This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch nightly
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 2a87c4cfc0 [BYOC][NNAPI] Add NNAPI backend for BYOC (#17385)
add a90fb8e2d9 [TIR][NarrowDataType] Bufferload's index should not inherit
bits constraint of value (#17411)
add 7fc8adcc7e [CI][Windows] Workaround for error in FindLLVM (#17409)
add 5648a8e114 [Runtime] Add property Module.is_device_module (#17407)
add 4e70e4a4ba [CUTLASS] Add FP8 gemm kernels (#17408)
No new revisions were added by this update.
Summary of changes:
cmake/modules/contrib/CUTLASS.cmake | 1 +
cmake/utils/FindLLVM.cmake | 9 ++
python/tvm/relax/vm_build.py | 2 +-
python/tvm/runtime/module.py | 4 +
src/runtime/contrib/cublas/cublas.cc | 6 +-
src/runtime/contrib/cutlass/fp8_gemm.cu | 95 +++++++++++++
src/runtime/contrib/cutlass/gemm_runner.cuh | 155 +++++++++++++++++++++
src/tir/transforms/narrow_datatype.cc | 14 +-
.../test_runtime_ndarray.py | 1 +
tests/python/contrib/test_cutlass.py | 107 ++++++++++++--
.../test_tir_transform_narrow_datatype.py | 17 +++
11 files changed, 394 insertions(+), 17 deletions(-)
create mode 100644 src/runtime/contrib/cutlass/fp8_gemm.cu
create mode 100644 src/runtime/contrib/cutlass/gemm_runner.cuh