it isnt developed fomr ground for DX12 or any low level API, it received a patch for DX12, it has nvidia biased (code and features) it doesnt really use of better CPU bound scenarios advantages on DX12.and doesnt supportt features like async shaders or explcit multiadapter
if you are against w10 and DX12 benefit and you want an alternative go with Khronos/AMD´s Vulkan API
It boosted 8120 and phenom 960t unlocked to 6 cores performance by like 10 fps when i tested under dx12 vs dx11. It absolutely works in rise of the tomb raider.
Well, that doesn't give me much confidence based on how terrible PC games have been lately.Originally Posted by BradleyW
It will, by a good margin as long as the deleopers put the work in to build fresh DX12 applications from the ground up with key targeted optimization techniques.
you could expect performance in crease in mgpus setups. because then there is a CPU bound scenario, aswell high end GPUs running on 1080 with fhih framerateOriginally Posted by corky dorkelson
So, pretty clear that DX12 is primarily helping those with weaker CPUs. I was hoping for better performance on higher end systems. I hope future games can harness the power a bit better.
you can be sure DICE will release a really well DX12 based gameOriginally Posted by corky dorkelson
Well, that doesn't give me much confidence based on how terrible PC games have been lately.
have you compared hitman to AotS? because computerbase shown that on hitman AMD CPUs on DX12 are closer to Intel probably because the game is less CPU demandingOriginally Posted by OneB1t
if you have AMD card it also helps with performance if application use AS
this can be 0-20% boost (for example in AOTS)
for my system i get about 0-50% from DX12 (as my cpu is bottleneck in alot of DX11 games) in CPU limited situations
with DX12 FX-8xxx will catch up with i5-6xxx which is very good for 3 year old CPU
but it seemsin this testing the HD 7970 results with Underclocked CPUs were mislabelled as GPU boundOriginally Posted by SoloCamo
It is offering better performance on higher end systems, it's further alleviating a cpu bottleneck and putting more load on the gpu. Issue is, we don't really have the gpu power at this point (in a single gpu configuration for reliability of testing's sake) to really see how much more it's helping. When I do cpu bound testing I drop it down way below 1080p.