Author: Matt Taylor
Date: 15:30:05 01/13/03
Go up one level in this thread
On January 13, 2003 at 17:59:50, Dann Corbit wrote: >On January 13, 2003 at 17:02:10, Sander de Zoete wrote: >[snip] >>It is >>possible that, even with a relatively thorough evaluation function, the nps >>(Nodes Per Second) rate could exceed 10,000,000. > >Since a bitboard is 8 bytes, that is 80 MB per second bandwidth, just moving the >data with *no computations*. It would be quite a remarkable achievement if you >could attain it. > >PCI-X can theoretically do 10x that, but we still have to deal with memory >access speed. 80 MB/sec is easy. In worst-case, a machine on a 100 MHz bus with 10T latency (like SDRAM) that always cache misses, the bus can still deliver 80 MB/sec. I think the real bottleneck would be the misjudgement of the speed of MMX. It is not as fast to respond as the integer units, though it maintains similar throughput. Using MMX for 64-bit arithmetic is not worthwhile as the same operations are available from the integer unit with lower setup costs. The only advantages include a minor gain in parallelism in hand-tweaked code and additional register space. -Matt
This page took 0.01 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.