Good Question Here,,

truehighroller

Distinguished
Oct 1, 2006
608
0
18,980
I have the following stuff that matters in my system first off.

GA-P35-DQ6
2 Western digital 160Mb 16Mb Cahe SATA Drives in Raid0
1900XTX
Current Processor E6750 @ 3.7Ghz
Crucial Ballistix Tracer DDR2 1066
I get a 3DMARK score of 7760~ with this setup now,,,.

My question is, if I get a Q6600 and keep my 1900xtx will this affect my total CPU score? Not my whole score but just my CPU score,,,.

I am just curious if the GPU affects the CPU score at all. I actually ran 3dmark with my GPU underclocked then ran it Overclocked and maybe , maybe , only noticed a 10 point difference on my total for my CPU score.
 

truehighroller

Distinguished
Oct 1, 2006
608
0
18,980

Bump for answer,, Please.
 

Thanatos421

Distinguished
Mar 26, 2007
549
0
18,990
No. You can run the test over and over and over, and everytime, you will most likely get a different CPU score. The CPU test in 3dMark doesn't touch your GPU at all. It simply renders the scene using raw CPU power.
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

I've always wondered about that. If I used integrated versus Tri-SLI 8800Ultra...

Common sense would say the cpu score for the TRI-SLI 8800Ultra would be higher on avg.
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



Your common sense says that a CPU score should be HIGHER with a different GPU?
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



Rendered by a CPU? :kaola:

I talked to Wireless, I think he saw a 150-200 point increase in CPU score from a 3870 to 8800 GTS. However as high as he was that was within the 5% error range.
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

So the graphics were rendered by the cpu? Don't you need some sort of GPU to render them?

I also thought GPU takes off load from CPU so...
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



GPU does take load off the CPU... But cmon, you get 1-2 FPS in that video. Does it look THAT advanced?

CPU is a general processor. It can do GPU calculations but at a much reduced rate. It's not specialized for anything. Just like a CPU can do ray-tracing.. albeit very slowly. Ray tracing cards running at much slower speeds but specialized for ray-tracing.

Same thing with physics. The CPU can do them (and currently does). However in the future they may offload more of that to a Physics processor which will be specialized and be able to do more calculations in less time.
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

Talk to people who used to play Pacman.


I know CPUs can do GPU calculations and all. So does that mean the GPU isn't helping the CPU at all during the CPU tests?
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



Maybe it is slightly, but if it is than I really think that validity of the CPU score would be in question. Maybe you could call me a purist on that.. but if it's a true CPU test the GPU should be irrelevant.

If the GPU does help in anyway I think that really invalidates the CPU Score part of 3D Mark06.

Oh and as far as PacMan, I meant advanced compared to the rest of the test... I still play Pong occasionally.. hell my NES is still hooked up. I love Tecmo Super Bowl.
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

So if you remove the GPU and do the tests, will it be more accurate?

I didn't start gaming until N64. That was my first gaming console ever. Well, unless you can call a TI83 a gaming console. :D
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



Lol.. Drug Wars?

And yes, I think if you isolate the GPU so that the CPU does all the work it would be a valid test of the CPU. I mean.. logically.. Should a GPU have any effect whatsoever on a CPU test?

When comparing a 3.0 GHz QX9650 versus a 3.2 GHz Q6600 should the Video card matter?
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

Yeah, when you're comparing frame rates in games. Usually they use the same video card to cross out any extra variables. Oh and they use the best video card out there too. Imagine using a 8400GS or something.

As for the CPU test, logically a GPU should NOT have any effect on a CPU test. BUT my twisted common sense tells me it does. After all, the graphics is being viewed through the monitor, which in turn is connected to the GPU. That means the CPU had to "talk" with the GPU in order to see anything on the monitor.

I seriously don't know anything about deeper stuff in computers and programing. I just, well, use logic.
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



I can see both sides... I guess my logic wants to tell me that GPU is irrelevant, but you are probably right.

And yes, you are right about using the same video cards... But at the same time I want to compare system X to system Y on Tom's. I have a QX9650 and an X1900XTX and you have a Phenom 9900 with Tri-SLI... Know what I mean?
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780

NO I DON'T.

Phenom...Phenom...PHENOM!

How about I have the QX9650 WITH the Tri-SLI and you have the Phenom 9900 with the X1900XTX?

Then I'll understand :kaola:
 

truehighroller

Distinguished
Oct 1, 2006
608
0
18,980
No I'm not eighteen... I'm twenty eight. I just say it out of habbit because I am a slave. :p I just say it to say it,, alot. Sorry I don't speak exactly like you in every way lol. Would the world be better if every one did? Any way , I was just wondering about the whole cpu/gpu thingy for awhile now and finally got around to posting it. I appreciate the input.
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780


Lol, I was giving Thanatos a hard time.