This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch refactor-s0
in repository https://gitbox.apache.org/repos/asf/tvm.git
discard 70bbe36645 fix lint and minor compile
discard 16603265fc lint
discard 3bb75d4cb1 Fix deleter to match latest
discard cc061117c0 Pass compile
discard 6127599320 clang format
discard 66f993be83 pass expr first stab
discard 130f39e5f8 Pass all runtime
discard 5df4dec4b2 pass object and runtime compile
discard 6c60fd30e1 Use specific import via using and not do global namespace
discard 5d3d586e94 pass basic disco session compile
discard ad6d5fbdf0 pass ndarray compile
discard 8e7df79382 pass packed func compile
discard 5315cb9a3f Fix NDArray compile
discard 6c16f46c9a Move over container on new object
discard b9e1fd8d25 First step, pass string compile
discard 02127d4039 CMake links ok
discard ac79f9e8cf [REFACTOR] Migration starting pt
discard acd5e5ca26 Fix ffi
discard 8503919cdb [FFI] Enhance the deleter and reflection to be aware of
padding
discard 68f2cead2c Cleanup and focus on the integrated version
discard 695ac175b9 [FFI] bool, device, tensor, raw str
discard 61249b46e7 [FFI] Initial reflection support
discard 00fd84b40d [FFI] Global function registry
discard f29e76f9f4 [FFI] Formalize SafeCall handling
discard 63c86e58a6 [FFI] Initial map support
discard d5b942f94e [FFI] Optional Support
discard 3076b1f260 [FFI] Introduce String support
discard 98e38b3368 [FFI][REFACTOR] Cleanup naming convention and lint
discard 38ffb87768 [FFI] Introduce array support
discard 67986d3369 [FFI] Improve Any coverage to include ObjectPtr
discard dfbf0e587e [FFI] Function support
discard 1ae001e701 [FFI] Any support
discard 89ccf3f967 [FFI] Object type hierachy cast and check support
discard 2a461fa42f [FFI] Add libbacktrace support
discard 45f4ff950f [FFI] ObjectRef based Error
discard efd253828e [FFI] Enable ObjectPtr and tests
discard ea2549e0a8 [FFI] Initial step to create ffi from the original tvm
source
discard 5b3ff33039 [FFI] Initial Code Commit
add cf6e61b285 [Relax] Allow ingesting Upsample module from torch.export
either using Size or Scale Factor argument (#17721)
add 6e8c367fd7 [Relax] Add torch exported program ingestion capability for
Tensor.detach(), Tensor.copy_, and aten.lift_fresh_copy (#17723)
add e74f7eee32 [Relax] Add support to ingest Tensor.expand_as() (#17724)
add 6afa62cbe1 Fix typos in comments and strings (#17736)
add 148737b1ff [IR] Compact Functor vtable (#17731)
add 36a92a4ae5 Remove Google Analytics (#17734)
add 37fe645f94 [Relax] Ingest Tensor.clamp from torch export (#17725)
add 363ebb453b [BF16] Support ndarray.asnumpy() to bfloat16 tensor
natively using ml_dtypes (#17739)
add 64dae9de9b [Relax][PyTorch] Support elu, celu, selu ops for
ExportedProgram importer (#17738)
add 6b89f9537c [Codegen][CUDA] Fix codegen of cast among vector bfloat16,
fp8 and fp4 (#17741)
add cd92392d34 [Refactor] Clean up Relay references in the codebase
(#17733)
new 73ea198a2a [FFI] Initial Code Commit
new af35e25d38 [FFI] Initial step to create ffi from the original tvm
source
new 1ab674c646 [FFI] Enable ObjectPtr and tests
new 8c2a94802d [FFI] ObjectRef based Error
new d3280ddb40 [FFI] Add libbacktrace support
new d3145e864a [FFI] Object type hierachy cast and check support
new 4d03e51f29 [FFI] Any support
new 854abf58be [FFI] Function support
new 9a5cca4645 [FFI] Improve Any coverage to include ObjectPtr
new 40b5ad587e [FFI] Introduce array support
new a2b5e9a726 [FFI][REFACTOR] Cleanup naming convention and lint
new 344ab18437 [FFI] Introduce String support
new 441ceae16e [FFI] Optional Support
new 514a798b2c [FFI] Initial map support
new 36f8f8f2b0 [FFI] Formalize SafeCall handling
new 463425f90b [FFI] Global function registry
new 27ffeda002 [FFI] Initial reflection support
new 7b7b1e7f1c [FFI] bool, device, tensor, raw str
new 4472f6ec3b Cleanup and focus on the integrated version
new 661cf0bd48 [FFI] Enhance the deleter and reflection to be aware of
padding
new 238342a200 Fix ffi
new 969b2192b3 [REFACTOR] Migration starting pt
new dce37a83b3 CMake links ok
new 84d7aaa81b First step, pass string compile
new b8a65dbab0 Move over container on new object
new e4dd9cc031 Fix NDArray compile
new 24fce43a82 pass packed func compile
new 125da2e573 pass ndarray compile
new d263641f10 pass basic disco session compile
new 8b0ac0deb6 Use specific import via using and not do global namespace
new fb62f9034d pass object and runtime compile
new fb524a6935 Pass all runtime
new 958938ee34 pass expr first stab
new eb0199921a clang format
new a16c253f0e Pass compile
new 69b4611a64 Fix deleter to match latest
new ffc1ec0ebd lint
new dbf7c372f3 fix lint and minor compile
This update added new revisions after undoing existing revisions.
That is to say, some revisions that were in the old version of the
branch are not in the new version. This situation occurs
when a user --force pushes a change and generates a repository
containing something like this:
* -- * -- B -- O -- O -- O (70bbe36645)
\
N -- N -- N refs/heads/refactor-s0 (dbf7c372f3)
You should already have received notification emails for all of the O
revisions, and so the following emails describe only the N revisions
from the common base, B.
Any revisions marked "omit" are not gone; other references still
refer to them. Any revisions marked "discard" are gone forever.
The 38 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
Summary of changes:
apps/ios_rpc/tvmrpc/RPCServer.mm | 30 +--
apps/ios_rpc/tvmrpc/TVMRuntime.mm | 2 +-
cmake/config.cmake | 6 +-
cmake/modules/CUDA.cmake | 6 +-
cmake/modules/contrib/ArmComputeLib.cmake | 2 +-
cmake/modules/contrib/BNNS.cmake | 3 +-
cmake/modules/contrib/DNNL.cmake | 7 +-
cmake/modules/contrib/Mrvl.cmake | 5 -
cmake/modules/contrib/TensorRT.cmake | 2 +-
docs/arch/index.rst | 8 +-
docs/arch/pass_infra.rst | 18 +-
docs/conf.py | 1 -
docs/contribute/document.rst | 2 +-
include/tvm/arith/iter_affine_map.h | 2 +-
include/tvm/ir/expr.h | 4 +-
include/tvm/ir/transform.h | 10 +-
include/tvm/ir/type.h | 4 +-
include/tvm/ir/type_functor.h | 1 +
include/tvm/meta_schedule/builder.h | 4 +-
include/tvm/node/functor.h | 28 ++-
include/tvm/relax/dataflow_pattern.h | 2 +
include/tvm/relax/dataflow_pattern_functor.h | 2 +-
include/tvm/relax/expr.h | 4 +-
include/tvm/relax/expr_functor.h | 3 +-
include/tvm/relax/op_attr_types.h | 2 +-
include/tvm/relax/struct_info_functor.h | 1 +
include/tvm/relax/type.h | 2 +-
include/tvm/target/virtual_device.h | 4 +-
include/tvm/tir/analysis.h | 2 +-
include/tvm/tir/block_dependence_info.h | 4 +-
include/tvm/tir/builtin.h | 2 +-
include/tvm/tir/expr_functor.h | 1 +
include/tvm/tir/op.h | 2 +-
include/tvm/tir/stmt_functor.h | 1 +
include/tvm/topi/transform.h | 4 +-
python/gen_requirements.py | 6 +-
python/setup.py | 6 -
python/tvm/__init__.py | 2 +-
python/tvm/contrib/cutlass/build.py | 4 +-
python/tvm/contrib/mrvl.py | 2 +-
.../msc/framework/tensorflow/frontend/translate.py | 2 +-
.../msc/framework/torch/frontend/translate.py | 2 +-
python/tvm/ir/transform.py | 2 +-
python/tvm/ir/type.py | 2 +-
python/tvm/meta_schedule/builder/builder.py | 4 +-
python/tvm/relax/dpl/pattern.py | 5 +-
.../frontend/torch/base_fx_graph_translator.py | 116 +++++++++++-
.../frontend/torch/exported_program_translator.py | 52 +++++-
python/tvm/relax/frontend/torch/fx_translator.py | 4 +-
python/tvm/runtime/ndarray.py | 7 +-
python/tvm/testing/utils.py | 42 -----
python/tvm/tir/analysis/analysis.py | 4 -
rust/tvm-rt/src/module.rs | 2 +-
src/ir/attr_functor.h | 1 +
src/ir/transform.cc | 2 +-
src/node/structural_equal.cc | 4 +-
src/relax/backend/contrib/cutlass/codegen.cc | 2 +-
src/relax/ir/dataflow_pattern.cc | 3 +-
src/relax/ir/dataflow_pattern_functor.cc | 2 +-
src/relax/ir/py_expr_functor.cc | 3 +
src/relax/ir/transform.cc | 1 -
src/runtime/const_loader_module.cc | 5 +-
src/runtime/contrib/dnnl/dnnl.cc | 1 -
src/runtime/contrib/dnnl/dnnl_json_runtime.cc | 2 +-
src/runtime/contrib/mrvl/mrvl_hw_runtime.cc | 4 +-
src/runtime/contrib/mrvl/mrvl_runtime.cc | 4 +-
src/runtime/contrib/tensorrt/tensorrt_builder.cc | 4 +-
src/runtime/contrib/tensorrt/tensorrt_ops.cc | 4 +-
src/runtime/contrib/tensorrt/tensorrt_ops.h | 7 +-
src/support/scalars.h | 4 +-
src/target/llvm/codegen_hexagon.cc | 2 +-
src/target/llvm/codegen_llvm.cc | 2 +-
src/target/source/codegen_cuda.cc | 58 ++++--
src/target/source/literal/cuda_half_t.h | 151 ++++++++++-----
tests/lint/cpplint.sh | 1 -
.../python/codegen/test_target_codegen_cuda_fp8.py | 49 ++++-
tests/python/codegen/test_target_codegen_llvm.py | 43 +----
.../contrib/test_hexagon/test_meta_schedule.py | 138 +-------------
.../contrib/test_hexagon/test_relax_integration.py | 2 +
tests/python/relax/test_from_exported_to_cuda.py | 205 ++++++++++++++++++++-
.../relax/test_frontend_from_exported_program.py | 185 ++++++++++++++++++-
tests/python/relax/test_frontend_from_fx.py | 64 +++++--
.../test_transform_merge_composite_functions.py | 3 -
tests/python/tvmscript/test_tvmscript_roundtrip.py | 3 +-
web/emcc/tvmjs_support.cc | 2 +-
web/emcc/wasm_runtime.cc | 2 +-
86 files changed, 931 insertions(+), 477 deletions(-)
