[Dev][TL] Merge Hopper and Pipeline Modifications (#213)

* Merge TL Update * submodule update * cutlass submodule update * Update submodule commit reference
microsoft · Oct 6, 2024 · 314b2a1 · 314b2a1
1 parent 988e782
commit 314b2a1
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 6 deletions.
diff --git a/.gitmodules b/.gitmodules
@@ -1,8 +1,8 @@
 [submodule "3rdparty/tvm"]
 	path = 3rdparty/tvm
-	url = https://github.com/LeiWang1999/tvm
-	branch = bitblas_tl
+	url = https://github.com/TileLang/tvm.git
+	branch = tilelang
 [submodule "3rdparty/cutlass"]
 	path = 3rdparty/cutlass
-	url = https://github.com/NVIDIA/cutlass.git
-	branch = v3.2.2
+	url = https://github.com/TileLang/cutlass
+	branch = tldev
diff --git a/3rdparty/cutlass b/3rdparty/cutlass
diff --git a/3rdparty/tvm b/3rdparty/tvm
+24 −0		include/cute/arch/mma_sm90.hpp
+766 −0		include/cute/arch/mma_sm90_gmma.hpp
+221 −0		include/cute/atom/mma_traits_sm90_gmma.hpp
+3 −0		.gitignore
+8 −0		include/tvm/tir/transform.h
+4 −1		python/tvm/contrib/nvcc.py
+16 −0		python/tvm/tir/transform/transform.py
+37 −10		python/tvm/tl/autotuner.py
+15 −3		python/tvm/tl/engine.py
+22 −2		python/tvm/tl/language.py
+33 −0		python/tvm/tl/transform.py
+16 −2		python/tvm/tl/utils.py
+10 −9		src/driver/driver_api.cc
+8 −0		src/tir/transforms/lower_opaque_block.cc
+381 −0		src/tir/transforms/thread_partial_sync.cc
+4 −3		src/tir/transforms/thread_storage_sync.cc
+13 −1		src/tl/ir.cc
+4 −4		src/tl/layout/gemm_layouts.cc
+8 −0		src/tl/op/builtin.cc
+16 −0		src/tl/op/builtin.h
+5 −5		src/tl/op/bulk_copy.cc
+10 −1		src/tl/op/gemm.cc
+8 −0		src/tl/op/reduce.cc
+1 −0		src/tl/op/reduce.h
+11 −2		src/tl/target/codegen.cc
+10 −0		src/tl/tl_templates/copy_sm90.h
+59 −19		src/tl/tl_templates/gemm_sm90.h
+2 −0		src/tl/tl_templates/reduce.h
+170 −0		src/tl/transform/inject_fence_proxy.cc
+242 −0		src/tl/transform/inject_mbarrier.cc
+1 −1		src/tl/transform/lower_hopper_intrin.cc
+321 −0		src/tl/transform/multi_version_buffer_rewriter.cc
+1 −1		src/tl/transform/pipeline_planning.cc
+849 −842		src/tl/transform/warp_specialized_pipeline.cc
+939 −0		src/tl/transform/warp_specialized_rewriter.cc
+81 −0		testing/mamba_triton.py
+233 −0		tl_pipeline/dp.py
+202 −0		tl_pipeline/gen_configs.py
+180 −0		tl_pipeline/generate_plan.py
+67 −0		tl_pipeline/graph.py
+29 −0		tl_pipeline/main.py
+134 −0		tl_pipeline/pipeline_transform.py
+142 −0		tl_scripts/dequant_gemm.py
+2 −3		tl_scripts/gemm_example.py
+51 −0		tl_scripts/gemm_hopper.py
+71 −0		tl_scripts/gemm_rs_example.py
+1,083 −80		tl_scripts/mamba_example.py
+4 −5		tl_scripts/mha_example.py
+191 −0		tl_scripts/mha_pipeline.py
+202 −0		tl_scripts/mha_pipeline_search.py
+0 −0		tl_scripts/profile_workloads.py
+88 −66		tl_scripts/retnet_example.py
+0 −54		tl_scripts/test.py
+254 −0		tl_scripts/torch_ref.py