NVidia and AMD driver performance review [HardOCP]

[Carfax83]

https://www.pcper.com/reviews/Graphics-Cards/PresentMon-Frame-Time-Performance-Data-DX12-UWP-Games

Didn't he use the release version of PresentMon? No idea if that makes any difference. Also like I said previously, PresentMon has a ton of data points and probably needs to be averaged out somewhat to make it more intelligible.

 

[Carfax83]

I think I figured out what was causing the stuttering in Deus Ex MD under DX12. I had the Steam overlay turned on. Apparently, it can cause issues with DX12. I remember this problem being patched out, but it's possible that being on the Creators preview build has caused it to resurface again..

 

[Bacon1]

Im real life AMD cards lose performance under DX12.

I just tested this again tonight and yes, I do still gain from DX12 over DX11.

There was also some big (3 or so) stutters when I first started playing in DX11 other than that the run went fine, just lower FPS overall. No stutters when using DX12.

This is from doing a loop through Prague, all in game. Tested 3 times per load. It doesn't seem like the registry key for async compute does anything, as I didn't notice any difference as you can see (0.1 fps difference)

3440x1440 VH / Custom settings

DX11: Avg 49.7 fps
DX12: Avg 53.8 fps (8% gain)

 

[Bacon1]

Didn't he use the release version of PresentMon? No idea if that makes any difference. Also like I said previously, PresentMon has a ton of data points and probably needs to be averaged out somewhat to make it more intelligible.

Thats what I've been trying to say forever. You get spikey data because its having multiple datapoints per second. For example for my testing:

1st: 3219 over 60 seconds -> 53.65 (53.7) entries per second
4th: 2983 over 60 seconds -> 49.71 (49.7) entries per second

Those also match up with the average FPS. So you get one entry per frame as expected.

There is dx11 vs dx12 with a 11 frame running average to help smooth them out.

 

[littleg]

nvm

 

[SirDinadan]

What a lively debate!
We are just about to start benchmarking the singleplayer part of DeusEx:MD with FCAT. Hopefully, we can add some interesting findings to this conversation, stay tuned!

 

[Bacon1]

What a lively debate!
We are just about to start benchmarking the singleplayer part of DeusEx:MD with FCAT. Hopefully, we can add some interesting findings to this conversation, stay tuned!

Do you work for a site?

 

[SirDinadan]

Do you work for a site?

I'm the main guy behind BoostClock.

 

[moonbogg]

Rise of the Tomb raider DX12 sends performance through the damn roof. I made it to the soviet installation and FPS tanked to under 40fps at times with DX11. Tried DX12 and FPS is pegged at 100fps (monitor limit) almost 100% of the time with dips into the 90s. CPU bottlenecking under DX11 is pretty insane in this game, but it doesn't make itself very visible until those special maps where FPS tanks for everyone under DX11. With a single GPU it might not be very noticeable either.

 

[Bacon1]

FPS tanked to under 40fps at times with DX11. Tried DX12 and FPS is pegged at 100fps (monitor limit) almost 100% of the time with dips into the 90s

Thats with 2x 980 Ti right? Very nice!

I wish more sites would test ROTTR / DXMD and others in DX11 vs DX12 with SLI/CFX. Also an updated mixed Polaris + Pascal update for AOTS mixed mGPU would be awesome to see. Most of the tests for these were all back in beta.

 

[moonbogg]

Yeah with two 980ti. I ignored DX12 for a while because all the sites told me it performed worse. What they failed to mention was that with two cards, the GPU headroom is massive and well able to take advantage of the CPU benefits of DX12. In my case it made a titanic difference. Seriously unexpected. Also, that's with the cards at stock settings. They outperform the cards in DX11 when they were overclocked because afterburner causes the game to crash in DX12, so I have to run them stock but the performance is still miles ahead despite that.

 

[Bacon1]

afterburner causes the game to crash in DX12

Try disabling the overlay, or reducing the OC by a bit (might not be stable OC when it tries to use other parts of the hw in DX12?) But yeah, the coverage for DX11 vs DX12 has been terribly lacking. Using an OC'd 6700k and then saying there isn't a benefit @ 1080p? :\

Most people would probably see a benefit of DX12 over DX11 if they tried it themselves, not to mention the games seem to run smoother with better frame pacing.

 

[moonbogg]

I think I did get it to work once by disabling the overlay, but the next time it didn't work. Honestly I'm not bothered by it since the game is capped out at 100 most of the time which is my monitor's max refresh anyway. Looking forward to playing more tonight. Game is actually fun and interesting.

 

[Dave2150]

Vast majority of Gamers doesnt have a 4.7GHz Core i7 Skylake

Im sure if you have a slower CPU like Core i3/5 or Phenom II/FX you will gain a lot with DX-12 over DX-11 in the vast majority of games.

For reference,

Core 2 Quad 9450 @ 3.2GHz + RX 480 8GB GTR Black

Deus Ex Mankind Divided 1080p medium preset (build in benchmark)

DX-11
Average = 39,6fps
Minimum = 27,1fps

DX-12
Average = 45,5fps
Minimum =32,7fps

Who the hell still games on a Core2? Less than 1% of the market?

DX12 so far is a complete failure. Vulkan is the only API I have hopes for, mainly due to Doom's superb implementation.

 

[tviceman]

I'm the main guy behind BoostClock.

You do some great testing. It's a shame you're using an AMD FX processor instead of an i7 processor when testing GPU frame times. When comparing cards it's important to eliminate any bottleneck whatsoever.

 

[Bacon1]

You do some great testing. It's a shame you're using an AMD FX processor instead of an i7 processor when testing GPU frame times. When comparing cards it's important to eliminate any bottleneck whatsoever.

I'd say its good though, because it shows more "real world" performance. We have countless i7 benchmarks but hardly any with i3, i5 or FX.

 

[Bacon1]

DX12 so far is a complete failure. Vulkan is the only API I have hopes for, mainly due to Doom's superb implementation.

You realize that DX12 and Vulkan are very similar right? DX12 is anything but a failure. Even when its added onto a DX11 based engine it does the same or better, especially on older CPUs

 

[AtenRa]

Who the hell still games on a Core2? Less than 1% of the market?

DX12 so far is a complete failure. Vulkan is the only API I have hopes for, mainly due to Doom's superb implementation.

I have no idea how many still use Core 2 Quads but on Steam Hardware Survey 20% of Intel users have 2.3 to 2.69GHz CPUs. Since Core i3 2100 SandyBridge was 3.1GHz, those are either Mobile CPUs or older Core 2 Duo/Quads. So for those Gamers DX-12 is a Godsend API that will increase their Gaming performance and allow them to play latest games that would be unplayable with DX-11 due to low CPU performance creating a lot of stuttering and high frame times.

 

[Dribble]

I have no idea how many still use Core 2 Quads but on Steam Hardware Survey 20% of Intel users have 2.3 to 2.69GHz CPUs. Since Core i3 2100 SandyBridge was 3.1GHz, those are either Mobile CPUs or older Core 2 Duo/Quads. So for those Gamers DX-12 is a Godsend API that will increase their Gaming performance and allow them to play latest games that would be unplayable with DX-11 due to low CPU performance creating a lot of stuttering and high frame times.

PC's with ancient cpu's are highly likely to be running windows XP or at best 7, and won't have DX12 capable graphics anyway.

 

[nathanddrews]

PC's with ancient cpu's are highly likely to be running windows XP or at best 7, and won't have DX12 capable graphics anyway.

Don't forget how many users around the globe were "tricked" into installing the free Windows 10 upgrade.

I've got 3 or 4 C2Q machines kicking around and I upgraded them all to W10 while it was free (technically still is free), but none of them have DX12/Vulkan compatible GPUs. When I find a good deal, I'll get a couple 460/1050 GPUs to give them one last upgrade. They still play MOBA/F2P/emulators well, so I don't see a reason to get rid of them yet. Tossing in low power Polaris/Pascal GPU will not only make them better for more modern games, but also modern emulators that use Vulkan/DX12, and provide HDMI 2.0 and HEVC/VP9 decode support. It's actually pretty amazing what a new GPU can do for an old platform, even if the CPU bottlenecks the GPU. Even if they don't play all games on ultra 60fps, they still have tremendous value.

 

[Vaporizer]

Again with the fake internal benches!

What you are posting is FAKE! internal benches! Under with fake results there is a gain in DX12 for AMD systems. Im real life AMD cards lose performance under DX12.

So stop with the worthless debunked fake internal benches of games! Its crap, its fake, its untrue, its a lie, its fraud!

Stop with the AMD worship! AMD cards lose DX12 performance under DX12 in actual gameplay in Deux Ex, Total War according to reputable online sources like computerbase and hardocp. Certainly tons more reputable than some random internet bench in fake internal benchmarks!

Also don't go into the DX12 debate with me, this is not about that and I'm not going to allow for trolling like that. This is about Nvidia and AMD performance under DX11/12 and I'm telling you that internal fake benches are not a proper way to measure performance.

No one is denying there is zero DX12 benefit overall, I'm denying that Deus EX and Total War in majority of cases lose performance under AMD with DX12, just like Nvidia does.

Nvidia also has significantly better performance under DX11, so there is no point testing in DX12 if they lose performance, no single person on earth would use an API that lowers their performance to play at.

Yeah this is the new Zeitgeist. Everything that does not fit to own agenda is qualified as fake. Next step is presenting alternative facts. I look forward to it...

 

[moonbogg]

DX12 so far is a complete failure.

Identical settings and everything. Only difference is DX11 vs DX12. The benchmarks are much better, but the in game performance is even more amazing with 2X the FPS. That's a doubling of framerate in this game dude.

DX11

DX12

DX11 (48fps)

DX12 (97fps)

 

[Krteq]

Is it SLi working in DX11?

 

[moonbogg]

Is it SLi working in DX11?

Yep, its working perfectly. The CPU bottleneck is so severe under DX11 that the GPU's are barely working at half capacity. This actually angers me really. This tells me that we are playing games that run like crap when they don't have to. I doubt DX11 is fully to blame here. CPU utilization in the game under DX11 is wretched and I see no excuse for it. I'm glad they fixed it with DX12, but there are plenty of DX11 games that use the CPU perfectly well and I see no reason for such bad coding in this game. Most of the time we think we need to upgrade our hardware, but I think we'd be absolutely shocked to find out just how much performance is left on the table due to lazy and inefficient game development.

 

[Bacon1]

Most of the time we think we need to upgrade our hardware, but I think we'd be absolutely shocked to find out just how much performance is left on the table due to lazy and inefficient game development.

Ding ding ding!

That's why I supported AMD with their Mantle initiative and glad to see we are finally getting DX12 / Vulkan games. Will be a few years before the benefits are truly there with better designed engines but at least it's progressing! Its amazing how well consoles run compared to many PC ports. Its laughable we have some indie games that run so bad as well. I'm sorry but your 2d side scroller shouldn't be hammering a $400+ video card