Author: Mathieu Pagé
Date: 08:38:16 08/15/05
Go up one level in this thread
>>So you are maintaining that this thing is twice as fast as an Xeon 2.8ghz ? >absolutely. Here is some data. I'm going to use one cpu on my xeon 2.8ghz box >to search kopec position 22 to 14 plies... Then I'll do the same on a single >opteron CPU. First the xeon: > > 14 52.77 -0.31 1. ... Bxe4 2. Bxe4 Qxc4 3. Qxc4 Rxc4 > 4. Nxb6 Nxb6 5. Bd3 Rc7 6. Nd4 Rd7 > 7. Nc6 a4 8. Bd4 Nfd5 > 14-> 1:35 -0.31 1. ... Bxe4 2. Bxe4 Qxc4 3. Qxc4 Rxc4 > 4. Nxb6 Nxb6 5. Bd3 Rc7 6. Nd4 Rd7 > 7. Nc6 a4 8. Bd4 Nfd5 > >and then the opteron: > > 14 26.58 -0.31 1. ... Bxe4 2. Bxe4 Qxc4 3. Qxc4 Rxc4 > 4. Nxb6 Nxb6 5. Bd3 Rc7 6. Nd4 Rd7 > 7. Nc6 a4 8. Bd4 Nfd5 > 14-> 43.28 -0.31 1. ... Bxe4 2. Bxe4 Qxc4 3. Qxc4 Rxc4 > 4. Nxb6 Nxb6 5. Bd3 Rc7 6. Nd4 Rd7 > 7. Nc6 a4 8. Bd4 Nfd5 > >So, 95 / 43.28 = 2.19X faster. > >Note that this is a 2.2ghz opteron, not 2.4, the 2.4 will be 10% faster still. >Also note that this is a 32 bit version of crafty on the xeon (the xeon is a 32 >bit architecture) vs a 64 bit version of crafty on the opteron. > >there is simply no comparison... Ok, i'm convinced. I did not tough the difference was that huge. Well, as I said I know only a little about AMD processors. >>btw I just saw you beat Diep, congratulations. I wonder how you manage that, I >>mean aren't you playing remotely? Unbelivable! > >Yes. Remotely, on a NUMA architecture with a program that knows nothing about >NUMA, and on a parallel NUMA box with 8 cpus with a program that won't scale to >8 cpus... > >amazing... > >:) Good luck for the end of the tournament. Mathieu Pagé
This page took 0 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.