This is an automated email from the ASF dual-hosted git repository.
github-actions[bot] pushed a change to branch nightly
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 81e62ecbb2 [TIRx][LLVM] Support scalable Ramp lowering (#19866)
add 120812e9ac [REFACTOR][Relax] Phase out PrimValue and Relax expression
wrappers (#19891)
add 08f7d9b984 [Relax] Use optional dtype for absent Relax dtype fields
(#19890)
add acb65f79c6 [ONNX] Use structural checks for composite frontend tests
(#19880)
add b00b85b30b [Frontend][ONNX] Fix structural tests for TVMScript checks
(#19894)
add 46a9a00c8e [DOCS][TIRX] Add in-kernel profiling (CudaProfiler)
tutorial (#19895)
No new revisions were added by this update.
Summary of changes:
docs/deep_dive/relax/tutorials/relax_creation.py | 2 +-
docs/get_started/tutorials/ir_module.py | 4 +-
docs/how_to/tutorials/cross_compilation_and_rpc.py | 4 +-
docs/how_to/tutorials/optimize_llm.py | 2 +-
docs/reference/api/python/ir.rst | 1 +
docs/reference/api/python/relax/op.rst | 10 +
docs/tirx/native_basics.rst | 1 +
docs/tirx/native_basics/cuda/profiling.rst | 310 +
include/tvm/ir/base_expr.h | 18 +-
include/tvm/ir/expr.h | 57 +-
include/tvm/ir/function.h | 8 +-
include/tvm/ir/module.h | 2 +-
include/tvm/ir/op.h | 14 +-
include/tvm/relax/attrs/create.h | 2 +-
include/tvm/relax/attrs/image.h | 4 +-
include/tvm/relax/attrs/linear_algebra.h | 2 +-
include/tvm/relax/attrs/manipulate.h | 6 +-
include/tvm/relax/attrs/nn.h | 12 +-
include/tvm/relax/attrs/sorting.h | 10 +-
include/tvm/relax/attrs/statistical.h | 2 +-
include/tvm/relax/expr.h | 109 +-
include/tvm/relax/expr_functor.h | 51 +-
include/tvm/relax/type.h | 18 +-
include/tvm/tirx/expr.h | 6 +-
python/tvm/backend/cuda/op.py | 7 +-
python/tvm/ir/__init__.py | 2 +-
python/tvm/ir/expr.py | 28 +-
python/tvm/ir/function.py | 4 +-
python/tvm/ir/module.py | 5 +-
python/tvm/ir/op.py | 4 +-
python/tvm/relax/__init__.py | 2 +-
python/tvm/relax/backend/metal/coreml.py | 5 +-
python/tvm/relax/expr.py | 57 +-
python/tvm/relax/expr_functor.py | 40 +-
python/tvm/relax/frontend/nn/extern.py | 6 +-
python/tvm/relax/frontend/nn/llm/kv_cache.py | 48 +-
python/tvm/relax/frontend/nn/modules.py | 4 +-
python/tvm/relax/frontend/nn/op.py | 12 +-
python/tvm/relax/frontend/onnx/onnx_frontend.py | 61 +-
.../tvm/relax/frontend/tflite/tflite_frontend.py | 12 +-
.../frontend/torch/base_fx_graph_translator.py | 26 +-
.../frontend/torch/exported_program_translator.py | 4 +-
python/tvm/relax/frontend/torch/fx_translator.py | 4 +-
python/tvm/relax/op/base.py | 2 +-
python/tvm/relax/op/builtin/builtin.py | 4 +-
python/tvm/relax/op/create.py | 73 +-
python/tvm/relax/op/manipulate.py | 34 +-
python/tvm/relax/op/memory/memory.py | 8 +-
python/tvm/relax/op/memory/view.py | 12 +-
python/tvm/relax/op/set.py | 12 +-
python/tvm/relax/op/vm/vm.py | 8 +-
python/tvm/relax/relax_to_pyfunc_converter.py | 8 +-
python/tvm/relax/script/builder/ir.py | 18 +-
python/tvm/relax/script/parser/parser.py | 27 +-
python/tvm/relax/testing/ast_printer.py | 3 -
.../fold_batch_norm_to_conv2d_for_inference.py | 2 +-
.../tvm/relax/transform/ipc_allreduce_rewrite.py | 2 +-
.../tvm/relax/transform/lazy_transform_params.py | 6 +-
python/tvm/relax/transform/legalize_ops/binary.py | 10 +-
python/tvm/relax/transform/legalize_ops/common.py | 6 +-
python/tvm/relax/transform/legalize_ops/create.py | 18 +-
.../tvm/relax/transform/legalize_ops/datatype.py | 4 +-
python/tvm/relax/transform/legalize_ops/index.py | 6 +-
.../relax/transform/legalize_ops/linear_algebra.py | 2 +-
.../tvm/relax/transform/legalize_ops/manipulate.py | 5 +-
.../relax/transform/lower_gpu_ipc_alloc_storage.py | 2 +-
python/tvm/relax/type.py | 2 +
python/tvm/relax/utils.py | 28 +-
src/arith/canonical_simplify.cc | 12 +-
src/arith/const_fold.h | 4 +-
src/arith/iter_affine_map.cc | 8 +-
src/backend/trn/transform/lower_trainium_layout.cc | 2 +-
src/ir/expr.cc | 7 +-
src/ir/module.cc | 4 +-
src/relax/analysis/type_analysis.cc | 35 +-
src/relax/backend/contrib/codegen_c/codegen_c.h | 2 +-
.../backend/contrib/codegen_json/codegen_json.h | 4 +-
src/relax/backend/contrib/cutlass/codegen.cc | 4 +-
src/relax/backend/contrib/tensorrt/codegen.cc | 11 +-
src/relax/backend/vm/codegen_vm.cc | 15 +-
src/relax/backend/vm/codegen_vm_tir.cc | 10 +-
src/relax/backend/vm/lower_runtime_builtin.cc | 10 +-
src/relax/backend/vm/vm_shape_lower.cc | 60 +-
src/relax/ir/block_builder.cc | 2 +-
src/relax/ir/dataflow_expr_rewriter.cc | 4 +-
src/relax/ir/dataflow_matcher.cc | 2 +-
src/relax/ir/dependent_type.cc | 11 +-
src/relax/ir/emit_te.cc | 2 +-
src/relax/ir/expr.cc | 19 -
src/relax/ir/expr_functor.cc | 21 +-
src/relax/ir/py_expr_functor.cc | 26 +-
src/relax/op/ccl/ccl.cc | 4 +-
src/relax/op/distributed/binary.h | 2 +-
src/relax/op/distributed/distributed.cc | 2 +-
src/relax/op/distributed/linear_algebra.cc | 6 +-
src/relax/op/distributed/nn.cc | 4 +-
src/relax/op/distributed/unary.h | 5 +-
src/relax/op/image/resize.cc | 20 +-
src/relax/op/memory/view.cc | 70 +-
src/relax/op/nn/convolution.cc | 48 +-
src/relax/op/nn/convolution.h | 2 +-
src/relax/op/nn/nn.cc | 18 +-
src/relax/op/op.cc | 58 +-
src/relax/op/op_common.h | 37 +-
src/relax/op/tensor/binary.cc | 5 +-
src/relax/op/tensor/create.cc | 99 +-
src/relax/op/tensor/create.h | 8 +-
src/relax/op/tensor/index.cc | 43 +-
src/relax/op/tensor/inspect.cc | 26 +-
src/relax/op/tensor/linear_algebra.cc | 10 +-
src/relax/op/tensor/manipulate.cc | 86 +-
src/relax/op/tensor/manipulate.h | 6 +-
src/relax/op/tensor/qdq.cc | 24 +-
src/relax/op/tensor/sampling.cc | 10 +-
src/relax/op/tensor/search.cc | 5 +-
src/relax/op/tensor/set.cc | 37 +-
src/relax/op/tensor/set.h | 4 +-
src/relax/op/tensor/sorting.cc | 15 +-
src/relax/op/tensor/sorting.h | 5 +-
src/relax/op/tensor/statistical.cc | 9 +-
src/relax/op/tensor/ternary.cc | 4 +-
src/relax/op/tensor/unary.cc | 12 +-
src/relax/script/printer/dependent_type.cc | 3 +-
src/relax/script/printer/distributed.cc | 5 +-
src/relax/script/printer/expr.cc | 8 -
src/relax/transform/adjust_matmul_order.cc | 12 +-
src/relax/transform/allocate_workspace.cc | 2 +-
src/relax/transform/alter_op_impl.cc | 2 +-
src/relax/transform/call_tir_rewrite.cc | 8 +-
src/relax/transform/combine_parallel_matmul.cc | 3 +-
src/relax/transform/compute_prim_value.cc | 19 +-
src/relax/transform/dataflow_inplace.cc | 4 +-
src/relax/transform/decompose_ops.cc | 8 +-
src/relax/transform/expand_matmul_of_sum.cc | 3 +-
src/relax/transform/fold_constant.cc | 4 +-
src/relax/transform/fuse_ops.cc | 10 +-
src/relax/transform/fuse_tir.cc | 10 +-
src/relax/transform/gradient.cc | 8 +-
src/relax/transform/infer_amp_utils.cc | 8 +-
src/relax/transform/lazy_transform_params.cc | 6 +-
src/relax/transform/lift_transform_params.cc | 2 +-
src/relax/transform/lower_alloc_tensor.cc | 8 +-
src/relax/transform/merge_composite_functions.cc | 2 +-
src/relax/transform/remove_unused_outputs.cc | 3 +-
src/relax/transform/remove_unused_parameters.cc | 2 +-
src/relax/transform/reorder_take_after_matmul.cc | 4 +-
src/relax/transform/rewrite_cuda_graph.cc | 10 +-
.../specialize_primfunc_based_on_callsite.cc | 8 +-
src/relax/transform/static_plan_block_memory.cc | 26 +-
src/relax/transform/to_mixed_precision.cc | 10 +-
src/relax/utils.cc | 10 +-
src/runtime/extra/contrib/tensorrt/tensorrt_ops.cc | 2 +-
src/runtime/vm/builtin.cc | 20 +-
src/s_tir/transform/inject_virtual_thread.cc | 2 +-
src/tirx/ir/expr.cc | 68 +-
src/tirx/ir/expr_functor.cc | 4 +-
src/tirx/op/op.cc | 4 +-
src/tirx/transform/flatten_buffer.cc | 2 +-
src/tirx/transform/lower_tirx_cleanup.cc | 2 +-
.../relax/backend/adreno/test_texture_network.py | 24 +-
...est_distributed_transform_propagate_sharding.py | 136 +-
.../test_distributed_tvmscript_printer.py | 2 +-
tests/python/relax/test_analysis_type_analysis.py | 2 +-
tests/python/relax/test_ast_printer.py | 11 +-
.../relax/test_backend_transform_shape_lower.py | 5 +-
tests/python/relax/test_bind_params.py | 2 +-
tests/python/relax/test_blockbuilder_core.py | 4 +-
tests/python/relax/test_codegen_cutlass.py | 2 +-
tests/python/relax/test_codegen_tensorrt.py | 2 +-
tests/python/relax/test_dataflow_inplace.py | 12 +-
tests/python/relax/test_dataflow_pattern.py | 23 +-
tests/python/relax/test_dataflow_rewriter.py | 4 +-
tests/python/relax/test_expr.py | 52 +-
tests/python/relax/test_expr_functor.py | 16 +-
tests/python/relax/test_frontend_dynamo.py | 2 +-
.../relax/test_frontend_from_exported_program.py | 23 +-
tests/python/relax/test_frontend_from_fx.py | 21 +-
tests/python/relax/test_frontend_nn_exporter.py | 2 +-
tests/python/relax/test_frontend_nn_modules.py | 29 +-
tests/python/relax/test_frontend_nn_op.py | 12 +-
tests/python/relax/test_frontend_onnx.py | 8949 +++++++++++++++-----
tests/python/relax/test_frontend_tflite.py | 93 +-
tests/python/relax/test_op_ccl.py | 20 +-
tests/python/relax/test_op_create.py | 66 +-
tests/python/relax/test_op_index.py | 120 +-
tests/python/relax/test_op_manipulate.py | 23 +-
tests/python/relax/test_op_qdq.py | 28 +
tests/python/relax/test_op_search.py | 2 +-
tests/python/relax/test_op_view.py | 12 +
tests/python/relax/test_relax_operators.py | 2 +-
.../python/relax/test_relax_to_pyfunc_converter.py | 4 +-
tests/python/relax/test_runtime_builtin.py | 6 +-
tests/python/relax/test_training_loss.py | 6 +-
.../relax/test_transform_adjust_matmul_order.py | 104 +-
.../relax/test_transform_canonicalize_bindings.py | 6 +-
tests/python/relax/test_transform_codegen_pass.py | 2 +-
.../test_transform_combine_parallel_matmul.py | 8 +-
.../python/relax/test_transform_convert_layout.py | 2 -
.../relax/test_transform_fuse_ops_by_pattern.py | 2 +-
tests/python/relax/test_transform_gradient.py | 10 +-
.../relax/test_transform_lazy_transform_params.py | 2 +-
.../relax/test_transform_lift_transform_params.py | 16 +-
.../test_transform_merge_composite_functions.py | 2 +-
.../relax/test_transform_to_mixed_precision.py | 14 +
tests/python/relax/test_tvmscript_parser.py | 30 +-
tests/python/relax/test_tvmscript_printer_relax.py | 29 +-
tests/python/relax/test_utils.py | 1 -
tests/python/relax/test_vm_build.py | 4 +-
tests/python/relax/test_vm_codegen_only.py | 4 +-
209 files changed, 9009 insertions(+), 3503 deletions(-)
create mode 100644 docs/tirx/native_basics/cuda/profiling.rst
