The new Nvidia GeForce 378.78 driver is out, and Nvidia is making some rather bold claims with this driver. DirectX 12 performance should have been increased big-time, this is the promised driver you guys have been hearing about. We have been testing with this driver for our GeForce GTX 1080 Ti review for a couple of days now and indeed noticed a perf increase for Hitman as well as Doom.
Hence in our review we updated all cards for these two benchmarks already in the GeForce GTX 1080 Ti review. Nvidia however makes the bold statement that average performance gains across all titles would be roughly 16%. NVIDIA states 33% on Tomb Raider; 16% on Hitman; Gears of War 4 sees a boost of 10%; Ashes of the Singularity makes do with 9%; and Tom Clancy’s The Division would get a 4% increase. Now Nvidia is comparing to the 1080 release driver (368.xx series from May last year), which we feel would not exactly be the right measure point. Let's compare to the previous WHQL driver 378.66.
| GeForce GTX 1080 @ WQHD | GeForce 378.66 | GeForce 378.78 | Difference % |
| Deus Ex : Mankind Divided | 69 | 70 | 1,43% |
| Hitman (2016) | 92 | 104 | 11,54% |
| Rise of the Tomb Raider | 93 | 94 | 1,06% |
| Gears of War 4 | 95 | 94 | -1,06% |
| Ashes of the Singularity | 86 | 88 | 2,27% |
| Sniper Elite 4 | 89 | 89 | 0,00% |
| Battlefield 1 | 95 | 96 | 1,04% |
| The Division | 72 | 75 | 4,00% |
| % difference = 100 - (x/y * 100) |
|||
We just grabbed these titles and put that to the test, the reality is that only Hitman makes a substantial enough difference in our testing. The Division indeed also sees a 4% increase in performance. Ashes also sees a bit of an increase, the rest remains within normal anomaly differences. We do have to state that we are using a fast clocked (8 cores @ 4.3 GHz) 16 thread processors (5960X) this might hinder the threading performance that DX12 optimizes so much, Have a peek at our results. We test at our standard quality settings used in all our graphics benchmark, we also test at the audience preferred resolution of 2560x1440 which you could argue a bit.
| 3Dmark Time Spy | Score | GPU | CPU |
| GeForce 378.66 WHQL | 7446 | 7181 | 9415 |
| GeForce 378.78 WHQL | 7521 | 7250 | 9550 |
3DMark Time Spy, testing the GeForce GTX 1080 in-between the two driver does yield into a better score. However the differential sticks are give or take 1% again, the limited perf increase might very well have to do with the fact that we use a spicy tweaked processor. That also always has been the reason that we use such hig-end platforms, as we do not want the processor to bottleneck the GPU. Hence we try to get rid of the system being CPU bound, which is exactly where DX12 normally tries to help out. Well, just sayin'
