onnxruntime
Improvements to the INT8 GEMM portion of the code for Power
#20595
Merged
Overview
Commits
13
Changes
View On GitHub
Commits
  • Faster S8 Packing routines.
    Chip Kerchner committed 1 year ago
  • Use vectors instead of individual element in MlasGemmQuantCopyPackB8x8 - up to 4x faster.
    Chip Kerchner committed 1 year ago
  • Remove comment and fix tab.
    Chip Kerchner committed 1 year ago
  • Merge branch 'main' into powerImprovements
    Chip Kerchner committed 1 year ago
  • Merge branch 'main' into powerImprovements
    Chip-Kerchner committed 1 year ago
  • Fix lint warnings.
    Chip Kerchner committed 1 year ago
  • Merge branch 'main' into powerImprovements
    Chip Kerchner committed 1 year ago
  • Merge remote-tracking branch 'origin_chip/powerImprovements' into powerImprovements
    Chip-Kerchner committed 1 year ago
  • Fix commenters suggestions and other lint warnings.
    Chip Kerchner committed 1 year ago
  • Unroll MlasQgemmStoreVectorMMA and vectorize other variables.
    Chip Kerchner committed 1 year ago
  • Merge remote-tracking branch 'origin/main' into powerImprovements
    Chip-Kerchner committed 1 year ago
  • Merge branch 'powerImprovements' of github.ibm.com:PowerAppLibs/onnx into powerImprovements
    Chip-Kerchner committed 1 year ago
  • More lint fixes.
    Chip-Kerchner committed 1 year ago
FAQ Terms Privacy Refunds Impressum
Loading