Re: [Mesa-dev] [PATCH] intel/fs: Optimize and simplify the copy propagation dataflow logic.

Hi, I got unexpected results with this, when testing it on BXT & SKL GT2. While performance in GpuTest Volplosion and GfxBench v4 Tessellation test improved slightly, performance in SynMark v7 CSDof and GpuTest v0.7 Piano dropped clearly. Piano dropped only on SKL, but there the drop was

Re: [Mesa-dev] [PATCH] intel/fs: Optimize and simplify the copy propagation dataflow logic.

Hi, On 20.12.2017 16:29, Eero Tamminen wrote: I got unexpected results with this, when testing it on BXT & SKL GT2. While performance in GpuTest Volplosion and GfxBench v4 Tessellation test improved slightly, performance in SynMark v7 CSDof and GpuTest v0.7 Piano dropped clearly. Piano

Re: [Mesa-dev] [PATCH] intel/fs: Optimize and simplify the copy propagation dataflow logic.

Matt Turner writes: > There's a comment near the top of the file describing the algorithm > and referencing Muchnick. I think with your changes both of those are > no longer accurate? I don't have a copy of Muchnick's book at hand right now, but the paragraph at the top of

Re: [Mesa-dev] [PATCH] intel/fs: Optimize and simplify the copy propagation dataflow logic.

On Tue, Dec 19, 2017 at 9:19 PM, Francisco Jerez wrote: There's a comment near the top of the file describing the algorithm and referencing Muchnick. I think with your changes both of those are no longer accurate? ___ mesa-dev

[Mesa-dev] [PATCH] intel/fs: Optimize and simplify the copy propagation dataflow logic.

Previously the dataflow propagation algorithm would calculate the ACP live-in and -out sets in a two-pass fixed-point algorithm. The first pass would update the live-out sets of all basic blocks of the program based on their live-in sets, while the second pass would update the live-in sets based