Author: Robert Hyatt
Date: 07:36:08 12/10/00
Go up one level in this thread
On December 09, 2000 at 23:48:39, Joshua Lee wrote: >> >>With the right kind of programming, the T932 will simply blow the doors (and >>the paint) off of any of those machines. Remember the simple math: >> >>each cpu can read 32 bytes per clock cycle, and write 16 bytes per clock >>cycle. That is 48 bytes, times 32 cpus, at a clock rate of 500mhz. I >>think that _any_ machine will choke when presented with that kind of data >>throughput requirement. :) > >so 48x32 1536 x 500Million if this is right that's >76.8 Trillion bytes i know from memory that the athlon bus at 266mhz is >1.6Gbytes per second bandwidth so (with ddr chipset ofcourse and probably a >1.2Ghz cpu......) > > >480 times as much if i am wrong in comparing it this way .... my 800mhz athlon >found a win in a cray blitz game (not on this T932 beast) and took 1hour >13minutes and 37 seconds 1,951,276,000nodes it would find this 480 times >faster???? 9.2 seconds it doesn't sound right but then again it is a cray..... In the game in question, CB found the right move quickly. But a timing problem let a worse move overwrite the correct move in very rare and hard to reproduce circumstances. The Cray is _very_ fast... yes... Of course, you have to remember that the machine used in that 1987 game is far slower than the T932 of today... and I am sure that I can put together a machine today that would be faster than the Cray of 1986 when applied to playing chess. ie an 8-way 700mhz machine would probably be faster than that machine.
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.