Generic CPU Benchmark

[SoulsCollective]

Yeah if he says so... just don't think multicore optimization works as well in x86.
Who still runs x86 anyways?

Erm...most of the people in this thread

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

*snip*

In fact, most of the top scores not done on 4GHz quads or Harpertown systems

[jcool]

Do you always work so hard to prove your points? No need to, really.
Guess you're right, something's off. Mainrig in sig does "only" 30,200 (with 24/7 settings and 1,5 weeks uptime tho...).
Should be more compared to some x86 scores.

[SoulsCollective]

Lol, little bit of quoting never hurt anyone

And yeah, compare your 4.1GHz Kentsfield with 4GB of DDR2-1100 to Post #16 - 3.7 Kentsfield with half the RAM operating at a lower speed gets 3k more - because it's running in x86 mode.

[roofsniper]

on the techpowerup thread they created a version that forces x86 mode even on 64 bit systems.

[jcool]

Lol, little bit of quoting never hurt anyone

And yeah, compare your 4.1GHz Kentsfield with 4GB of DDR2-1100 to Post #16 - 3.7 Kentsfield with half the RAM operating at a lower speed gets 3k more - because it's running in x86 mode.

And yet, that performance issue doesn't seem to affect my Octocore.
Or do you think it'd do 80k on x86

[SoulsCollective]

Seeing as you haven't tried the Harpertown on an x86 OS, we have no idea It'll be difficult to find one that supports eight cores properly, though, so the point is rather moot.

[jcool]

XP Pro should do it.. no?
I'm not gonna reinstall for the sake of a benchmark though

[roofsniper]

on the techpowerup thread they created a version that forces x86 mode even on 64 bit systems.

no one see this? the performance should be very close if not equal of a true 32 bit OS.

[jcool]

I can't find it. Link?

[SLK320]

9850BE @ 3,41GHz.

[roofsniper]

I can't find it. Link?

http://forums.techpowerup.com/showthread.php?t=73579

[Particle]

Man I'm so lazy. I wanted to release a 32-bit locked version but never did.

[Particle]

Here's an x86 locked version straight from my compiler. It also shows the relative performance of each thread.

[speedfreak86]

Here is the result in XP x64 and Vista x64 respectively. A very nice gain from a simple switch to x86.

[jcool]

Thanks particle, maybe I'll run it later tonight to see how big the differences are. Next goal should be a proper 64bit version

[roofsniper]

Here is the result in Vista x64 and XP x64 respectively. A very nice gain from a simple switch to x86.

hmmm vista gives better results. thats a surprise.

[speedfreak86]

hmmm vista gives better results. thats a surprise.

Pardon that, I flopped them. Vista got worse results, but at the same time, my xp is running barebones and Vista was with everything running in the background. I would say that they are actually fairly similar. The difference is probably in that last two percent of core 1 you see going to run the OS during the test and in test to test variation.

[roofsniper]

Pardon that, I flopped them. Vista got worse results, but at the same time, my xp is running barebones and Vista was with everything running in the background. I would say that they are actually fairly similar. The difference is probably in that last two percent of core 1 you see going to run the OS during the test and in test to test variation.

i recently reinstalling windows xp and i ran the test on a clean installation and no performance difference from before. and with using the amd fusion program to turn off processes whenever i want i didn't see a change either. but the amd program gave a more consistant score. usualy my score will range around 40 kcus but with fusion on it will stay about the same with a range of about 5-10.

[jcool]

Some more Mhz + x86 mode bench = ridiculous speed

[roofsniper]

Some more Mhz + x86 mode bench = ridiculous speed

well its impossible to beat that score. 8 cores at 4.15 with 920 mhz fb ram. and why is cpu reading odd voltages lately? it says ur cpu has .072 volts. i have seen it say ridiculous numbers for many others as well.

[Titan7171]

a little overclocking

[jcool]

well its impossible to beat that score. 8 cores at 4.15 with 920 mhz fb ram. and why is cpu reading odd voltages lately? it says ur cpu has .072 volts. i have seen it say ridiculous numbers for many others as well.

Sure it's possible. Just give me a pair of X5460 or 5470.. need higher mult
CPUs are only at 1,27V Vcore, nowhere near max frequency. Only the bus is maxed out

CPU-Z always misread Harpers, my E5420's show a similar number. Clovertowns as well, dunno. Just not optimized for server-grade hardware I guess.
Everest shows the correct Vcore, however.

[roofsniper]

CPU-Z always misread Harpers, my E5420's show a similar number. Clovertowns as well, dunno. Just not optimized for server-grade hardware I guess.
Everest shows the correct Vcore, however.

i saw bad readings on some X2s as well.
http://forums.techpowerup.com/attach...6&d=1224657297

[Zucker2k]

Q9550 @ 4.1Ghz, 2x2GB G.Skills 8500s @ 1160MHZ, Vista x64, utorrent downloading in background.

[Calmatory]

I don't like since it hogs the resources. I would've loved to inspect how it uses memory, e.g. CPU cache accces, hit/miss rates, branch success rates, retired instructions, DRAM access etc. :|

10K KCU's with 2.4GHz Athlon64 X2.