feat: merge dcu branch features

cf6e11c9 · qisan · 3f27f85a · d0436b7b · cf6e11c9 · cf6e11c9
Commit cf6e11c9 authored Feb 05, 2026 by qisan
20 changed files
--- a/src/tl_templates/hip/debug.h
+++ b/src/tl_templates/hip/debug.h
--- a/src/tl_templates/hip/gemm.h
+++ b/src/tl_templates/hip/gemm.h
--- a/src/tl_templates/hip/hip_fp8.h
+++ b/src/tl_templates/hip/hip_fp8.h
--- a/src/tl_templates/hip/ldsm.h
+++ b/src/tl_templates/hip/ldsm.h
--- a/src/tl_templates/hip/reduce.h
+++ b/src/tl_templates/hip/reduce.h
--- a/src/tl_templates/hip/threadblock_swizzle.h
+++ b/src/tl_templates/hip/threadblock_swizzle.h
--- a/src/transform/align_dynamic_shared_memory_allocations.cc
+++ b/src/transform/align_dynamic_shared_memory_allocations.cc
--- a/src/transform/annotate_device_regions.cc
+++ b/src/transform/annotate_device_regions.cc
--- a/src/transform/annotate_read_only_params.cc
+++ b/src/transform/annotate_read_only_params.cc
--- a/src/transform/annotate_warp_group_reg_alloc.cc
+++ b/src/transform/annotate_warp_group_reg_alloc.cc
--- a/src/transform/arg_binder.cc
+++ b/src/transform/arg_binder.cc
--- a/src/transform/arg_binder.h
+++ b/src/transform/arg_binder.h
--- a/src/transform/atomicadd_vectorize.cc
+++ b/src/transform/atomicadd_vectorize.cc
--- a/src/transform/atomicadd_vectorize.h
+++ b/src/transform/atomicadd_vectorize.h
--- a/src/transform/cluster_planning.cc
+++ b/src/transform/cluster_planning.cc
--- a/src/transform/common/assume.cc
+++ b/src/transform/common/assume.cc
--- a/src/transform/common/assume.h
+++ b/src/transform/common/assume.h
--- a/src/transform/common/attr.h
+++ b/src/transform/common/attr.h
--- a/src/transform/common/collector.h
+++ b/src/transform/common/collector.h
--- a/src/transform/common/loop_fusion_utils.h
+++ b/src/transform/common/loop_fusion_utils.h