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 b07c6c1e75 [Relax] Add symbolic shape support to BasePyModule for
dynamic tensor operations (#18288)
add 865d8f079c [CUDA] Support NVTX in CUDA 13 (#18300)
add da7b68d820 [Python] Fix runtime tensor import (#18299)
add 8f658cc3c6 [FFI][REFACTOR] Refactor python ffi call mechanism for perf
(#18302)
No new revisions were added by this update.
Summary of changes:
cmake/utils/FindCUDA.cmake | 2 +-
ffi/CMakeLists.txt | 1 +
ffi/include/tvm/ffi/container/tensor.h | 65 ++-
ffi/python/tvm_ffi/cython/base.pxi | 55 ++-
ffi/python/tvm_ffi/cython/function.pxi | 538 +++++++++++++--------
ffi/python/tvm_ffi/cython/tensor.pxi | 100 ++--
ffi/python/tvm_ffi/cython/tvm_ffi_python_helpers.h | 447 +++++++++++++++++
ffi/scripts/benchmark_dlpack.py | 16 +
ffi/src/ffi/extra/testing.cc | 2 +-
python/tvm/exec/disco_worker.py | 2 +-
python/tvm/testing/runner.py | 4 +-
11 files changed, 990 insertions(+), 242 deletions(-)
create mode 100644 ffi/python/tvm_ffi/cython/tvm_ffi_python_helpers.h
