The Impact Of Switching To Linux 4.3 + Mesa-11.1/LLVM-3.8 On Ubuntu 15.10

[phoronix]

Phoronix: The Impact Of Switching To Linux 4.3 + Mesa-11.1/LLVM-3.8 On Ubuntu 15.10

Yesterday I posted some performance results of a Radeon R9 290 tested on Ubuntu 15.04 and Ubuntu 15.10 out-of-the-box. In this article are some numbers when upgrading the Ubuntu 15.10 installation to use the non-standard Linux 4.3 Git kernel as well as Mesa 11.1-devel Git that's built against LLVM 3.8 SVN for the newest open-source AMD Linux experience...

http://www.phoronix.com/scan.php?page=news_item&px=Mesa-LLVM-Up-Ubuntu-15.10

[microcode]

Phoronix: The Impact Of Switching To Linux 4.3 + Mesa-11.1/LLVM-3.8 On Ubuntu 15.10

Yesterday I posted some performance results of a Radeon R9 290 tested on Ubuntu 15.04 and Ubuntu 15.10 out-of-the-box. In this article are some numbers when upgrading the Ubuntu 15.10 installation to use the non-standard Linux 4.3 Git kernel as well as Mesa 11.1-devel Git that's built against LLVM 3.8 SVN for the newest open-source AMD Linux experience...

http://www.phoronix.com/scan.php?pag...p-Ubuntu-15.10

It's good to see the new PTS in action on Phoronix, good work.

[ElectricPrism]

In simple concise words - please summarize - what is the conclusion based on this new data about R9 290 with the latest software.

[Veto]

Why the remarkable regressions?

[microcode]

Why the remarkable regressions?

I guess they just don't test performance before merging into master.
It'd be nice if they had a CI setup which would build mesa then run a representative selection of phoronix tests; and before it gets to master.

[Namenlos]

I guess the performance regression of Valley is because the now possible tessellation costs a lot of processing power. No clue about GPU Test though.

[dungeon]

Maybe things regressed because i switched to Catalyst and stopped bisecting mesa I can recommend at first checking if Michael results are reproducible, maybe it is something on his own.

Maybe things regressed because Ubuntu losses one mesa maintainer too, so yeah man changing can be a difference too

And i of course i didn't check what PMD do with his PPA

I guess the performance regression of Valley is because the now possible tessellation costs a lot of processing power. No clue about GPU Test though.

I don't think Valley use tessellation, only Heaven does.

[marek]

Is there a way to run a test and get the result on the output, e.g.:
- run PTS / gputest / pixmark volplosion at 1920x1080
- it prints one number: average fps

I'd like to do this in an automated manner, i.e. without any user interaction.

Thanks.

[Michael]

Is there a way to run a test and get the result on the output, e.g.:
- run PTS / gputest / pixmark volplosion at 1920x1080
- it prints one number: average fps

I'd like to do this in an automated manner, i.e. without any user interaction.

Thanks.

Hi Marek,

Look at the batch-setup/batch-run options (or internal-run) that will get you having it fully-automated. http://www.phoronix-test-suite.com/d...test-suite.pdf

To get just the single result, there's many ways to go about doing that, if you're just doing it on a single system and don't care about anything else, probably just easiest sending the output through sed or similar command.

[Michael]

Also look at build-suite to select GpuTest Pixmark 1920x1080 (and any other tests) you want and then that suite's name is what you'd pass to phoronix-test-suite batch-run xxx. Feel free to let me know if any questions or issues. Tom might also be able to answer some of your questions as I think he had some sort of deployment going for his OpenCL tests.