[PATCH] D19894: AMDGPU: Fix not counting shader input registers

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Fri May 6 18:08:07 PDT 2016


arsenm added a comment.

I found a problem with the generated input registers. If you have a case like
(<3 x i32> inreg, i64 inreg, i32 inreg), it correctly selects s0, s1, s2 for the vector, s[4:5] for the i64, and the final i32 picks the s3 in the alignment gap between the vector and i64. I'm guessing that changing the order this way will break something


http://reviews.llvm.org/D19894





More information about the llvm-commits mailing list