Init the main branch for aiter

85762c1a · Xiaowei.zhang · ae0b3521 · 85762c1a · 85762c1a · 85762c1a
Commit 85762c1a authored Apr 15, 2026 by Xiaowei.zhang
20 changed files
--- a/csrc/include/opus/README.md
+++ b/csrc/include/opus/README.md
--- a/csrc/include/opus/logo.png
+++ b/csrc/include/opus/logo.png
--- a/csrc/include/opus/opus.hpp
+++ b/csrc/include/opus/opus.hpp
--- a/csrc/include/pos_encoding.h
+++ b/csrc/include/pos_encoding.h
--- a/csrc/include/py_itfs_common.h
+++ b/csrc/include/py_itfs_common.h
--- a/csrc/include/quant.h
+++ b/csrc/include/quant.h
--- a/csrc/include/quant_common.cuh
+++ b/csrc/include/quant_common.cuh
--- a/csrc/include/quant_utils.cuh
+++ b/csrc/include/quant_utils.cuh
--- a/csrc/include/quick_all_reduce_base.h
+++ b/csrc/include/quick_all_reduce_base.h
--- a/csrc/include/rmsnorm.h
+++ b/csrc/include/rmsnorm.h
--- a/csrc/include/rocm_ops.hpp
+++ b/csrc/include/rocm_ops.hpp
--- a/csrc/include/rope.h
+++ b/csrc/include/rope.h
--- a/csrc/include/smoothquant.h
+++ b/csrc/include/smoothquant.h
--- a/csrc/include/topk_plain.h
+++ b/csrc/include/topk_plain.h
--- a/csrc/include/topk_transform.h
+++ b/csrc/include/topk_transform.h
--- a/csrc/include/torch/mha_batch_prefill.h
+++ b/csrc/include/torch/mha_batch_prefill.h
--- a/csrc/include/torch/mha_bwd.h
+++ b/csrc/include/torch/mha_bwd.h
--- a/csrc/include/torch/mha_fwd.h
+++ b/csrc/include/torch/mha_fwd.h
--- a/csrc/include/torch/mha_v3_bwd.h
+++ b/csrc/include/torch/mha_v3_bwd.h
--- a/csrc/include/torch/mha_v3_fwd.h
+++ b/csrc/include/torch/mha_v3_fwd.h