On Wed, Mar 18, 2015 at 4:27 PM, Connor Abbott <[email protected]> wrote: > On Wed, Mar 18, 2015 at 2:58 PM, Matt Turner <[email protected]> wrote: >> On Thu, Mar 12, 2015 at 9:46 PM, Matt Turner <[email protected]> wrote: >>> Here are some ideas I think might be reasonable GSoC ideas. >>> >>> - GLSL linking in NIR >>> - Would allow us to stop doing optimizations and other expensive >>> things on GLSL IR >> >> Ian said this should wait until everything is using NIR so that we >> don't have two linkers. :) > > Yeah, and it also won't make sense to do it until we can get the NIR > optimizations up to par with the GLSL ones so that we can disable all > the GLSL ones after linking and not get a regression. This would > involve porting all the GLSL optimizations to NIR, which now that I > think about it actually seems like a quite doable GSoC project -- it's > a pretty well-defined task and there are lots of little optimizations > to translate so it's easy to make progress, although porting all of > them over might be too much for one summer.
Yeah, grabbing a handful of optimizations and porting them seems like a good GSoC project to me. There's also the NIR checklist which people could grab stuff off of. --Jason _______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
