Repository navigation

feat(tf32): add TF32 TensorCore GEMM kernel achieving 27 TFLOPS - #45

Merged

m96-chan merged 23 commits into

mainfrom

feature/v0.2.3-tf32-tensorcore

Dec 14, 2025

Merged

feat(tf32): add TF32 TensorCore GEMM kernel achieving 27 TFLOPS#45
m96-chan merged 23 commits into
mainfrom
feature/v0.2.3-tf32-tensorcore

Commits on Dec 13, 2025
- wip(tf32): add TF32 TensorCore GEMM kernel (correctness bug)
  
  m96-chan
  and
  claude
  committed
  b64806c
- fix(tf32): WMMA store_matrix_sync alignment bug for N % 8 != 0
  
  m96-chan
  and
  claude
  committed
  914b58f
- docs(CLAUDE.md): add kernel development workflow
  
  m96-chan
  and
  claude
  committed
  baa31de
- wip(tf32): 2-stage cp.async pipeline with transposed B (correctness bug)
  
  m96-chan
  and
  claude
  committed
  6e06319
- wip(tf32): BK=32 kernel with manual epilogue fix
  
  m96-chan
  and
  claude
  committed
  2eb35cc
- wip(tf32): BK=16 with Bs[BN][BK] col-major layout
  
  m96-chan
  and
  claude
  committed
  b10fa7e
- wip(tf32): user rewrite v3 - 51KB smem, 0 spills
  
  m96-chan
  and
  claude
  committed
  8ee0e82
- wip(tf32): G3 kernel - 40KB smem, BK=16
  
  m96-chan
  and
  claude
  committed
  0be67dc
- wip(tf32): cp.async for both A and B, row_major fragments (6 TFLOPS)
  
  m96-chan
  and
  claude
  committed
  10967f8
- wip(tf32): revert to simplified kernel structure
  
  m96-chan
  and
  claude
  committed
  0b1345e
- wip(tf32): add launch function (correctness broken)
  
  m96-chan
  and
  claude
  committed
  a2a70dd
- wip(tf32): restore 32 TFLOPS kernel (correctness bug)
  
  m96-chan
  and
  claude
  committed
  007f732
- wip(tf32): 44 TFLOPS kernel (correctness still broken)
  
  m96-chan
  and
  claude
  committed
  b256ecb
- wip(tf32): WMMA row_major×row_major verified working
  
  m96-chan
  and
  claude
  committed
  2fe874a
- docs(tf32): add WMMA 16x16x8 fragment mapping from dump_fragments
  
  m96-chan
  and
  claude
  committed
  20b78b1
- fix(tf32): correct C fragment mapping for PTX mma.sync m16n8k8
  
  m96-chan
  and
  claude
  committed
  7d01fb9
- feat(tf32): correct cp.async pipeline achieving 27 TFLOPS
  
  m96-chan
  and
  claude
  committed
  1d69de4
- docs(readme): add v0.2.3 TF32 benchmark comparison table
  
  m96-chan
  and
  claude
  committed
  8c79f98
- docs(readme): fix benchmark numbers with actual cuBLAS data
  
  m96-chan
  and
  claude
  committed
  db89f94
Commits on Dec 14, 2025
- perf(tf32): optimize kernel with A fragment hoisting (+1.35 TFLOPS)
  
  m96-chan
  and
  claude
  committed
  da41bf7
- docs: add TF32 kernel optimization summary
  
  m96-chan
  and
  claude
  committed
  d00b446
- docs: remove optimization summary md (moving to Issue #41)
  m96-chan
  committed
  ea3700f
- fix(lint): remove extraneous f-string prefixes
  
  m96-chan
  and
  claude
  committed
  a1c8f3c

Commits

Commits on Dec 13, 2025

Commits on Dec 14, 2025