Author: Vincent Diepeveen
Date: 14:42:55 01/19/01
Go up one level in this thread
On January 19, 2001 at 17:24:31, Jorge Pichard wrote: >On January 19, 2001 at 16:11:23, Vincent Diepeveen wrote: > >>Hello, >> >>I did a small comparision between 1 Ghz FCPGA cpu mounted on a >>slot1 motherboard Asus P3v4x with a converter stepping cCo. >>It has 133Mhz 2-2-2 SDRAM >> >>I compared with my OWN slot1 dual 800PIII-cumine bx supermicro motherboard. >>In my motherboard there is only 100Mhz 3-3-3 SDRAM. >> >>At those huge speeds of processors SDRAM makes a big difference! >> >>HOWEVER, when i ran the same version with the same amount of hashtables >>(40mb for this test) and also on the 1 Ghz machine, >>then the difference was ONLY 9% speed. >> >>So the 1Ghz machine WAS ONLY 9% faster as a single 800Mhz of mine. >> >>Now we can complain about the OS and about a few programs that >>ran on both machines, but those hardly eated system time. >> >>Speed difference on paper should be 30% of so, and there is a HUGE >>gap between 9% and 30%. >> >>What did INTEL do to the FCPGA processors, and do others have the >>same results there as i have here? >> >>Unless this speed difference was because of some mistake like >>a bug in the comparision at any point, i am VERY dissappointed!! >> >>Greetings, >>Vincent > >Sorry to hear that, but after you spent some considerable amount of Money just >to increase your program by a few rating points. I hope that others don't make >the same mistake as you did. > >Pichard. Well diep is dual, so a single cpu is never going to outgun a dual. I didn't spent the money. The machine is not my own. It's of course a fast machine, not bad to spend money on a fast machine anyway. The machine when running with 150Mhz memory and not connected to internet when running and so on, just diep running is now 14.4% faster as my PIII800. So that's 5% win for faster memory and 100% sure no other conditions as that. I don't know how much the RAM part of this is and what part the not being connected to internet part is (as 1 internet program ran on the background idling in the first test where i had 9% difference). Anyway, personally i would have expected more like something as 25 to 30% difference between this 1Ghz and 800Mhz P3, whereas it is in reality only have of it. I'm simply missing a performance increase from around 10 to 15%, which is a LOT. Perhaps buy AMD stocks/shares? I remember first K7 at 1 Ghz were also dissappointing in januari 2000, it appeared they had clocked down the L2 cache from 1/2 processor speed to 1/3 processor speed. Then instead of being 20% faster as a PII at the same clock they were only 5% or something faster as a PII virtually at 1Ghz (bus speeds of RAM not counted in that experiment). So intel seems more or less with P3 to be at the same point where AMD was in januari 2000. Greetings, Vincent
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.