Merge tag 'v0.8.2' into v0.8.2-dev

469e903b · zhuwenwen · 389ebcf7 · 25f560a6 · 469e903b · 469e903b
Commit 469e903b authored Mar 28, 2025 by zhuwenwen
20 changed files
--- a/csrc/core/math.hpp
+++ b/csrc/core/math.hpp
@@ -7,8 +7,3 @@ inline constexpr uint32_t next_pow_2(uint32_t const num) {
  if (num <= 1) return num;
  return 1 << (CHAR_BIT * sizeof(num) - __builtin_clz(num - 1));
 }
-
-template <typename T>
-inline constexpr std::enable_if_t<std::is_integral_v<T>, T> ceil_div(T a, T b) {
-  return (a + b - 1) / b;
-}
\ No newline at end of file
--- a/csrc/cpu/attention.cpp
+++ b/csrc/cpu/attention.cpp
--- a/csrc/cpu/cache.cpp
+++ b/csrc/cpu/cache.cpp
--- a/csrc/cpu/cpu_types.hpp
+++ b/csrc/cpu/cpu_types.hpp
--- a/csrc/cpu/cpu_types_arm.hpp
+++ b/csrc/cpu/cpu_types_arm.hpp
--- a/csrc/cpu/cpu_types_vxe.hpp
+++ b/csrc/cpu/cpu_types_vxe.hpp
--- a/csrc/cpu/cpu_types_x86.hpp
+++ b/csrc/cpu/cpu_types_x86.hpp
--- a/csrc/cpu/pos_encoding.cpp
+++ b/csrc/cpu/pos_encoding.cpp
--- a/csrc/cpu/quant.cpp
+++ b/csrc/cpu/quant.cpp
--- a/csrc/cuda_utils.h
+++ b/csrc/cuda_utils.h
--- a/csrc/custom_all_reduce.cu
+++ b/csrc/custom_all_reduce.cu
--- a/csrc/custom_all_reduce.cuh
+++ b/csrc/custom_all_reduce.cuh
--- a/csrc/custom_all_reduce_test.cu
+++ b/csrc/custom_all_reduce_test.cu
--- a/csrc/cutlass_extensions/epilogue/scaled_mm_epilogues_c2x.hpp
+++ b/csrc/cutlass_extensions/epilogue/scaled_mm_epilogues_c2x.hpp
--- a/csrc/cutlass_extensions/epilogue/scaled_mm_epilogues_c3x.hpp
+++ b/csrc/cutlass_extensions/epilogue/scaled_mm_epilogues_c3x.hpp
--- a/csrc/cutlass_extensions/gemm/collective/sm90_mma_tma_gmma_ss_warpspecialized_fp8_blockwise_scaling.hpp
+++ b/csrc/cutlass_extensions/gemm/collective/sm90_mma_tma_gmma_ss_warpspecialized_fp8_blockwise_scaling.hpp
--- a/csrc/cutlass_extensions/vllm_cutlass_library_extension.py
+++ b/csrc/cutlass_extensions/vllm_cutlass_library_extension.py
--- a/csrc/dispatch_utils.h
+++ b/csrc/dispatch_utils.h
--- a/csrc/layernorm_quant_kernels.cu
+++ b/csrc/layernorm_quant_kernels.cu
--- a/csrc/moe/moe_ops.h
+++ b/csrc/moe/moe_ops.h