[PC World] 290X As Fast As Titan X in DX12 Enabled 3DMark?
本帖最後由 Stiggy930 於 2015-3-27 11:19 編輯
AMD drivers has been holding back the GPU for ages!!! Dx 12 and Vulkan comes to rescue.
After further processing the graphs, nVidia's driver seem to be as bad on draw call processing on dx 11.
The test works by tasking the GPU to draw something on the screen. This instruction goes through the API, whether it is DX11, DX12, or AMD's Mantle. The less efficient the API is in handling these "draw calls" from the CPU to the GPU, the fewer objects can be drawn on the screen. 3DMark rapidly ramps up the draw calls and objects until the frame rate drops under 30 frames per second (fps).
This translate to:
R290X: 13474728 draw calls @ 30 fps
Titan X: 13419183 draw calls @ 30 fps
Included Iris Pro 5200 to shown Dx 12 improvement.
Iris Pro: 2126150 draw calls @ 30 fps
IMPORTANT NOTE: The above figure is number of draw call processed. NOT fps, NOT 3D mark final score.
下面有張係Intel iris pro ...
danny9428 發表於 2015-3-27 10:23
CPU is the same, except for Iris Pro test.
This is testing the API and driver and GPU. How many draw calls they can "push" to the GPU and GPU able to process.
According to test, if you reduce the number of draw calls on each GPU, the fps will go up. So if you reduce the draw call on 290X to Titan X level, the fps of 290X should go up by 1 to 2 fps. Hence they think 290x may be faster than Titan X in dx 12 level. BUT, it may be nVidia's driver not being optimized for dx 12 yet. AMD has a head start with mantle.
This is an assumption based upon the test method. Does not indicate the (final) performance and the actual outcome.
3DMark and Microsoft point out that the new feature test is not a tool to compare GPUs but an easy way to gauge a single PC's performance and API efficiency. Don't use it to compare PC Y with PC X, nor as a GPU test: This is all about how your particular PC configuration performs when running DX11, and how that same PC configuration performs when running DX12.
I have stated pretty much the same thing in my first post. This does not indicate the (final) performance of any GPUs. This is more like an experiment and should be treated as such in any situation.
However, my previous statement is a assumption based on their test methods. Their test method is to see how many draw calls each configuration can take and still give 30 fps. So if you reduce the draw calls, in theory, the GPU process less and fps goes up. I think this is a fair and logical assumption. So if you reduce the draw call of X to the level of Y, you can have a sense on the performance differences.