X.Org GSoC 2010 With KMS, Cairo, Etc. What Was Done?

[phoronix]

Phoronix: X.Org GSoC 2010 With KMS, Cairo, Etc. What Was Done?

Two days ago there were some ATI R300 GLSL compiler optimizations made to the open-source Radeon driver stack by Tom Stellar as part of his Google Summer of Code project as a student developer to improve the R300 GLSL compiler support. This code is still living in a branch outside of Mesa, but some progress has been made. Though the R300 GLSL compiler optimization work was just one of five projects that were part of the 2010 Google Summer of Code for X.Org. Here's a few words on the other student projects...

http://www.phoronix.com/vr.php?view=ODc3OA

[agd5f]

FWIW, Tom has committed quite a few r300 compiler improvements since he started. The most recent code you mentioned is just his most recent experimental improvements which will be committed when he feels they are ready. The article implies that his changes gsoc haven't been merged, when that can't be farther from the truth:
http://cgit.freedesktop.org/mesa/mes...hor&q=stellard

[HokTar]

Wow, this _is_ a long list. For some reason it never *felt* like that for me so now I'm stunned!

Nice work, hands down.

[MostAwesomeDude]

Tom's code was merged prior to the end of his summer. He did a fantastic job, and his patches continue to come in. He's definitely a regular developer with commit access to Mesa.

Matt's KMS driver is sitting in my kernel tree. He and I were going to take winter break as an opportunity to hack on more KMS drivers and hopefully create some code suitable for merging into mainline.

[Michael]

Tom's code was merged prior to the end of his summer. He did a fantastic job, and his patches continue to come in. He's definitely a regular developer with commit access to Mesa.

Matt's KMS driver is sitting in my kernel tree. He and I were going to take winter break as an opportunity to hack on more KMS drivers and hopefully create some code suitable for merging into mainline.

Thanks for the notice, will update the article.

[mattst88]

There, however, appears to be no recent activity on this KMS driver or in producing the KMS driver documentation (we haven't found any yet) nor does the DRM Glint driver appear ready for any upstream integration in the Linux kernel.

Thanks for the notice, will update the article.

I don't get it. Why didn't you just email us with questions? I'd have been happy to explain that I've started graduate school and haven't had nearly enough time to do anything outside of homework but that my driver and documentation are still on my list of things to do.

I've seen other people ask similar questions. Clearly everyone is available via email, and almost definitely they'd rather (even be happy to) answer some questions than see things written about their work like "What Was Done? Sadly, Not Too Much."

Moreover, getting information straight from the source(s) makes for better articles.

Anyway, here's the status:

It works for my local Permedia 3 (VX1 PCI) setup at a hardcoded 1024x768 resolution, but is untested on really anything else.

The TODO list is:

• I2C/DDC for querying supported modes
• Support for at least Permedia 2, 2V, and 4
• Support for DVI, DFP, and OpenLDI for the SGI 1600SW LCD
• Hardware cursor
• a real memory manager
• Hardware accelerated routines to replace the cfb helpers

From there, maybe it can go upstream. Maybe I'll need to do some work to the DDX and libdrm.

[ianmacfarlane]

I don't get it. Why didn't you just email us with questions? ...

+12 to what he said - this is really poor journalism Michael.

[Henri]

I don't get it. Why didn't you just email us with questions? I'd have been happy to explain that I've started graduate school and haven't had nearly enough time to do anything outside of homework but that my driver and documentation are still on my list of things to do.

I've seen other people ask similar questions. Clearly everyone is available via email, and almost definitely they'd rather (even be happy to) answer some questions than see things written about their work like "What Was Done? Sadly, Not Too Much."

Moreover, getting information straight from the source(s) makes for better articles.

Indeed. But it has probably been said often enough already.

[russofris]

We refer to this phenomenon as "news entertainment", much like we refer to WWE wrestling as "sports entertainment". It does serve to interest readers, who are then free to research and reach their own conclusions.

[marek]

The article really looked like there had been nearly nothing done.