feat: merge dcu branch features

cf6e11c9 · qisan · 3f27f85a · d0436b7b · cf6e11c9 · cf6e11c9
Commit cf6e11c9 authored Feb 05, 2026 by qisan
20 changed files
--- a/examples/warp_specialize/example_warp_specialize_gemm_barrierpipe_stage2.py
+++ b/examples/warp_specialize/example_warp_specialize_gemm_barrierpipe_stage2.py
--- a/examples/warp_specialize/example_warp_specialize_gemm_copy_0_gemm_1.py
+++ b/examples/warp_specialize/example_warp_specialize_gemm_copy_0_gemm_1.py
--- a/examples/warp_specialize/example_warp_specialize_gemm_copy_1_gemm_0.py
+++ b/examples/warp_specialize/example_warp_specialize_gemm_copy_1_gemm_0.py
--- a/examples/warp_specialize/example_warp_specialize_gemm_copy_gemm_0_1.py
+++ b/examples/warp_specialize/example_warp_specialize_gemm_copy_gemm_0_1.py
--- a/examples/warp_specialize/example_warp_specialize_gemm_softpipe_stage2.py
+++ b/examples/warp_specialize/example_warp_specialize_gemm_softpipe_stage2.py
--- a/examples/warp_specialize/test_example_warp_specialize.py
+++ b/examples/warp_specialize/test_example_warp_specialize.py
--- a/format.sh
+++ b/format.sh
--- a/images/MatmulExample.png
+++ b/images/MatmulExample.png
--- a/images/MatmulExample.svg
+++ b/images/MatmulExample.svg
--- a/images/logo-row.svg
+++ b/images/logo-row.svg
--- a/images/mha_performance_h100.png
+++ b/images/mha_performance_h100.png
--- a/images/op_benchmark_a100_wq_gemv.png
+++ b/images/op_benchmark_a100_wq_gemv.png
--- a/images/op_benchmark_consistent_gemm_fp16.png
+++ b/images/op_benchmark_consistent_gemm_fp16.png
--- a/images/op_benchmark_h100.png
+++ b/images/op_benchmark_h100.png
--- a/images/op_benchmark_mi300_fp16_gemm_normalized_latency.png
+++ b/images/op_benchmark_mi300_fp16_gemm_normalized_latency.png
--- a/maint/gemm_v2/correctness_evaluation.py
+++ b/maint/gemm_v2/correctness_evaluation.py
--- a/maint/gemm_v2/correctness_evaluation_sm70.py
+++ b/maint/gemm_v2/correctness_evaluation_sm70.py
--- a/maint/gemm_v2/correctness_evaluation_tcgen05.py
+++ b/maint/gemm_v2/correctness_evaluation_tcgen05.py
--- a/maint/gemm_v2/latency.py
+++ b/maint/gemm_v2/latency.py
--- a/maint/gemm_v2/latency_gemm.py
+++ b/maint/gemm_v2/latency_gemm.py