remove cum_offsets from get_block_shape_and_split_kv_block (#2913)

* remove padding_offsets from get_padding_offset.cu

* remove padding_offsets from get_padding_offset.cu

* remove padding_offsets from get_padding_offset.cu

* remove cum_offsets from get_block_shape_and_split_kv_block

* remove cum_offsets from get_block_shape_and_split_kv_block

custom_ops/gpu_ops/append_attn/get_block_shape_and_split_kv_block.cu

@@ -194,12 +194,12 @@ get_max_len_kv_ernel(int *max_seq_lens_out, const int *seq_lens_this_time,
 std::vector<paddle::Tensor> GetBlockShapeAndSplitKVBlock(
     const paddle::Tensor &seq_lens_encoder,
     const paddle::Tensor &seq_lens_decoder,
-    const paddle::Tensor &seq_lens_this_time, const paddle::Tensor &cum_offsets,
+    const paddle::Tensor &seq_lens_this_time,
     const int encoder_block_shape_q, const int decoder_block_shape_q,
     const int group_size, const int block_size,
     const int decoder_step_token_num) {
   auto stream = seq_lens_encoder.stream();
-  int bsz = cum_offsets.shape()[0];
+  int bsz = seq_lens_encoder.shape()[0];
   auto max_len_tensor =
       GetEmptyTensor({8}, paddle::DataType::INT32, seq_lens_encoder.place());
   GetMaxLen(seq_lens_decoder, seq_lens_this_time, seq_lens_encoder,
@@ -335,8 +335,7 @@ std::vector<paddle::Tensor> GetBlockShapeAndSplitKVBlock(
 std::vector<paddle::DataType> GetBlockShapeAndSplitKVBlockInferDtype(
     const paddle::DataType &seq_lens_encoder_dtype,
     const paddle::DataType &seq_lens_decoder_dtype,
-    const paddle::DataType &seq_lens_this_time_dtype,
+    const paddle::DataType &seq_lens_this_time_dtype) {
-    const paddle::DataType &cum_offsets_dtype) {
   return {
       paddle::DataType::INT32, paddle::DataType::INT32, paddle::DataType::INT32,
       paddle::DataType::INT32, paddle::DataType::INT32, paddle::DataType::INT32,
@@ -347,8 +346,7 @@ std::vector<paddle::DataType> GetBlockShapeAndSplitKVBlockInferDtype(
 std::vector<std::vector<int64_t>> GetBlockShapeAndSplitKVBlockInferShape(
     const std::vector<int64_t> &seq_lens_encoder_shape,
     const std::vector<int64_t> &seq_lens_decoder_shape,
-    const std::vector<int64_t> &seq_lens_this_time_shape,
+    const std::vector<int64_t> &seq_lens_this_time_shape) {
-    const std::vector<int64_t> &cum_offsets_shape) {
   std::vector<int64_t> dynamic_shape = {-1};
 
   return {dynamic_shape,
@@ -365,8 +363,7 @@ std::vector<std::vector<int64_t>> GetBlockShapeAndSplitKVBlockInferShape(
 }
 
 PD_BUILD_STATIC_OP(get_block_shape_and_split_kv_block)
-    .Inputs({"seq_lens_encoder", "seq_lens_decoder", "seq_lens_this_time",
+    .Inputs({"seq_lens_encoder", "seq_lens_decoder", "seq_lens_this_time"})
-             "cum_offsets"})
     .Outputs({paddle::Optional("encoder_batch_ids"),
               paddle::Optional("encoder_tile_ids_per_batch"),
               paddle::Optional("encoder_num_blocks"),

custom_ops/gpu_ops/cpp_extensions.cc

@@ -234,7 +234,7 @@ paddle::Tensor InitSignalLayerwiseFunc(const paddle::Tensor &kv_signal_metadata,
 std::vector<paddle::Tensor> GetBlockShapeAndSplitKVBlock(
     const paddle::Tensor &seq_lens_encoder,
     const paddle::Tensor &seq_lens_decoder,
-    const paddle::Tensor &seq_lens_this_time, const paddle::Tensor &cum_offsets,
+    const paddle::Tensor &seq_lens_this_time,
     const int encoder_block_shape_q, const int decoder_block_shape_q,
     const int group_size, const int block_size,
     const int decoder_step_token_num);

fastdeploy/model_executor/layers/attention/append_attn_backend.py

@@ -145,7 +145,6 @@ class AppendAttentionBackend(AttentionBackend):
             forward_meta.seq_lens_encoder,
             forward_meta.seq_lens_decoder,
             forward_meta.seq_lens_this_time,
-            forward_meta.cum_offsets,
             metadata.encoder_block_shape_q,
             metadata.decoder_block_shape_q,
             self.num_heads // self.kv_num_heads,

fastdeploy/model_executor/layers/attention/flash_attn_backend.py

@@ -151,7 +151,6 @@ class FlashAttentionBackend(AttentionBackend):
             forward_meta.seq_lens_encoder,
             forward_meta.seq_lens_decoder,
             forward_meta.seq_lens_this_time,
-            forward_meta.cum_offsets,
             metadata.encoder_block_shape_q,
             metadata.decoder_block_shape_q,
             self.num_heads // self.kv_num_heads,

fastdeploy/model_executor/layers/attention/mla_attention_backend.py

@@ -173,7 +173,6 @@ class MLAAttentionBackend(AttentionBackend):
             forward_meta.seq_lens_encoder,
             forward_meta.seq_lens_decoder,
             forward_meta.seq_lens_this_time,
-            forward_meta.cum_offsets,
             metadata.encoder_block_shape_q,
             metadata.decoder_block_shape_q,
             self.num_heads // self.kv_num_heads,

fastdeploy/model_executor/layers/attention/ops/get_block_shape_and_split_kv_block.py

@@ -28,7 +28,6 @@ def get_block_shape_and_split_kv_block(
     seq_lens_encoder: paddle.Tensor,
     seq_lens_decoder: paddle.Tensor,
     seq_lens_this_time: paddle.Tensor,
-    cum_offsets: paddle.Tensor,
     encoder_block_shape_q: int,
     decoder_block_shape_q: int,
     group_size: int,
@@ -55,7 +54,6 @@ def get_block_shape_and_split_kv_block(
             seq_lens_encoder,
             seq_lens_decoder,
             seq_lens_this_time,
-            cum_offsets,
             encoder_block_shape_q,
             decoder_block_shape_q,
             group_size,