Hi
There's something I don't understand - why was the CPU overclocked in ALL the tests? I would like to know the usage of a normal CPU (something between 2 and 3 GHz?) while playing HD videos, not that of a 4 GHz beast.
@d2kx
The test shows todays Linux performance and the AMD GPUs have lostOf course you can boot Win to get different results, that's not the point.
Hi
There's something I don't understand - why was the CPU overclocked in ALL the tests? I would like to know the usage of a normal CPU (something between 2 and 3 GHz?) while playing HD videos, not that of a 4 GHz beast.
I also take issue with the CPU overclocking, but since all tests were done with the same speeds it's not such a big deal.
Last edited by Melcar; 06-26-2008 at 06:22 PM.
These cards are great cards, I think within the next 3 months it should be the card to get since the drivers will mature. Although when crossfire comes out soon and say its nearly as effective as Windows, it might be a great solution. Although my expectations of crossfire performance on Linux isn't very high at the moment but we'll see.
Yes, for games, but HD content playback should show something else - high usage on a normal CPU because there's no GPU support enabled in the drivers, yet. The overclocked CPU, in this case, invalidates the test, as there is no logic in using so much (unrealistic) computational power just to show that 'There is no UVD2 (Unified Video Decoder 2) support on Linux at this time'. They should have used something at 1.8 - 2 Ghz. and say 'look, the CPU usage jumps to 50 - 60% while playing this movie (average CPU) because there's not hardware acceleration'.
im not really sure what you expect catalyst drivers to do about h264 either.. its not like they can magically speed up H264 decoding, you would need support for it in ffmpeg, or another h264 decoder.
and theres no way in hell the ffmpeg project will include support for that stuff unless it were to be open...
Is there a phoronix test suite profile I can use to compare against my system? I've seen past articles where you provide the profile parameter that needs to be entered for comparison.