Author: Robert Hyatt
Date: 11:19:55 01/16/00
Go up one level in this thread
On January 16, 2000 at 10:33:08, Vincent Lejeune wrote: >On January 15, 2000 at 11:27:10, Robert Hyatt wrote: > >>On January 15, 2000 at 08:31:26, David Blackman wrote: >> >>>On January 15, 2000 at 01:20:28, Robert Hyatt wrote: >>> >>>>It is time for my next project. Today I finished the PO for a new beowulf >>>>cluster machine here. This machine will have 8 nodes, with each node being >>>>a quad xeon 550mhz machine. The nodes will be connected by a gigabit/sec >>>>switch. And no, it won't be crafty's permanent machine. But look for it to >>>>do some interesting matches on ICC later this year when I get to the >>>>distributed search. :) >>> >>>Bob Hyatt does distributed processing? That's something not many people >>>expected. >> >> >>I figured everyone knew I did this. I had a distributed version of Cray >>Blitz. >> >> >>> >>>Distributed processing is something not many programs have used well, or even at >>>all. I think there was something called Sun Phoenix back around 1985? Was >>>Shaeffer involved? >>> >> >> >>There was sun phoenix. And cilkchess. And Waycool. And several others. >>All running on "cubes" using message passing protocols. It is doable. DB >>ran on a message-passing cluster (SP) don't forget. >> >> >> >>>And Deep Blue, but that thing had such an enormous amount of speed it is hard to >>>say if the distributed processing was used well or not. I think Hsu published >>>that he got 35x speedup with 100 cpus. I suspect that's not as easy to do as it >>>sounds, but it also is not as good as you'd like. >>> >>>The few other examples haven't exactly set the world on fire. Bob Hyatt has been >>>one of the people pointing out that efficient use of distributed processing is >>>difficult or impossible. >>> >>>Bob was one of the first and still one of the most successful with small scale >>>SMP (but maybe not for much longer ...). And Cray Blitz was probably the only >>>program to make good use of a vector unit. Maybe after a bit of tuning and >>>experimenting, we will see an efficient distributed processing chess program. >> >> >>I hope so. 32 xeon processors offer a lot. 64 later this year will offer >>more. :) > >I'm just reading a news : Hitachi plans quick move to 32-way Itanium servers at >http://www.eetimes.com/story/OEG20000113S0014 >I have automatically make a mental association between CPU, computer chess, >distributed power Sounds interesting but I'll bet it will cost a fortune. IE the 8xquad machines is costing us about 120K total. I'll bet that 32-way machine ends up at one million or so...
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.