public Access UNIX, Brookline, MA References: <5tfkn8$8uh@src-news.pa.dec.com> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u8m2t$9o2$1@shell3.ba.best.com> <5uk5be$quq$1@kirin.wwa.com> <5um370$k2u$1@decius.ultra.net> Date: Fri, 5 Sep 1997 02:39:04 GMT Lines: 45 Xref: nntp1.ba.best.com alt.sys.pdp10:3262 jmfbahxx@ma.ultranet.com writes: >Right. You guys were on 3-5 when I started working at DEC which was July, >1971. I think it was later in the year (or maybe the beginning of the >1972) when you moved to 5-5. IIR, there wasn't enough space for the second >processor to implement the master/slave configuration. Remember the >slanting floors on 3-5? I was back in the mill a few weeks ago (it is really sad seeing it so empty... I found the places where my office(s) used to be, all there are are impressions in the rugs of the partitions). I was helping someone find their way around the mill, and specifically showed the area on 3-5. Starting at the bldg.5 end of 3-5, the parition walls come to the top of my head, and although they are level for the length of the building, about 30 feet down the hall, the floor buckles enough for me to easily look into an office over the partitions. >> >>The Group later moved to the top floor of Building 5 before departing for >>Marlboro in 1973. >> >>Which reminds me (for some strange reason), how many others remember >>(accurately) the origin of the name Leprechaun Lounge for a conference >>room in the Marlboro facility? I'll bet the story has been through a few >>permutations over the years. >> >I remember how the aisles got their names but, for the life of me, I can't >recall the naming of the lounge. Neither do I... and I worked on ML5-5 from 1978 to 1988 Megan Gentry Former RT-11 Developer +--------------------------------------+-------------------------------------+ | Megan Gentry | tcp/ip (work): gentry@zk3.dec.com | | Unix Support and Engineering Group | or: gentry@rusure.enet.dec.com | | Digital Equipment Corporation | (non-work): mbg@world.std.com | | 110 Spitbrook Rd. ZK03-2/T43 | URL http://world.std.com/~mbg/ | | Nashua, NH 03062 | "Still real-time after all these | | (603) 881 1055 | years." | +--------------------------------------+-------------------------------------+ Article 3264 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newsxfer3.itd.umich.edu!newbabylon.rs.itd.umich.edu!sarr From: sarr@engin.umich.edu (Sarr J. Blumson) Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: 5 Sep 1997 14:11:09 GMT Organization: University of Michigan, Ann Arbor Lines: 20 Message-ID: <5up3tt$hfv$1@newbabylon.rs.itd.umich.edu> References: <5tfkn8$8uh@src-news.pa.dec.com> <5u8m2t$9o2$1@shell3.ba.best.com> <5uk5be$quq$1@kirin.wwa.com> <5um370$k2u$1@decius.ultra.net> Reply-To: sarr@umich.edu NNTP-Posting-Host: sinshan.engin.umich.edu Xref: nntp1.ba.best.com alt.sys.pdp10:3264 In article <5um370$k2u$1@decius.ultra.net>, wrote: >[unfamiliar newsgroups trimmed] > >In article <5uk5be$quq$1@kirin.wwa.com>, > jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: >> >>I often wish I had a map of The Mill with the building numbers on it. I >>remember where that meeting was held which helps with chronology. The >>group started on the first floor of Building 12 (I think) Because, as I understand it, the monster Bryant disk drives originally planned for the 10 _had_ to be on the first floor. The Mill flooring couldn't hold them anywhere else. -- -------- Sarr Blumson sarr@umich.edu voice: +1 313 764 0253 home: +1 313 665 9591 ITD, University of Michigan http://www-personal.umich.edu/~sarr/ 535 W William, Ann Arbor, MI 48103-4943 Article 3269 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!feta.direct.ca!newsfeed.direct.ca!news.he.net!news.pagesat.net!decwrl!pa.dec.com!depot.mro.dec.com!nntpd.lkg.dec.com!alingo.zk3.dec.com!werme From: werme@alingo.zk3.dec.com (Eric Werme) Newsgroups: alt.sys.pdp10,alt.folklore.computers,comp.sys.dec.micro Subject: Re: Dec 40th anniversary web pages Date: 5 Sep 97 12:04:22 GMT Organization: Digital Equipment Corporation Lines: 25 Message-ID: References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> Reply-To: werme@zk3.dec.com NNTP-Posting-Host: alingo.zk3.dec.com Xref: nntp1.ba.best.com alt.sys.pdp10:3269 alt.folklore.computers:91573 comp.sys.dec.micro:3272 inwap@best.com (Smith and O'Halloran) writes: > [1]Timeline at [2]Digital.com > * 1977 [16]ARPAnet comes to the PDP-10. The text in the timeline may be more accurate, "DIGITAL is the first computer company to connect to the ARPAnet. The connection is made via a PDP-10." I have no idea who did that, even though I left DEC in 1978. The PDP-10, of course, was on the net much earlier as several TENEX systems (possibly all of them) and a few TOPS-10 systems. The early code was written at Harvard, C-MU got involved around 1970/1. I took over the C-MU work in 1972 or so. After the main Harvard devo went to Xerox PARC (1973?) and I went to DEC (1974), Rutgers put a KI-10 on the net with our help (1975). I believe I installed the first FTP client in 9/1972 just before leaving for a vacation in Europe for a couple weeks. When I got back there were servers to talk to, and I found my client didn't work. Fixed quickly, but I'm not sure if I installed the first _working_ client. Ed Taft at Harvard wrote the TOPS-10 FTP server. Later he wrote the server for the "new" Telnet protocol and I wrote the client. -- <> Eric (Ric) Werme <> DEC's back! (As an officially approved <> <> <> nickname, we're really still DIGITAL.) <> Article 3268 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!agate!bh From: bh@anarres.CS.Berkeley.EDU (Brian Harvey) Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: 5 Sep 1997 19:37:19 GMT Organization: University of California, Berkeley Lines: 9 Message-ID: <5upn1f$bac@agate.berkeley.edu> References: <5tfkn8$8uh@src-news.pa.dec.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> NNTP-Posting-Host: anarres.cs.berkeley.edu Xref: nntp1.ba.best.com alt.sys.pdp10:3268 jmfbahxx@ma.ultranet.com writes: >Just as a side note: I don't remember KAs having paging hardware; is >somebody confusing a software implementation with hardware innovation? BBN built a hardware pager for the KA-10 as part of an ARPA-funded project to bring virtual memory to the PDP-10. The other part of the project was TENEX, which ran on these modified KA-10s (and later on other models). At that time Dan Murphy was at BBN. TENEX came with him to DEC, where it transmogrified into TOPS-20. Article 3271 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!newsfeed1-hme1!newsfeed.internetmci.com!18.181.0.26!bloom-beacon.mit.edu!senator-bedfellow.mit.edu!usenet From: Tom Knight Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: 05 Sep 1997 18:46:48 -0400 Organization: Massachvsetts Institvte of Technology Lines: 11 Sender: tk@watt Message-ID: References: <5tfkn8$8uh@src-news.pa.dec.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> <5upn1f$bac@agate.berkeley.edu> NNTP-Posting-Host: watt.ai.mit.edu X-Newsreader: Gnus v5.2.25/XEmacs 19.14 Xref: nntp1.ba.best.com alt.sys.pdp10:3271 bh@anarres.CS.Berkeley.EDU (Brian Harvey) writes: > BBN built a hardware pager for the KA-10 as part of an ARPA-funded > project to bring virtual memory to the PDP-10. The other part of the > project was TENEX, which ran on these modified KA-10s (and later on > other models). At that time Dan Murphy was at BBN. TENEX came with > him to DEC, where it transmogrified into TOPS-20. And well before BBN did this, paging hardware for the KA-10 was built in the MIT AI Lab by Jack Holloway. It was used for running ITS. Article 3279 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!news-out.communique.net!communique!decius.ultra.net!d4 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Sun, 07 Sep 97 11:45:56 GMT Organization: UltraNet Communications, Inc. Lines: 81 Message-ID: <5uu4ok$tfv$1@decius.ultra.net> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu><3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> NNTP-Posting-Host: d4.dial-21.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3279 Mark Crispin sent me the following text and I am reprinting it here with his permission. On Fri, 5 Sep 1997 jmfbahxx@ma.ultranet.com wrote: > >As it was explaned to me, Tenex came after TOPS-10, and it was written > >for KA-10's with paging hardware added. Tops-20 was adapted from it, > >and could run TOPS-10 with a compatibility layer. > TOPS10 _never, never, never_ ran under TOPS20. There existed a > compatibility program, called PA1050 (I think) that allowed programs that > did TOPS10-style UUOs to run in a TOPS20 operating system environment. > What PA1050 did was take the arguments to the UUOs (or monitor calls) and > convert them into arguments that TOPS20's JSYS calls would understand. So far, so good. PA1050 got its start on Tenex. When a program did an MUUO, if the "enable 10/50 compatibility" bit was set (and it normally was), then the monitor would automatically load PA1050.SAV and give it the trap. TOPS-20 version: When a program did an MUUO other than JSYS, if the "enable TOPS-10 compatibility" bit was set (and it normally was), then the monitor would automatically load PA1050.EXE and give it the trap. Once PA1050 was loaded, it would get all subsequent MUUOs directly. PA1050 lived in the high end of the address space, at about 700000. > For instance, there wasn't such a thing as a LOOKUP monitor call on >TOPS20. I'm not sure about this, but I don't think there any such >thing as a CALLI UUO equivalence on TOPS20. LOOKUP and CALLI were both implemented by PA1050. JSYS isn't much different in concept from CALLI, except that the AC argument was ignored by JSYS. > Just as a side note: I don't remember KAs having paging hardware; is > somebody confusing a software implementation with hardware innovation? Paging hardware for the KA was built by both BBN and Systems Concepts. Systems Concepts pagers were used on ITS machines. BBN pagers were used on all KA Tenex systems as well as the WAITS system at Stanford. BBN pagers had what was later called "KL paging" or "microcode paging" in pretty much its full glory (or the Model A equivalent thereof: direct/share/indirect pointers, SPT, CSTs, etc.) and added JSYS as a machine instruction. JSYS was a pure code JSR; the effective address pointed to a word in which the left half contained a pointer to a place to save the old PC, and the right half pointed to the new PC, hence: JSYS [RET,,SUBR] called a subroutine that would look something like: SUBR: ... JRSTF @RET If E was less than 1000, then it would get the JSYS pointer from physical memory (I think that it added 1000, so JSYS 23 uses physical 1023 -- I'd have to check my Tenex manuals to be sure) and leave user mode. If E was greater than 1000, then it would get the pointer from user space and stay in user mode ("user mode JSYS"). Tenex on the KI used the KI's pager, and JSYS was implemented as an MUUO. The KI pager was very poor, so KI Tenex did not perform all that well in terms of timesharing over KA Tenex, but it did crunch numbers a bit better. KI Tenex systems were often dual-CPU, and found more favor in the AI community than with more timeshared environments. TOPS-20 implemented the BBN pager (and later went beyond it on Model B). JSYS was still implemented as an MUUO. From time to time, I've thought of implementing JSYS in microcode on my KS just to see if it would do any better. User mode JSYS fell out of favor after KA Tenex, but even the latest versions of TOPS-20 still implemented it to support old software that uses it. It's very slow, since it's an MUUO. -- Mark -- Article 3280 of alt.sys.pdp10: Path: nntp1.ba.best.com!not-for-mail From: inwap@best.com (Smith and O'Halloran) Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: 7 Sep 1997 15:08:50 -0700 Organization: Chez INWAP (people, computers, cats) Lines: 20 Message-ID: <5uv8li$qpc$1@shell3.ba.best.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> <5uu4ok$tfv$1@decius.ultra.net> NNTP-Posting-Host: shell3.ba.best.com Xref: nntp1.ba.best.com alt.sys.pdp10:3280 In article <5uu4ok$tfv$1@decius.ultra.net>, wrote: >Mark Crispin sent me the following text and I am reprinting it here with >his permission. >>When a program did an MUUO, if the "enable 10/50 compatibility" bit was >set (and it normally was), then the monitor would automatically load >PA1050.SAV and give it the trap. With one notable exception. If the MUUO was GETTAB [CALLI 41] and the contents of the accumulator was the right value to get the operating system type, then TOPS20 would provide a return value without invoking PA1050. This was so that programs could determine at run time whether they should use MUUOs or JSYS for the remainder of the program. Late in the game for TOPS10 7.0x, I got Jim Flemming to add a code for TYMCOM-X, the PDP-10 operating system used by Tymshare, to the list of legal return values. -Joe -- INWAP.COM is Joe and Sally Smith, John and Chris O'Halloran and our cats See http://www.inwap.com/ for "ReBoot", PDP-10, and Clan MacLeod. Article 3286 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!vncnews!HSNX.wco.com!news.wco.com!not-for-mail From: "Henry W. Miller" Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Mon, 08 Sep 1997 10:00:59 -0700 Organization: Cisco Systems Lines: 70 Message-ID: <34142F4A.46C7@cisco.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu><3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> <341398B7.4B3@cisco.com> <5v0p86$nd6$1@decius.ultra.net> NNTP-Posting-Host: hmiller.mistic.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (Win95; I) Xref: nntp1.ba.best.com alt.sys.pdp10:3286 jmfbahxx@ma.ultranet.com wrote: > > In article <341398B7.4B3@cisco.com>, > "Henry W. Miller" wrote: > > > >jmfbahxx@ma.ultranet.com wrote: > >> > >> [unfamiliar newsgroups snipped] > >> > >> In article <01bcb4cd$721363e0$1d49cf8d@wardch>, > >> "Chris Ward" wrote: > >> > > >> > > >> >Richard M. Alderson III wrote in article > >> >... > >> >> In article <5u50u7$iiu$1@shell3.ba.best.com> inwap@best.com > >> >> (Smith and O'Halloran) writes: > >> >> > >> >> >* 1976 [15]TOPS-20 was the indirect descendent from TOPS-10 (direct > >> >> > descendent from BB&N's TENEX operating system). > >> >> > >> >As it was explaned to me, Tenex came after TOPS-10, and it was written > >> >for KA-10's with paging hardware added. Tops-20 was adapted from it, > >> >and could run TOPS-10 with a compatibility layer. > >> > > >> > >> Who explained this to you? I really would like to know so that these > >> misconceptions can be corrected. > >> > >> TOPS10 _never, never, never_ ran under TOPS20. There existed a > >> compatibility program, called PA1050 (I think) that allowed programs > >> that did TOPS10-style UUOs to run in a TOPS20 operating system > >> environment. What PA1050 did was take the arguments to the UUOs (or > >> monitor calls) and convert them into arguments that TOPS20's JSYS calls > >> would understand. > >> > > > > I ***THINK*** that he meant that it could run (most) TOPS10 > >programs. > > He may have meant that but it surely isn't how I read it. My dander > just got het up; old rivalries and all. I didn't intend for my > message to stomp on the person...just the notion. :-) > I supported them both; I loved them both. TOPS10 was better in some environments; TOPS20 was better in others. > >(Surely not all - anyone besides me ever try to launch > >TOPS10 DAEMON on TOPS20?) > > > > ROTFL. It would have never occurred to me. Now I'm trying to remember > what DAEMON did. It was considered to be the swappable part of the > monitor. You don't, by any chance, remember what happened? Now you've got > me curious. > The system hung in a strange and bizarre way that I could not explain. All of the subsystems were unresponsive. I finally had to break into the Front-End console and halt the system. And, actually, over the years, I think that DAEMON evolved to mainly a super mode user program, that portions of TOPS10 would call and then block until it's return. Maybe that's what happened when I tried running it under TOPS20 - I don't recall to what extent, if any, PA1050 "supported" the DAEMON. call. > /BAH -HWM Article 3293 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!news-peer.sprintlink.net!news-sea-19.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d6 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Tue, 09 Sep 97 10:55:11 GMT Organization: UltraNet Communications, Inc. Lines: 88 Message-ID: <5v3ai0$oud$1@decius.ultra.net> References: <5tfkn8$8uh@src-news.pa.dec.com> <5tsm8d$sa0$1@newbabylon.rs.itd.umich.edu><3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <01bcb4cd$721363e0$1d49cf8d@wardch> <5up1a3$bso$1@decius.ultra.net> <341398B7.4B3@cisco.com> <5v0p86$nd6$1@decius.ultra.net> <34142F4A.46C7@cisco.com> NNTP-Posting-Host: d6.dial-19.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3293 In article <34142F4A.46C7@cisco.com>, "Henry W. Miller" wrote: >jmfbahxx@ma.ultranet.com wrote: >> >> In article <341398B7.4B3@cisco.com>, >> "Henry W. Miller" wrote: >> > >> >jmfbahxx@ma.ultranet.com wrote: >> >> >> >> [unfamiliar newsgroups snipped] >> >> >> >> In article <01bcb4cd$721363e0$1d49cf8d@wardch>, >> >> "Chris Ward" wrote: >> >> > >> >> > >> >> >Richard M. Alderson III wrote in >> >> >article >> >> >... >> >> >> In article <5u50u7$iiu$1@shell3.ba.best.com> inwap@best.com >> >> >> (Smith and O'Halloran) writes: >> >> >> >> >> >> >* 1976 [15]TOPS-20 was the indirect descendent from TOPS-10 (direct >> >> >> > descendent from BB&N's TENEX operating system). >> >> >> >> >> >As it was explaned to me, Tenex came after TOPS-10, >> >> >and it was written >> >> >for KA-10's with paging hardware added. Tops-20 >> >> >was adapted from it, >> >> >and could run TOPS-10 with a compatibility layer. >> >> > >> >> >> >> Who explained this to you? I really would like to know so that these >> >> misconceptions can be corrected. >> >> >> >> TOPS10 _never, never, never_ ran under TOPS20. There existed a >> >> compatibility program, called PA1050 (I think) that allowed programs >> >> that did TOPS10-style UUOs to run in a TOPS20 operating system >> >> environment. What PA1050 did was take the arguments to the UUOs (or >> >> monitor calls) and convert them into arguments that TOPS20's JSYS >> >> calls >> >> would understand. >> >> >> > >> > I ***THINK*** that he meant that it could run (most) TOPS10 >> >programs. >> >> He may have meant that but it surely isn't how I read it. My dander >> just got het up; old rivalries and all. I didn't intend for my >> message to stomp on the person...just the notion. :-) >> > > I supported them both; I loved them both. TOPS10 was better in >some environments; TOPS20 was better in others. Understood. My whole being was dedicated to keeping TOPS10 viable. There wasn't any time to learn to love TOPS20 although I respected the work and, generally, liked the people. It's a psychological thing that I'll probably never completely get rid of. :-) > >> >(Surely not all - anyone besides me ever try to launch >> >TOPS10 DAEMON on TOPS20?) >> > >> >> ROTFL. It would have never occurred to me. Now I'm trying to remember >> what DAEMON did. It was considered to be the swappable part of the >> monitor. You don't, by any chance, remember what happened? Now you've >> got me curious. >> > > The system hung in a strange and bizarre way that I could not >explain. All of the subsystems were unresponsive. I finally had to >break into the Front-End console and halt the system. And, actually, >over the years, I think that DAEMON evolved to mainly a super mode user >program, that portions of TOPS10 would call and then block until it's >return. Maybe that's what happened when I tried running it under >TOPS20 - I don't recall to what extent, if any, PA1050 "supported" the >DAEMON. call. Gad. I hope not other than a POPJ P, . Maybe it was one of the SYSERR aspects of DAEMON. And, you're right, it did end up as a very priviledged user mode program. I don't know what the original plans were for a swappable monitor segment. Hmmm. I can't remember who wrote it. I know that Cliff Romash did a bit of work on it (time frame = early Marlboro years). /BAH Article 3287 of alt.sys.pdp10: Newsgroups: alt.folklore.computers,vmsnet.pdp-11,alt.sys.pdp11,alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!uunet!in4.uu.net!uucp5.uu.net!world!mch From: mch@world.std.com Subject: Re: What "SOS" is. Message-ID: Organization: The World Public Access UNIX, Brookline, MA References: <5ukng7$i6g$1@sue.cc.uregina.ca> <5uopad$96d$1@decius.ultra.net> <5up8aa$4aj$1@nntp.ucs.ubc.ca> <5ustht$3bv$1@decius.ultra.net> Date: Mon, 8 Sep 1997 17:42:37 GMT Lines: 18 Xref: nntp1.ba.best.com alt.folklore.computers:91691 vmsnet.pdp-11:5145 alt.sys.pdp11:2822 alt.sys.pdp10:3287 In article <5ustht$3bv$1@decius.ultra.net>, John Santos wrote: >I first learned TECO on a PDP-11/45 running some strange, heavily hacked >1975 version of UNIX. I wonder if it is still available somewhere, for >UNIX in general? Yes, you can have TECO on many UNIX versions... ftp://ftp.mindlink.net/pub/teco ftp://ftp.usc.edu/pub/teco In fact, this message was edited using a verison of TECO running under IRIX 5.3 -- Mark Henderson -- mch@squirrel.com, mch@world.std.com -- NIC: MCH14 PGP/MIME key 1024/B2667EFF - 5A 93 7D 29 EB 63 84 09 EA A2 AA 6C FA C5 A6 21 unstrip for Solaris, MIND LINK crypto archive, TECO, computer security links, change-sun-hostid, Sun NVRAM/hostid FAQ - http://www.squirrel.com/squirrel/ Article 3288 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newsfeeds.sol.net!ix.netcom.com!netcom15!alderson From: alderson@netcom15.netcom.com (Richard M. Alderson III) Subject: Re: More on the pdp-10 emulator In-Reply-To: shoppa@alph02.triumf.ca's message of 6 Sep 1997 15:38:03 GMT Message-ID: Sender: alderson@netcom15.netcom.com Reply-To: alderson@netcom.com Organization: NETCOM On-line services References: <01bcbac1$0b089ea0$50c906c6@hsnewman.ccsi.com> <5urtcr$h80$1@nntp.ucs.ubc.ca> Date: Mon, 8 Sep 1997 18:03:32 GMT Lines: 9 Xref: nntp1.ba.best.com alt.sys.pdp10:3288 The PDP-10 emulator on the Gates Alpha is the Harrenstien KL10B version. Oddly enough, we received an order for a Toad-1 from Paul Allen about that time. -- Rich Alderson You know the sort of thing that you can find in any dictionary of a strange language, and which so excites the amateur philo- logists, itching to derive one tongue from another that they know better: a word that is nearly the same in form and meaning as the corresponding word in English, or Latin, or Hebrew, or what not. --J. R. R. Tolkien, alderson@netcom.com _The Notion Club Papers_ Article 3289 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!newsfeed1-hme1!newsfeed.internetmci.com!206.72.192.13!news.albany.net!news.wizvax.net!dbit.com!wilson From: wilson@dbit.com (John Wilson) Newsgroups: alt.sys.pdp10 Subject: Re: More on the pdp-10 emulator Date: 8 Sep 1997 18:29:48 GMT Organization: D Bit, Troy, NY Lines: 17 Message-ID: <5v1g6s$1lm$1@news.wizvax.net> References: <01bcbac1$0b089ea0$50c906c6@hsnewman.ccsi.com> <5urtcr$h80$1@nntp.ucs.ubc.ca> NNTP-Posting-Host: dbit.dbit.com Xref: nntp1.ba.best.com alt.sys.pdp10:3289 In article , Richard M. Alderson III wrote: >The PDP-10 emulator on the Gates Alpha is the Harrenstien KL10B version. > >Oddly enough, we received an order for a Toad-1 from Paul Allen about that >time. Actually it might not be that odd, the DECtapes that Paul Allen's guy brought me to read were old enough that they probably dated back to when Gates and Allen were using a 10 at Harvard, so maybe they both want to play with their old toys. I see Paul Allen's web page has a pointer to Inwap so I suspect he's got the Toad because he's interested, not just because he's trying to preserve some boring old application. It's great to see those two messing around with stuff from the days before programmers forgot how to program! John Wilson D Bit Article 3295 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newshub1.home.com!news.home.com!feta.direct.ca!newsfeed.direct.ca!newsfeed1-hme1!newsfeed.internetmci.com!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: 9 Sep 1997 14:01:41 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 41 Message-ID: <5v3ks5$a94$2@kirin.wwa.com> References: <5tfkn8$8uh@src-news.pa.dec.com> <3402352B.76BC@cisco.com> <5u50u7$iiu$1@shell3.ba.best.com> <5u8m2t$9o2$1@shell3.ba.best.com> <5uk5be$quq$1@kirin.wwa.com> <5um370$k2u$1@decius.ultra.net> NNTP-Posting-Host: poolf1-055.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3295 In article , mbg@world.std.com says... > >>>Which reminds me (for some strange reason), how many others remember >>>(accurately) the origin of the name Leprechaun Lounge for a conference >>>room in the Marlboro facility? I'll bet the story has been through a few >>>permutations over the years. >>> > >>I remember how the aisles got their names but, for the life of me, I can't >>recall the naming of the lounge. > >Neither do I... and I worked on ML5-5 from 1978 to 1988 > When the PDP-10 Programming Group moved from 3-5 to 5-5 in 1971 or 1972, Allan Pommer was put in charge of the logistics. He helped layout the offices, cubes, and conference rooms. One of the things he did was pick out the colors. He came around one day with one of those paint company color chip cards to show us the colors he had picked. I seem to recall there were three conference rooms in the new layout, and Allan picked out three different colors for them. One was a very pale green the paint company called Leprechaun. Allan named the three conference rooms after the colors he picked, calling the green one the Leprechaun Lounge. I can't remember the names of the other two. The regular Monday Morning Monitor Group meetings were always held in the Leprechaun Lounge. When we moved to Marlboro, for some reason the name followed. The place where the meetings were held continued to be called the Leprechan Lounge, even though the walls were no longer Leprechaun. After a while the Monitor Group meetings were moved to a different room, but this time the name didn't follow. Does anyone know if there's still a Leprechaun Lounge at DEC? Come to think of it, is there still a Marlboro facility at DEC? -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3301 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!europa.clark.net!131.103.1.102!news2.chicago.cic.net!iagnet.net!128.255.40.11!news.uiowa.edu!not-for-mail From: jones@pyrite.cs.uiowa.edu (Douglas W. Jones,201H MLH,3193350740,3193382879) Newsgroups: alt.sys.pdp10,alt.folklore.computers,alt.sys.pdp8 Subject: The DEC PDP-3 Date: 10 Sep 1997 15:25:04 GMT Organization: The University of Iowa Lines: 26 Message-ID: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> NNTP-Posting-Host: pyrite.cs.uiowa.edu Xref: nntp1.ba.best.com alt.sys.pdp10:3301 alt.folklore.computers:91799 alt.sys.pdp8:2074 In my recently acquired copy of "digital Building Blocks, computers, systems" published by DEC in 1960, the following paragraph is repeated at the end of each of a number of sections: DEC Programmed Data Processor-3 (PDP-3) -- a large scale, solid state, general purpose digital computer with the same high speed and high reliability as the PDP-1. Also a single address, single instruction, stored program ma- chine, PDP-3 operates in parallel on 36 bit numbers. The main storage is coincident current magnetic core mod- ules of 4096 words each, with a memory cycle time of 5 microseconds. This version is verbatim from the end of DEC Publication number B-3000 dated 11/60; the same text is at the end of DEC Publication B-100 dated 7/60, C-4000A dated 11/60, A-710 dated 8/60. So, while DEC may never have built a PDP-3, and while only one may have been built, by a customer, DEC did advertise it as a product, if only briefly! The spiral bound set of DEC publications I got includes a whole section of material on the PDP-1, including photos, but nothing but the above paragraph on the PDP-3. Doug Jones jones@cs.uiowa.edu Article 3307 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!news-dc-10.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!199.232.56.18!news.ultranet.com!d8 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Thu, 11 Sep 97 10:44:24 GMT Organization: UltraNet Communications, Inc. Lines: 17 Message-ID: <5v8imc$ps7$1@decius.ultra.net> References: <5tfkn8$8uh@src-news.pa.dec.com> <5v0p86$nd6$1@decius.ultra.net> <5v5gm5$l1e$1@shell3.ba.best.com> <5v5uvu$98h$1@decius.ultra.net> <5v7c18$si5@agate.berkeley.edu> NNTP-Posting-Host: d8.dial-15.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3307 In article <5v7c18$si5@agate.berkeley.edu>, bh@anarres.CS.Berkeley.EDU (Brian Harvey) wrote: >jmfbahxx@ma.ultranet.com writes: >> Look up the word daemon in a >>dictionary. That's what JMF told the filename came from. > >By the time of the PDP-10, the name "daemon" for user-mode kernel helpers >was well established on other systems. I'm pretty sure CTSS had one. Oh, I wasn't claiming that TOPS10 had coined the term. I was just speculating that the letters didn't come from cybercrud. Programs were called daemon if they worked in the background doing nefarious things that nobody understood and had the power to make life really miserable if one wasn't good [emoticon with a twinkle in it's eye]. /BAH Article 3310 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!newsfeed.internetmci.com!newsfeed.internetmci.com!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10,alt.folklore.computers,alt.sys.pdp8 Subject: Re: The DEC PDP-3 Date: 11 Sep 1997 13:59:32 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 38 Message-ID: <5v8tg4$4sf$1@kirin.wwa.com> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> NNTP-Posting-Host: poolf2-059.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3310 alt.folklore.computers:91857 alt.sys.pdp8:2080 In article <5v8mnc$2cf$1@kirin.wwa.com>, jeverett@wwa.DEFEAT.UCE.BOTS.com says... > >In article <34172B67.54569811@stoneweb.com>, carl.friend@stoneweb.com says... >> >> We all know how the PDP-6 was the "father" of the PDP-10, and set >>the tone for the instruction set; > >It did more than "set the tone" for the PDP-10's instruction set, it was >identical. The first instruction I recall appearing on the -10 that wasn't on >the -6 was the JFFO, but I don't remember if that was on the KA, or the KI. > Bob Clements, who lurks here but never posts to avoid spam, emailed to remind me that 1)the JFFO first appeared on the KA, and 2)there were significant changes to the suite of floating point instructions and data format between the -6 and the KA. I'd forgotten about the floating point changes, probably because in years of writing MACRO code, I don't think I ever wrote a floating point instruction. Bob should know, he and Alan Kotok were the logic designers on the KA. Although Dick Gruen had business cards that gave his position as "Factotum", Bob was the real factotum at DEC. In addition to his design work, he also spent some time in the Monitor Group. After he had left DEC for BBN, we spent some time at one of the Monitor Group meetings trying to come up with an RCC STOPCD in "honor" of Bob. Since he had spent some time working with SCNSER, Don Lewine suggested it would only be right and proper that the RCC STOPCD be in that module. Eventually we came up with "range check chunks", and so it was. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3311 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!newsfeed.internetmci.com!newsfeed.internetmci.com!198.60.22.3!xmission!rivie From: rivie@rivie.daautah.com (Roger Ivie) Newsgroups: alt.sys.pdp10,alt.folklore.computers,alt.sys.pdp8 Subject: Re: The DEC PDP-3 Date: 11 Sep 1997 16:23:46 GMT Organization: XMission Internet (801 539 0900) Lines: 34 Message-ID: References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> Reply-To: rivie@cc.usu.edu NNTP-Posting-Host: logan38.modem.xmission.com X-Newsreader: slrn (0.8.8.2 UNIX) Xref: nntp1.ba.best.com alt.sys.pdp10:3311 alt.folklore.computers:91861 alt.sys.pdp8:2081 In article <34172B67.54569811@stoneweb.com>, Carl R. Friend wrote: >Douglas W. Jones,201H MLH,3193350740,3193382879 wrote: >> >> So, while DEC may never have built a PDP-3, and while only one may >> have been built, by a customer, DEC did advertise it as a product, >> if only briefly! > > If anybody here has any info about the PDP-3, it'd be interesting >to see it. In a late '50s or early '60s "Encyclopedia of Computing" or somesuch (I don't have the exact reference handy), there is a two-page blurb on the PDP-1 and the PDP-3. The blurb concentrates on the PDP-3 and mentions the PDP-1 only as an 18-bit version of the PDP-3. The only thing resembling technical information in the blurb is a drawing of the instruction word. I don't have the photocopy I made handy, so I can't reproduce it at the moment. The volume in question had two-page blurbs on just about everything starting with ENIAC and going to some Ramo-mumble 5-processor mainframe. It was really interesting, but I didn't think I could get away with swiping it... BTW, unless I listened incorrectly, Art Bell's guest on Monday claimed the PDP-1 was designed by space aliens. -- Roger Ivie Design Analysis Associates 75 West 100 South Logan, UT 84321 mailto:rivie@daa-utah.com phoneto:(801)753-2212 Article 3323 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!europa.clark.net!206.229.87.25!news-peer.sprintlink.net!news.sprintlink.net!Sprint!ix.netcom.com!news From: JCGreen@ix.netcom.com (John C Green Jr) Newsgroups: alt.sys.pdp10 Subject: JFFO (Was Re: The DEC PDP-3) Date: Sat, 13 Sep 1997 00:23:41 GMT Organization: Netcom Lines: 36 Message-ID: <5vcmgh$sqq@sjx-ixn10.ix.netcom.com> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> <5v8tg4$4sf$1@kirin.wwa.com> NNTP-Posting-Host: sjc-ca4-03.ix.netcom.com X-NETCOM-Date: Fri Sep 12 5:24:49 PM PDT 1997 X-Newsreader: Forte Free Agent 1.0.82 Xref: nntp1.ba.best.com alt.sys.pdp10:3323 jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: >jeverett@wwa.DEFEAT.UCE.BOTS.com >says... >> >>The first instruction I recall appearing on the -10 that >>wasn't on the -6 was the JFFO, but I don't remember if >>that was on the KA, or the KI. > >Bob Clements, who lurks here but never posts to avoid spam, >emailed to remind me that 1)the JFFO first appeared on the >KA, and 2)there were significant changes to the suite of >floating point instructions and data format between the -6 >and the KA. I'd forgotten about the floating point changes, >probably because in years of writing MACRO code, I don't >think I ever wrote a floating point instruction. Bob should >know, he and Alan Kotok were the logic designers on the KA. JFFO was not in the original instruction set of the KA which was first shipped in DEC 1967. It was was added in 1968 as the 366th instruction in honor of 1968 being a leap year. Just kidding. Although it was mentioned. I had the pleasure of having the office (5-5 Pole 41A) adjacent to the double office RCC and AK shared in the Summer of '68 when JFFO was added. It was done at the request of a prospect who was trying to code a minimal execution time 'population count' routine and promised to become a customer if it was available. I don't remember who it was or whether they actually bought a machine. - - - RP10 (controller for RP02) Project Engineer 21483 Old Mine Rd (408)353-1870 Los Gatos CA 95030 JCGreen@ix.netcom.com Article 3328 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news-pull.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!207.241.0.194!news.wwa.com!not-for-mail From: jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) Newsgroups: alt.sys.pdp10 Subject: Re: JFFO (Was Re: The DEC PDP-3) Date: 13 Sep 1997 13:26:56 GMT Organization: WorldWide Access (tm) - Midwestern Internet Services http://www.wwa.com Lines: 63 Message-ID: <5ve4b0$jsr$1@kirin.wwa.com> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <5v8mnc$2cf$1@kirin.wwa.com> <5v8tg4$4sf$1@kirin.wwa.com> <5vcmgh$sqq@sjx-ixn10.ix.netcom.com> <5vd0v8$dhl@bonkers.taronga.com> NNTP-Posting-Host: digi1-002.wwa.com Mime-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII X-Newsreader: WinVN 0.99.8 (x86 32bit) Xref: nntp1.ba.best.com alt.sys.pdp10:3328 In article <5vd0v8$dhl@bonkers.taronga.com>, peter@taronga.com says... > >In article <5vcmgh$sqq@sjx-ixn10.ix.netcom.com>, >John C Green Jr wrote: >>I had the pleasure of having the office (5-5 Pole 41A) adjacent >>to the double office RCC and AK shared in the Summer of '68 >>when JFFO was added. It was done at the request of a prospect >>who was trying to code a minimal execution time 'population >>count' routine and promised to become a customer if it was >>available. I don't remember who it was or whether they >>actually bought a machine. > >That would have been the NSA, right? Wrong. What follows is an excerpt from a 1992 posting by Bob Clements (RCC). --------------Beginning of quoted material--------------------- JFFO was added very late in the development cycle, at the request of a potential customer. The customer was a European telephone exchange maker. They wanted to use it to scan for changed bits representing phone line occupancies. This change was, of course, a new incompatibility with the predecessor machine, the PDP-6, so we were reluctant to add the instruction. We asked Marketing how many machines we would sell if we added it. They said "At least 5 for a testbed, and hundreds if they decide to go ahead with the design." So we added it. It was 1968, which was a leap year. I wrote the blurb for the internal Sales Newsletter, proclaiming that in honor of Leap Year, we were increasing the number of instructions in the machines instruction set from 365 to 366. I still have a copy of that newsletter somewhere, I think. End result: The potential customer never bought a single machine. -------------End of quotation----------------------------------- The above reminds me, during the early years of the -10, DEC marketing seemed to always be seeing potential sales "if only we could implement (your feature here)". My entry into the Monitor Group was via one of these schemes. Someone thought we could sell x number of machines if only we had a Real Time Monitor, so a group we formed to create one. I was Group Leader. My "group" consisted of Jim Flemming and Kerry Bensman. The object was to create an "uber operating system" that would preempt the running Monitor (not yet named TOPS-10) to provide service to some external stimulus. After several months of design and coding, but before we could even get to debugging, Marketing identified yet another way to sell systems. Jim was pulled off the project to work on the ORELA (Oak Ridge Electron Linear Accelerator) machine, I was pulled off to turn LOADER into a "virtual memory" loader so several customers who were "loaned" memory so that they could link the monitor would either be forced to pay for it or give it back. I don't recall what happened to Kerry. At the end of my little project, which produced a horrible abortion known as MONLOD (MONitor LOaDer), I was absorbed by the monitor group. First task; implement extended RIBs. -- jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) http://www.wwa.com/~jeverett ^^^^^^^^^^^^^^^ Things have gotten so bad I feel the need to disguise my email address. And I don't like this explanation because I just hate long signatures. Article 3319 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!cam-news-hub1.bbnplanet.com!news.bbnplanet.com!news-feed1.tiac.net!news.ultranet.com!d1 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: The DEC PDP-3 Date: Fri, 12 Sep 97 11:25:46 GMT Organization: UltraNet Communications, Inc. Lines: 27 Message-ID: <5vb9g5$3tq$1@decius.ultra.net> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> NNTP-Posting-Host: d1.dial-12.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3319 [newsgroups snipped] In article <5v8mnc$2cf$1@kirin.wwa.com>, jeverett@wwa.DEFEAT.UCE.BOTS.com (John Everett) wrote: >In article <34172B67.54569811@stoneweb.com>, >carl.friend@stoneweb.com says... >> >> We all know how the PDP-6 was the "father" of the PDP-10, and set >>the tone for the instruction set; > >It did more than "set the tone" for the PDP-10's instruction set, it was >identical. The first instruction I recall appearing on the -10 that wasn't >on the -6 was the JFFO, but I don't remember if that was on the KA, or the >KI. > Well, I've looked for my instruction set card and now can't manage to find it, so I can't answer the question for sure. There's a TW story connected with JFFO. Tony always said that, when writing code, he constantly looked for a legitimate opportunity to be able to use a JFFO instruction. One day, he came back from the hardware lab (I can't remember if he was coding the RH20 or the RP05s) and announced to Jim that he got to write a JFFO. He looked like he was on cloud nine. /BAH Article 3329 of alt.sys.pdp10: Path: nntp1.ba.best.com!nntp2.ba.best.com!thvv.vip.best.com!user From: thvv@best.com (Tom Van Vleck) Newsgroups: alt.sys.pdp10 Subject: Re: Dec 40th anniversary web pages Date: Sat, 13 Sep 1997 07:12:13 -0700 Organization: Multicians Lines: 12 Message-ID: References: <5tfkn8$8uh@src-news.pa.dec.com> <5v0p86$nd6$1@decius.ultra.net> <5v5gm5$l1e$1@shell3.ba.best.com> <5v5uvu$98h$1@decius.ultra.net> <5v7c18$si5@agate.berkeley.edu> NNTP-Posting-Host: thvv.vip.best.com X-Trace: 874159944 6314 (none) 206.86.0.12 X-url: http://www.best.com/~thvv/multics.html Xref: nntp1.ba.best.com alt.sys.pdp10:3329 In article <5v7c18$si5@agate.berkeley.edu>, bh@anarres.CS.Berkeley.EDU (Brian Harvey) wrote: >jmfbahxx@ma.ultranet.com writes: >> Look up the word daemon in a >>dictionary. That's what JMF told the filename came from. > >By the time of the PDP-10, the name "daemon" for user-mode kernel helpers >was well established on other systems. I'm pretty sure CTSS had one. Yup. Term attributed to M. J. Bailey circa 1964. See http://www.best.com/~thvv/mgd.html#daemon in the Multics glossary. Article 3336 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!su-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!cam-news-hub1.bbnplanet.com!news.bbnplanet.com!news-feed1.tiac.net!news.ultranet.com!d15 From: jmfbahxx@ma.ultranet.com Newsgroups: alt.sys.pdp10 Subject: Re: The JFFO instruction [Was Re: The DEC PDP-3] Date: Sun, 14 Sep 97 11:13:44 GMT Organization: UltraNet Communications, Inc. Lines: 72 Message-ID: <5vghi9$30e$2@decius.ultra.net> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> <5vb9g5$3tq$1@decius.ultra.net> <3419E809.6BB2FA03@stoneweb.com> <5ve023$uei$1@decius.ultra.net> NNTP-Posting-Host: d15.dial-18.mbo.ma.ultra.net X-Newsreader: News Xpress Version 1.0 Beta #4 Xref: nntp1.ba.best.com alt.sys.pdp10:3336 [article dropped in with Mark Crispin's permission] In article <5ve023$uei$1@decius.ultra.net>, jmfbahxx@ma.ultranet.com wrote: >In article <3419E809.6BB2FA03@stoneweb.com>, > "Carl R. Friend" wrote: >>jmfbahxx@ma.ultranet.com, in article >>nr. <5vb9g5$3tq$1@decius.ultra.net>, wrote: >>> >>> Tony always said that, when writing code, he constantly looked for a >>> legitimate opportunity to be able to use a JFFO instruction. One >>> day, he came back from the hardware lab (I can't remember if he was >>> coding the RH20 or the RP05s) and announced to Jim that he got to >>> write a JFFO. He looked like he was on cloud nine. >> >> I'd imagine that JFFO might be one of the less-commonly used >>instructions, but they _were_ used. >> >> I remember a particularly nasty failure in a KI back at ADP that >>only affected the JFFO and was data-sensitive; the machine would pass >>all the diags with margins active happily until the proverbial cows >>came home. I had to find the problem on a live system (UGH!). I >>think I still have the failed module here someplace. Sadly, all my >>notes are long gone... >> >UGH? That must have been fun. The reason that TW was so happy was >that he was a very bitty-bytey programmer. He never could justify >using the JFFO if there was a better, faster instruction or method to use. > >I'd be interested in how people used the instruction. Get out your >old rusty memories, polish them up [pun definitely intended], and >see if you can describe your technique. [grinning emoticon here] > The TOPS-20 IMAP server used JFFO three times, to translate between a bitmask of flags into an index into a vector of pointers to the flag names. I was particularly proud of that program. It was the last major assembly language program that I ever wrote, and perhaps the only "true" native TOPS-20 program ever written. There was no JOBDAT or section 0, not even a vestigal trace; it compiled and lunk in section 1 using PSECTs and PDVs. From its inception, it was designed to run in extended addressing; it made good use of OWGBPs and MOVSLJ [I later regretted that decision, since that meant it couldn't run on my 2020...]. Of course, it used the infamous MACSYM macros -- there was not a single JRST to be found. Alas, it was deployed only on a couple of TOPS-20 machines; SUMEX-AIM and SIMTEL20 are all that come to mind. I sometimes think that IMAP could have saved the PDP-10 -- it was a *great* IMAP engine, much better than UNIX -- but IMAP didn't start taking off until the mid-1990s, nearly 10 years too late. I doubt that anyone other than me used its on SIMTEL20. It never went beyond RFC 1176 support (basic IMAP2); if I were to bring up an IMAP server on a TOPS-20 system today I would probably think more about building the UNIX server than doing the work to write a MIME parser in assembly language. In any case, it was a loving tribute to the most wonderful machine architecture ever designed. I'll be nice to you TOPS-10 and ITS folks and not say "and the best operating system". ;-) [Note from /BAH: Oh, you can say it, I just would argue good-naturedly with the opposing view :-). BTW, what the hell does a semi-colon mean in these damn smileys? It's taken me two years just to cope with a colon.] You can still get its source, on: ftp://ftp.cac.washington.edu/mail/tops-20/mapser.tar.Z Article 3346 of alt.sys.pdp10: Newsgroups: alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!news.maxwell.syr.edu!ix.netcom.com!netcom3!alderson From: alderson@netcom3.netcom.com (Richard M. Alderson III) Subject: Re: The JFFO instruction [Was Re: The DEC PDP-3] In-Reply-To: jmfbahxx@ma.ultranet.com's message of Sun, 14 Sep 97 11:13:44 GMT Message-ID: Sender: alderson@netcom3.netcom.com Reply-To: alderson@netcom.com Organization: NETCOM On-line services References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> <5vb9g5$3tq$1@decius.ultra.net> <3419E809.6BB2FA03@stoneweb.com> <5ve023$uei$1@decius.ultra.net> <5vghi9$30e$2@decius.ultra.net> Date: Mon, 15 Sep 1997 17:52:46 GMT Lines: 35 Xref: nntp1.ba.best.com alt.sys.pdp10:3346 In article <5vghi9$30e$2@decius.ultra.net> jmfbahxx@ma.ultranet.com writes: >[article dropped in with Mark Crispin's permission] >The TOPS-20 IMAP server... >... perhaps the only "true" native TOPS-20 program ever written. There was no >JOBDAT or section 0, not even a vestigal trace; it compiled and lunk in >section 1 using PSECTs and PDVs. >From its inception, it was designed to run in extended addressing; it made >good use of OWGBPs and MOVSLJ [I later regretted that decision, since that >meant it couldn't run on my 2020...]. Of course, it used the infamous MACSYM >macros -- there was not a single JRST to be found. I was going to argue about "the only 'true' native" by pointing to the XKL version of Tops-20 itself (no section 0, etc.), but a case can be made against it since it was *not* designed to run in extended addressing from its start, a fact we have had reason to curse for years... >In any case, it was a loving tribute to the most wonderful machine >architecture ever designed. I'll be nice to you TOPS-10 and ITS folks and not >say "and the best operating system". ;-) >[Note from /BAH: Oh, you can say it, I just would argue good-naturedly >with the opposing view :-). BTW, what the hell does a semi-colon mean >in these damn smileys? It's taken me two years just to cope with a colon.] That's a wink, ma'am. I often use ";->" which is a sardonic grin and a wink. (Or perhaps it's a smirk and a wink?) -- Rich Alderson You know the sort of thing that you can find in any dictionary of a strange language, and which so excites the amateur philo- logists, itching to derive one tongue from another that they know better: a word that is nearly the same in form and meaning as the corresponding word in English, or Latin, or Hebrew, or what not. --J. R. R. Tolkien, alderson@netcom.com _The Notion Club Papers_ Article 3340 of alt.sys.pdp10: Path: nntp1.ba.best.com!news1.best.com!newsxfer3.itd.umich.edu!newsfeed.internetmci.com!164.67.42.145!nntp.info.ucla.edu!128.32.155.1!agate!bh From: bh@anarres.CS.Berkeley.EDU (Brian Harvey) Newsgroups: alt.sys.pdp10 Subject: Re: The DEC PDP-3 Date: 14 Sep 1997 21:38:21 GMT Organization: University of California, Berkeley Lines: 10 Message-ID: <5vhlgd$kbf@agate.berkeley.edu> References: <5v6e4g$1g7a$1@flood.weeg.uiowa.edu> <34172B67.54569811@stoneweb.com> <5v8mnc$2cf$1@kirin.wwa.com> <5vb9g5$3tq$1@decius.ultra.net> NNTP-Posting-Host: anarres.cs.berkeley.edu Xref: nntp1.ba.best.com alt.sys.pdp10:3340 jmfbahxx@ma.ultranet.com writes: >Tony always said that, when writing code, he constantly looked for a >legitimate opportunity to be able to use a JFFO instruction. I distinctly remember a DECUS session some time around 1976 at which the TOPS-10 folks proudly announced that they had dramatically sped up SCNSER by using JFFO on all those tty input mode bits. If I remember correctly, L M Ericsson was the non-customer that wanted JFFO in the first place. Article 3347 of alt.sys.pdp10: Newsgroups: alt.sys.pdp8,alt.folklore.computers,alt.sys.pdp10 Path: nntp1.ba.best.com!news1.best.com!nntprelay.mathworks.com!cam-news-hub1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newsfeed.internetmci.com!169.132.11.200!news.idt.net!ix.netcom.com!netcom3!alderson From: alderson@netcom3.netcom.com (Richard M. Alderson III) Subject: Re: HP 2100A In-Reply-To: jhunt@callnet.com's message of 14 Sep 1997 17:16:59 GMT Message-ID: Sender: alderson@netcom3.netcom.com Reply-To: alderson@netcom.com Organization: NETCOM On-line services References: <5v6ep9$1bv2$1@flood.weeg.uiowa.edu> <5vh66b$2a6@taro.futuris.net> Date: Mon, 15 Sep 1997 18:00:53 GMT Lines: 17 Xref: nntp1.ba.best.com alt.sys.pdp8:2099 alt.folklore.computers:92021 alt.sys.pdp10:3347 In article <5vh66b$2a6@taro.futuris.net> jhunt@callnet.com (Mr. Computer Wiz) writes: >A PDP-6 or any other PDP model computer (on the other hand) are welcome too, >while I'm at it. Let me know, my PDP-8/I needs company. It's gonna be a little hard to find a PDP-6; only 23 were ever made, and I think most were junked decades ago. The last one that I know of was installed at the Stanford AI Lab until the fall of 1984, when it was taken to DECUS for the 20th Anniversary event, then shipped to the Computer Museum in Boston. It was apparently sold on little chunks by the folks there, instead of receiving the caring treatment it deserved. -- Rich Alderson You know the sort of thing that you can find in any dictionary of a strange language, and which so excites the amateur philo- logists, itching to derive one tongue from another that they know better: a word that is nearly the same in form and meaning as the corresponding word in English, or Latin, or Hebrew, or what not. --J. R. R. Tolkien, alderson@netcom.com _The Notion Club Papers_