minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 13 Packages Projects Releases Wiki Activity

[CPU] Optimize image preprocessor performance for Qwen2VLImageProcessorFast (#15168)

Browse Source
This commit is contained in:
Ma Mingfei
2026-03-18 22:18:15 -07:00
committed by GitHub
parent 62d7454976
commit 687d9eb66f
3 changed files with 402 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
sgl-kernel/csrc/cpu/common.h
+2 -1
View File
@@ -127,6 +127,8 @@ namespace {
#define CHECK_DIM(d, x) TORCH_CHECK(x.dim() == d, #x " must be a " #d "D tensor")
#define CHECK_EQ(a, b) TORCH_CHECK((a) == (b), "CHECK_EQ(" #a ", " #b ") failed. ", a, " vs ", b)
#define CHECK_GT(a, b) TORCH_CHECK((a) > (b), "CHECK_GT(" #a ", " #b ") failed. ", a, " vs ", b)
#define CHECK_GE(a, b) TORCH_CHECK((a) >= (b), "CHECK_GE(" #a ", " #b ") failed. ", a, " vs ", b)
template <bool is_only_lastdim_contiguous>
static inline void CHECK_INPUT_SHAPE_DTYPE(const at::Tensor& tensor, const at::IntArrayRef sizes, at::ScalarType st) {
@@ -138,7 +140,6 @@ static inline void CHECK_INPUT_SHAPE_DTYPE(const at::Tensor& tensor, const at::I
CHECK_INPUT(tensor);
}
}
#define CHECK_GE(a, b) TORCH_CHECK((a) >= (b), "CHECK_GE(" #a ", " #b ") failed. ", a, " vs ", b)
// [NB] Parallel Routines
//