On Wed, Dec 21, 2016 at 8:26 PM, Timothy Arceri <[email protected]> wrote: > There is a single regression in loop unrolling which is: > > loops HURT: shaders/orbital_explorer.shader_test GS SIMD8: 0 -> 1 > > However the loop is huge so it seems reasonable not to unroll it. It's > surprising that GLSL IR does unroll it.
Heh, that's funny. Orbital Explorer's GS doesn't have a loop in the GLSL source. I believe it appears in the IR because we implement "switch" with a loop (see commit 73dd50acf6d244979c2a657906aa56d3ac60d550). If that is the case, I am surprised that we're able to unroll it since it has multiple breaks. That would be interesting to investigate, but Orbital Explorer is certainly not a case we should optimize for. :) _______________________________________________ mesa-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/mesa-dev
