Code

F16 intrinsics standalone#14

Code

Closed

Narsil wants to merge 15 commits intosarah-quinones:mainsarah-quinones/gemm:mainfrom LaurentMazare:f16_intrinsics_standaloneLaurentMazare/gemm:f16_intrinsics_standaloneCopy head branch name to clipboard

+1,342-24

Commits

Commits on Jul 1, 2023

• Vectorize the inner loop of f16's pack_generic.

  LaurentMazare

  committed
  a9f610e

  Copy full SHA for a9f610e

• Remove the old version.

  LaurentMazare

  committed
  4b189f2

  Copy full SHA for 4b189f2

Commits on Jul 16, 2023

• Wasm simd ops (f32 only).

  Narsil

  committed
  926d13a

  Copy full SHA for 926d13a

• Fix multiplier.

  Narsil

  committed
  57922f0

  Copy full SHA for 57922f0

• Adding a parallelism bench.

  Narsil

  committed
  8d205b6

  Copy full SHA for 8d205b6

Commits on Jul 17, 2023

• Merge remote-tracking branch 'narsil/wasm_simd' into f16-vec-plus-wasm-simd

  LaurentMazare

  committed
  e91056e

  Copy full SHA for e91056e

Commits on Jul 20, 2023

• Fixing large multi-threading (-40% improvement for parallelism

  Show description for 0da2128
  Narsil

  committed
  0da2128

  Copy full SHA for 0da2128

• Fix.

  Narsil

  committed
  c1a5b31

  Copy full SHA for c1a5b31

Commits on Jul 26, 2023

• Fix tests.

  Narsil

  committed
  76ea6bd

  Copy full SHA for 76ea6bd

• Format.

  Narsil

  committed
  b11ea6f

  Copy full SHA for b11ea6f

• Merge pull request #2 from Narsil/rayon3

  Show description for f577d01
  LaurentMazare

  authored
  f577d01

  Copy full SHA for f577d01

• Merge pull request #3 from LaurentMazare/f16-vec-plus-wasm-simd

  Show description for c03b453
  LaurentMazare

  authored
  c03b453

  Copy full SHA for c03b453

Commits on Aug 1, 2023

• Using m1 intrinsics for f16xf16

  Narsil

  committed
  c7a1ceb

  Copy full SHA for c7a1ceb

• Removing black box.

  Narsil

  committed
  a8f0280

  Copy full SHA for a8f0280

• Cleanup.

  Narsil

  committed
  c2d2173

  Copy full SHA for c2d2173