What's All This About Gallium Arsenide?, lecture by Seymour Cray
HTML-код
- Опубликовано: 10 сен 2024
- What's All This About Gallium Arsenide?, lecture by Seymour Cray. The video was recorded in November 1988.
From University Video Communications' catalog:
"Seymour Cray discusses the history of supercomputing and the evolution of the supercomputer industry, reviews the development of the Cray-1 and Cray-2, and discusses the new role of gallium arsenide in the Cray-3 and Cray-4. He then answers questions from the audience."
Lot number: X6636.2013
Catalog number: 102741363
I was employed by Cray Computers of Colorado Springs from July 1991 through February 1992 (after they failed to produce the Cray 3 for Lawrence National Labs in California. The development wedge still works today at NOA in Boulder, CO (I believe). We worked 3 shift (rotational; 1 week days, 1 week afternoon and 1 week nights). I did not enjoy the schedule but was privileged to have worked directly under Seymour. He worked 6:30 am through 11-12pm so he's with all three shifts, reporting directly to Seymour and our supervisor, John Scarborough.
The devices were GAs (0--5VDC while the silicon memory was 0-+5vdc.
I worked there too. I was there for three years up to the day they shut the doors.
There's an anecdote of Steve Jobs running in to Seymour Cray in the latter half of the eighties. Jobs said to Cray: "We are using a Cray supercomputer to create the next generation Macintosh", to which Cray responded: "That's funny, because I'm using a Macintosh to create the next generation Cray supercomputer".
Such a shame Seymour didn't speak publicly more often - he seemed ot have an extraordinary sense-of-humour.
One of the best parts is around 33:00 where he talks about startups, and how people who are unhappy can leave and start their own company, and occasionally find the pot of gold. He did this a few times over his long career. A real pioneering spirit he was.
I like how bright people of this era always made a point of saying when the government was involved. It really stuck out back then. In the present age, its a surprise when whichever government you happen to suffer from isn't in control. God help us.
This was awesome! There is so little material of Seymour. What a great find.
"I had this powerful computing tool..." the circular 10" slide rule. Top of the line. "If you had one, you probably had some social problems in college." ... too funny!!... love this
I enjoyed this lecture. I did not know S. Cray had died in 1996 until I watched this video.
Seymour died in a motor vehicle accident. Someone ran into his vehicle. What a terrible waste.
What a charming gentleman.
A very rewarding hour with plenty of detail. I had believed until now that R & D of gallium arsenide never happened at all.
Gallium is also a -Secret- metal used in the mix with Pu to make it machinable, when making Thermonuclear Weapons parts, on the Supercomputer made by Mr Cray that was being developed, to calculate and simulate Thermonuclear Weapons, without needing to test them in the atmosphere again.
The Joy on Mr Crays face looks almost unbearable! He is a personal Hero
Well, I mean I could mention Gadolinium....
Seymour Cray is the Seymour Cray of Supercomputing! During the 1980s, the Brazilian State controlled oil company, Petrobrás, wanted to get a Cray III but the State Dept. vetoed the sale. They went after a newfangled Fujitsu machine that used massive parallel computing.
Now I have in my pocket an old Samsung Galaxy S7 that is more powerful than a Cray III. (not much more, if I'm not mistaken)
"Now I have in my pocket an old Samsung Galaxy S7 that is more powerful than a Cray III. (not much more, if I'm not mistaken)"
Your point?
Thank you for posting this very educational and informative.
Cray was a genius but I think he hyperfocused in one direction via Pushing more power towards as fast as possible using TTL towards the improvement of PVP performance and did not look beyond. CMOS processors became cheaper and could hold much more complex designs as mask sizes shrank. Eventually so much math hardware fit on a LSI CMOS processor to way outperform the traditional designs.
You have GPU chips that can do 6 teraflops of DP math and push over 800GB/s. Amazing how far we have gone.
For a guy that started with a 10" circular slide ruler as his powerful computer tool I'm highly impressed with what he accomplished.
When Cray started they had supercomputers that required over 18,000 thousand vacuum tubes, 7200 crystal diodes and 10000 capacitors and they took up over 1800 square feet of space.
He lead the computer industry a long way from where we started.
No way we would be where were at today without him.
He had the right approach for his time. The CDC 6600 and the Cray-1 were important machines, and successful. You can’t expect one person to make every innovation.
Fascinating and informative from a brilliant man. Innovator extraordinaire!
This video is underliked :) underviewed and OVERPOWERED :)
A real engineer!
A great talk on the early history of Cray supercomputers and the approaches to increasing the scalar performance. Strange now how most of our devices around are already well past the GHz range that was causing so many problems for Cray when this video was recorded.
Plus, Seymour is a great story teller.
The clock is two-phased, as he said, which effectively doubles the speed requirements (or "tolerance") of the circuitry. Also, it depends on what amount of work you actually accomplish on each cycle, or half-cycle for that matter.
I'm not quite sure how to compare this to the technology today. Btw, we've been up to 4+ GHz in desktops but went back to half of that in favour of more parallelism.
Broadband fibre cable 5 gee.
re: "Strange now how most of our devices around are already well past the GHz range"
Yes, also note, here, in 2023 where we are, kind of plateaued, again, but, new semiconductor families are being looked at beyond GaAs now ... btw, former alum at TI's GaAs Facility in Dallas Texas circa late 1990's.
Very interesting to say the least. :)
Awesome
7:20 Lecture starts.
His speech pattern reminds me a bit of Tom Hanks.
Ge Transistors 20:20
1:01:47 the staredown 😂 "get back to your seat"
CPU Galaxy sent me.
1:06:26 Hey, Steve!
HI, I'm John Rollwagen. Meet my brother, Eric Climbstair. Get it?
"But will it run Pac-Man?" (This being long before Crysis)
Or DOOM
Put at 1.25x speed. Makes it a lot more bearable
seymour would appreciate that, he liked everything fast...
00:03:01 NSA mentioned in 1988 - oops
I worked for Cray in the 1980s. NSA was one of our "secret sites" which were never mentioned by name, but NSA ran a recruiting ad which contained a photo of a Cray-1.
Great! He just needs a glass of water
Who calls capacitance capacity lol?
Some do. Interchangeable.
@@James_Bowie not ppl who talk about circuits and devices all day :p
It’s the old style term. Capacitors used to be called condensers so capacity made a little more sense then.
I invented the abacus.