Is my new video card being bottlenecked?

[protokiller]

I am getting 7950 marks in 3dmark 06. System specs are:

Pentium Dual core e2220@2.4GHz
Geforce 9600GSO (768mb)
2GBram
Vista

If I overclock the 9600GSO's core from 550-700mhz the marks only go up to the low 8000s

 

[Zecow]

Its your CPU

 

[yonef]

It is not the CPU!
how fast do you think 9600 is ?! This is prety much good result for that card;

 

[daskrabbe]

No, you're not being bottlenecked.
However, it is a very easy overclock to 3ghz and you'll most likely gain a few fps in addition to a generally faster computer.

 

[Zecow]

It is not the CPU!
how fast do you think 9600 is ?! This is prety much good result for that card;

Not fast enuff..
What I mean is... its the CPU bottleneck.

 

[protokiller]

The reason I question the CPU is because most people with 9600GSO's get in the 8000s on stock settings and get over 10,000 3d marks once they overclock the card.

 

[yipsl]

It's your CPU. It's one of the relatively low end Intel duals for the budget market.

I used to have an Athlon X2 4600+ with my 3870x2 and only got around 9400 or so in 3DMark06. When I upgraded to triple core 8750, I went up to around 12000 or so. The clock speed for both processors is 2.4 gigahertz, but the 8750's extra core and higher IPC made a difference.

Xbit Labs has a review of Core 2 based Pentiums compared to several Core 2 Duo's, the Athlon X2 6000+ and the 8750:

http://www.xbitlabs.com/articles/cpu/display/core2duo-e7300-pdc-e5200.html

 

[Zecow]

Upgrade to an e7200 or e8400 and OC it above 3Ghz. Then the next thing you know... GPU bottleneck.. lol

as mentioned before.. its the CPU

 

[stridervm]

It's fine actually. I would be more suprised if you got about 9000 points in 3DMark06, since 3DMark is primarily CPU bound. Due to the CPU test included in the final score.

 

[eklipz330]

really low end cpu, overclock it to 3.2ghz, and you should be on your way!

 

[yipsl]

How are people noobs? Everyone knows that 3DMark06 is CPU reliant, so I linked to a review of his CPU and others in the same market niche from Intel and AMD. That provides real world data that's not synthetic.

I should have mentioned that resolution affects CPU bottlenecks. At higher resolutions, the CPU can keep up and the GPU reaches it's full potential. The 9600GSO is a weak card compared to an 8800gt, but it's not all that bad for games. It's not like it's an 8400gs or 3650.

It has 96 shaders, better than the 9600gt's 64 but not as good as the 8800gt's 112:

http://www.guru3d.com/article/geforce-9600-gso-386-mb-review-point-of-view/2

These benchmarks aren't that bad:

http://www.guru3d.com/article/geforce-9600-gso-386-mb-review-point-of-view/8

It seems playable up to resolutions of 1600 x 1200. I'd say his Pentium e2220 is a weaker CPU than his card is a weaker budget gaming card.

Look at World in Conflict. Not too bad either:

http://www.guru3d.com/article/geforce-9600-gso-386-mb-review-point-of-view/10

50 fps at 1600 x 1200.

3DMark06 is 8746 in this round of test. Note that they used an X6300 Extreme Conroe core CPU, so we see the card can do better with a better CPU.

He can probably overclock to get better performance, or go to a higher resolution if he doesn't want to overclock his Allendale beyond the stock 2.4. Note that he didn't mention what resolution he games at, though 3DMark06's stock resolution is 1280 x 1024.

 

[NewLCD123]

If he wants high scores in benchmarks, of course he needs to overclock! Otherwise he should be happy with stock scores.

 

[yipsl]

Yes, but synthetic benchmarks aside, he's bound to be bottlenecked with that CPU in actual games. That's why I linked to real game results using the same card plus a much better CPU. Those he can compare to his in game results.

I mentioned in my post that the addition of extra cores makes a difference in the synthetic benchmark. No one benchmarks even a budget or mainstream card with a low end CPU, they all use high end duals or quads from Intel for their benchmarking platform. That was the point I was trying to make in showing the improvement with extra cores from my experience.

Maybe Futuremark's products should just be banned from discussion and people would just ask why they're getting such low framerates in Crysis or Call of Duty 4? That would please the purists who think that a CPU can't be a bottleneck in a synthetic benchmark.