AMD's Turks GPUs Work On Open-Source, Sort Of

[phoronix]

Phoronix: AMD's Turks GPUs Work On Open-Source, Sort Of

Back on Tuesday, AMD officially rolled out their "Turks" graphics processors with the launch of the Radeon HD 6570 and Radeon HD 6670 graphics cards. On Wednesday the Phoronix review of the Sapphire Radeon HD 6570 was published under Ubuntu Linux, but using the proprietary Catalyst driver. Open-source testing wasn't done at that time due to only having the graphics card since Monday. But do these new AMD Turks GPUs work with the open-source Linux driver stack, including Gallium3D?..

http://www.phoronix.com/vr.php?view=OTM2Mg

[Jonimus]

So have you also submitted a bug report on your issues, or are you hoping that all the devs read this blog and know exactly what all your log files and such after the crash were?

Reporting on things that are or are not working is nice, but actually doing things about it is another. And if you already have submitted a bug report a link to it so readers can check for progress and contribute if they are also experiencing it.

Its great that you are reporting on these things but if you are going through these tests take the time to inform the people who can actually improve things about it and do it through the proper channels. Blogs are not bug trackers and it seems this one is often treated as if it were. So next time a game crashes during testing with the FOSS drivers, do the dev's a favor and submit a bug through the proper channels so they can fix it.

That said its great that you are informing the general public of the current status of the FOSS drivers and I hope you continue to report on it.

[V!NCENT]

@Jonimus,
The is no 'hoping' in AMD FLOSS devs are reading this (they do).

I think they know what's up

Great progress on Compiz working out of the box (meaning at least textures and geometry). This was different for the HD4870x2 and the HD5770 I got (the former died on me shortly).

Great progress and thanks for the great work! ^^,

[bridgman]

@Jonimus,
The is no 'hoping' in AMD FLOSS devs are reading this (they do).

I think they know what's up

Last time I asked I got the impression that neither agd5f nor the other devs had seen this problem on their systems yet.

[agd5f]

I haven't run into this on my BTC boards, but I think the following patch should fix the issues:
http://git.kernel.org/?p=linux/kerne...5f5eaca56f5cd0

[agd5f]

FWIW, no one has yet filed a bug about this. I've just heard rumors of problems on phoronix.

[Jonimus]

FWIW, no one has yet filed a bug about this. I've just heard rumors of problems on phoronix.

This is my point exactly, whether the devs read the blog or not is irrelevant, the issues should still be submitted through the proper channels so that they cannot get lost and it is 100% certain that the devs know the issue exist and can fix it.

[Michael]

This is my point exactly, whether the devs read the blog or not is irrelevant, the issues should still be submitted through the proper channels so that they cannot get lost and it is 100% certain that the devs know the issue exist and can fix it.

I don't have enough time to do proper bug reports on every issue I come across.

[Jonimus]

I don't have enough time to do proper bug reports on every issue I come across.

Oh I have no doubt that time is an issue, but submitting reports for showstopping issues would save the devs a lot of headache and allow for them to be fixed much quicker. I'm sure there would be a way to have PTS save the log files from what ever benchmark you are running and then all it takes is a short description such as what is already contained in your blog post and then submitting it under the proper component. You are in a great position to be improving the state of mesa and the opensource world with your vast array of hardware and configurations you can test. All it would take is a little extra work to make a huge difference.

Like I said I think its great that you are keeping Linux users up to date on the state of Linux graphics and overall performance but it would be even better if you put a little effort into helping improve that state.

[Michael]

Oh I have no doubt that time is an issue, but submitting reports for showstopping issues would save the devs a lot of headache and allow for them to be fixed much quicker. I'm sure there would be a way to have PTS save the log files from what ever benchmark you are running and then all it takes is a short description such as what is already contained in your blog post and then submitting it under the proper component. You are in a great position to be improving the state of mesa and the opensource world with your vast array of hardware and configurations you can test. All it would take is a little extra work to make a huge difference.

Like I said I think its great that you are keeping Linux users up to date on the state of Linux graphics and overall performance but it would be even better if you put a little effort into helping improve that state.

PTS already saves logs and they are uploaded to OpenBenchmarking.org when the user allows it to do so.... e.g. http://openbenchmarking.org/system/1...%20Black%20P67 the multiple log files as an example. The problem though is when all of the tests crash and none of the tests successfully finish, no logs are saved since there's no results to save.