Game Tales: Cray YMP (2010) | Hacker News Reader

Game Tales: Cray YMP (2010)

rome.ro

58 pointsthe-enemy18 comments

I knew a guy who worked at one of the national labs that had its own Cray supercomputer, in a computer room with a big observation window that visitors could admire it through.

Just before a tour group came by, he hid inside the Cray, and waited for them to arrive. Then he casually strolled out from the Cray, pulling up the zipper of his jeans, with a relieved expression on his face.

It seems to me that if you wanted a multi head-capable supercomputer in the 90s, SGI is exactly who you would buy from. A rackmount Onyx or Onyx2 could support lots of graphics heads, at least 8 CPUs, tons of memory, and it was a fairly normal/popular UNIX. Was there anyone doing this sort of work on Cray machines?

Yeah, I was a bit surprised reading this, myself. $500k for a Y-MP does seem like a nice price - I assume they were getting a machine that had been previously loaned to someone who was upgrading to something massively multiprocessor.

But it seems like the SGI Challenge architecture makes more sense in 1995. An Onyx would be half the price of that Y-MP, with superior graphics. A Power Challenge GR would as performant as the Y-MP, again with superior graphics, though I suspect about the same price. And anything from SGI has better OS and development tools than UNICOS did at the time. Plus they're easier to install, power and cool.

I suspect that moving over to Win32 probably conferred a lot of advantages, developing on the same system that is your target market. I wonder if they would have ultimately done that even if the Cray deal was on the table.

> An Onyx would be half the price of that Y-MP, with superior graphics.

As someone who used both: it also would be a massive downgrade in terms of reliability and system stability. The Onyx I had access to had regular problems with various system hangs and lockups.

Interesting. I'm sorry to hear that; I actually never used an Onyx but I went from a Cray 2 to a Cray X-MP to a Power Challenge. I was skeptical about Irix and the Challenge, mostly because a bunch of consumer-ish processors seemed like a downgrade, but it turned out to be both fast and rock solid for my workload. That said, I was a consumer, not an administrator, so it's very possible that our admins kept the reliability issues from affecting me.

What an awesome cynical ending: "John decided that developing on Windows 95 was the way to go."

I think John Romero is just glossing over the details since it's a (cray) Cray story. The PC was id's primary target platform and Carmack surely knew that Win32 is where things were headed. He ended up moving to NT as his development platform. Here's his .plan from a little later that talks about this in some detail:

http://rmitz.org/carmack.on.operating.systems.html

There's a few more answers about choosing Windows on a Slashdot QA

https://games.slashdot.org/story/99/10/15/1012230/john-carma...

Yep, also interesting, I ended up not linking it since it seemed a bit too far from the events Romero talks about. It's also a great reminder of how shaky OS X's early years were.

This sort of decision is probably why John Carmack is now CTO of multi-billion dollar corporation Oculus VR, and John Romero is... umm... well, I think he lives in Ireland now?

Mind you, you might also like to count how many awesome genre-defining games Carmack worked on after Romero got the boot...

how many awesome genre-defining games Carmack worked on after Romero got the boot

Not that many, really, after their collaboration ended.

Whoosh.

Neither did really.

but then ID games were always more about the tech then the games themselves.

While reading over the backlog of filfre.net articles i recall running into a quote about how game developers are notorious for forgetting everything about story and such when new tech comes along. Resulting in games that are basically spruced up tech demos for the first few years.

we can kinda see the shift with Quake3 vs Half-Life.

Q3 had the better engine, but HL had a solid story and presentation.

Meaning that by that time the world of 3D engines had matured, and people were less likely to be wowed by higher polygon counts and more lights.

Yeah, I'm not blaming Carmack for making that decision. It just sounds a little bit like Romero didn't think too highly of Windows 95.

Why do you think I should count Carmack games? I have nothing against the guy and I'm fully aware of his accomplishments. I was merely commenting what Romero wrote.

There seems to be a massive tone shift towards there end there.

The rest of the article is all "we" but towards the end it is "I released Quake", "I quit".

Funny thing is that i seem to recall the gaming press back then putting the blame squarely on Romero for Quake languishing in development hell. This because he kept changing his mind about the theme of the game, resulting in the hodgepodge of map styles and enemies we encounter.

not sure if the link is broken but the article is about using the CS-6400 for a build machine, not the YMP. Generally the YMP would make a bad "build machine" - parallel vector Cray systems were slow as shit doing scalar things like compiling code. I was fortunate enough to work on 3 pre-release 6400 series machine, they were not supercomputers per se (at least in our opinion) but rather large Sun Solaris servers capable of up to 64 CPUs (a lot for the day); this line eventually turned into the E10K from Sun IIRC. We had 3 of them and once the bugs got worked out, we were able to keep them running.

I wish we would have the option to buy superior computers nowadays. 10 GHz 16 core hundreds of GB RAM in a desktop tower. Innovation kind of stopped, now no one invests on single core performance anymore. Notebook CPUs are still at two core and less than 3GHz. People come up with myth that Hz doesn't matter - yes the memory and everything else got faster, but faster clock speed still matters. Maybe AMD's new CPU is a wake up call to release some new better CPUs. It's time to rethink the Pentium3/PentiumM/Core/i7-architecture, maybe give a different architecture idea (like Pentium 4) a new spin.

Or that it got a lot harder/more expensive to further improve single core performance the further we go along, rather than intel/amd being lazy.

Reply on news.ycombinator.com