1 file changed
+1
-1
lines changed- 3rdparty/mscclpp/include/common.h+107
- 3rdparty/mscclpp/include/msccl.cuh+323
- 3rdparty/mscclpp/include/msccl.h+494
- CMakeLists.txt+10-3
- cmake/config.cmake+5
- cmake/modules/LibInfo.cmake+2
- cmake/modules/contrib/MSCCLPP.cmake+50
- cmake/modules/contrib/Mrvl.cmake+1
- cmake/utils/Utils.cmake+2-5
- docker/Dockerfile.demo_mrvl+28
- docs/how_to/deploy/mrvl.rst+53-36
- include/tvm/relax/expr.h+117-73
- include/tvm/relax/transform.h+4-1
- include/tvm/runtime/memory/memory_manager.h+6-2
- include/tvm/script/ir_builder/tir/ir.h+2-1
- python/tvm/contrib/mrvl.py+172
- python/tvm/contrib/rocm.py+2-1
- python/tvm/dlight/gpu/general_reduction.py+27-8
- python/tvm/relax/backend/dispatch_sort_scan.py+33-24
- python/tvm/relax/transform/transform.py+5
- python/tvm/relax/utils.py+82-48
- python/tvm/relay/op/contrib/mrvl.py+35-45
- python/tvm/relay/op/strategy/arm_cpu.py+12
- python/tvm/script/ir_builder/tir/ir.py+5-2
- python/tvm/topi/arm_cpu/arm_utils.py+98-1
- python/tvm/topi/arm_cpu/conv2d.py+39-4
- python/tvm/topi/arm_cpu/conv2d_gemm.py+59-75
- python/tvm/topi/nn/conv2d.py+6-1
- src/arith/analyzer.cc+3-7
- src/arith/const_int_bound.cc+3
- src/arith/int_set.cc+6
- src/arith/rewrite_simplify.cc+20
- src/arith/scalable_expression.cc+14
- src/arith/scalable_expression.h+13
- src/contrib/msc/core/ir/graph_builder.cc+3-6
- src/contrib/msc/core/transform/set_expr_layout.cc+6-14
- src/relax/analysis/well_formed.cc+12-20
- src/relax/backend/contrib/utils.cc+1-1
- src/relax/ir/dataflow_matcher.cc+23-6
- src/relax/ir/expr.cc+8
- src/relax/op/op_common.h+82-21
- src/relax/op/tensor/binary.cc+86-26
- src/relax/training/utils.cc+2-5
- src/relax/transform/fuse_ops.cc+58-25
- src/relax/transform/fuse_tir.cc+1-3
- src/relax/transform/gradient.cc-2
- src/relax/transform/utils.h+2-1
- src/relay/backend/contrib/mrvl/codegen.cc+137-94
- src/relay/backend/contrib/mrvl/compiler_attr.cc-1
- src/relay/backend/utils.cc+3-1
- src/runtime/contrib/mrvl/mrvl_base64.h+78
- src/runtime/contrib/mrvl/mrvl_runtime.cc+32-6
- src/runtime/contrib/mrvl/mrvl_sw_runtime_lib.cc+175
- src/runtime/contrib/mrvl/mrvl_sw_runtime_lib.h+45
- src/runtime/contrib/mscclpp/allreduce.cu+184
- src/runtime/cuda/cuda_device_api.cc+4-3
- src/runtime/disco/bcast_session.cc+1-1
- src/runtime/relax_vm/builtin.cc+3-6
- src/script/ir_builder/tir/ir.cc+5-5
- src/script/printer/relax/binding.cc+2-2
- src/script/printer/relax/function.cc+1-2
- src/script/printer/relax/tir.cc+4-3
- src/support/libinfo.cc+12-2
- src/te/operation/create_primfunc.cc+8-7
- src/tir/transforms/ir_utils.cc+11-2
- src/tir/transforms/storage_rewrite.cc+7
- src/tir/transforms/vectorize_loop.cc+11-2
- tests/python/arith/test_arith_rewrite_simplify.py+25
- tests/python/codegen/test_target_codegen_aarch64.py+35
- tests/python/contrib/test_mrvl/infrastructure.py+49-1
- tests/python/contrib/test_mrvl/test_mrvl.py+31-18
- tests/python/dlight/test_gpu_general_reduction.py+149
- tests/python/relax/test_backend_dispatch_sort_scan.py+38
- tests/python/relax/test_codegen_cublas.py+2-8
- tests/python/relax/test_codegen_cudnn.py+1-8
- tests/python/relax/test_codegen_cutlass.py+1-8
- tests/python/relax/test_codegen_tensorrt.py+11-2
- tests/python/relax/test_contrib_vllm.py+1-1
- tests/python/relax/test_expr_functor.py+1-1
- tests/python/relax/test_op_binary.py+86-20
- tests/python/relax/test_op_nn_convolution.py+4-4
- tests/python/relax/test_op_search.py+2-2
- tests/python/relax/test_transform_codegen_pass.py+6-4
- tests/python/relax/test_transform_legalize_ops_binary.py+533-1
- tests/python/relax/test_vm_builtin.py+29-2
- tests/python/runtime/test_runtime_device_api.py+52
- tests/python/tir-schedule/test_tir_schedule_split_fuse.py+43-43
- tests/python/tir-transform/test_tir_transform_inject_ptx_async_copy.py+2-2
- tests/python/tir-transform/test_tir_transform_split_host_device.py+23-2
- tests/python/tir-transform/test_tir_transform_vectorize.py+97-55
- tests/python/topi/test_topi_conv2d_nhwc.py+35-12
- tests/python/tvmscript/test_tvmscript_parser_tir.py+15
- tests/scripts/release/PRERELEASE_NOTES.md-24
- tests/scripts/release/make_notes.py+4
- web/emcc/wasm_runtime.cc+17
- web/package-lock.json+2-2
- web/src/runtime.ts+26-6
0 commit comments