The Performance summary charts

Here is the summary chart of 25 games and 1 synthetic test of our latest benchmarking suite using the latest drivers with the GTX 780 Ti in the first column, the 290X in the second, and the GTX Titan in the third. The highest settings are always chosen and it is DX11 or DX12 with the settings at ultra or maxed and they are identically high or ultra across all platforms.

The benches are run at 1920×1080, 2560×1440, and at 3840×2160.  All results, except for Firestrike, show average framerates and higher is always better. We also show minimum frame rates where they make sense on the same line as the averages, but using slightly smaller regular unbolded italicized font to differentiate them.780tiVs290xVsTitan-FIXWe can see the performance of the GTX 780 Ti is a bit lower than the 290X for many of the newer games, although they do trade blows in some games, and in a few, the GTX 780 Ti is faster.  We also note the Titan lags well behind except for a few situations where the 6GB of vRAM makes a difference at higher resolutions.

Now we go back in time and highlight our 2013 original benchmarks for the launch of the GTX 780 Ti.

2013 benchmarks – GTX 780 Ti/3GB vs. 290X/4GB vs. the Titan/6GB

Here are the benches from our original ABT GTX 780 Ti Launch evaluation chart that was published on November 7, 2013.  In this chart, the stock GTX 780 Ti results are shown in the first column, the 290X Uber OC (1030MHz) is in the second, the overclocked 780 Ti is shown in the 3rd benchmark column, and the GTX Titan performance results are shown in the last column.

Launch-780-Ti

The newly released GTX 780 Ti was generally faster than the brand new R9 290X, and significantly faster than the Titan.

2013 vs. 2016 benchmarks – GTX 780 Ti/3GB vs. 290X/4GB vs. the Titan/6GB

Now we test 13 of the above games with the same 3 cards using the latest drivers compared with the 2013 drivers as in the original launch chart above.

780tiVs290xVsTitan-launch-FHere is another way to look at the same chart – comparing the same 3 cards at the same clockspeeds and settings using 2013 drivers as on the launch chart directly against the 2016 drivers:

Even though we are comparing 2013 with 2016 using a slightly upgraded platform, we generally see small improvements in 2016 over 2013 for the GTX 780 Ti.  What is surprising is that the Titan may appear neglected, but not the GTX 780 Ti. Titan’s performance has regressed in a few games.  We also generally see larger performance improvements for the 290X than the GTXes, with a big improvement for the 290X in Metro Last Light.  Unfortunately for this game, it just means that AMD has optimized for the benchmark.  We used to test it with PhysX On, but now there is no performance difference between On or Off in the benchmark.  In actually playing the game, the 290X still runs at about 35fps at 2560×1600 as it always has, and PhysX cannot be set in-game or the framerates would still tank completely.  We also see that a recent driver or patch broke 290X performance in the Secret World.

However, generally, both the 780 Ti and the R9 290X have had improvements since 2013, even for the old games.  What we have seen is that AMD has continued to optimize for newer games for the 290X as it reuses the same architecture for the 390X.

There is no doubt that the R9 290X with its 4GB of vRAM, is a generally faster GPU for today’s games than the 3GB vRAM-equipped GTX 780 Ti although they are still competing in the same class. There are only a few of the games that run poorly on the GTX 780 Ti because of its 3GB of vRAM compared with either 4GB or 6GB, and this is only at the highest resolution.  In these cases, tweaking a few settings downward generally make the newest games run decently.

The Big Picture 2016

Here is the Big Picture for 2016:

Let’s head for our conclusion.