Author: Slater Wold
Date: 20:12:00 09/05/02
Go up one level in this thread
On September 05, 2002 at 23:00:32, Dann Corbit wrote: >>I understand what you're getting out, but I do not agree. Simply because the >>definition of "relative speedup" is "the ratio of the >>serial run time of a parallel application for solving a problem on a >>single processor, to the time taken by the same parallel application >>to solve the same problem on n processors". It's all about "run time" and less >>about "run parameters". IMO. >> >>As long as both runs were using the *same exact* settings, I think all would be >>fair. >> >> >>Also, I simply used 'st 60' in Crafty. A *lot* of positions were thrown out >>because a.) they were solved at root or b.) the search time was less than 60 >>seconds. >> >>WAC is probably not an "optimal" suite to use, because 99% of the positions are >>solved so easily. If anyone wants to put something together for me that suits >>me better, I would greatly appreciate it. >> >>Going over 600 position logs is eating all my time at the moment. ;) > >The problem in this case is: >What *exactly* are you measuring? How are you calculating the speedup? I am measuring everything. I wasn't planning on "calculating" anything. Simply providing results that prove to be interesting. I was looking for ideas on a new (or just improved way) of determining *computer chess* speedup. However, looking at these logs, it's not going to be possible, IMO. >I doubt if you can do it accurately. I am too, now. >Two different "11-ply" searches can be drastically different. Mine were. That was the point. ;) >I am pretty sure that just NPS will give a wildly wrong answer. Or may. I'm >not sure. I am sure that I don't trust most common sense sorts of measurements >unless they have a way to compensate for parallel effects (such as improved hash >table utilization). This is the part that amazes me, immensely. 1 CPU: NPS: 800k NTS: 120,000,000 TTS: 150 seconds 2 CPUs: NPS: 1300k NTS: 150,000,000 TTS: 115 seconds *This is all hypothetical, of course.* It *always* takes more nodes to solve a problem using more than 1 CPU, than it does using just 1 CPU. It *always* takes less time to solve a problem using more than 1 CPU, than it does using just 1 CPU. And that would be more/less what I am trying to "coorelate" here; Nodes To Solve vs Time To Solve.
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.