Results: Crysis 3
We’re forced to dial down Crysis 3’s System Spec setting from Very High to High to maintain the average frame rates we want at 1920x1080. Nevertheless, six of our seven tested cards fall within a 10 FPS range.
One particularly demanding passage knocks the GeForce GTX 760 down under 30 FPS. However, we remain happy with a >45 FPS average.
The Radeon HD 7950 doesn’t dip below 30 FPS, but the same card with Boost does, so this is likely an artifact of our manual benchmark and run-to-run variation. Even the GeForce GTX 770, which appears to avoid the dip one-third of the way into our test, drops lower than most other cards toward the end.
Curiously, Nvidia’s cards get hit by significant variance between consecutive frames. This is particularly strange in light of the results we gathered in our GeForce GTX 770 review, which showed the 770’s 99th percentile outcome at 3.5 ms. Suddenly, under the 320.39 driver, the 95th percentile is up to 10.42 ms. Because this benchmark requires manual input, it’s possible that a slightly different path explains the difference. But the GeForce GTX 760 and 670 also exhibit higher-than-expected numbers, while the 660 and 660 Ti are under 2 ms.
Re-read the conclusion in question below. He doesn't say it is faster, he says this card will replace Don's recommendation for best $250 card and displace the 7950 Boost. ie. Don won't be recommending a $300 card that trades blows or barely beats a $250 card. If both were to end up $250, things change.
quote - "A quick reference to Best Graphics Cards For The Money: June 2013 shows that Don is currently recommending the Tahiti-based Radeon HD 7870 for $250. With almost certainty, the GeForce GTX 760 will take that honor next month, displacing the Radeon HD 7950 with Boost at $300 in the process."
test, given it does so well for the other CUDA tests, especially iRay and Blender?
Btw, I don't suppose you could include 580 SLI results for the game tests? ;)
Or do you have just the one 580?
My only gripe with the 760 is the misuse of a model number which allows one to
infer it should be quicker than older cards with 'lesser' names (660, etc.) when
infact it's often slower. I really wish NVIDIA would stop releasing products that
exhibit such enormous performance overlap. Given the evolutionary nature of
GPUs, and the time that has passed since the 600s launched, one might
reasonably expect a 760 to beat the 670 too, but it never does. To me, the
price drop is the only thing it has going for it. The endless meddling with shader
numbers, clocks, bus width, etc., creates an utter muddle of performance
response depending on the game. One really has to judge based on the
individual game rather than any general product description or spec summary.
I just hope Skyrim players with 660s don't upgrade on the assumption newer
model names mean better performance, but I expect some will.
Amazing performance at 250$. The 265bit memory interface does wonders for GK104.
Now I am wondering if there will even be a GTX760ti, while there is a large enough gap in the product stack, I have a feeling there is a chance there may not be a "ti" version.
Anyone know more?