UPDATED! - My Tests are In! The new 81.26 drivers DO use both cores! More than 70% 2x core usage.

[johnnqq]

NEVER. amd>intel as long as they're prices are lower!!

 

[fbrdphreak]

we definitely need benchies to verify this. Perhaps run 3DMark01 or something for a CPU-bound test?

 

[Velk]

Originally posted by: fbrdphreak
we definitely need benchies to verify this. Perhaps run 3DMark01 or something for a CPU-bound test?

Benches to verify what ?

 

[SuperStrokey]

wow this makes me want to go dual core if thats really true. Any single core results?

 

[fbrdphreak]

Originally posted by: Velk

Originally posted by: fbrdphreak
we definitely need benchies to verify this. Perhaps run 3DMark01 or something for a CPU-bound test?

Benches to verify what ?

That the two cores are ACTUALLY improving performance. First of all, I'm a bit skeptical that a higher reported usage means that the two cores are actually being used. Even if they are, how do we know they influence performance without performance tests? Jason did a great job, but we need performance verification.

 

[Polish3d]

Make sure its done on a 7800GTX too if possible as I read that the optimizations are only for the 7000 series...may be inaccurate though

 

[formulav8]

OK, I ran a couple benches. I used versions: 77.72, 78.01, and 81.26 drivers. Below are my results.

*************************

*81.26 Driver Results*

3DMark2001: 22,191

Aquamark: 63,645

------------------------

*78.01 Driver Results*

3DMark2001: 21,346

Aquamark: 60,734

------------------------

*77.72 Driver Results*

3DMark2001: 21,215

Aquamark: 60,845

*************************

So, although the results are not huge, there ARE almost a 5% performance increase in the use of the 81.26 drivers which uses some of the 2nd cpu core.

So, a free performance boost for Dual Cpu/Core users is not to shabby at all



Jason

 

[fbrdphreak]

Cool. Whenever you get bored, how about 3DM05 with CPU score too Thanks Jason!!

 

[blckgrffn]

Very nice! Too bad those damn things are still out of my price range! Never spend more than $175 on a CPU, personally

Thanks for all the work!

Nat

 

[formulav8]

Will do, I just got done installing version 2003 and 2005. But, I won't be able to run them until tomorrow. I am heading to bed in a few minutes


Jason

Originally posted by: fbrdphreak
Cool. Whenever you get bored, how about 3DM05 with CPU score too Thanks Jason!!

 

[fbrdphreak]

Night

 

[Velk]

Originally posted by: fbrdphreak

Originally posted by: Velk

Originally posted by: fbrdphreak
we definitely need benchies to verify this. Perhaps run 3DMark01 or something for a CPU-bound test?

Benches to verify what ?

That the two cores are ACTUALLY improving performance. First of all, I'm a bit skeptical that a higher reported usage means that the two cores are actually being used. Even if they are, how do we know they influence performance without performance tests? Jason did a great job, but we need performance verification.

Scroll up. I already posted that earlier in this thread, was wondering if you missed it, or were talking about hyperthreading.

 

[fbrdphreak]

I posted that before Jason posted his benchies

 

[stevty2889]

I just installed the 81.26 drivers as well. X2 rig in my sig.

3Dmark05 results: % over 77.77
77.77 drivers 7342 80.40 7379 81.26 drivers 7509 ~2.3% increase

3Dmark03 results:
77.77 drivers 15481 80.40 15511 81.26 drivers 15899 ~2.7% increase

3Dmark01 results:
77.77 drivers 24331 80.40 25774 81.26 drivers 27080 ~11% increase

PCmark05 results:
77.77 drivers 5195 81.26 drivers 5714 ~10% increase

Aquamark:
80.40 drivers 79743 81.26 drivers 91,879 ~15% increase
15,699 GFX
11,075 CPU
91.88 FPS
27,658,712 TPS

Also ran it on my Pentium-M based system, but I was getting horrible artifacts..
3Dmark05:
77.77 3303 81.26 3289

Aquamark:
77.77 58050 81.26 58464

 

[orangat]

The 3dmark scores don't seem to move much.

 

[fbrdphreak]

What I'm wondering is if the 11% on 3DM01 is CPU or GPU performance increase? Perhaps include detailed results (i.e. sub-scores)

 

[Keysplayr]

I would love to see if this has any effect on Hyper Threaded systems.

 

[Avalon]

Originally posted by: orangat
The 3dmark scores don't seem to move much.

3dmark01 was more based on system wide performance, whereas 03 and 05 are almost completely GPU derived, with 03 being slightly less so.

 

[Topweasel]

I Don't think that its using the second core the way everyone thinks it is. I think that these drivers are doing some process off loading, eventually they would have to stop doing this as programs actually start using the second core. It also means that CPU intensive program won't prove a thing, as it not the CPU feeding the video card like normal, but the CPU doing some video work. Its like the good old days of software 3D.

 

[Duvie]

I can confrim Stevty's results in 01...I had same 10% and I was running less ram speed, and slwoer by50mhz cpu wise..It was compared to drivers beofre 77.x...so while it could have endedup being 14% it was likely due to some enhancements done between that set of drivers and the 77.x's....

 

[gsellis]

I have obviously missed something in this issue with the drivers not working with dual cores. Did I follow this right, CPU utilization went UP and now on both cores? If so, why is this good? What am I missing here? I thought you said it was 50% and went to 70% ( or did you mean there was a 70% improvement on the 50% it used to use?)

I would want CPU utilization to be balanced and go down.

And yes, Perfmon, using a Log file. Runs in the background until you turn it off, tunable to what ever interval you want. Better tool for a performance monitor.

 

[Ackmed]

Whats with all this 3dmar and Aquamark crap? Show us some real games results please. At the highest res you can do. None of this 640x480 stuff... 1600x1200+would be great, thanks.

 

[formulav8]

Ackmed, We are making the CPU the bottleneck to make sure its the drivers using the 2nd core thats increasing the performance and not general driver optimizations. Plus, all my benches were run at 1024x768 not 640x480 as you supposed.

I do NOT know how to bench BF2 at the moment or KOTOR or any of those 'other' games. So, what I posted, is the best I can do at the moment. Sorry it doesn't meet your standards. I just did what I was able to do at that moment.

---------

To whoever was wondering earlier about eh 50% up to 70% cpu usage I try to explain alittle below


When I mentioned 50% or so with the 77.72 and 78.01 drivers and then 70% with the 81.26 drivers, that is the TOTAL system usage. That is the balance between the total of both cores.

So, the 77.72 and 78.01 drivers were definitely NOT using he 2nd core to do much of anything while playing the games since the total system usage was only 50% or so which would be the performance of 1x core.

BUT, 81.26 drivers DOES show that they aRE using up to 20% of the 2nd core to do some type of rendering.

Sorry if I don't make sense Maybe I will try later.


Jason

 

[rise]

thanks for the work formulav8. :thumbsup:

if you use FRAPS you can hit f11(?) during game play and it will record the fps for 60 seconds and save it to a text file.

 

[UltraWide]

My Pentium D 820 buy does not seem as silly anymore. very exciting news!