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 f4cdb3f3ef [Relax][PyTorch] Add support for ge, gt, le, mod, ne ops 
(#17664)
     add cc2f0796ba [RELAX][BYOC] OpenCLML offload support for Relax (#17654)
     add 06def23ad1 [Refactor] Improve TargetHasSVE function with optional 
target handling (#17666)
     add bbd77ace31 Pick up vector length from 'zvlXXXb' (RVV) mattr for riscv 
(#17641)

No new revisions were added by this update.

Summary of changes:
 cmake/modules/contrib/CLML.cmake                   |   4 +-
 python/tvm/relax/backend/__init__.py               |   2 +-
 .../backend/{cpu_generic => adreno}/__init__.py    |   3 +-
 python/tvm/relax/backend/adreno/clml.py            | 618 +++++++++++++++++
 .../tvm/relax/backend/{rocm => adreno}/pipeline.py |  52 +-
 python/tvm/relax/backend/gpu_generic/__init__.py   |   1 +
 python/tvm/relax/pipeline.py                       |  10 +
 python/tvm/relax/transform/__init__.py             |   1 +
 .../fold_batch_norm_to_conv2d_for_inference.py     | 102 +++
 python/tvm/target/__init__.py                      |   1 +
 python/tvm/target/codegen.py                       |  17 +
 python/tvm/target/target.py                        |   7 +-
 python/tvm/testing/utils.py                        |  33 +
 src/arith/scalable_expression.cc                   |  12 +-
 src/arith/scalable_expression.h                    |   2 +-
 src/relax/backend/contrib/clml/codegen.cc          | 358 ++++++++++
 src/relax/backend/contrib/utils.cc                 |  11 +
 src/relax/backend/contrib/utils.h                  |  26 +-
 src/runtime/contrib/clml/clml_runtime.cc           | 181 +++--
 src/runtime/contrib/clml/clml_runtime.h            |  17 +-
 src/runtime/contrib/clml/clml_utils.cc             |  11 +
 src/runtime/contrib/clml/clml_utils.h              |   2 +
 src/target/llvm/codegen_llvm.cc                    |  26 +-
 src/target/llvm/llvm_instance.cc                   |  50 +-
 src/target/llvm/llvm_instance.h                    |   2 +-
 src/target/llvm/llvm_module.cc                     |  13 +
 tests/python/codegen/test_target_codegen_riscv.py  |  49 ++
 .../test_nnapi => relax/backend/clml}/conftest.py  |  32 +-
 tests/python/relax/backend/clml/mod_utils.py       | 728 +++++++++++++++++++++
 .../python/relax/backend/clml/test_clml_codegen.py | 505 ++++++++++++++
 .../backend/clml/test_op_exec_clml_codegen.py      | 329 ++++++++++
 tests/python/relax/backend/clml/utils.py           |  90 +++
 .../test_transform_fold_batch_norm_to_conv2d.py    | 144 ++++
 tests/python/target/test_riscv_features.py         |  66 ++
 tests/scripts/task_config_build_gpu.sh             |   1 +
 tests/scripts/task_python_adreno.sh                |   9 +
 tests/scripts/unity/task_python_relax.sh           |   3 +
 37 files changed, 3356 insertions(+), 162 deletions(-)
 copy python/tvm/relax/backend/{cpu_generic => adreno}/__init__.py (90%)
 create mode 100644 python/tvm/relax/backend/adreno/clml.py
 copy python/tvm/relax/backend/{rocm => adreno}/pipeline.py (59%)
 create mode 100644 
python/tvm/relax/transform/fold_batch_norm_to_conv2d_for_inference.py
 create mode 100644 src/relax/backend/contrib/clml/codegen.cc
 create mode 100644 tests/python/codegen/test_target_codegen_riscv.py
 copy tests/python/{nightly/test_nnapi => relax/backend/clml}/conftest.py (56%)
 create mode 100644 tests/python/relax/backend/clml/mod_utils.py
 create mode 100644 tests/python/relax/backend/clml/test_clml_codegen.py
 create mode 100644 tests/python/relax/backend/clml/test_op_exec_clml_codegen.py
 create mode 100644 tests/python/relax/backend/clml/utils.py
 create mode 100644 
tests/python/relax/test_transform_fold_batch_norm_to_conv2d.py
 create mode 100644 tests/python/target/test_riscv_features.py

Reply via email to