DX12 / Vulkan on a FX 8350 and Fury Nano [PCGamesHardware / PCGH]

[Bacon1]

http://www.pcgameshardware.de/DirectX-12-Software-255525/Specials/Direct-X-11-Vulkan-1218975/

 

[tamz_msc]

So depending on the implementation, it is basically a hit or miss, but more importantly, if done right, then it is great for the vast majority of Steam users who game at less than 1080p.

This is something most review sites would never tell you.

 

[Bacon1]

So depending on the implementation, it is basically a hit or miss, but more importantly, if done right, then it is great for the vast majority of Steam users who game at less than 1080p.

This is something most review sites would never tell you.

Yeah wish they would have shown 1080 as well, surprised they skipped and went 720p / 1440p. Would be nice for a pentium or i3 test as well to see how those differ. Hopefully some other sites do similar testing. I know computerbase used to do i7 / FX 83?? testing with their dx12 articles which was nice. Sadly most sites just test with an i7 @ 4.5+

 

[Concillian]

So depending on the implementation, it is basically a hit or miss, but more importantly, if done right, then it is great for the vast majority of Steam users who game at less than 1080p.

This is something most review sites would never tell you.

Vulkan and DX12 are basically ways of optimizing CPU usage in graphics intensive applications.

This means that a typical power machine with an i7 and maybe even overclocked will see little benefit, since they are already not very constrained by CPU performance, but the benefit will increase on machines with a lesser CPU.

Reviews of graphics cards want to show the differences between graphics cards... so they try to minimize limitations in the rest of the system by using top end CPUs, fast RAM, etc... This amplifies any differences from the graphics cards, which is often what people want to see when they go to a graphics card review. I think people who are familiar with DX12 and Vulkan and what they do are not surprised to see a minimal benefit on most reviews (which typically use a 4 or 6 core i7), but a larger benefit on a low spec CPU. You are correct that reviews do not make this kind of thing obvious, but I don't think it's intentional... It's not an easy thing to convey that there's an interaction between the CPU and GPU and resolution. People want to think one dimensionally (GPU vs. GPU) rather than multi-dimensionally. The reality is that the system as a whole has to be considered.

In general, with a decent CPU, low resolution is more CPU limited (but usually at a minimum FPS number that is well over a threshold where you can tell a difference) and high resolution is more GPU limited. Since DX12 and Vulkan are essentially the same thing as using a better CPU, you see the benefit most at lower resolutions and on low spec CPUs.

 

[bystander36]

Vulkan and DX12 are basically ways of optimizing CPU usage in graphics intensive applications.

This means that a typical power machine with an i7 and maybe even overclocked will see little benefit, since they are already not very constrained by CPU performance, but the benefit will increase on machines with a lesser CPU.

Reviews of graphics cards want to show the differences between graphics cards... so they try to minimize limitations in the rest of the system by using top end CPUs, fast RAM, etc... This amplifies any differences from the graphics cards, which is often what people want to see when they go to a graphics card review. I think people who are familiar with DX12 and Vulkan and what they do are not surprised to see a minimal benefit on most reviews (which typically use a 4 or 6 core i7), but a larger benefit on a low spec CPU. You are correct that reviews do not make this kind of thing obvious, but I don't think it's intentional... It's not an easy thing to convey that there's an interaction between the CPU and GPU and resolution. People want to think one dimensionally (GPU vs. GPU) rather than multi-dimensionally. The reality is that the system as a whole has to be considered.

In general, with a decent CPU, low resolution is more CPU limited (but usually at a minimum FPS number that is well over a threshold where you can tell a difference) and high resolution is more GPU limited. Since DX12 and Vulkan are essentially the same thing as using a better CPU, you see the benefit most at lower resolutions and on low spec CPUs.

I don't know if this is just a glitch with Hitman, but here, it was found that DX12 with a slower CPU gave a boost in FPS, but had stuttering issues. There was no issue in DX11. This may be an issue with DX12, or maybe it's just that dev's aren't using it properly yet.

http://www.tomshardware.com/reviews/presentmon-performance-directx-opengl-vulkan,4740-3.html

 

[AtenRa]

I don't know if this is just a glitch with Hitman, but here, it was found that DX12 with a slower CPU gave a boost in FPS, but had stuttering issues. There was no issue in DX11. This may be an issue with DX12, or maybe it's just that dev's aren't using it properly yet.

http://www.tomshardware.com/reviews/presentmon-performance-directx-opengl-vulkan,4740-3.html

In both graphs above DX-12 has less Unevenness (Less and lower gray line in X axes).

 

[bystander36]

In both graphs above DX-12 has less Unevenness (Less and lower gray line in X axes).

Those gray lines that go above the dotted lines are spikes and causing stuttering. If you read the article, they also notice it while gaming in DX12 on the AMD RX 480 (not in Nvidia's case). This is one game, so this might not be a trend, so further testing would be needed.

Note: the gray blocks below the dotted line shows slight unevenness. The spikes above the dotted line, shows noticeable stutters. Stuttering is worse than slight unevenness. As noted by the author, below the dotted line is pretty unnoticeable, above is very noticeable.

Edit: I've noticed quite a few other reviews which notice a similar trend in regards to DX12 having a lot more stuttering.