Ashes of the Singularity benchmark test with GTX 970 & 355.82

hello guys i managed to test dx 12 with Ashes of the Singularity
with driver version 355.82 and here is the results and as expected GPU performance on dx12 is worse than dx 11 :bang:

CPU test

with DX 11


with DX 12


Full system test

DX 11


DX 12

 

http://www.guru3d.com/news-story/nv...12-benchmark-to-disable-certain-settings.html

looks when one game uses a resources tier 3 or asynchronous shaders the result is horrible to nvidia... so pascal should support all this resources... but in maxwell forget it...

 

I have the benchmark but no DX 12 driver yet......

 

If the game and drivers support is good, there no reason for a decrease in performance, even with lack of AS supported . Dx12 "should" be faster .

Give it sometime for them to make adjustments , it is going to be slow IMO as far as 100% support .

 

I thought the whole point of DX12 is to not need drivers getting in the way...

 

doesn't appear to be using much of the gpu nor cpu.

 

As it seems nvidia is food for today and hungry for tomorrow, if nvidia can't show a real world proof of dx12 advantage on their cards i will really regret buying my gtx 980, i was going to buy a r9 290X
(back in february) but i got this 980 at the same price thanks to a friend of mine, i tought was a good deal but now seeing that benchmark and seeing how 290X surpasses 980ti in that dx12 benchmark, i'm really having my doubts now maybe i should have bought that 290X or waited for the next months to buy a r9 390X.

 

This is a hardware problem. 10 years of continuous driver support won't do jack for you :roll:

 

Thats in DX11, as for DX12 OSD isnt working so we cant see how is the CPU/GPU usage.

Its really a bummer that first DX12 benchamark had not give us anything they promised us.

 

Still waiting for a true DX12 benchmark ....

 

No its not . The whole premise of Dx12 is to get low level access and features of "said" card .
It more up to Dev now to get full support of HW whether its Nvidia or AMD .
What Dx12 will show IMO, is drastic shifts depending on how good Dev's do on there game .

One thing is true and has been said many times as far as efficiency , from Dx11 to Dx12 . AMD will see big gains , but that is just there driver in Dx11 .

 

Yeah, alot of people seem to think the driver is no longer important.

I didn't know that we had no way of monitoring DX12, that is far from ideal when trying to look for software/hardware problems.

 

and from where did you download it?
I want to try it too :nerd:

 

You can see the usage on the G19 keyboard LCD if you have set it up in Afterburner, by the way.

My Titan X GPU is used at 98-99% in the DX12 benchmark with full details at 1920x1080 ("Crazy+4x MSAA + 4xSGSSAA forced via Nvidia Control Panel), but my CPU is only being used 20-40% per core.
The performance is slightly worse on DX12 compared to DX11. This Nitrous engine just won't be a good show for Nvidia, and AMD will likely shine in it (my AMD-based laptop does very well in DX12 although DX11 is really bad), although I do expect some improvements for Nvidia cards until it's ready for release, which is still a while away.

It just is what it is. DX12 will require game devs to optimize their game better for various hardware, and we'll see who's really good at that. Some will work better for AMD cards, some better for Nvidia, some might even work better on Intel, depending on what the games will be optimized for.

 

I'm pretty sure that DX12 will eventually run at least as good as DX11 on GF. There are no reasons why it would be slower.

 

Too early for dx12 to even care tbh.

 

http://www.guru3d.com/news-story/nvidia-will-fully-implement-async-compute-via-driver-support.html

maybe this save the maxwell´s performance in dx12...

but the best solution looks wait by pascal with tier 3 and supported asynchronous shaders by hardware

 

Ashes hardly even uses async compute at all so if you're referring to Maxwell's performance in this benchmark then the reason why it's slower in DX12 than in DX11 lies elsewhere.

 

Well I found it,

this is only gpu test (full system test), seems Im around on par with 970GTX, gpu was @ stock boost 1150mhz, cpu @ 4.6Ghz.


DX11 @ 355.85Quadro



DX12 @ 355.85Quadro



Although I had a bit higher quality temporal AA, dx11 @ultra, dx12 @ high instead of default medium.

 

wrong...

http://www.guru3d.com/news-story/nv...12-benchmark-to-disable-certain-settings.html