Remembering the Cray-1 The triumphant return of This Old Box! You begged and pleaded, and we listened. Ok, ok… nobody said anything — but your indifference speaks volumes to our hearts! Apathy is the seed of enthusiasm. Or something. This Old Box In this installment of TOB, we’re taking another gander at one of the big boys of computing. And, oh …
A long time ago in a galaxy far far away some of the lab rats (aka graduate students) that "worked" in the lab I (really) worked in did some remote batch entry to the CRAY at NCAR (this was in the 70's). It was a really big deal, and we used an exciting "high speed" (1200 bps) modem to get the data across. One "feature" that we all were amazed about was the "terraBIT" memory they had there. I never got to see it, but I was told that it was HUGE. Now we fast forward to the present, and think nothing of 128Gbyte (equivalent storage!) disk drives that can be had for less than $100 and hold in the palm of ones hand. My how times have changed.
Ah, that's beautiful. I never saw a Cray 1, but I did get a good close up look at a Cray X-MP sometime around 1985. An animator friend was doing some work at Digital Productions, he said I should come over and take a tour. They had just finished production on "The Last Starfighter" and were working on a film for artist Matt Mullican. I got a good look at some convoluted animation software by Symbolics, and listened to their technicians griping that no matter how powerful a Cray they had, the animators would invariably choose the most computationally intensive effects, so all that CPU power was useless, their current productions were still taking 4 hours to render one frame, just like it took 4 hours in all their previous work in years past, before they bought the Cray. The 4hr/frame scenes I saw rendering would run in realtime today using modern graphics cards, but back then, this sort of graphics rendering seemed like a damn miracle.
Anyway, on my tour around the office, I saw a framed computer graphic image on the wall, it caught my eye because I'd seen before in one of my first computer graphics textbooks. It was a pen plotter artwork of outlines of little birds, swirling in a pattern that looked like a magnetic field, it looked like it was produced with early 1970s technology. As I paused to admire the print, someone walked by and noticed me noticing it. I mentioned that I'd seen this in my textbook and I had always admired it, and where did he get it? He smiled and said, "I made it myself." I suddenly realized, I'm talking to John Whitney Jr.
"The seat was sometimes used for naps by sleepy installers. But this begs the nine million dollar question: has anyone ever got lucky on top of this super computer?"
I can only think of how many technicians, who currently toil in the large data centers today, would love to have a soft bench surrounding their rack of blade servers to repose upon as they wait for their installs to complete. As for getting "lucky" how about security not stopping you from bringing in a folding camping chair for the long haul install tasks in these places.
"The seat was sometimes used for naps by sleepy installers. But this begs the nine million dollar question: has anyone ever got lucky on top of this super computer?"
I suppose in the Virtualisation Realm, and with the new XTCray models ..... http://www.cray.com/products/xt4/index.html ..... the question now would have evolved to ... "Has anyone ever got lucky on top as and/or with a super computer?
Great article, Austin. IT is not what you know, IT is how you share it which defines One from a Zero and being a Loser in a Game Created for Win Winning with a Right Royal Scaleable Processing Architecture.
*If Computer Systems were designed around Successful Love Algorithms, they wouldn't fail at all. In fact, they would grow in Stature and in Protective Stealth, hence Paris, in tribute to Venus and her mounds. An Immaculate Addiction which Scales to Unbelievable Heights .
"Processor: 80 MHz".
There's so much wrong with that I don't know where to begin. Like "the damn thing IS the processor!" Or "does nobody care about minor details such as instruction set architecture, programming model etc etc anymore?" Or "so you would describe the engine of a car by quoting the maximum RPM?" Or...
(goes on muttering about things not being what they used to, computer architecture being dead etc)
On a school trip with about five other students I visited the Shell site in Wythenshaw. We got a full site tour, I got a close look at their Cray-1 Super Computer. This would be about 1982.
I was already aware of its reputation and was very chuffed to get up close. And yes, I got a look inside the C shape, and sit on the bench. On another floor they had 50 IBM PC's hooked up to the Cray to handle communication with storage, that being 10 fixed Winchester stacks, and another 15 demountable stacks. Seem to recall they were about 25MB's each.
We also visited an enormous tape storage room and I got to load a tape into one on to a tape drive. It had an automatic sliding door to reveal the spindles, and sucked the end of the tape in using air.
Chris, Manchester
Please, please, please, I beg of you to keep up the good work on the 'this old box' articles.
Happy? :)
That thing is a dinosaur, no 2 ways about it. But, no 2 way about it either, it's cool. The piped suit is 'debatable' but hey, that guy knows what all those wires do. He could go the the Arthur Dent school of fashion and still get big respect.
Would love to hear more on some of the 80's designs that came to a dead end when 'IBM compatible' sent development into fast forward. IBM caused great things to happen there but left us with a platform that is definitely lacking in some areas.
cheers
In 1987 I was fortunate enough to be involved in a project at Los Alamos national Laboratory - no, not nukes, but a project to determine very precisely the gravitational constant using measurements taken down a borehole in the Greenland Ice Cap. My job was to delineate the topography of the rock beneath the ice using a lot of very closely spaced radar profiles.
The main thing I remember was the total awe with which I realized that no merely did I have a Cray at my disposal, but that I had no less than THREE at my disposal - I think it was one X-MP and two Y-MPs! That was really mind-numbing, going to work in the morning and deciding which Cray to log in to! And that was just the ones outside the security barrier... Of course, I now have more power available on my desk-top, but back then it made possible things like geometric correction of radar profiles in minutes rather than hours; given the experimental nature of much of the programming, being able to test changes to the program in minutes rather than hours made the development cycle realistic.
Black helicopters given the connection with LANL!
In the Early-mid 80s I worked for a UK-based Oil company with offices in Moorgate, developing graphics tools for reservoir simulations and we had a Cray Y-MP installed. They had to strengthen the floor to get it into the building and it was a treat to be allowed to sit on the seats! I should have a photo around somewhere - I'll have to find it!
As for "getting lucky" on it... not a chance, although I would not have put it past our operators at all!
Our IT manager had a "fun" trick he used to pull to impress techies and others... he'd telnet onto a Cray 1 in another office, our Y-MP, another Y-MP in Aberdeen and a Cray-2 in Texas. Of course he couldn't DO anything on these boxes - they were all set up as batch processors for various engineering simulation systems or seismic survey signal processing.
according to http://www.cs.man.ac.uk/~toby/writing/PCW/cray.htm
-------------
There are many legends about Seymour Cray. John Rollwagen, a colleague for many years, tells the story of a French scientist who visited Cray's home in Chippewa Falls. Asked what were the secrets of his success, Cray said "Well, we have elves here, and they help me". Cray subsequently showed his visitor a tunnel he had built under his house, explaining that when he reached an impasse in his computer design, he would retire to the tunnel to dig. "While I'm digging in the tunnel, the elves will often come to me with solutions to my problem", he said.
-------------
We had a Cray-1 at Bell Labs, and we always showed it off to guests. I also learned to program on a CDC 6600 in 1972, so Cray has long been a hero of mine.
There's a famous story, about the Cray-2 actually. Apple bought a high end Cray machine to help design their next Mac. Cray told them that this was amusing, because he was using a Mac to design the Cray-2.
Previously, Cray was famous for doing his design on large sheets of paper. He was remarkably low tech.
knowing very well the mentality of the techies of the time, and the total lack of the ability to talk to members of the oposit sex, the only sexual encounters performed on a cray would be flying solo with a knuckle shuffle.... imagining what it would be like to interface and login to a W.O.M.A.N.
where's my jacket....
We still have our old Cray-1 in our data center foyer. A vertical clear plastic panel covers one segment so one can peer into the inards.
The leather clad seating was always useful late at night while developing code for the Cray-1. However, one needed to be careful when throwing down the briefcase onto the seating as it could impact the mainframe and cause electrical shorts issues.
As we were developing OS code enahancements we started with (paid for) 2MB, then went to 4MB and finally 8MB of central memory as it was so expensive.
We placed our Cray-1 in a room with a small window that had a blind on the inside. During production the blind was up for all employees to gawk at the Cray as it crunched on the numbers. At night when Admins and code developeres were swarming about we lowered the blind.
Your TOB article brings back fond memories.
I love this story I was once told by someone in the intelligence business. Apparently the folks at Fort Meade have been through a fair number of Crays in their time, and pretty much kept the company in business so they could keep getting more. This meant they had a lot of sway with Seymour, and could ask for special favours.
One thing they used a lot in their programs is the popcount algorithm - it counts the number of bits set to 1 in a binary number. It's fairly trivial to write a quite efficient implementation of popcount, but the best you can get is one CPU tick per 1 bit - so for 64 bit chunks it will take 32 ticks on average. The Ft Meade chaps asked very nicely, and so Seymour added popcount to the Cray instruction set - so that doing a popcount on a 64-bit chunk would only ever take 1 tick. Nice.
Now, wind things on a bit, and the Japanese supercomputer companies figure they can try going after the US govt's supercomputer business. They show off their shiny new boxen to the US guys, who are a little surprised to see that the instruction set includes popcount! They ask "why did you include popcount in the instruction set?", and the Japanese reply "Well... we don't really know why. The Cray has it in its instruction set, so we figured we might as well include it too. We've no idea what it's for, but somebody must want it...".
They didn't get the contract... :-)
In about 1992 I had a 486 with similar specs to the Cray 1. Albeit with only 32 bits. But fifteen years after that my desktop PC's frequency is over thirty times faster with 200 times more memory.
Still no love seat, however. And the office chair just doesn't cut it.
PS I had a friend who made a killing buying Cray stock. I unfortunately....didn't.
...in Geneva when he came to DONATE a free Cray to our hole in the ground. I think it was the X-MP running UNIX. well it was entirely free - we just had to pay the million dollar annual maintenance contract! I asked Seymour for a job, but I wasn't quite American enough - he did mention that Gallium Arsenide was the future!
it still might be !
It was 1977 and the LAST computer that Apple ACTUALLY MADE themselves and didn't contract out to the lowest bidder was the Apple ][. Shortly there after, the Apple II+ and then the IIe were totally contracted out to a Canadian manufacturer (down to the PCBs), who then subcontracted manufacturing our to a Japanese plant (don't believe me, look at the back of any II+ or IIe!)
Ever since EVERYTHING Apple has sold has be "Sourced Out" to a low bidder out of the US for basic level manufacturing (PCB and Logic Boards) up through the WHOLE THING! Indeed the first 1984 Macs were not made in the US and while Apple did some "assembly" of Power Macs (ALL components made in Taiwan and China) in Roseville CA, EVERY Mac portable has been made and DESIGNED by someone else starting with Sony / IBM in Japan through 1999! The G3 iMac was completely designed and manufactured by Hyundai, who also brought you those POS eMachines. Now Apple shit is made in Commie China at slave labor level Sweatshops. Apple was the FIRST to outsource in the US computer industry, and that is a fact.
I worked for Apple 1991 thru 1998, I know about the truth the Apple Kool Aid Drinkers and Apple Media Hacks like to hide from.
Besides the special popcount instruction (which BTW was also very usful for determining disk space allocations in the disk allocation bit maps) there was the BMX or bit-matrix multiply. I don't exactly recall if this WAS in the Cray-1 but it certainly was in later Crays. This was a hardware feature that allowed a 64x64 bit matrix to be manipulated very quickly - again presumably for the Govt. spook sites.
The Cray OS was named COS until UNICOS came along during the 1980s.
We've had Crays since the Cray-1 days and have had every model except Cray C90 and the Cray-2. I would say the Cray-2 was the most visually exciting model. Today we have the liquid cooled Cray X1 with some 4096 GB physical memory (512,000 times the Cray-1's max memory size of 8 MB). The Cray X1 and X1e (X1e was the follow-on model to Cray X1) are the last of a kind - last of the big Cray Iron. Today we have the Cray XT4, XT5 and XT5h(or X2) - all come either air-cooled or chilled water cooled.
I recall when a Japanese research site bought a Cray T932. They were impressed that they were able to complete all of their 1 year's worth of research in just one month on the T932.
Crays have come a long way in 31 years. From being king of the hill in supercomputing to just one of the numerous HPC players today. They certainly had a good run.
Was it the Cray that gave rise to the word "supercomputing" ?
I worked on the Scope OS for the CDC 6600 and later worked on vectorization on the Cray-1. Great machines at the time, but we only need to look at the floating point processors of the Intel x86 cpu's to see the ghost of Cray's legacy. The floating point registers and instructions look very much like the CDC 6400.
I was nowt but a glint in my fathers eye when this goliath of a machine was invented but that did not stop me jumping up and down like a nutter and screaming "IT'S A F*CKING CRAY 1, SWEEEEEEEEEEEEEEEEEEEEEET!" when I was last in the Science Museum over the summer. However I think it was roped off so I was unable to test out the seat :( ...still, it was amazing to see what was once deemed a 'super computer' and yet have a mobile in my pocket with a faster proc, bigger storage and probably more memory.
Any UK peeps that have never seen one should head down to the science museum in london and have a butchers.
The Itanium happens to have a population count instruction. It has a few bit-manipulation instructions but no real bit matrix multiply, though. (Vector operations with stride, used to multiply large matrices, are another type of instruction that had triggered export controls.)
Also, I have an old Datamation that discusses the CDC 6600 and the IBM 360/195; it has the word "Supercomputer" on the cover, in answer to another post.
Elves? I've only ever heard of bugs and gremlins when related to all things computer. Metaphors are a useful tool for techs trying to explain what was the fix for the last outage.
User: "Why was the network down?"
Tech: "Gremlins."<no problem at this point, users know gremlins>
User: "What fixed it?"
Tech: "Elves."
User: "Ummm....Elves?"
Tech: "We had to clear the LSA cache by rebooting one of the layer three switches because of a known bug for OSPF interfaces in the firmware we are currently running."
User: "Ummm....what"?
Tech: "Elves"
User: "Right, er, thanks".
But back in those days there were probably more women in the computer biz than there are now, especially in the Operations side. Sadly we didn't have a Cray, but all good BOsFH know that you need to be able to lock the computer room door from the inside sometimes.
Don Mitchell: Oh yeah, you triggered an old Cray memory. Yes indeed, Apple did buy a Cray to design their computers. I remember attending a lecture by Apple's Advanced Technology Group about this. They said they were using the Cray during the day to design custom chips, but at night, they turned the Cray over to the ATG for experimentation.
The ATG lecturer said that soon desktop computers would have the power of a Cray, and they intended to find out what you could do with all that power in a single-user computer. Then they made startling predictions (for 1985) about what the personal computer would be like, based on a Cray. It would have astonishing features like a CPU that ran over 1 gigaflop, a gig of RAM, 1 gigabyte disk storage, 1 megapixel 24 bit color displays, 1mbps networking, etc etc. Those things all sounded astonishing because a system like that cost millions of bucks. But if they had to spend millions just to do some experiments to figure out where the future of computing would go, they had to do it.
Alas the ATG was disbanded shortly after Jobs returned, but their influence is still felt widely throughout Apple and the entire industry. And their predictions of the future of personal computing came true even faster than they expected.
Apparently the French atomic energy agency had a brand new CDC 6600 that refused to operate correctly and the service guys from CDC had been unable to locate the problem. The French government was threatening to throw the machine out, so CDC sent Seymore over to rescue the machine. After looking at all of the tests that had been performed by the techs, he closed himself in a room (not with the computer, and with no schematics), emerging a few hours later to tell them that the problem was a particular transistor on a particular board in a particular part of the machine -- and he was right. When asked how he had arrived at the solution, he said it was the only possible answer given the test results and the structure of the machine. His understanding of the machines he created was legendary...
With the follow-on to the Cray-1 came the Cray XMP which could be configured with an external Solid State Device (known as the SSD). This SSD was housed in a huge massive box some 8' tall and several feet long and was liquid cooled. It had up to 4 VHSP (Very High-Speed) channels into the XMP's main memory. The SSD could be used to extend main memory, one or more SSD file systems and/or file system cache (known as ldcache). It was extremely beneficial for out-of-core solutions where the main memory just wasn't large enough to hold an applications data.
No matter, my point is that today we are starting to see SSD-type devices becoming common place in the desktop. Sizes are up to 128 GB and expensive but the smaller ones such as 16, 32 and 64 are within desktop prices. It may even be that Apple's next model will come standard with SSD rather than a HD or could be an option. This is another example of how far technology has come since the Cray-1 and Cray XMP days - some 25 - 30 years.
So what do others predict will be around on the desktop in 2040 ?
Number crunching job been running for something like 3 days Only had around an hour left to go. Me being the bored tape monkey (computer op) decided to play around with one of the terminals.
Oh there's a command called "debug" I'll give that a go.
Two secs later the Cray starts counting down until it shuts down
Got quite a bollocking over that one. If memory serves me correct, if only I'd known about the command "bugoff" all could have been saved.
Never seen someone making out on one (after all, weren't they all in 24hr sites?)
It's in building 24, and you can see it from the outside. I've been in their computer room, which is the size of a football field (or at least several tennis courts, end-to-end). At the very back are the current IBM mainframes, which are the size of a filing cabinet. All of that computing power has been reduced from the entire room to a dozen machines. The tape room is still full, though.
What will we have for computers in 2040? Really good question, and what kind of an operating system will we have? Will we actually have a reliable operating system? I worked as a tech on the Celerity box with an NCR 32032 dual processor rig. It was a supercomputer in 1983, and had 32Mb. Now my notebook has a dual core CPU and 4Gb of RAM. Is the response of the OS that much faster? No, not really.
I think that in 2040 we will have optical computing. The computer interface will have become more of an extension of us, and there won't be a screen. The information will be either projected against our retina, or placed directly into our optic nerves or visual cortex. We will be able to communicate with the computer by thinking ideas at it. There won't be any external wires or jacks, as all external communication will be RF. I wonder how easy it will be to hack somebody's brain.
(The Nap, that is; I never got lucky at work.) I worked for Cray. On a particular all-nighter, which ended too late to bother with going home, I tried to curl up on the cushions for a couple of hours-- the lounge and offices had only chairs, and I really wanted to be HORIZONTAL for a few minutes.
But it sucked. First of all, With the room full of other computers as well (Data General, DEC, IBM) we had a COLD hurricane of air conditioning blowing hard-- very cold, and very noisy. And although the padding was OK to sit on, for a moment, it wasn't thick enough to lie on-- my hip and shoulder bones complained a lot when I tried. And finally, there's a lot of vibration-- I don't know if the "shaking" parts were mostly power supply related or freon pumping related, but it wasn't at all relaxing. I shortly crawled back to my office and slumped into my chair until morning officially arrived.
as a young kid doing work experience at the rutherford laboratories i saw such a beast running - from memory it was a nice shade of red - wouldn't have looked out of place at dfs.
it was a james bond style setup - there was a giant room filled with big reel to reel tape drives and a few thousand of the tapes hanging from racks waiting to be queued up.
AC says: "It's fairly trivial to write a quite efficient implementation of popcount, but the best you can get is one CPU tick per 1 bit - so for 64 bit chunks it will take 32 ticks on average."
No, actually you can do it much much faster than this even on a regular CPU. Google "HAKMEM" for this and other PDP-era bit twiddling goodness.
As other people have drifted off-topic a bit, i'll throw in my 2p.
I went on a school trip to the Regional Computing Centre at Manchester University where they had just taken delivery of their second CDC 7600. I was told a story about how the 7600 was too heavy for the floor. Unfortunately they only discovered that *after* it had been installed. Someone crawled into the floorspace with a couple of car jacks to take the weight until they could get it sorted properly.
On the same trip we were shown some older IBM boxes that used genuine oil-cooled core storage (out of date even then) and a 1/2" tape drive which used vacuum columns instead of rollers to take up the slack tape. Some techie had written a program which made the tape move up a down in the column creating different toned resonances which made it play a tune. All the kids thought that was great.
We finally managed to get head office approval to buy it. It had taken a long time. A very long time, and then we placed the order. This was delayed by the question, "Well, what color do you want it?". We had no idea there was a choice. We wasted three weeks deciding that black would fine!
After it was installed we started having loads of problems. It would crash, and the dump was always corrupted (yes, we used to get and read AND UNDERSTAND!) core dumps. The Cray engineer was baffled. His home support was baffled. The designers were baffled. Eventually we traced the times when it got trashed to when a certain female operator was on shift. She would walk past the machine and lovingly touch the side - only to earth the static from her nylon underware in to the machine. We had to ask that females kept away from the machine. The union went mad and complained that this was blatant sexist discrimination and that they would ballot for strike action. I just wish I had been in the meeting when the head of the data centre explained to the union rep what the problem was, and that the alternative was to have the duty sysop check that all the females were wearing silk or cotton underware! No, before anyone gets at me I am seriously not making this up, but I will protect the site from embarrasment!!!
I also remember being the client lead trying to hook a VAX 11 up to the Cray so we could load jobs and get back results fast. Up until then we had a serial line to a an IBM Mainframe, but the Vax was going to be Massbus connected. Imagine, as far as the Vax was concerned, the Cray was just a peripheral. On the big day when we wired everying up, it all went horribly wrong. As soon as the Vax tried to access the Cray, the Cray powered down. I had to wait until 2am to get a Cray senior engineer on site. I was well impressed when he had me connect a pulse oscilloscope on to the Massbus and we watched the signals going down the line. After 2 hours he took the back panels of the VAX and showed me where two pins on the backplane were a little bit bent and touched from time to time. Not enough to crash the VAX but enough to crash the Cray. Fixed with a gentle tweak by a pair of pliers and never a problem since then.
Overall I worked on Cray-1 and Cray-XMP. One of the other sites had a Cray-2 but I only saw it, never go access to it. We were talking about a Y-MP when I left the company. Ahh those were the days, when a super computer looked like a real super computer.
I visited Rutherford labs to see a Cray XMP as part of a College IT course I was doing in 1987.
It's computing time was shared between various universities and faculties and at the time we visited, it was crunching through data concerning some sort simulation in space. The Cray was surrounded by screens, each showing it's current CPU load, Each one showing at 99.9% or thereabouts. Wonderful stuff!
When I was at university in the early 80's, taking computer architecture classes, we learned that the original Cray-1 was built up from extremely simple gates - 4 and 5 input ECL NAND gates and very small ECL RAMs. Imagine that, building up vector processing arithmetic units using only simple gates! Hierarchical design and modularity conquer all, I guess.
The Cray's were certainly iconic machines. In front of me on my desk is a framed circuit board with the following caption:
"This Cray-1 S printed circuit board is a genuine part of the second supercomputer ever built. The computer arrived at Kirtland Air Force Base in 1980 and was used by the Phillips Laboratory (...) it was used to perform calculations, simulations and computer modeling to determine the effects of nuclear weapons detonations (...) on military systems and activities."
A great little souvenir that I picked up from the gift shop of the National Atomic (bomb) Museum in Albuquerque, at that time (pre-9/11)you had to get clearance into the air force base to visit the museum. I believe it is a memory card, it carries a bank of 32 surface mounted chips with an 'F' logo and the serial number 10470, plus other supporting chips.
Ah, yes, the Cray-1 at NCAR. In the spring of 1979, the local IEEE chapter had a "visit the Cray" tour at NCAR. The big thing at the time, I recall, was standing inside the "C" and mumbling something about having Scotty beaming you up. Knowing what know now about static, I'm surprised the NCAR folks even considered allowing it to happen.
About this time I had started work at HP. A little later that year (after the NCAR trip), my operation manager came to me and asked me to write up a fake P.O. for a Cray. Whipping up the P.O. and pulling numbers out of my... imagination, I delivered it. I think the bottom line was about $10 million or so. Seems the manager had just obtained a new rubber stamp and wanted to try it out. The P.O. came back stamped, "I Can't Possibly Sign This!"
Too bad. Would have loved to have run SPICE simulations on it.
Oh, and puh-leeeeeez keep up with the This Old Box articles! I've loved every one!
Cray-1, X-MP, Y-MP (first GFLOP in 1989)
then a split on the product line
Vector/Scalar - C90, T90, J90, SV1, SV1ex (the SVs were SGI's aspirations after buying Cray in 1996 for some $750M and 3 yrs later selling it off for $50M)
MPP (1000's of processors) - T3D, T3E, T3E-1200 (first sustained TFLOP), T3E-1350
then product line collapsed to one and continued on as...
High Bandwidth MPPs - Cray X1, X1e, XT3, XT4, XT5/XT5h, ...... onto PFLOP hardware maybe by 2010
Renton or Everett? I was contracted doing desktop and terminal hardware swap for awhile in the mid 90's at the Everett facility. I was assigned from roughly D-F of the main assembly building. Didn't work the Glass Palace much. Didn't see most of their computer design assets (other than CATIA terminals) or some of the CAD/CAM and assembly automation.
Briefly got over to Longacres while they were working on the 777 flight simuator. That was a pretty awesome setup.
The company I work at have a fair few old machines, all about the same sort of spec as the machine that Andy found. They're all around 486dx33's with a couple of megs of ram. They play audio through special audio cards which are old full length ISA cards. I'd go as far as saying that they're more reliable than their Windows NT equivalents.
Rob
Sorry I forgot to beg for this article. Reminds me of the days I had to sort punch cards, play with IBM 3090/S38/S36/AS400Model-B20-40-60 and Harry (Harris computer). I still have some 8inch IML boot disks left in case I ever find another....
Harry was the best: in winter times, this was the first system we'd boot. Why? Well, after power on it would produce a nice warm place to sit on top. The harddisk (amazing 10MB storage) took about 3 minutes to speed up. You could not boot the CPU before the HD was fullly functional. I hated the 3 minutes extra cold. 1988 was nice.
The AS/400's were cool too, but lacked a convenient place to warm up. Probably this is why they are still running......
During those days, I actually dreamed of having -any- cray available. SUPERcomputer!
Take care, more of these 'This old box' wanted!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
