NEW PC Setup

[OS-Wiz]

Yes, when you're done OCing, simply click the Default button to return to stock clocks. Also remember to set the nvidia control panel settings back to stock for best eye-candy when gaming.

You can't run with a small window with the basic version of 3dm11. Use Heaven 3.0 instead: Download Heaven DX11 Benchmark | Unigine: real-time 3D engine (game, simulation, visualization and VR)
Initially just use the settings in the pic below while you're learning to OC your card.

The youtube video you found is OK.

I've also copied in a bench I ran on my 680s in SLI so you can see the settings I used. But every card is different and my settings might not work for you. Remember to manually set the fan speed at max when you are OCing. ALWAYS go in small steps while you're getting use to OCing the card and watch temps closely.

 

[rio1]

Ok so I test ran the Precision test for 12min and very stable at the setting on the pic. I tried several other higher settings and anything over GPU clock offset of 140MHZ it will crash in the middle of test. The memory clock offset doesn't seem to do much so I left it at 400MHZ. Now im gone try to do some benchmarks.

 

[OS-Wiz]

Looking good so far ...

 

[rio1]

3dmark vantage

Score
P40923 3DMarks

Graphics Score
54910

CPU Score
23197




NVIDIA GeForce GTX 690 video card benchmark result - Intel Core i5-3570K Processor,ASUSTeK COMPUTER INC. P8Z77-V PRO score: P40923 3DMarks

 

[OS-Wiz]

3dmark vantage

Score
P40923 3DMarks

Graphics Score
54910

CPU Score
23197




NVIDIA GeForce GTX 690 video card benchmark result - Intel Core i5-3570K Processor,ASUSTeK COMPUTER INC. P8Z77-V PRO score: P40923 3DMarks

Nice score, well done.
Was PPU off, PhysX on CPU, not GPU?
CPU overclocked?
For us to help we need to see all your measurements, like this, with PPU off and on:

 

[Hameister]

I've been following along. Interesting to see how the 680/690 cards OC, vs. the 580s.

I'm also realizing that the CPU contributes far more to the final graphics score than I previously thought. It becomes obvious when I compare my Vantage scores to rio's. Really startling.

 

[rio1]

I have the free version of Vantage so I cant change any options. I checked the options and PPu is off.PhysX is on Autoselect. CPU is overclocked @ 4313Mhz .
Ham, do you now believe that my CPU is the bottleneck?

 

[Hameister]

...Ham, do you now believe that my CPU is the bottleneck?....

Yes, I now believe that is the case. I must admit I am very surprised.

While I was certainly aware of the part the CPU plays, in this, and other benchmarks, I always believed that the major role was that of the GPU, with any modern state of the art CPU, having an effect of about 10% to 15%.

I now see that the CPU's role, is significantly more.

I've attached 3 screenshots below.

The GTX 460 was disabled. Only the 580 cards were used, and only because you are also running dual GPUs. The characteristics of the 580 cards remained the same throughout all the runs, they were never touched. Only the operating characteristics of the CPU was changed for each run.

For the first run, you can see that I severely crippled the CPU, by completely disabling Hyper-Threading, as well as shutting down 2 cores. This left me with a basic 4 core, threadless processor, as close to yours as I could get. I noticed that you were over clocked to 4.3GHz, so I dropped mine down to 4.2GHz. My score was P36510 vs. your P40923. The 4,000 to 5,000 point variance is about what I'd expect.

Next, I opened all 6 cores, but kept Hyper-Threading shut down. The still crippled i7 3930k, allowed a score of P42701. That huge increase in score was a surprise!

Finally, I enabled Hyper-Threading, which now allowed the CPU to operate to it's full design capabilities, which produced a score of P45697! A full 9,187 point variance from the original P36510, simply by changing the characteristics of the CPU! Pretty amazing stuff.

This proved to me that the CPU plays a far greater role in these benchmarks than I originally thought, and yes, you are correct, about the i5 3570k bottlenecking your score.

However, putting all the benchmarking aside, and getting into the real world of every day use, these numbers don't matter very much, because that 690 will rip through anything you throw at it for sometime to come.

 

[rio1]

Yeah I'm surprised as well. Thought these benchmarks score more on the graphics side, but it looks like it 50/50. Well, I'm very happy with the difference from my old gts 450 to this monster!

 

[Hameister]

Yeah I'm surprised as well. Thought these benchmarks score more on the graphics side, but it looks like it 50/50. Well, I'm very happy with the difference from my old gts 450 to this monster!

Absolutely! Benchmarks are fine for testing hardware, to insure it's working properly. Beyond that, people put far too much emphasis on benchmarks.

My tired ol' 580s are still slicing through any software available today. I don't expect them to start showing their age until Crysis 3 is released next year. By then, it will be time for a pair of GTX 780 cards. I can get about 3 years out of a pair of cards. As soon as the first game comes along that I cannot max out, it's time for new cards.

Then my son gets one, and my grandson gets the other, ha!