Author: Robert Hyatt
Date: 10:36:42 07/21/00
Go up one level in this thread
On July 21, 2000 at 11:45:37, Amir Ban wrote: >On July 21, 2000 at 11:15:24, Robert Hyatt wrote: > >>On July 21, 2000 at 03:54:59, Amir Ban wrote: >> >>>On July 20, 2000 at 22:49:46, Robert Hyatt wrote: >>> >>>>On July 20, 2000 at 21:59:03, Chris Carson wrote: >>>> >>>>>. >>>> >>>> >>>>Is there some reason why I can't quote DB's MAX NPS? 2M-2.5M is Junior's >>>>MAX NPS. It doesn't run that fast all the time. >>> >>>No. 2-2.5 MNPS was typical (and I think it is clear from the KC page that gave >>>that number). The highest number I saw was 3.8M. >>> >>>Amir >> >> >>It is still "raw" which was my point. Hsu has factored out the SMP search >>overhead nodes from his 200M value, which means his number is not comparable to >>ours. >> >>My numbers on the quad xeon go from about 500K to 1.5M. Raw. Multiply by >>roughly 3.2/4.0 to get an equivalent number to compare to Hsu's... > >I don't care about your point and I'm not involved in this argument. It's just >that you are inventing stuff as you go and I wish you wouldn't do that when >talking about my program. > >Thanks. > >Amir What did I invent? Do you report RAW NPS or effective NPS? I assumed your 2.5M was a max number, as nobody said any different. I was wrong. I assume your NPS is a raw number (total nodes searched by all threads divided by the total elapsed time used to search them) which is apparently right. But I didn't invent _anything_ that I know of...
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.