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 bf0607bd31 [VTA] Provide zero-initialization for VTAGenericInsn
(#13698)
add 07a5a9eadb [Tensorize][runtime] Add support for AMX(Advanced Matrix
Extensions) through Tensor intrinsics (#13642)
add 048028b72b [BugFix][TVMScript] Fix the roundtripability of intrinsic
pow (#13692)
add 721f1151b1 [BugFix][Runtime] Fix Incorrect node information (#13693)
add 39dbce1f9c [microTVM] Build standalone_crt with cmake instead of
makefile (#13600)
add 21d7968b61 [microTVM] Fix MacOS build with USE_MICRO=ON (#13711)
No new revisions were added by this update.
Summary of changes:
CMakeLists.txt | 2 +
cmake/config.cmake | 3 +
cmake/modules/LibInfo.cmake | 1 +
cmake/modules/StandaloneCrt.cmake | 256 ++++++++++++---------
cmake/modules/contrib/{CUTLASS.cmake => AMX.cmake} | 10 +-
include/tvm/runtime/crt/crt.h | 0
python/tvm/contrib/debugger/debug_result.py | 24 +-
python/tvm/relay/op/strategy/x86.py | 9 +-
python/tvm/script/ir_builder/tir/ir.py | 4 +-
python/tvm/tir/__init__.py | 2 +-
python/tvm/tir/op.py | 22 ++
python/tvm/topi/x86/dense.py | 167 ++++++++++++--
python/tvm/topi/x86/dense_alter_op.py | 10 +-
python/tvm/topi/x86/tensor_intrin.py | 226 +++++++++++++++++-
python/tvm/topi/x86/utils.py | 7 +
src/runtime/contrib/amx/amx_config.cc | 135 +++++++++++
src/runtime/thread_storage_scope.h | 5 +
src/support/libinfo.cc | 5 +
tests/python/contrib/test_amx.py | 126 ++++++++++
tests/python/relay/test_op_level1.py | 47 ++++
tests/python/unittest/test_runtime_graph_debug.py | 26 ++-
tests/python/unittest/test_tvmscript_roundtrip.py | 9 +
22 files changed, 929 insertions(+), 167 deletions(-)
copy cmake/modules/contrib/{CUTLASS.cmake => AMX.cmake} (75%)
mode change 100755 => 100644 include/tvm/runtime/crt/crt.h
create mode 100644 src/runtime/contrib/amx/amx_config.cc
create mode 100644 tests/python/contrib/test_amx.py
