- User Since
- May 6 2015, 7:43 PM (251 w, 1 d)
Apr 3 2018
I committed this patch with canPeel() check before setting PeelCount in Hexagon Target. Thank you very much Max for your Review.
Apr 2 2018
Thank you very much Max for your review. It is very much appreciated.
Thank you very much Max for your review. I tried to zero UP.PeelCount before every return from computeUnrollCount that stands before computePeelCount and it causes test/Transforms/LoopUnroll/pr33437.ll to fail. This test expects to peel a loop first and than unrolling it.
Mar 30 2018
Hello Max, This assert(canPeel(L)) causes several tests to fail when I set PeelCount in Hexagon target. Apparently there is another path that is taken that calls peelLoop() and does not go through computePeelCount(). So it looks like the second check for canPeel in needed. Here is a test case: opt -march=hexagon -O3 -S < test.ll
- test.ll ---
target datalayout = "e-m:e-p:32:32:32-a:0-n16:32-i64:64:64-i32:32:32-i16:16:16-i1:8:8-f32:32:32-f64:64:64-v32:32:32-v64:64:64-v512:512:512-v1024:1024:1024-v2048:2048:2048"
target triple = "hexagon"
Mar 29 2018
Mar 28 2018
Thanks Max. Please note that this patch also fixes a bug as well. Without this patch, the value passed by the flag -unroll-peel-count was ignored.
Mar 27 2018
I addressed the comments from Max and Florian.