https://www.shacknews.com/article/83924/nvidia-competes-with-mantle-with-its-latest-drivers
Hold on, so if I download these new drivers what games will get performance increases? It seems to imply (or outright say) that these performance increases should affect all DX9, 10, and 11 games to some degree or another. I guess 3% is kind of cool, and they go higher, but it seems that most games aren’t getting too much performance increases. Being that it’s free and just because of the drivers I guess it’s pretty cool still.
See this article
Sounds like games other than the ones tested might see improvements as well, and that the improvements will vary a bit based on hardware.
I wonder if this will have the biggest benefit with a weak CPU, similar to Mantle?
Might test the driver on a PC with a GTX460 + i7 950, using Thief and Bioshock Infinite as benchmarks (and will look forward to using it when I get my GTX770 back… it might have WHQL certification by then).
Didn’t realize they were beta drivers until I went to install them. Oh well. I did a few benchmarks with Thief, Sleeping Dogs, and Hitman Absolution.
Thief Old Drivers:
Min: 19.9
Max: 68.6
Avg: 50.3
Thief New Drivers:
Min: 36.5
Max: 69.7
Avg: 51.3
Sleeping Dogs Old Drivers:
Min: 46.1
Max: 63.8
Avg: 61.2
Sleeping Dogs New Drivers:
Min: 52.3
Max: 63.8
Avg: 61.3
HMA Old Drivers:
Min: 40.78
Max: 60.31
Avg: 48.27
HMA New Drivers:
Min: 42.80
Max: 60.31
Avg: 49.41
Hitman Absolution seems to be stupid but the other two pretty much only show a raising of the minimum framerate and I don’t know if those are likely to vary anyway depending on the run. If they’re not then that’s great I guess. Got a 6 fps increase on my minimum framerate in Sleeping Dogs and a 16 fps increase on my minimum framerate in Thief. These benchmarks are pretty much worse case scenarios too. Neither of those games run that poorly when actually playing the game. Shame I couldn’t do Arkham Origins and Metro LL but I didn’t have them installed.
Edit: Holy shit. Playing Sleeping Dogs right now to see how the framerate is while playing. Turned the frame limiter/vsync off to see how high and stuff it is. It’s rock solidly locked at 63 fps without budging at all. I remember my framerate often going into the 40s and 50s while driving but I’ve played for like 10 minutes with walking around and shooting and the game just stays at 63 fps. Now that’s something that these drivers actually do that’s noticeable.
That’s very interesting.
Why would it lock at 63 exactly?
Well if you look at my benchmark that’s about the max framerate I got. So while the drivers might not have increased my maximum framerate they help keep it there. I never looked at my framerate while playing that wasn’t locked so I don’t know if it was also 63 there but I remember the framerate dropping while driving and that didn’t happen the entire time I played just now.
The point of these new low-level drivers/APIs is to increase performance stability rather than dramatically increase maximum framerate.
Actually Mantle is advertising itself as increasing overall performance so AMD is at least. Nvidia’s benchmarks were more focused on how much the maximum/average framerate increased as well.
I installed these new 337 beta drivers on my Geforce 210 and it won’t boot. I can boot in safe mode but windows always hangs on the welcome screen.
Someone I know has a GeForce GT440 that refuses to boot if the drivers are updated to anything more recent than 285.62.
Had to do a system restore (or possibly repair install) to get it working again.
Kind of want to try out Thiaf and some other games to see if I have anymore cool things happen like with Sleeping Dogs.
Edit: Tried Crysis for the hell of it and I swear my performance is better. I still get the occasional drop but it’s not as bad as I remember it, and I just beat it a few weeks ago. For example my framerate doesn’t drop nearly as much as I remember it doing when zooming in with the binocs. I don’t have any scopes to try right now though, as they also gave me performance issues. And I was using a custom config file with performance saving stuff too and still had these issues, but now I’m running entirely maxed and haven’t experienced what I remember happening. Wish I had benchmarked it before, or at least remembered better. This is great.
RESULTS (Nvidia GeForce GTX460 1GB, Intel Core i7 950, 6GB DDR3 RAM @ 1600MHz):
Bioshock Infinite (1920x1080, Ultra DX11, normal DOF, 3 runs)
335.23
Min 8.47
Avg 36.71
Max 67.35
337.50
Min 8.04
Avg 36.90
Max 62.56
Metro 2033 (1920x1080, AAA, Very High, Advanced DOF 3 runs)
335.23
Min 4.93
Avg 18.00
Max 61.48
337.50
Min 2.34
Avg 18.00
Max 68.21
Metro 2033 (1920x1080, 4xMSAA, Very High, Advanced DOF 3 runs)
335.23
Min 3.56
Avg 14.00
Max 61.33
337.50
Min 4.29
Avg 15.33
Max 70.89
THIEF (1920x1080, max settings, no AA, 64-bit)
335.23
Min 2.65
Avg 22.65
Max 36.45
337.50
Min 3.10
Avg 23.05
Max 35.45
NOTE: not enough VRAM, need >1GB for “normal” textures and higher at 1920x1080
THIEF (1920x1080, low settings, no AA, 64-bit)
335.23
Min 23.70
Avg 35.35
Max 51.65
337.50
Min 22.30
Avg 36.05
Max 52.85
Minimal to no improvement on this system, with these games. Probably due to the GPU being the bottleneck…
I noticed after I updated that my Skyrim hung around 29-35 FPS, and thats with almost max graphics and about 30 or so mods
Is that lower or higher than normal?
higher methinks
Well was just wondering because 30 fps seems low in Skyrim. Guess the mods he has could make it run worse though. And it depends on his specs.
I hope AMD brings that kind of optimization to their Directx drivers.