Tweet
For clarity, I wasn't saying that the shader compiler difference accounted for all of the performance delta, just that it accounted for a big enough chunk that the difference between drivers didn't var as much between apps on Kaveri as on a dGPU, where you either have more shader cores or such sucky performance nobody looks at performance deltas.
-
Test signature -
Well, i am not sure about perfect scalable performance delta between dGPU and Kaveri
as both use TTM (designed because of dGPUs) down to the same mapped BOs in userspace mesa.
Comment
-
How's the stutter? That's a lot more important real-world question than simple framerate.
Reshuffling everything in memory makes a lot worse gaming experience than just drawing 30 FPS, and in that article about the new memory manager, I remember the new memory management code was less stuttery than the old one-or Catalyst.Comment
-
Stutter is supportedOriginally posted by A Laggy Grunt View Post
Too me both frame rate and less stutter are valuable, well both can be awfull sometimes.Comment
-
Maybe because with an APU you're more than likely memory constrained (iGPU<->DDR3), but not bus constrained (iGPU<->CPU), and you're almost guaranteed to be running at the same clock between radeon and Catalyst, whereas with a dGPU you may or may not be (especially if the dGPU is throttling due to temperature). Seems like APU's iGPUs don't get as hot as a dGPU, besides--whether that's because the RAM is elsewhere or because the APU has better cooling or a lower TDP, I don't know.Originally posted by bridgman View PostInteresting... the performance offset between all-open and Catalyst seems more consistent between applications than it has been on dGPU.
Not sure what that means yet but it seems like useful informationComment
-
Hi Michael
Could you run The 7870k against this test
Would be interesting to see how the 7870k does against discrete cards and Intel igp graphics.
Thanks
Comment
Comment