[PATCH] D24683: [DAGCombine] Generalize build_vector -> vector_shuffle combine for more than 2 inputs

Dorit Nuzman via llvm-commits llvm-commits at lists.llvm.org
Wed Oct 5 23:55:44 PDT 2016

Previous message: [PATCH] D25302: [AMDGPU] Promote uniform (i1, i16] operations to i32
Next message: [PATCH] D24683: [DAGCombine] Generalize build_vector -> vector_shuffle combine for more than 2 inputs
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

dorit added a comment.

+1  
Just FYI - this patch gives 15% improvement in one of the Geekbench workloads on Haswell (compared to enabling interleaving for X86 without this patch) :-)


https://reviews.llvm.org/D24683

Previous message: [PATCH] D25302: [AMDGPU] Promote uniform (i1, i16] operations to i32
Next message: [PATCH] D24683: [DAGCombine] Generalize build_vector -> vector_shuffle combine for more than 2 inputs
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

More information about the llvm-commits mailing list