Difference between revisions of "Talk:Socket 7 benchmark results"

Revision as of 01:26, 25 February 2013

RacoonRider, I noticed that some of your articles are nowhere linked on the main page or even other articles. If no one can find your articles, they won't be of much use to anyone. Just saying :) D1stortion (talk) 19:52, 21 February 2013 (EST)

Thanks, I'll take that into account :) RacoonRider (talk) 20:22, 21 February 2013 (EST)

Could you please check the config of your benchmark machine once more? It is a bit odd that I get 223 MB/s mem speed whereas you get 157 MB/s on a very similar machine (this is 42% higher!). At least it represents the few percent different towards SDRAM usage quite well. The same performance behavior is also shown by Mad Lunatics score for the 430FX with P200. Enigma (talk) 02:17, 24 February 2013 (EST)

I guess, that's because SpeedSys 4.75 uses abstract "memory score", not actual bandwith. If you look at memory speed curves, they're practically identical. I'll find 4.78 and rerun the test. RacoonRider (talk) 20:54, 24 February 2013 (EST)

Speedsys 4.75 and 4.78 results are confirmed to match. Memory bandwith is confirmed to be a function of FSB, not CPU speed. Bandwith was tested with the same CPU on different FSB speeds (50MHz, 60MHz, 66MHz, 75MHz). Results are in the table. RacoonRider (talk) 22:00, 24 February 2013 (EST)

Could you please check the Pentium table for the Pentium 150 MHz. Shouldn't it be that the FSB states 50 MHz and multiplier 3x ? Enigma (talk) 01:26, 25 February 2013 (EST)