[llvm-dev] Vectorization width not correct using #pragma clang loop vectorize_width (original) (raw)
hameeza ahmed via llvm-dev llvm-dev at lists.llvm.org
Thu Sep 20 15:40:59 PDT 2018
- Previous message: [llvm-dev] Vectorization width not correct using #pragma clang loop vectorize_width
- Next message: [llvm-dev] Vectorization width not correct using #pragma clang loop vectorize_width
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Thank you.
I am working on a machine with greater vector widths. How to enable the emission of greater and different vector widths in loop codes through pragma ? and automatically.
On Friday, September 21, 2018, Friedman, Eli <efriedma at codeaurora.org> wrote:
On 9/20/2018 2:15 PM, hameeza ahmed wrote:
Hello, I m trying to set vector width using #pragma clang loop vectorizewidth(32) but i m getting width 8 for the following kernel; i m getting following output when i compiled;
*clang -O3 correlation.c -Rpass=loop-vectorize -emit-llvm -march=knl -S -o 1.ll correlation.c:38:9: remark: vectorized loop (vectorization width: 8, interleaved count: 4) [-Rpass=loop-vectorize] _for (j = 0; j < M; j++) ^ *_ _With AVX-512, an instruction can operate on at most 8 double-precision_ _lanes. The vectorizer recognizes that, and interleaves the loop so you get_ _8*4==32 scalar iterations per iteration of the vectorized loop._ _-Eli_ _--_ _Employee of Qualcomm Innovation Center, Inc._ _Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project_ -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20180921/745710b9/attachment.html>
- Previous message: [llvm-dev] Vectorization width not correct using #pragma clang loop vectorize_width
- Next message: [llvm-dev] Vectorization width not correct using #pragma clang loop vectorize_width
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]