Slashdot Mirror
Building A Homemade Chess Supercomputer
nado writes "There's a new article on Chessbase.com which has
GM John Nunn showing you his chess-orientated PC upgrade to a double Xeon system, with some Fritz benchmarks." Elsewhere in the article, John Nunn discusses the unique computer needs for chess computation: "One of the problems with currently available processors is that they are not particularly well suited to the integer calculations used for chess. A Pentium 4 will be slower at chess than a Pentium 3 of an equivalent clock speed."
This is sorta true, but the P4 was an excercize to see where AMD would go. By making chips that they knew would be really fast once they got the foundation down, they drove AMD to finding a way out. My guess is that Intel was trying to drive them into a corner since they couldn't build an Itanium-compatible processor without a license, but instead AMD knew when they were beat and decided to move on to a higher playing field.
Like this delicate game of chess, Intel's next move is uncertain. While the P4 has what is needed to smoke the Athlon for years (just as long as they keep tweaking the predication engine and improve on branch prediction's accuracy), it can't really compete with Opteron. Neither can Itanium. Intel just hasn't invested enough in the future since they were ruling the present. I even read somewhere that they had started Williamette and Itanium (forgot the codename) at nearly the same time back in 95? but neither really caught their supervisors eyes since they were more than profittable already. So in short, Intel's game of chess has been too passive for too long. And it's not time to look back on the P3 and say what is good... the P4 is something completely new, like the pentium one was so long ago. Give it some time and it will vastly out perform the P3 comparitively, but you have to realize that the P3 is something like 6 years old, and the Athlon even older than that.... It's time people stop trolling on how past processors were faster comparitively and move on to making the new processors faster. Borrowing from the old is ok.. Look at Banias for example.. but we seriously need to worry about the future more. But thanks for trolling on by.
"Victory means exit strategy, and it's important for the President to explain to us what the exit strategy is." G.W.Bush
IIRC, the P3 was the end of the Pentium Pro core family. The P4 is the beginning of a new core. So, yes, the first P4s were not the best chips and the P3s of the same time were better buys. But do you remember the P-Pro? Give them a few years of refinement and then judge the core.
-B
See my earlier post, asking how old FritzMark is, because the article says that it only uses one processor - ie: It's not a multi-threaded app.
"God, root, what is difference?" - Pitr, userfriendly
I just turned up a dual Xeon 2.4 rack-mount server for work and it's BIOS mentioned warned us to turn off Hyperthreading for anything other than Windows XP or Linux 2.4 (yeah, mention of Linux in BIOS! :).
:)
Anyways, since I am using linux 2.4, two hyperthreaded Xeons look like four processors to the box, I"m sure it's not the same performance of for seperate processors, but I'm hopeing it's at least slightly better then two non Xeons
The writer of the article wrote that for Windows he prefers 2000 over XP. I am curious if XP (or Linux 2.4) and thus Hyperthreading might help his already built computer with a bit more performance...
Please send all UCE to scally@devolution.com so I can f
Pentium 4 clock speed vs. performance discussion...
Seconds before G3, G4 or PPC970 is mentioned:
3...
My home desktop machines both have ECC memory. I never open the boxes. Haven't had a crash on either the Windows 2000 machine or the QNX machine in over a year.
Why is it that no one has written a chess benchmarking program for the mac (ie *nix)?
I mean, for number crunching and math and calcs, the mac seems to rule close to the top...
just my 2cents
"an eye for an eye only makes the whole world blind"
And after all of that, you can go buy a P4 or an Athlon and they will both perform and work just about the same, in just about every task.
You summed it up so eloquently. Thankyou.
Of course there's always the notion that we should be using asynchronous logic now, anyway; That logic has sped up to the point where it is not actually important to have everything happen on the same clock, but instead more useful to have it occur as rapidly as possible.
As for your automotive analogy, the primary reason that it seems that doing things the Japanese way is practical is that you keep the weight down, which is good from the standpoint that your handling improves and you simply have less weight to push around. On the other hand, a large engine need not necessarily be heavier than a small one. The primary advantage today (it seems to me) is that it ends up being cheaper on gas to run the smaller motor, but of course the more power you use, the more fuel you throw down the thing. Larger cars can be much lighter now than they used to be, though, what with aluminum getting cheap and high strength steel being readily available, not to mention that monocoque technology has moved along nicely with all this computer modeling.
Anyway aside from that digression; I don't think either company really has the win here, just like small engines and large engines are interesting for different reasons, though you can certainly get more power out of larger engines... It only becomes more expensive at a certain point. Of course, Intel's processors are artificially expensive, simply because people pay for them; AMD's are as well, though to a lesser extent. Silly analogies :(
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
not sure if this is off-topic...mod me thus if you must.
I play free internet chess at the Free Internet Chess Server. Find them at...you guessed it: www.freechess.org.
All you CLI guys out there will love the fact that using a graphical client is optional! For those of us who are sane, there are a handful of graphical boards available to complement the irc-ish interface that allows people to find opponents.
It's fairly popular already, but I sure wouldn't mind a bigger crowd...cause all the guys on there kick my arse consistently. I've got a whopping 1300-something rating right now, and I'm already 0-3 for the night...sheesh.
peace.
@ASP.NET's parent-teacher meeting: "Little Johnny.NET is very bright, but he doesn't play well with others."
Therefore, chess programs try to estimate the best moves, based on attack patterns and history. They are quite good at this and take a correct estimation for the best move in most cases.
This means that adding processors/threads does not give a lineair speedup by far. Although the second processor does in fact help a lot, it is estimated that more than eight processors makes no sense at all, and even decreases performance (hash table locking). Google "alpha beta search parallel" if you want to know more.
Ah, yes. You bring up a great point.
Performance, as I have it defined in my head is simply the time it takes to complete a task. The lower the better. For this guy's chess application, I think he is more concerned with absolute performance irrespective of total energy consumed.
Your definition includes power efficiency as a consideration. This is a worthy metric that is not lost on the engineers at Intel and AMD - let me assure you. There's a very talented team of engineers in Haifa, Israel building very power efficient designs. Banias is only a taste of what's to come.
Actually Othello is very much a solved problem. Logistello beat the Othello world champion 5-0 in 199something, so your pocket PC can play better Othello than any human hands down.
:-)
Lucky for you, most othello implementations are total crap.
-josh