Effect of CPU on 3dmark06

[Nicocat]

What kind of effect would increasing the processor have on a 3dMark06 score, considering this starting system:

HD 3870x2 graphics
Athlon64 X2 6400+ at 3.4ghz
2 gig of RAM
Windows XP
Etc, etc.

The CPU score is negligible as it stands, but is the processor holding back, say, the video card? I get around 12k 3dmarks, and that seems low from hearing other, comparable systems' scores

 

[birdman1016]

with my q6600 and a x2 i got 15k

 

[mactronix]

My understanding of it is that 3dM06 is very heavy on the cpu and you need a lot of power differance processor wise to make much of a differance.
You could always do a search on the Orb and compare your score to those with a similar system but bigger/better CPU and work out that way if you think its worth upgrading.
Mactronix

 

[seabreeze]

Hi, what's the break up of the score?

3DMark Score
SM 2.0 Score
SM 3.0 Score
CPU Score
Graphics Tests
1 - Return to Proxycon
2 - Firefly Forest
CPU Tests
CPU1 - Red Valley
CPU2 - Red Valley
HDR Tests
1 - Canyon Flight (SM 3.0)
2 - Deep Freeze (SM 3.0)

CPU can have a big effect on 3D Mark 06 results, the following is a 8800GTS320 with different CPUs, same hardware (yes, changes in FSB to suit processor):
Cel 347 @ 3GHz 4334 marks, SM 2.0 = 1865, SM 3.0 = 2614, CPU = 761.
P4 631 @ 3GHz 5178 marks, SM 2.0 = 2370, SM 3.0 = 3023, CPU = 895.
E6850 @ 3GHz 9726 marks, SM 2.0 = 3970, SM 3.0 = 4044, CPU = 2614.

Note that Tom's Hardware has SM 2.0 as 4315 and SM 3.0 as 4568 for this card, so this means my GTS ran at the following percentages compared to the 4315/4568 scores due to CPU bottleneck:
Cel 347 SM 2.0 43%, SM 3.0 57%.
P4 631 SM 2.0 55%, SM 3.0 61%.
E6850 SM 2.0 92%, SM 3.0 89%.

So even though they all ran at 3GHz on the same rig with the same hardware, it can just show how CPU clock speed is not everything. To put it another way, 3.4GHz on your Athlon is not the same as 3.4GHz on a Core 2 Duo or Quad.

If you're comparing your score with other Athlon64 X2 @ 3.4GHz + 3870x2 + 2GB RAM + XP systems and you're finding it low, maybe the break up of the 3D Mark score will help explain it.

If you're comparing your score with other systems that are running different CPUs at or around 3.4GHz, then it could well be apples and oranges.

Check your SM 2.0 and 3.0 scores against Tom's first to see how much there is to gain. If they are within 10% of those scores, then CPU may not have much to do with it. More than that it could very well the CPU holding things back (a little or a lot).

I'm getting 14200 marks with 2 x 8800GT512s SLIed, but the CPU score is 2676. It's the same CPU as above with the GTS320 that scored 2614. Only 62 CPU points more, but over 4470 3D marks more.

While not Athlon/3870x2, hope this helps anyway.

 

[bmadd]

SM2 5584
SM3 7464
CPU 2439
13036Marks

thats my score the best i can with an athlon x2

specs are different then sig. this is what i used Nicocat

6000+ @ 3402Ghz (14x243) HTT @ 1215mhz @ 1.55V
Ram @ 972 MHZ 4-4-4-15-18 2.26V (4x1gig)
3870x2 @ 910 core/1100 Memory with CCC8.2
Vista x64

 

[dev1se]

ya cant beat quad core at the minute in 3DMark06... streets ahead of the rest.

 

[seabreeze]

^ for sure.

 

[crusoe74]

My Q6600 @ stock with a 8800GTS 512 get 's just under 12k. Most benchies have this set up at about 13.5K but all benchies have o'ced the Q6600 to 3.0Ghz+

 

[jay2tall]

My Q6600 @ stock with a 8800GTS 512 get 's just under 12k. Most benchies have this set up at about 13.5K but all benchies have o'ced the Q6600 to 3.0Ghz+

Interesting because my poor little e6400 OCed(3.2GHz) with a 8800GTS 512 gets 13.2K so really the fact that its a Quad doesn't make a big dif, I'd chalk the dif up to the L2 cache size more than anything.

A better CPU will obviously give you a better CPU score in 3DM06 as well as opening up your video card a little more and give better frames. A better CPU is going to open up the road a little more for that video card to deliver. Everyone will say "OHHH that CPU will bottleneck your GPU". In a sense it will but it basically just hinders it's potential a bit. The faster CPU you have the less it will be restricted. Once you get up to the higher end CPU's that gain levels off and are marginal.

Besides, who really cares what you get on 3DMark anyways. I rather be getting great frames on a game than anything. So what if you have a renegade class 1000HP hot rod that gets 8 seconds in the 1/4 mile. I bet my Honda Accord performs better on the streets than that thing ever well. Sorry for the odd analogy.

 

[crusoe74]

Oh, and that score is with Vista

 

[Reynod]

NVidia and Intel will both tell you a different story on this ... several articles around.

GPU vs CPU ... lets focus on Gaming scores tho.

Interesting if someone throws a few benchies up here so we can see.

 

[firebird]

3dmark06 on my system:

e6850 @3.0 (333x9) with 3870x2 stock = 14,400
e6850 @3.777 (420x9) with 3870 stock = 16,699

 

[jay2tall]

Why do we need 3Dmark to look good anyways? you can't play and beat 3Dmark? I can score 200 in 3dMark as long as my COD4 stays around 60FPS I'm happy as a clam.

 

[firebird]

3dMark has it uses. If I want to reproduce a standardized benchmark in order to gage system performance, than I use either 3DMark or Crysis benchmark. I can tell what effect the changes I've made to my system have on overall performance.

I agree that a benchmark like 3DMark06 shouldn't be used to judge how well my PC will play The Witcher or COD4. The point shouldn't be for bragging rights, just a tool for analyzing performance. Maybe if we had a benchmark that combined specific runs through the top 5 most popular games...

 

[jay2tall]

Maybe if we had a benchmark that combined specific runs through the top 5 most popular games...

I WOULD FREAKING LOVE THAT!

I agree with the rest as well. I used 3Dmark to see if what i am doing is making a difference. I know with my video card my score went up like 200 points but gained anywhere from 1-5FPS on certain games.

 

[babybudha]

with my q6600 and a x2 i got 15k

at stock? or at what speed? (in regards to the Q6600)