Author: Bernhard Bauer
Date: 06:22:38 09/17/99
Go up one level in this thread
On September 15, 1999 at 23:00:27, Robert Hyatt wrote:
>On September 15, 1999 at 16:57:32, James T. Walker wrote:
>
>>I've noticed that the benchmarks of crafty vary with each version. What do they
>>represent? What is the significance of the "Total elapsed time" vs the "Raw
>>nodes per second". If the total elapsed time is greater and so is the Raw nodes
>>per second, what is the meaning? For example, at Dann Corbit's site I found
>>Crafty1.exe which appears to be Crafty 16.18. I noticed its Raw nodes per
>>second is lower than the Crafty 16.18 at Bob's site but so is the Total elapsed
>>time. Which is actually faster? Can someone please explain.
>>Jim Walker
>
>
>Like everything in crafty, this is experimental. Jason wrote that code and
>has been trying to turn it into more meaningful information. The current
>version took too long so I toned it down a bit... but it still isn't exactly
>what is needed...
I ran Crafty's benchmark for version 16.8 by starting the execulable and typing
bench. I used no crafty.rc, so no special hash settings. I ran bench with
1. Dann Corbit's executable
2. Bob Hyatt's executable
3. My executable
all on a 450 MHz pentium.
I got
nodes nps sec time_to-ply
Dann Corbit 61519180 249065 247 2.591
Bob Hyatt 71022354 280720 253 2.529
My own 191067196 250744 762 0.8399
Question: Why does my own result differ so much in "nodes". I used MSVC++ 6.0.
Remark: This version looks really god to me. Perhaps it is the strongest ever.
Kind regards
Bernhard
Bernhard
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.