I'm pretty sure that AMD's APUs will do a lot better in Dx12 than Intel's iGPUs since AMD's APUs have more Dx12 HW accelerated instructions than Intel's iGPUs. Intel will have to do more stuff on CPU than AMD. On external GPUs the story changes obviously. Well... check this post: http://forums.guru3d.com/showpost.php?p=5132768&postcount=12

Of course GCN based APU's will benefit a lot from DX12 because of the feature level and extra optional function support.

Well... check this post: http://forums.guru3d.com/showpost.php?p=5132768&postcount=12

That doesn't say much. They're just doing what they enjoy doing.

So, the i7-4790K boosts to 4.4 GHz? So, how come my i5-3570K+R9-280X under Windows 10 at 4.3 GHZ, has 20 percent higher draw call performance compared to the i7-4790K boosting to 4.4 GHz (not forgetting it has higher IPC anyway)? The only reasonable answer would be that the test wasn't done using 15.7.1.

That's what I said originally 🙂

Not surprised either. Who actually thought that Dx11 performance would change from Win8.1 to Win10? The real difference comes with Dx12 vs Dx11, where Dx12 hardware will shine against older Dx11 hardware, and where CPU performance will be a bit less important than with Dx11. That's what AMD is talking about their "magic" stick, their underpowered CPUs will be able to shine better in games thanks to Dx12 (even more the APUs).

Yep, exactly my thoughts. AMD's processors are very weak in single-threaded, DX12 should change the landscape since their octa-cores are not that bad in multithreaded.

Yep, exactly my thoughts. AMD's processors are very weak in single-threaded, DX12 should change the landscape since their octa-cores are not that bad in multithreaded.

The reason AMD CPU's will clearly benefit is because IPC matters in CPU overhead, the quicker the IPC the more work the GPU can do, if we talk about draw calls. Since Intel have a higher IPC, CPU overhead is still a problem but the GPU can perform better, that's why lately you see AMD CPU's being far behind in games, FPS wise. When DX12 is here, CPU bound will be shifted over to GPU bound. That means more innovation will be needed for GPU's to equal it out and at the same time developers will be able to put CPU's to more use because it will no longer cause bottlenecks in the game. GTA 5 is a good example when you up extended viewing scaling etc, FPS is hit a lot because the CPU has to do much more work and that bottlenecks the driver thread. Project CARS is another example. AMD GPU's were being bottlenecked. That's why it performed much better on Windows 10 because of the earlier CPU overhead optimizations. Intel will benefit too as games don't just require a GPU but a CPU too. As I said above, developers will be able to use the CPU more for more complex things like physics and AI. CPU performance will matter there.

On windows 8.1 i was using 15.7, on windows 10 15.7.1. Also regarding the "lower than 3570K" performance. I locked my CPU to 3.1Ghz. So the 4790K is running at 3.1Ghz, and not 4Ghz. As the 4Ghz, reported is the stock clock without boost. at 3.1Ghz its only pulling max 48W. 🙂 My apologies for not clarifying the CPU clocks. I see no reason to be running my CPU and full when i only need 80%.. maybe in a year or so i'll need the rest of the speed.

Are you positive the clocks were locked to 3.1GHz with Windows 10? It still doesn't make sense why there's such a big difference.

Doesn't make sense. 700k with Windows 8.1 and 15.7 Nearly 1.2m with Windows 10 and 15.7 Both with CPU clock at 3.1GHz. Hmm. Off to ponder.

Maybe this is a dumb request/statement but it'd be kind of sweet to see Win 7 up in those charts too.

Was going to say the same thing. I upgraded from Win 7 to 10 for obvious reasons. Would like to see a comparison of the two.