Those are the words of truth. However, even with Catalyst Linux desktop is quite usable.
Yeah, words of the truth. Difference is - if you use nvidia, you use linux 100% for everything. I like this truth much better. If Iīd prefer windows, I would use it 100% for everything - this is possible with amd & nvidia, but reverse possible only on nvidia. Less choice with AMD.
Wooohooo, to me it's all progressing nicely, it ain't perfect but there's a limit to what can be done by just a few people.
The floss drivers are working way better for me than catalyst, though I don't care about or use opengl.
The performance of normal gtk or qt is still lower than what I'd think is possible, and seems slower than nicely implemented software rendering (unless you're trying to do highres alpha blending).
Originally Posted by Qaridarium
its really sad michael@phoronix do not understand computer benchmarking
synthetic benchmarks like unigine heaven are invalid!
benchmarks like Quake3 are invalid because NO ONE USE MORE THAN 200FPS!
200fps only because stereoscopic view and 100fps per eye.
Micheael really need to chance his benchmarks in a way that they are usefull.
this means Unigine-Oil-rush instead of unigine heaven
this means doom3 instead of quake3
this means all results over 200fps market as invalid result. (no joke its not valid!)
more compute tests like openCL-Bitcoin a raytracing realtime engine openCL test.
more wine tests with engines like TA-spring.
and wine test with ARMA2-Free edition-
The benchmarks are a bit crappy but really there aren't too many options.
The main gripe I have is the lack of any analysis, or practical conclusions.
For example there wasn't (or I missed it, which admitedly is possible) a test of the benefits of HT on compilation time, and in my experience it seems pretty significant.
When something weird happens the guys at anandtech, techreport, etc. usually try to undertstand what's going on, explain it or at least give some guesses.
Then again on linux you get very weird results just about all the time, and this is probably one of the greater weaknesses of linux (though at least we have lots of ways to fix stuff).
I'm guessing the lack of analysis is due to lack of time, so we should probably try doing some benchmarks on our own and submit them so Michael has a bit of breathing room.
All in all he is doing a lot of work, which I really really appreciate.