biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[MISC] Replace c10::optional with std::optional (#11730)

Browse Source 
Signed-off-by: Lu Fang <lufang@fb.com>
This commit is contained in:
Lu Fang
2025-01-04 17:20:34 -08:00
committed by GitHub
parent 47831430cc
commit 4068f4b5b5
24 changed files with 136 additions and 136 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
csrc/cpu/attention.cpp
View File

@@ -386,7 +386,7 @@ void paged_attention_v1_impl_launcher(
torch::Tensor& out, torch::Tensor& query, torch::Tensor& key_cache,
torch::Tensor& value_cache, int num_kv_heads, float scale,
torch::Tensor& block_tables, torch::Tensor& seq_lens, int max_seq_len,
const c10::optional<torch::Tensor>& alibi_slopes) {
const std::optional<torch::Tensor>& alibi_slopes) {
int num_seqs = query.size(0);
int num_heads = query.size(1);
int head_size = query.size(2);
@@ -459,7 +459,7 @@ void paged_attention_v1(
torch::Tensor& out, torch::Tensor& query, torch::Tensor& key_cache,
torch::Tensor& value_cache, int64_t num_kv_heads, double scale,
torch::Tensor& block_tables, torch::Tensor& seq_lens, int64_t block_size,
int64_t max_seq_len, const c10::optional<torch::Tensor>& alibi_slopes,
int64_t max_seq_len, const std::optional<torch::Tensor>& alibi_slopes,
const std::string& kv_cache_dtype, double k_scale, double v_scale,
const int64_t tp_rank, const int64_t blocksparse_local_blocks,
const int64_t blocksparse_vert_stride, const int64_t blocksparse_block_size,
@@ -702,7 +702,7 @@ void paged_attention_v2_impl_launcher(
torch::Tensor& tmp_out, torch::Tensor& query, torch::Tensor& key_cache,
torch::Tensor& value_cache, int num_kv_heads, float scale,
torch::Tensor& block_tables, torch::Tensor& seq_lens, int block_size,
int max_seq_len, const c10::optional<torch::Tensor>& alibi_slopes) {
int max_seq_len, const std::optional<torch::Tensor>& alibi_slopes) {
int num_seqs = query.size(0);
int num_heads = query.size(1);
int head_size = query.size(2);
@@ -781,7 +781,7 @@ void paged_attention_v2(
torch::Tensor& tmp_out, torch::Tensor& query, torch::Tensor& key_cache,
torch::Tensor& value_cache, int64_t num_kv_heads, double scale,
torch::Tensor& block_tables, torch::Tensor& seq_lens, int64_t block_size,
int64_t max_seq_len, const c10::optional<torch::Tensor>& alibi_slopes,
int64_t max_seq_len, const std::optional<torch::Tensor>& alibi_slopes,
const std::string& kv_cache_dtype, double k_scale, double v_scale,
const int64_t tp_rank, const int64_t blocksparse_local_blocks,
const int64_t blocksparse_vert_stride, const int64_t blocksparse_block_size,

10
csrc/cpu/quant.cpp
View File

@@ -359,7 +359,7 @@ void int8_scaled_mm(torch::Tensor& c, // [M, OC], row-major
const torch::Tensor& b, // [IC, OC], column-major
const torch::Tensor& a_scales, // [1] or [M]
const torch::Tensor& b_scales, // [1] or [OC]
const c10::optional<torch::Tensor>& bias // [OC]
const std::optional<torch::Tensor>& bias // [OC]
) {
CPU_KERNEL_GUARD_IN(cutlass_scaled_mm)
// Checks for conformality
@@ -442,8 +442,8 @@ void int8_scaled_mm_azp(torch::Tensor& c, // [M, OC], row-major
const torch::Tensor& a_scales, // [1] or [M]
const torch::Tensor& b_scales, // [1] or [OC]
const torch::Tensor& azp_adj, // [OC]
const c10::optional<torch::Tensor>& azp, // [1] or [M]
const c10::optional<torch::Tensor>& bias // [OC]
const std::optional<torch::Tensor>& azp, // [1] or [M]
const std::optional<torch::Tensor>& bias // [OC]
) {
CPU_KERNEL_GUARD_IN(cutlass_scaled_mm_azp)
// Checks for conformality
@@ -561,7 +561,7 @@ void int8_scaled_mm_azp(torch::Tensor& c, // [M, OC], row-major
void static_scaled_int8_quant(torch::Tensor& out, // [..., hidden_size]
const torch::Tensor& input, // [..., hidden_size]
const torch::Tensor& scale,
c10::optional<torch::Tensor> const& azp) {
std::optional<torch::Tensor> const& azp) {
CPU_KERNEL_GUARD_IN(static_scaled_int8_quant)
TORCH_CHECK(input.is_contiguous());
TORCH_CHECK(out.is_contiguous());
@@ -590,7 +590,7 @@ void dynamic_scaled_int8_quant(
torch::Tensor& out, // [..., hidden_size]
const torch::Tensor& input, // [..., hidden_size]
torch::Tensor& scale, // [..., 1]
c10::optional<torch::Tensor> const& azp) {
std::optional<torch::Tensor> const& azp) {
CPU_KERNEL_GUARD_IN(dynamic_scaled_int8_quant)
TORCH_CHECK(input.is_contiguous());
TORCH_CHECK(out.is_contiguous());

6
csrc/cpu/torch_bindings.cpp
View File

@@ -9,14 +9,14 @@ std::string init_cpu_threads_env(const std::string& cpu_ids);
void int8_scaled_mm(torch::Tensor& c, const torch::Tensor& a,
const torch::Tensor& b, const torch::Tensor& a_scales,
const torch::Tensor& b_scales,
const c10::optional<torch::Tensor>& bias);
const std::optional<torch::Tensor>& bias);
void int8_scaled_mm_azp(torch::Tensor& c, const torch::Tensor& a,
const torch::Tensor& b, const torch::Tensor& a_scales,
const torch::Tensor& b_scales,
const torch::Tensor& azp_adj,
const c10::optional<torch::Tensor>& azp,
const c10::optional<torch::Tensor>& bias);
const std::optional<torch::Tensor>& azp,
const std::optional<torch::Tensor>& bias);
TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
// vLLM custom ops