Code

AVX Q4_0 and Q8_0 sgemm#6891

Code

Merged

ggerganov merged 12 commits intoggml-org:masterggml-org/llama.cpp:masterfrom netrunnereve:sgemm-avxCopy head branch name to clipboardMay 8, 2024

+56-21

Commits

Commits on Apr 23, 2024

• basic avx implementation

  netrunnereve

  committed
  86d1d84

  Copy full SHA for 86d1d84

• style

  netrunnereve

  committed
  257391a

  Copy full SHA for 257391a

Commits on Apr 24, 2024

• combine denibble with load

  netrunnereve

  committed
  9facb0f

  Copy full SHA for 9facb0f

• reduce 256 to 128 (and back!) conversions

  netrunnereve

  committed
  dee9566

  Copy full SHA for dee9566

Commits on Apr 25, 2024

• sse load

  netrunnereve

  committed
  063a31f

  Copy full SHA for 063a31f

• Merge branch 'ggerganov:master' into sgemm-avx

  netrunnereve

  authored
  e97c0fd

  Copy full SHA for e97c0fd

• Update sgemm.cpp

  netrunnereve

  authored
  fb80f13

  Copy full SHA for fb80f13

• Merge branch 'ggerganov:master' into sgemm-avx

  netrunnereve

  authored
  330b3bc

  Copy full SHA for 330b3bc

Commits on Apr 30, 2024

• merge

  netrunnereve

  committed
  8916954

  Copy full SHA for 8916954

• oops

  Show description for ae0b5ea
  netrunnereve

  committed
  ae0b5ea

  Copy full SHA for ae0b5ea

Commits on May 1, 2024

• Merge branch 'ggerganov:master' into sgemm-avx

  netrunnereve

  authored
  8af6f9c

  Copy full SHA for 8af6f9c

Commits on May 8, 2024

• Merge branch 'ggerganov:master' into sgemm-avx

  netrunnereve

  authored
  f87fc83

  Copy full SHA for f87fc83