kernel

Commit History

  1. Build

    danieldk HF Staff
    ba53cf7

  1. Use `-static-global-template-stub=false` nvcc option for Marlin

    danieldk HF Staff
    e384bb2
  2. Sync updates for CUDA 13 compat

    danieldk HF Staff
    70da85f

  1. Update build.toml

    danieldk HF Staff
    7c5923b
  2. Update flake

    danieldk HF Staff
    83332ce

  1. Build (aarch64-linux)

    danieldk HF Staff
    150f8c2
  2. Build (x86_64-linux)

    danieldk HF Staff
    daf6c87
  3. Prepare for Torch 2.8

    danieldk HF Staff
    eabe7c2

  1. Build (x86_64)

    danieldk HF Staff
    b65f8ab
  2. Build (aarch64)

    danieldk HF Staff
    4dcf20d
  3. Fix undefined symbol on CUDA 11.8⁠

    danieldk HF Staff
    3313895
  4. Fix absolute imports

    danieldk HF Staff
    c516610

  1. Build (aarch64)

    danieldk HF Staff
    229f047

  1. Build (x86_64)

    danieldk HF Staff
    82ffd1f
  2. Update kernel builder flake

    danieldk HF Staff
    dfa7d18
  3. Export Marlin and quantization utilities

    danieldk HF Staff
    2f56f34
  4. Fixup platform FP8 data type query

    danieldk HF Staff
    a965094

  1. Build (AArch64)

    danieldk HF Staff
    a02ac19

  1. Set CUDA minimum versions for hopper/blackwell cutlass kernels

    danieldk HF Staff
    3370704
  2. Build (x86_64)

    danieldk HF Staff
    c73c09e

  1. Temporarily disable ROCm

    danieldk HF Staff
    d5fd23d

  1. Sync to vLLM 20250627

    danieldk HF Staff
    8aa00a3

  1. Sync on vLLM 20240402

    danieldk HF Staff
    d26f884

  1. Build (aarch64)

    danieldk HF Staff
    34013ad
  2. Build

    danieldk HF Staff
    cfc95fb
  3. Update flake

    danieldk HF Staff
    0435ccb
  4. Sync capabilities with upstream

    danieldk HF Staff
    59b2fef

  1. Build

    danieldk HF Staff
    32721c3
  2. Add support for ROCm

    danieldk HF Staff
    3c8bb73

  1. Add `flake.lock`

    danieldk HF Staff
    6d36a16
  2. Update for latest kernel-builder (add revision)

    danieldk HF Staff
    2e9b7c5

  1. Update for build.toml changes

    danieldk HF Staff
    116e562

  1. Update README.md (#1)⁠

    danieldk +1: lysandre HF Staff
    Verified signature
    64f4463

  1. Build (Torch 2.6)

    danieldk HF Staff
    95272c7

  1. Build

    danieldk HF Staff
    329999a
  2. Export `ScalarType`/`scalartypes`

    danieldk HF Staff
    b2e431d
  3. Update flake.lock

    danieldk HF Staff
    613a874

  1. Build

    danieldk HF Staff
    d636d53
  2. Sync with vLLM⁠

    danieldk HF Staff
    0da5bf5

  1. Build

    danieldk HF Staff
    0bc428e
  2. Adjust for latest kernel-builder changes

    danieldk HF Staff
    f44baaa

  1. Build

    danieldk HF Staff
    2e44e8c

  1. Simplify `flake.nix`

    danieldk HF Staff
    88883b3

  1. Build

    danieldk HF Staff
    d62ac09
  2. Expose ops (handy for tests etc.)

    danieldk HF Staff
    e3a7455

  1. Add a `flake.nix` for convenience

    danieldk HF Staff
    be218ea

  1. Build

    danieldk HF Staff
    a6c77d7
  2. Add full Marlin support and tests for Marlin/CUTLASS

    danieldk HF Staff
    165b25c

  1. Import CUTLASS tests and add missing scaled mm with zp signature

    danieldk HF Staff
    2dd62c9
  2. Add missing `scalar_type.hpp`

    danieldk HF Staff
    9d045c3