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."
I think. W00T! Anyway, Go is actually a much more interesting and elegant game, for both playing and programming. A neural network program is active in the computer Go scene. Dedicated hardware would be interesting for Go, as it is essentially a simple matrix--maybe a very elegant computer player could be made.
-Libertarian secular transhumanist
Does this actually surprise anyone? The P4 was only an exercise in marketing by Intel - redesign the chipset so it can be clocked nice and high (so it appeals to the average consumer) and to hell with the performance...
I'm not sure this should have been said:
... than a Pentium 3 of an equivalent clock speed."
"A Pentium 4 will be slower at chess than a Pentium 3 of an equivalent clock speed."
That's too easy to be distorted
I'm sure a marketing group or some such, for intel competitors or even PPC, will say
"A Pentium 4 will be slower
And then use it to justify their own means.
Hmmm?
Software to examine chess games would be a perfect example of the major performance improvements to be had with multi-threading. A new thread per processor, with each thread examining different possible move paths, would give dramatic speed gains.
"God, root, what is difference?" - Pitr, userfriendly
It's pretty simple really, the guy is not a computer nerd like us. Note that from the article he says his brother is really the one that built it, and that putting a chip on a motherboard is tricky.
So this Brit (who's REALLY good at chess) put together a machine that overall isn't all that stunning, specifically to play chess.
Let me get this straight: he didn't select a purpose-designed processor, he didn't even do a survey of available processors (forget including non-Intel architecures) to see which would give him the best integer performance for the task, he doesn't consider chipset, he doesn't consider memory architecture, he's willing to accept one hardware-caused crash per month, he seems to think that configuring a machine and having his brother put it together is "building" one, and thinks that a purpose-built machine should be able to accept the OS and data (read: disk contents) from a previous machine without hiccough. While perhaps interesting to the chess afficionados, I fail to see the relevance on Slashdot.
Why are we seeing this article instead of something on any one of the serious chess machines? Why is this article more newsworthy than, say, Anandtech or SharkyExtreme or Tom's Hardware's pick for the baddest machine you can currently build? Just because a Grand Master did it?
To be fair, I have great respect for anyone who can attain the Grand Master level -- that's something I'll never do in my lifetime. He's clearly shown tremendous talent and devotion to chess, and my hat is off to John Nunn for that. But he's a computer harware expert? A supercomputer architect? Are we at the start of a new series of Slashdot articles on computers of the Rich and Famous? What's next, diet tips from RMS? Health advice from Linus? The EFF Cookbook?
Put my fist through my alarm clock with its ding-dong death inside my ear. - The Blackjacks.
Price (as mentioned) and performance in specific applications, if applicable.
"Current platform" is of course also a big reason to stick with one or the other when upgrading, though that might be a little bit more relevant for people upgrading on the Athlon-line, since AMD stuck with SocketA for a long time while Intel enjoyed going from Slot-1/whatever to SocketX/Y/Z forcing motherboard replacement between processors.
Belief is the currency of delusion.
I expect this might be a different picture tomorow, with the much rumored anouncment of the G5@2GHz.
"A language that doesn't affect the way you think about programming, is not worth knowing" - Alan Perlis
If he's concerned about reliability and is having problems convincing his vendor that he's getting hardware errors, he should get ECC memory.
While he's got him on the phone, he should ask the vendor where he can get one of these "equivalent" Pentium III's. I didn't know PIII's came in 3Ghz these days.
The whole point of the differing Pentium 4 architecture is that it scales well with clockspeed; and with the introduction of Hyperthreading on the newer chips, The P4 has really come into its own as far as performance.
Comparing nonexistent chips with existing chips is kind of a pointless exercise.
The theory of relativity doesn't work right in Arkansas.
That is true.
But if you can include special purpose chips (ASICs) in any comaparison, then general purpose processors won't be the best at any single task. An ASIC can be made for any program that can be written, and it'll run that program faster than any CPU made by the same process.
For instance: want to know what is the best CPU for performing matrix multiplication? An ASIC. What's the best CPU for rendering 3D images? An ASIC (like the ones used in modern video cards - they're ASICs of a sort). What's the best CPU for playing Quake III - Arena? An ASIC with the Quake III program encoded in its logic gates... You see how comparisons like this can be silly.
The only thing general-purpose processors are best at is running many different programs.
Disclaimer: IANAL. This post is, however, legal advice, and creates an attorney-client relationship.
That the software doesn't (seem) to exist to use a cluster instead.
No, really, this isn't one of the "imagine a Beowolf of these..." posts. Here's my point: For the cost of just one of the *processers* that he bought, you can build an *entire machine*, happily running an AthlonXP 2700+. An ENTIRE MACHINE. So, for the cost of the two processers, you've got two machines. For the cost of the SuperMicro motherboard and chassis, you can build two MORE machines. With the cost for the rest of the stuff, there's a fifth machine thrown in to boot.
So, what will be faster - a dual 2.8 GHz Xeon, or 5 AthlonXP 2700+ machines? My money's on the cluster, for this particular application. The Xeon machine has 533 MHz of total memory bandwidth, split between two processers, effectively 266 MHz each. The AthlonMP systems, with 333 MHz each, would have a combined bandwidth of 1,665 MHz - about three times that of the Xeon system.
To make it better, the Athlon is MUCH better than the P3 OR the P4 for integer work, which makes me wonder why he would choose the P4 in the first place. Furthermore, not only does the Athlon do much more in a clock cycle than a P4, you'd have a combined clock speed of 10.8 GHz with the Athlons instead of the 5.6 GHz of the Xeons. Twice the clock speed, AND more work per cycle!
Now, of course, being able to actually USE that clock speed would be dependent upon actually transmitting the messages back and forth, and efficiently dividing the work between the machines. In this sort of situation, where for any one point in time, there would be a great deal of possibilities to compute, it would seem like it would divide up very well.
steve
Oh, you're not stuck, you're just unable to let go of the onion rings.
People have always associated clockspeed with performance, even though we got our first taste of the higher performance at lower clock speed with the Pentium 60/66 which stomped on the 486/100.
That first Pentium was a flop (the table look up bug didn't help either). AMD were selling a 486 at 133Mhz, so they can't claim they weren't in the Mhz war themselves.
The Pentium 4 in my opinion is a piece of crap. Thankfully they have increased the frontside bus to 800Mhz, but that is still far behind the 3Ghz clock speed. I prefer SMP systems, so (with Intel Xeon) it only gets worse!
On the other hand the "Banias" cpu for mobiles is very interesting, and I wouldn't mind two in SMP configuration running at 2Ghz, now that would rock! Funny how Intel back peddled with the Mhz issue with this CPU.