Code

perf: i8 * i8 -> fp32 GEMM Boost#187

Code

Merged

yirongjie merged 11 commits intoUbiquitousLearning:mainUbiquitousLearning/mllm:mainfrom chenghuaWang:mainchenghuaWang/mllm:mainCopy head branch name to clipboardNov 22, 2024

+1,076-35

Commits

Commits on Nov 11, 2024

• feat: per-tensor and per-channel impl

  chenghuaWang

  committed
  416f82d

  Copy full SHA for 416f82d

• feat: boost i8 * i8 -> fp32 matmul

  chenghuaWang

  committed
  6d24b7a

  Copy full SHA for 6d24b7a

• feat: matmul_i8

  chenghuaWang

  committed
  ca3bab0

  Copy full SHA for ca3bab0

Commits on Nov 15, 2024

• Merge branch 'UbiquitousLearning:main' into main

  chenghuaWang

  authored
  c0644fb

  Copy full SHA for c0644fb

• feat: conv2d and im2col algorithm

  chenghuaWang

  committed
  366d02f

  Copy full SHA for 366d02f

• Merge branch 'main' of https://github.com/chenghuaWang/mllm

  chenghuaWang

  committed
  b18ac24

  Copy full SHA for b18ac24

Commits on Nov 16, 2024

• Merge branch 'UbiquitousLearning:main' into main

  chenghuaWang

  authored
  8396a31

  Copy full SHA for 8396a31

Commits on Nov 17, 2024

• fix: refactor matmul_i8

  chenghuaWang

  committed
  7a4ed0c

  Copy full SHA for 7a4ed0c

Commits on Nov 21, 2024

• Merge branch 'UbiquitousLearning:main' into main

  chenghuaWang

  authored
  0962b81

  Copy full SHA for 0962b81

Commits on Nov 22, 2024

• fix: remove demo_gemm_q40_perf

  chenghuaWang

  committed
  e3302ee

  Copy full SHA for e3302ee

• Merge branch 'main' into main

  chenghuaWang

  authored
  ff0b4c5

  Copy full SHA for ff0b4c5