dgrdsv
Member
Well we all have, at least for Nvidia.
Since no side can proof or state facts for the future, there isn't much left than waiting for the upcoming DX12 titles.
It depends on the use case, you could kill off round trips between the processors, saving additional latency and idle-times, you can build new mechanism living inside the GPU.
Intel gave away the performance numbers for their simple asteroid demo:
![]()
DICE and Confetti presented methods for triangle culling and a visibility buffer, one key foundation are the expanded capabilities coming with ExecuteIndirect.
Beginning with page 22:
http://frostbite-wp-prd.s3.amazonaws.com/wp-content/uploads/2016/03/29204330/GDC_2016_Compute.pdf
Also page 22:
http://www.conffx.com/Visibility_Buffer_GDCE.pdf
SM6 may also come to DX11.3/4 but the explicit resource managament, the command-list-generation, ExecuteIndirect and the new binding model are only available under DX12, as long I didn't get anything wrong:
https://msdn.microsoft.com/en-us/library/windows/desktop/dn859252(v=vs.85).aspx
I agree, the variance under AotS is too high for precise arguments but unfortunately there isn't much choice.
At least there is also Time Spy:
http://www.legitreviews.com/3dmark-time-spy-benchmark-dx12-async-compute-performance-tested_184260
10,6 % for the 380X Nitro, 12,1% with the RX 480 and 12,8% for Fury X.
I would be interested in cases where you could see smaller gains from Async Compute on the RX 480 but there are currently only two applications which allow comparisons between Async Compute on/off and are publicly tested.
There might be a third with Rise of the Tombraider since there is a Async Compute option in the config but I didn't find results for it.
Fortunately with Gears of War 4 there will be an upcoming title which also allows Async Compute on/off comparisons.
Yeah, you're right, bindless resources are D3D12 only feature, still it's mostly a CPU optimization as your example shows.
As for 480's async gains being lower - I don't really collect these cases but I've seen them a couple of times in benchmarks on 480's launch, you can usually spot them with 480 getting less performance gain from switching to DX12 when compared to 390 and Fury.