Nvidia’s GTX 760 arrives to take direct aim at the HD 7950
Performance summary charts & graphs
Here are the summary charts of 25 games and 5 synthetic tests. The highest settings are always chosen and it is DX11 when there is a choice; DX10 is picked above DX9, and the settings are ultra or maxed. Specific settings are listed on the Main Performance charts. The benches are run at 1920×1080 and 2560×1600 with separate charts to make comparisons easier, as well as dividing games up into easy-to-read charts by their DX pathway and by resolution.
Here is the main summary chart that we call “The Big Picture” as it includes all 15 configurations that we tested and with all 30 benchmarks:
All results, except for Vantage, Firestrike and 3DMark11, show average framerates and higher is always better. In-game settings are fully maxed out and they are identically high or ultra across all platforms.
[**BattleForge had flashing textures with the HD 7970 that was not present in the other cards, including the HD 6990]Before we post the graphs, let’s break up the charts a bit. First up, lets compare the GTX 760 to the GTX 660 Ti, the GTX 670, GTX 680, GTX 780 as well as to the Fermi GTXes 560 Ti, 570 and 580.
In every single case, the GTX 760 is faster than the GTX 660 Ti and overclocked it even generally outperforms the GTX 670 that it replaces not to mention that it dominates over the Fermi-based last generation GTX 560 Ti which also launched at $249, as well as the GTX 570 and the GTX 580.
It is interesting to see the ($249 launch price) GTX 560 Ti compared with the $249 GTX 760 as there is a huge improvement that Nvidia claims.
Finally, let’s see the GTX 760 at stock and at our maximum +110MHz core/+400MHz memory overclocked clocks, compared with AMD’s HD 7970 which is about ten percent faster than the Boost edition of the HD 7950
The $250 GTX 760 beats the $350 HD 7970 in almost one third of our 25 games. Looking at all of the benchmarks, overclocking the GTX 760 as many of Nvidia’s board partners will do, nearly shifts the performance ratio in its favor as overclocking just +110MHz makes the GTX 760 the near-equal of the HD 7970. As Nvidia’s fourth single-GPU card in their line-up, even coming in cheaper than the $300 AMD HD 7950 Boost Edition may also prove disruptive.
Let’s look at the individual charts. As usual, we begin with the synthetics.
Futuremark & Heaven synthetic tests
3DMark 2013 Firestrike is Futuremark’s latest DX11-only benchmark, 3D Mark 11 is also DX11 and Vantage is DX10. Unfortunately, scores are completely meaningless when they are presented in this way but they do offer supporting data to accompany our game benches. Vantage provides some interesting data on the GTX 760. It seems that the memory bandwidth made a solid improvement in bandwidth over the GTX 660 Ti. Here are all of the synthetics compared to all of our benchmarked cards:
Synthetic tests are interesting but they are not necessarily indicative of real world gaming performance. In all three cases, the GTX760 “wins” over the mainstream cards.
Next up, let’s look at DX9/DX10 games.
DX9/DX10 Games
We test fairly demanding DX9 games: Serious Sam 3, BFE and Alan Wake at completely maxed out settings. We have also added Borderlands 2 with high PhysX, and the Witcher with maxed out settings including “ubersampling”. Aliens: Colonial Marines is our latest benchmark. The original Crysis is the only DX10 game that we are benching.
Now on to DX11 games.
DX11 Games
Most of our testing emphasizes DX11 games. Since the charts get too long, we break them up into charts of 6 or 7 games each.
Now the newer DX11 games
Now let’s check out our newest DX11 games:
We ran Sniper Elite V2 with the maximum in-game settings including 2.25X SuperSampling which is incredibly demanding but looks awesome compared to running with it off.
PhysX
AMD has no answer to PhysX except perhaps TressFX which involves hair, although they have promised physics support for years. Nvidia has developed their own proprietary PhysX which is used in some games to create extra immersion with particle, fluid and other physics calculations that previously usually required a separate video card just dedicated to geometry and PhysX calculations.
Due to enhancements in Kepler’s SMX architecture as well as higher clocks, it is now easier to run PhysX on high with the GTX 770 than with the older cards such as the Fermi GTX 580. Unfortunately, the way the official benchmark in Borderlands 2 is set up does not allow for an accurate comparison of PhysX on versus PhysX off.
For a proper PhysX comparison, we turn to Metro: Last Light and we can see the performance hit each of our configurations take.
As you can see the GTX 760 is significantly faster than the Fermi cards and especially the GTX 560 Ti which becomes a slideshow at 2560×1600. Of course, the Radeon cannot run PhysX so the CPU has this task which really makes it unplayable in the game although it appears optimized enough to run the benchmark.
Metro: Last Light makes great use of PhyxX and it is a shame to play it without it. In both cases, turning on PhysX, although affecting the frame rate, it is enough to play the game with fully maxed out details and AA (but not SSAA) with our GTX 760 at 1920×1080, especially when overclocked. Playing with SSAA enabled in Metro 2033: Last Light looks spectacular but is reserved for only for SLI at 2560×1600, and for the very fastest single-GPU cards at 1920×1080.
The HD 7970 has improved smoothness in running the Metro: Last Light benchmark and game, although playing with PhysX disabled is a must outside of just running the benchmark.
Let’s head for our conclusion.
