Skip to content

Commit

Permalink
[CMake] FlashInfer bump and cmake changes
Browse files Browse the repository at this point in the history
  • Loading branch information
MasterJH5574 committed May 15, 2024
1 parent 2e4254e commit ce58d63
Show file tree
Hide file tree
Showing 2 changed files with 12 additions and 2 deletions.
2 changes: 0 additions & 2 deletions CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -956,8 +956,6 @@ if (USE_FLASHINFER STREQUAL "ON")
message(STATUS "Build with FlashInfer")
set(FLASHINFER_TVM_BINDING ON)
set(FLASHINFER_TVM_HOME ${PROJECT_SOURCE_DIR})
set(FLASHINFER_ENABLE_FP8 OFF)
set(FLASHINFER_ENABLE_BF16 OFF)
set(FLASHINFER_PREFILL OFF)
set(FLASHINFER_DECODE OFF)
set(FLASHINFER_PAGE OFF)
Expand Down
12 changes: 12 additions & 0 deletions cmake/config.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -444,6 +444,18 @@ set(USE_GTEST AUTO)
# Need to have USE_CUDA=ON
set(USE_CUTLASS OFF)

# Whether to enable FlashInfer or not.
set(USE_FLASHINFER OFF)
# Options for FlashInfer kernel compilation.
set(FLASHINFER_ENABLE_FP8 OFF)
set(FLASHINFER_ENABLE_BF16 OFF)
set(FLASHINFER_GEN_GROUP_SIZES 1 4 6 8)
set(FLASHINFER_GEN_HEAD_DIMS 128)
set(FLASHINFER_GEN_KV_LAYOUTS 1)
set(FLASHINFER_GEN_POS_ENCODING_MODES 0 1)
set(FLASHINFER_GEN_ALLOW_FP16_QK_REDUCTIONS "false")
set(FLASHINFER_GEN_CASUALS "false" "true")

# Enable to show a summary of TVM options
set(SUMMARIZE OFF)

Expand Down

0 comments on commit ce58d63

Please sign in to comment.