[PATCH] D124284: [SLP]Try partial store vectorization if supported by target.

Alexey Bataev via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Apr 25 12:56:10 PDT 2022


ABataev added a comment.

Increased priority, some numbers for long run tests:
**Regressions**

  Metric: exec_time
  test-suite :: External/SPEC/CINT2006/401.bzip2/401.bzip2.test  26.98     27.49      1.9%
  test-suite :: External/SPEC/CFP2017rate/526.blender_r/526.blender_r.test  94.46     95.73      1.3%
  test-suite :: External/SPEC/CFP2017speed/644.nab_s/644.nab_s.test  63.61     64.22      1.0%

**Gains**

  test-suite :: External/SPEC/CFP2017speed/619.lbm_s/619.lbm_s.test 234.34    231.17     -1.4%
  test-suite :: External/SPEC/CFP2017rate/519.lbm_r/519.lbm_r.test  27.07     26.70     -1.4%
  test-suite :: External/SPEC/CINT2017speed/631.deepsjeng_s/631.deepsjeng_s.test  59.34     57.95     -2.3%
  test-suite :: External/SPEC/CINT2017rate/531.deepsjeng_r/531.deepsjeng_r.test  45.51     44.42     -2.4%
  test-suite :: External/SPEC/CINT2006/464.h264ref/464.h264ref.test  36.46     35.40     -2.9%
  test-suite :: External/SPEC/CINT2006/445.gobmk/445.gobmk.test  57.70     55.93     -3.1%
  test-suite :: External/SPEC/CINT2006/473.astar/473.astar.test  57.02     53.86     -5.5%
  test-suite :: External/SPEC/CINT2017rate/520.omnetpp_r/520.omnetpp_r.test  45.46     40.61    -10.7%
  test-suite :: External/SPEC/CINT2017speed/620.omnetpp_s/620.omnetpp_s.test  51.19     41.87    -18.2%

Geomean is still -100%, which means that still the performance with the patch is better than before.
Tested for O3 <https://reviews.llvm.org/owners/package/3/> + LTO, generic CPU.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124284/new/

https://reviews.llvm.org/D124284



More information about the llvm-commits mailing list