Daniel Seagraves wrote: That could be very tricky if you attempt to read it out -- reading core is destructive, so if rig up an interface and you don't capture the data the first time, you might not get a second chance. People on this list have talked about some film that changes polarization or color or something in the presence of magnetic fields. Core is orders of magnitude larger than the bit domains of floppy disks, so using such a film to read out the core state non-destructively seems like a good idea. I haven't read the thread linked to above, so maybe this is already under consideration.

If each core is magnetized or not, isn't there a way to measure that independently of actually applying current to the wires? Today's (Hall effect?) devices are fairly sensitive, no? You could disassemble the stack and move a small sensor over each core. Maybe that's not possible here. - John

http://hackaday.com/2008/11/08/pcjr-25-years-later/

Christian Liendo wrote: Hmm, too early to say whether that's a shame or not. I mean it's sad of course, but it remains to be seen what the new folk do with the assets. I think I saw something the other day about IBM looking to acquire Sun, but I didn't chase that up - there's going to be none of the old names left soon, though! cheers Jules

William Donzelli wrote: That's what I was thinking... IBM is about the oldest name left in the computer industry. There's Unisys still around, barely. Groupe Bull. Who else? Peace... Sridhar

Too new. -- Will

In article <49D3C5D2.8090803 at gmail.com>, Sridhar Ayengar <ploopster at gmail.com> writes: Only 20 years earlier than SGI. -- "The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download <http://www.xmission.com/~legalize/book/download/index.html> Legalize Adulthood! <http://blogs.xmission.com/legalize/>

Sridhar> I wouldn't consider SGI or Sun to be "early". Sridhar> Or DEC or DG for that matter. DEC got into computers in, what, 1959? That's a bit later than IBM and Univac, but it seems to be close to the time when CDC and RCA did. DG of course is much later, they were started by the guy whose design for the PDP-11 was rejected. When did HP first make computers? I remember one in the mid 1970s. SGI and Sun are very new by comparison with any of these guys. paul

He died. -- Will

William Donzelli wrote: (chuckle) Why thank you, Captain Obvious! -- --- Dave Woyciesjes --- ICQ# 905818 --- AIM - woyciesjes --- CompTIA A+ Certified IT Tech - http://certification.comptia.org/ --- HDI Certified Support Center Analyst - http://www.ThinkHDI.com/ "From there to here, From here to there, Funny things are everywhere." --- Dr. Seuss

But your name is Will, not Lee. -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- Honk if you're illiterate! -------------------------------------------------

On Thu, 2 Apr 2009, Ian King wrote: . . . or the guy whose boat sank near Antikythera

NEC, athough their older machines are pretty obscure. Same with Hitachi. -- Will

I think Siemens just badge-engineers machines now. At least the big stuff is actually Fujitsu inside, I think. -- Will

Well, yes, but this could get hairy, figuring out who to include. You could say the same thing about GE, for example - they still make computers for their own products, but no computers per se. And then there is the whole defense industry. General Mills? Rockwell-Collins? I would say to keep it confined to older corporations that made general purpose computers a long time ago, and still make general purpose computers today (and by that I would mean anything from micros to supers). -- Will

Chuck Guzis wrote: Burroughs + Univac (Sperry) = Unisys CDC was chopped up. Some of the pieces survive. They switched from high-performance computing to storage sometime in the '80s, and that group became Imprimis, which was bought out by Seagate at some point. Their business services division still exists and is called Ceridian, I believe. Their financial services division became Travelers Group (as in Travelers Insurance), which is now Citigroup (as in the gigantic bank). (Although the insurance-specific businesses were spun back off at some point. Suitness.) GE and Honeywell's computer businesses were sold to Groupe Bull, where they survive. They still make a very small number of traditional mainframes, but most of their customers now use systems under emulation on Itanium processors. (Bull NovaScale) NCR exited the computer business not all that long ago, while they were still an AT&T division, I believe. RCA abandoned their computers and sold the division to Sperry sometime in the early '70s. Peace... Sridhar

On Wed, 1 Apr 2009, Sridhar Ayengar wrote: Back in January of '94 I was involved with moving off of DPS-6 Minicomputer hardware running GCOS-6 onto HP-9000's running GCOS-6 under an emulator running on HP-UX v7.something. Prior to that I worked at a site running on DPS-8's. A part of me likes the fact that GCOS-8 is still around, another part is horrified at the level of vendor lock in. It was not a "user friendly" environment. Zane

On Wed, 1 Apr 2009, Paul Koning wrote: Sort of!?!?!?! They're who bought Honeywell's computer division, and they make systems running GCOS-8 (though it runs on 64-bit Itanium 2 systems now rather than 36-bit DPS-8 varients). Zane

Paul Koning wrote: What do you mean by "sort of"? Bull is the scion of the whole GE->Honeywell->Bull thing. They were making electronics, but when did they start making computers? Peace... Sridhar

William Donzelli wrote: Indeed they were. Peace... Sridhar

Paul Koning wrote: Was that a Nixdorf at that point? Weirdness. Peace... Sridhar

On Wed, 01 Apr 2009 15:53:10 -0400 Sridhar Ayengar <ploopster at gmail.com> wrote: Siemens makes everything from trivial electronic components to atomic plants. They started to build transistorized computers in the 1950'ies. (SIEMENS 2002) Siemens bought the Zuse KG company in 1967 together with BBC, where BBC had bought Zuse KG in 1964. Nixdorf started in the 1950'ies and was swallowed by Siemens in 1990. Today, AFAIK, they sell the high end stuff from Fujitsu and self build (and engineered) PeeCees. Fujitsu started to build computers in the 1950'ies. NEC was mentioned already. -- tsch??, Jochen Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/

On Wed, Apr 01, 2009 at 07:50:15PM -0400, Roy J. Tellason wrote: Oh yes, and made as in "built their own stuff" not as in "slapped their label on it". Examples: - BS2000 mainframes: basically a slightly modified S/390 hardware with something similiar to OS/390 (I only saw that stuff from a distance), dead - SINIX Unix servers: Their own hardware architecture, MIPS CPUs and a (IMHO) truly horrible Unix, dead - various x86 PCs, using their own board designs and production lines (IIRC still in production) Regards, Alex. -- "Opportunity is missed by most people because it is dressed in overalls and looks like work." -- Thomas A. Edison

I don't suppose this was an April 1st joke? Not that SGI has a market cap over a few million lately anyway.

On Thu, Apr 02, 2009 at 09:59:46AM +0200, Jochen Kunz wrote: Ah, I only saw the MIPS machines, as sold to "special customers" a few years ago. Regards, Alex. -- "Opportunity is missed by most people because it is dressed in overalls and looks like work." -- Thomas A. Edison

Well, we knew what you meant, but we had to be picky anyway... It would be a shame to see Sun go - I hope the DOJ would kill that merger off. But, huge corporations coming and going is just a way of life. -- Will

And HP/UX is probably next to go. My guess, anyway, after the final nail comes in PA-RISC's coffin. -- ------------------------------------ personal: http://www.cameronkaiser.com/ -- Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com -- Cleanliness is next to impossible. -----------------------------------------

Yes, but how much 'stray' flux is there outside each core? Thinking a lot larger, one of the features of torroidal (mains) transformers is the relatively low external flux, so less interference nad the like. I would have thought the static stray flux from a core in a memory plane would be very difficult to detect. -tony

Tony Duell wrote: That's certainly true of normal read-write core storage. However, there was also a read-only form of core storage, known by the names "core rope memory", "wire braid memory", etc. In this form of core memory, the magnetization of the cores is not used to store the data, but rather the pattern of threading. The cores don't even need to exhibit the hysteresis loop needed for read-write core storage. Each core stores multiple bits of data; a typical system may store 64 bits per core. This was advantageous where it is undesirable for the data to be changed (e.g., a fixed program or data table), and where mass and volume are at a premium. The drawback was that if you decided you wanted to change the data, the most cost-effective way to do so was to replace the entire core assembly. This form of memory was used for the program storage of the Apollo Guidance Computer. I have no idea whether it was used for the LVDC, but it certainly wouldn't surprise me. Eric

Chuck Guzis wrote about core rope memory: Not really. You'd have thousands of very fine wires, all looking identical, that pass through some cores and go around others (on average perhaps 33 through each core), and packed incredibly densely. Even with very high resolution scans from multiple angles I don't think you could expect to extract the data visually. If you want to try the experiment, get the control store from a PDP-9 (DEC G920 module, 64 word by 36 bit) or the control store from and HP 9100 calculator (09100-69501 for 9100A, 09100-69551 for 9100B, either 64 words by 29 bits). Those should be much easier to read, as they didn't have to be packaged at maximum density for use on a spacecraft. I've looked at the 9100 control store module a few times, and I think reading it electrically would be *much* easier than trying to do it optically. Eric

Chuck Guzis wrote: The three ROS (Read Only Storage) technologies developed for the original System/360 processors were: CCROS - Card Capacitor Read Only Storage - used etched metalized IBM cards, either paper or mylar. Programmed using a normal card punch. Used in 360/30. 750 ns cycle time. BCROS - Balanced Capacitor Read Only Storage - similar but used two capacitors per bit for better SNR and higher speed, and did not use a normal card punch. Used in 360/65, with 90 ns access time and 200 ns cycle time. TROS - Transformer Read Only Storage - similar in operation to core rope memory. Used mylar film tapes with metalization that either passed between or around square holes for each of the transformer cores. Used in 360/20 and 360/40. 240 ns access time, 625 ns cycle time. Based on T.L. Dimond's "Dimond ring" transformer read-only storage first described in 1951. See: "No. 5 Crossbar AMA Translator," T. L. Dimond, Bell Laboratories Record Volume 29 Number 2 pp.62-68, February 1951 "Card Capacitor - A Semipermanent, Read Only Memory", H.R. Foglia, W.L. McDermid, H.E. Petersen, IBM Journal of Research and Development, Volume 5 Issue 1 pp. 67-68, January 1961 "Printed Cards for the Card Capacitor Memory", J.W. Haskell, IBM Journal of Research and Development, Volume 6 Issue 4 pp. 462-463, October 1962 "The Design of Transformer (Dimond Ring) Read Only Stores", D.M. Taub, B.W. Kington, IBM Journal of Research and Development, Volume 8 Issue 4 pp. 443-459, September 1964 "Design of a Printed Card Capacitor Read-Only Store", J.W. Haskell, IBM Journal of Research and Development, Volume 10 Issue 2 pp. 142-157, March 1966 "The 'Braid' Transformer Memory", W.H. Aldrich, R.L. Alonso, IEEE Transactions on Electronic Computers Volume 15 Issue 4 pp. 502-508, August 1966 "A Balanced Capacitor Read-Only Storage", S. A. Abbas, J. K. Ayling, C. E. Gifford, R. G. Gladu, T. C. Kwei, W. J. Taren, IBM Journal of Research and Development, Volume 12 Issue 4 pp. 307-317, July 1968 _IBM's 360 and Early 370 Systems_, E.W. Pugh, L.R. Johnson, J.H. Palmer, MIT Press 1991.

Chuck Guzis wrote: _Microprogramming: Principles and Practice_, Samir S. Husson, Prentice Hall, 1970

There are four independent flight computers involved in the Apollo program. The LVDC is the Launch Vehicle Digital Computer and was designed by IBM. It flew the Saturn booster. The Command Module Computer and Lunar Guidance Computer (CMC/LGC) are the Apollo Guidance Computer (AGC) from MIT. This is the one most people think of when they think "Apollo Capsule Computer". There was an additonal backup computer on the LM called the Abort Guidance System, or AGS. It was made by TRW. The one being discussed is the LVDC. As far as I can tell, all of the LVDC software was in the possession of IBM when it was broken up, and nobody seems to know where they went after that. On Nov 10, 2008, at 8:56 PM, Randy Dawson wrote:

http://computermuseum.informatik.uni-stuttgart.de/dev/wang700/wang700_ro m.html For more information visit http://oldcalculatormuseum.com/wang720.html

At least in the case of the HP9100, I am pretty sure it was tested _electrically_. One of the HP journals mentions some spcial-purpose test rigs used at the factory that compared the core-on-a-rope microcode store and the PCB main store with known-working examples. That said, there is a (temporarily) destructive visual method that would work on the HP9100 microcode sotre, I think. Namely to desolder each address wire from the PCB at one end (HP were kind enough to label the endpoints of these wires, so you can note down half the address at this point), then carefully untrhead it from the cores, ntoign which cores it loops through, and finally note down the other hald of the address from where it conencts to. Do that for all 64 wires and you have a 'dump' of the ROM. Finally, of course, replace and re-thread all the wires to restore the ROM contents. -tony

On 10 Nov 2008 at 22:37, dwight elvey wrote: Putting to bed the idea of the "core rope" program store, the 1964 document on the launch computer is pretty clear that this is ordinary read-write core storing the program: http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19730063841_19730 63841.pdf In particular, item 2-37 begins "Reading a ferrite-core memory destroys the information in the memory." It then goes on to describe the restore part of a read operation. No mention of "core rope" is made. The document, all 200-some pages contains a great deal of detail on the computer, including memory operation and organization. Cheers, Chuck