EngineX-Ascend/enginex-ascend-910-vllm
10
0
Fork 1
Code Issues Pull Requests Actions Projects Releases Wiki Activity

init v0.11.0rc0

Browse Source
This commit is contained in:
luopingyi
2025-10-14 10:38:28 +08:00
parent 67afd0ea78
commit 66dc16f966
278 changed files with 28130 additions and 11708 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
csrc/torch_binding_meta.cpp
View File

@@ -40,7 +40,7 @@ std::tuple<at::Tensor, at::Tensor> rotary_embedding_meta(
at::Tensor &positions,
at::Tensor &query,
at::Tensor &key,
int64_t head_size,
int64_t head_size,
at::Tensor &cos_sin_cache,
bool is_neox) {
auto num_tokens = positions.sym_numel();
@@ -86,9 +86,9 @@ at::Tensor sgmv_expand_meta(at::Tensor &x, at::Tensor &weight, at::Tensor &lora_
} // namespace vllm_ascend
namespace {
// Register the meta implementations of the custom kernels for symbolic tracing, this will also
// Register the meta implementations of the custom kernels for symbolic tracing, this will also
// the custom kernel been captured into aclgraph
TORCH_LIBRARY_IMPL_EXPAND(_C, Meta, ops) {
TORCH_LIBRARY_IMPL_EXPAND(CONCAT(_C, _ascend), Meta, ops) {
// Rotary embedding meta implementation
ops.impl("rotary_embedding", &vllm_ascend::meta::rotary_embedding_meta);
// Masked input and mask meta implementation
@@ -99,4 +99,4 @@ namespace {
ops.impl("sgmv_expand", &vllm_ascend::meta::sgmv_expand_meta);
}
}
}