Skip to content

【Hackathon 10th Spring No.53】[Feature][KVCache] Support head-wise SWA cache recycle in ResourceManagerV1 [cf] #7717

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
bob-cloudforge wants to merge 6 commits into
base: develop
Choose a base branch
from
+1,795 −34
Open

【Hackathon 10th Spring No.53】[Feature][KVCache] Support head-wise SWA cache recycle in ResourceManagerV1 [cf] #7717

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
18ffd94
feat(kvcache): support head-wise SWA recycle
bob-cloudforge May 4, 2026
ddb08f1
fix(kvcache): report head-wise free_gpu_block_num as logical blocks (…
bob-cloudforge May 6, 2026
67e50c0
fix(append_attn): document SWA-sentinel block_id==-1 guard contract (…
bob-cloudforge May 6, 2026
aebcd96
fix(kvcache): use float div in available_gpu_resource
bob-cloudforge May 6, 2026
263db00
fix(t53): mirror available_gpu_resource defensive form from PR2
bob-cloudforge May 6, 2026
584e033
fix(t53): available_gpu_resource use is-not-None guard, drop dead leg…
bob-cloudforge May 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions benchmarks/yaml/eb45-21b-a3b-32k-bf16-kv50-512s.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,10 @@
# T53 bench workload — KV-bound (not slot-bound); gate: FD_HEAD_WISE_KV_CACHE=1
# max_num_seqs raised to 512 so the KV pool, not the slot count, is the bottleneck.
# kv_cache_ratio lowered to 0.50 to shrink the pool and accelerate KV pressure.
# Use with: INPUT_LEN=8192 OUTPUT_LEN=4096 REQUEST_RATE=8
#
max_model_len: 32768
max_num_seqs: 512
kv_cache_ratio: 0.50
tensor_parallel_size: 1
max_num_batched_tokens: 32768
20 changes: 18 additions & 2 deletions custom_ops/gpu_ops/append_attention.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,7 @@ void AppendAttentionKernel(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_tables,
const paddle::optional<paddle::Tensor>& block_tables_headwise,
const paddle::Tensor& encoder_batch_ids,
const paddle::Tensor& encoder_tile_ids_per_batch,
const paddle::Tensor& encoder_num_blocks,
Expand Down Expand Up @@ -155,6 +156,7 @@ void AppendAttentionKernel(
batch_id_per_token,
cu_seqlens_q,
block_tables,
block_tables_headwise,
lambda_batch_ids,
lambda_tile_ids_per_batch,
cache_quant_type_str,
Expand Down Expand Up @@ -488,6 +490,9 @@ std::vector<paddle::Tensor> AppendAttention(
const paddle::optional<paddle::Tensor>& q_norm_weight,
const paddle::optional<paddle::Tensor>& k_norm_weight,
const paddle::optional<paddle::Tensor>& sinks,
const paddle::optional<paddle::Tensor>&
block_tables_headwise, // logical 3D, physical rank-2 [max_num_seqs *
// local_kv_heads, max_blocks_per_head]
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -595,6 +600,7 @@ std::vector<paddle::Tensor> AppendAttention(
batch_id_per_token,
cu_seqlens_q,
block_tables,
block_tables_headwise,
encoder_batch_ids,
encoder_tile_ids_per_batch,
encoder_num_blocks,
Expand Down Expand Up @@ -700,6 +706,9 @@ std::vector<paddle::Tensor> AppendAttentionWithOutput(
const paddle::optional<paddle::Tensor>& q_norm_weight,
const paddle::optional<paddle::Tensor>& k_norm_weight,
const paddle::optional<paddle::Tensor>& sinks,
const paddle::optional<paddle::Tensor>&
block_tables_headwise, // logical 3D, physical rank-2 [max_num_seqs *
// local_kv_heads, max_blocks_per_head]
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -753,6 +762,7 @@ std::vector<paddle::Tensor> AppendAttentionWithOutput(
batch_id_per_token,
cu_seqlens_q,
block_tables,
block_tables_headwise,
encoder_batch_ids,
encoder_tile_ids_per_batch,
encoder_num_blocks,
Expand Down Expand Up @@ -871,6 +881,7 @@ std::vector<std::vector<int64_t>> AppendAttentionInferShape(
const paddle::optional<std::vector<int64_t>>& q_norm_weight_shape,
const paddle::optional<std::vector<int64_t>>& k_norm_weight_shape,
const paddle::optional<std::vector<int64_t>>& sinks_shape,
const paddle::optional<std::vector<int64_t>>& block_tables_headwise_shape,
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -937,6 +948,7 @@ std::vector<paddle::DataType> AppendAttentionInferDtype(
const paddle::optional<paddle::DataType>& q_norm_weight_dtype,
const paddle::optional<paddle::DataType>& k_norm_weight_dtype,
const paddle::optional<paddle::DataType>& sinks_dtype,
const paddle::optional<paddle::DataType>& block_tables_headwise_dtype,
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -1024,6 +1036,7 @@ std::vector<std::vector<int64_t>> AppendAttentionWithOutputInferShape(
const paddle::optional<std::vector<int64_t>>& q_norm_weight_shape,
const paddle::optional<std::vector<int64_t>>& k_norm_weight_shape,
const paddle::optional<std::vector<int64_t>>& sinks_shape,
const paddle::optional<std::vector<int64_t>>& block_tables_headwise_shape,
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -1083,6 +1096,7 @@ std::vector<paddle::DataType> AppendAttentionWithOutputInferDtype(
const paddle::optional<paddle::DataType>& q_norm_weight_dtype,
const paddle::optional<paddle::DataType>& k_norm_weight_dtype,
const paddle::optional<paddle::DataType>& sinks_dtype,
const paddle::optional<paddle::DataType>& block_tables_headwise_dtype,
const float rms_norm_eps,
const std::string& compute_dtype,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -1140,7 +1154,8 @@ PD_BUILD_STATIC_OP(append_attention)
paddle::Optional("kv_signal_data"),
paddle::Optional("q_norm_weight"),
paddle::Optional("k_norm_weight"),
paddle::Optional("sinks")})
paddle::Optional("sinks"),
paddle::Optional("block_tables_headwise")})
.Outputs({"fmha_out"})
.Attrs({
"rms_norm_eps: float",
Expand Down Expand Up @@ -1203,7 +1218,8 @@ PD_BUILD_STATIC_OP(append_attention_with_output)
paddle::Optional("kv_signal_data"),
paddle::Optional("q_norm_weight"),
paddle::Optional("k_norm_weight"),
paddle::Optional("sinks")})
paddle::Optional("sinks"),
paddle::Optional("block_tables_headwise")})
.Outputs({"fmha_out_out"})
.SetInplaceMap({{"fmha_out", "fmha_out_out"}})
.Attrs({
Expand Down
8 changes: 8 additions & 0 deletions custom_ops/gpu_ops/append_attn/append_attention_c16_impl.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,6 +44,7 @@ void CascadeAppendAttentionC16Kernel(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -109,6 +110,7 @@ void CascadeAppendAttentionC16Kernel(
batch_id_per_token,
cu_seqlens_q,
block_table,
block_table_headwise,
batch_ids,
tile_ids_per_batch,
num_blocks,
Expand Down Expand Up @@ -156,6 +158,7 @@ CascadeAppendAttentionC16Kernel<paddle::bfloat16, paddle::bfloat16>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -204,6 +207,7 @@ CascadeAppendAttentionC16Kernel<paddle::bfloat16, paddle::float8_e4m3fn>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -251,6 +255,7 @@ template void CascadeAppendAttentionC16Kernel<paddle::bfloat16, int8_t>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -298,6 +303,7 @@ template void CascadeAppendAttentionC16Kernel<paddle::float16, paddle::float16>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -346,6 +352,7 @@ CascadeAppendAttentionC16Kernel<paddle::float16, paddle::float8_e4m3fn>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down Expand Up @@ -393,6 +400,7 @@ template void CascadeAppendAttentionC16Kernel<paddle::float16, int8_t>(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const int num_blocks,
Expand Down
2 changes: 2 additions & 0 deletions custom_ops/gpu_ops/append_attn/append_attention_kernel.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,6 +47,7 @@ void CascadeAppendAttentionKernel(
const paddle::Tensor& batch_id_per_token,
const paddle::Tensor& cu_seqlens_q,
const paddle::Tensor& block_table,
const paddle::optional<paddle::Tensor>& block_table_headwise,
const paddle::Tensor& batch_ids,
const paddle::Tensor& tile_ids_per_batch,
const std::string& cache_quant_type_str,
Expand Down Expand Up @@ -86,6 +87,7 @@ void CascadeAppendAttentionKernel(
batch_id_per_token,
cu_seqlens_q,
block_table,
block_table_headwise,
batch_ids,
tile_ids_per_batch,
num_blocks,
Expand Down
Loading