[Kernel][Misc] Use TORCH_LIBRARY instead of PYBIND11_MODULE for custom ops (#5047)

This commit is contained in:
bnellnm
2024-06-09 16:23:30 -04:00
committed by GitHub
parent 5d7e3d0176
commit 5467ac3196
55 changed files with 833 additions and 451 deletions
+14 -12
View File
@@ -420,12 +420,13 @@ void paged_attention_v1_impl_launcher(
void paged_attention_v1(
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 block_size,
int max_seq_len, const c10::optional<torch::Tensor>& alibi_slopes,
const std::string& kv_cache_dtype, float kv_scale, const int tp_rank,
const int blocksparse_local_blocks, const int blocksparse_vert_stride,
const int blocksparse_block_size, const int blocksparse_head_sliding_step) {
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,
const std::string& kv_cache_dtype, double kv_scale, const int64_t tp_rank,
const int64_t blocksparse_local_blocks,
const int64_t blocksparse_vert_stride, const int64_t blocksparse_block_size,
const int64_t blocksparse_head_sliding_step) {
TORCH_CHECK(kv_scale == 1.0f);
TORCH_CHECK(blocksparse_vert_stride <= 1,
"CPU backend does not support blocksparse attention yet.");
@@ -738,12 +739,13 @@ void paged_attention_v2_impl_launcher(
void paged_attention_v2(
torch::Tensor& out, torch::Tensor& exp_sums, torch::Tensor& max_logits,
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,
const std::string& kv_cache_dtype, float kv_scale, const int tp_rank,
const int blocksparse_local_blocks, const int blocksparse_vert_stride,
const int blocksparse_block_size, const int blocksparse_head_sliding_step) {
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,
const std::string& kv_cache_dtype, double kv_scale, const int64_t tp_rank,
const int64_t blocksparse_local_blocks,
const int64_t blocksparse_vert_stride, const int64_t blocksparse_block_size,
const int64_t blocksparse_head_sliding_step) {
TORCH_CHECK(kv_scale == 1.0f);
TORCH_CHECK(blocksparse_vert_stride <= 1,
"CPU backend does not support blocksparse attention yet.");