ROUND 2 : 3DMark11
Finally a benchmark that utilizes GPU! We will see here whether FX bottlenecks or not while overclocked to 4.8 Ghz. The score to really look at is GPU score (as this directly relates to fps of the rendered scenes), but because the total score also heavily relies on GPU score (especially in the Extreme Preset) it is also a good measure.
3DMark11 Performance Preset:
AMD FX 8150 @ 4.8 Ghz
6990 OC @ 990/1500 Mhz
Compared Results (with several 6990 OC's) :
As you can see the graphics score of my OC'd 6990 does not fall systematically behind intel rigs with similar GPU OC's. My GPU Score of 12046 is a clear winner over the rest of the rigs tested., however with combined and physics scores also put into consideration FX falls behind with a total score of only 10318.
The most noted comparison is that with the i5 2500k at 5.35 Ghz with a 6990 @ 1000/1420. Although it manages to squeeze out slightly higher combines/physics score, it still seems to bottleneck in GPU scores. The only intel cpu coming close to FX GPU score is the 3960x.
It should be noted that the OC on the 6990 does play a role in GPU score, so take these results with a grain of salt. a 930 Mhz OC is still 7% below a 990 Mhz OC, but nevertheless we can determine that FX does not heavily bottleneck when it is overclocked to 4.8 Ghz. What about Extreme Preset?
3DMark11 Extreme Preset:
AMD FX 8150 @ 4.8 Ghz
6990 OC1 @ 880/1250Mhz
6990 OC2 @ 990/1500 Mhz
- - - OC1 - - - 6990 @ 880/1250Mhz
- - - OC2- - - 6990 @ 990/1500 Mhz
Comparison (from Hexus.net) :
- - - OC1/OC2 - - -
The most noteable comparison is between my stock 6990 @ 880/1250 paired with my AMD FX 8150 @ 4.8 Ghz, and their stock 6990 with the exact same clocks paired with a 980x @ 3.6 Ghz Turbo. The difference in score is mostly due to a difference in Graphics Score, as the 980x generally destroys the FX in physics and combined results even at stock.
This tells us alot about whether FX bottlenecks or not, and the answer seems to be NO, atleast when comparing to a 980x @ 3.6 Ghz. The difference in score is roughly 5%, where the FX is slightly favoured.
When my 6990 is pushed to its stable limits @ 990/1500 Mhz, my score jumps an additional 13%. That is, for a 12/20% (clock/memory) overclock on my 6990. It is clear that an AMD FX 8150 does not bottleneck on Extreme Preset.
Look here : http://amdfx.blogspot.com/2012/01/look-at-this.html for comparisons to 1100t @ 4.2 Ghz and i7 2600k Stock.
These are examples where their GPUS are being bottlenecked. My Stock OC1 (880/1250) Graphics score actually manages to beat a 6990 @ 950/1450 on an 1100t @ 4.2 Ghz. (meaning higher fps). Here is an example where OCing a 6990 will not result in much benefit. (ie. the bottleneck is around that CPU frequency)
It is also interesting to see that OC'd my 6990 is the clear winner against the 980x @ 3.6 Ghz 580 SLI @ stock in the Extreme Preset.
It should be noted that the drivers I used were Catalyst 12.1 beta drivers, and those used in the HEXUS test were 11.4. The difference in 3DMark 11 scores should be negligible however. Also the 3DMark11 version used for my Performance Preset Results is 1.03, while that of the Extreme Preset Results is 1.02.