Author: Robert Hyatt
Date: 18:14:25 01/27/00
Go up one level in this thread
On January 27, 2000 at 19:19:26, Tom Kerrigan wrote: >On January 27, 2000 at 17:37:28, Robert Hyatt wrote: > >>The cray is a GP machine. The Xeon is a GP machine. At equal clock periods, >>the Cray is roughly 200 times faster. If you had asked me beforehand, I would >>not have guessed that... > >You're contradicting yourself. In one breath, you say that Hsu picked some >arbitrary high number. In the next breath, you're saying that number might be >super-optimized, vectorized Cray assembly instructions. That's like saying Hsu >made his arbitrary HIGH number as LOW as he possibly could. It makes absolutely >no sense. > No... what I said was that I knew what Cray Blitz did internally, and how it worked. And I was _still_ surprised by how much slower the PC would execute the code. If you had asked me, I would not have guessed that large a ratio with equal clock speeds. If I had written a paper, I would not have guessed that high. Yet, I could actually get the _real_ number by running the test. Hsu didn't have that luxury unless he wanted to first write the code, then count the instructions. So the point is that even being very familiar with how something works doesn't mean that accurately guessing anything about how it would look on a totally different platform is easy to do... >Here's the other thing that makes no sense: if Hsu just picked some high number, >why didn't he make it higher? He picked some number that is definitely feasable >on a fast PC. Instead, he could have just as easily made it 10 times bigger, and >I wouldn't even be having this debate with you. What is the difference between 40K and 400K instructions? Significant enough to make the debate go away. What about 4K vs 40K? Would we say "OK, 4K can be done in a PC, since everyone is hitting 4K already?" IE where is the threshold of doable? Only a factor of 10? Only a factor of 100? I don't think it really matters that much... > >I simply can't accept the fact that his estimate is arbitrary. There's just no >reason to. > >>Not at all. I don't think "the number" is important. I think the "bigness" >>is... > >Right, I don't care that it's EXACTLY 40k. It might as well be 20k or 80k >(similar "bigness"). The point is that all of these are possible on a PC. My point. And 80K is still ugly on a PC. 100K is ugly. 40K is ugly. And I consider them equally probable and equally ugly.. > >>To an extent. But I can answer that with one AND. How many for a non-bitboard >>program? Wouldn't the non-bitboarder and I come up with _different_ estimates >>for the same thing. And wouldn't we _both_ be right? And wrong? > >It doesn't matter HOW he thought he could compute the term, only only matters >that it's POSSIBLE. But it _does_ matter. Otherwise how do you do the computation? And are you thinking in terms of what is easy in hardware? Have you ever written a piece of code, thinking when you started that it would be 2K lines of code, but ended up 20K? I would never have guessed that Crafty would end up at nearly 50K and still growing. > >Another possibility is that he wrote a function that did similar work to DB's >evaluation function. It might produce total garbage, and not use any of the >"correct" weights, but it would have "some of these terms" and "some of those >terms" and he could draw conclusions from it. No doubt. But others are tricky yet not obvious. My 'pawn lever' code is a lot more complex than I would have guessed. It is a lot more complex than if I had designed it in hardware. As I said, 40K might be perfectly correct. It might be off by a factor of 10 either way. Without actual code, it is im- possible to say... > >You seem to accept everything Hsu says as gospel, except for this ONE estimate. >You're coming up with every excuse to invalidate this statistic, even to the >point of suggesting that it's an arbitrary big number. Why? Don't you have some >faith in Hsu and his work and what he writes? Maybe it IS a hard estimate to >make, but isn't it possible that he went about it in an intelligent way? > >-Tom Anything is possible. I simply take that "number" as a "SWAG" to make a point that a _lot_ of things are going on inside DB's processor. If _I_ had written that article, I would probably have guessed "low" to be conservative, since everyone seems to want to jump all over any significant claims he makes...
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.