Skip to content

[Executor] Fixed the issue of CUDA graph execution failure caused by different branches during decoding #3223

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

Merged
gongshaotian merged 5 commits into from
Aug 8, 2025
Merged

[Executor] Fixed the issue of CUDA graph execution failure caused by different branches during decoding #3223

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
e190ce6
彻底解决解码切块问题
littledgg Aug 5, 2025
80715b1
update C8 and C4 kernel
littledgg Aug 6, 2025
9cd514c
fix problem
littledgg Aug 7, 2025
f7604dc
fix with pre-commit
littledgg Aug 7, 2025
3300038
retain branch for mtp
littledgg Aug 7, 2025
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
323 changes: 135 additions & 188 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 @@ -1061,207 +1061,154 @@ void MultiQueryAppendAttention(
if (!is_decoder) {
chunk_size = static_cast<uint32_t>(encoder_max_partition_size);
}
const int num_chunks = div_up(max_dec_len, chunk_size);

const int num_chunks = div_up(max_seq_len, chunk_size);
dim3 grids(num_blocks_x_cpu, num_chunks, kv_num_heads);
dim3 blocks(32, num_warps);

if (num_chunks <= 1) {
auto nosplit_kv_kernel =
multi_query_append_attention_warp1_4_kernel<NV_TYPE,
false,
GROUP_SIZE,
CAUSAL,
num_warps,
NUM_WARP_Q,
NUM_WARP_KV,
HEAD_DIM,
BLOCK_SIZE,
num_frags_x,
num_frags_z,
num_frags_y,
OUT_NV_TYPE,
ENABLE_PREFILL>;
if (smem_size >= 48 * 1024) {
cudaFuncSetAttribute(nosplit_kv_kernel,
cudaFuncAttributeMaxDynamicSharedMemorySize,
smem_size);
}

nosplit_kv_kernel<<<grids, blocks, smem_size, stream>>>(
reinterpret_cast<NV_TYPE *>(const_cast<T *>(qkv.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_k.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_v.data<T>())),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(smooth_weight.get().data<T>()))
: nullptr,
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
batch_ids.data<int>(),
tile_ids_per_batch.data<int>(),
cu_seqlens_q.data<int>(),
block_table.data<int>(),
max_seq_len,
max_dec_len,
max_block_num_per_seq,
scale,
quant_max_bound,
quant_min_bound,
in_scale,
chunk_size,
nullptr,
nullptr,
nullptr,
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
speculate_max_draft_token_num);
phi::Allocator::AllocationPtr tmp_workspace, tmp_m, tmp_d;
if (is_decoder) {
tmp_workspace = allocator->Allocate(
phi::SizeOf(qkv.dtype()) *
static_cast<size_t>(bsz * num_chunks * num_heads * HEAD_DIM));
tmp_m = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(bsz * num_chunks * num_heads));
tmp_d = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(bsz * num_chunks * num_heads));
} else {
phi::Allocator::AllocationPtr tmp_workspace, tmp_m, tmp_d;
if (is_decoder) {
if (ENABLE_PREFILL) {
tmp_workspace =
allocator->Allocate(phi::SizeOf(qkv.dtype()) *
static_cast<size_t>(token_num * num_chunks *
num_heads * HEAD_DIM));
tmp_m = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(token_num * num_chunks * num_heads));
tmp_d = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(token_num * num_chunks * num_heads));
} else {
tmp_workspace = allocator->Allocate(
phi::SizeOf(qkv.dtype()) *
static_cast<size_t>(bsz * num_chunks * num_heads * HEAD_DIM));
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads * HEAD_DIM));
tmp_m = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(bsz * num_chunks * num_heads));
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads));
tmp_d = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(bsz * num_chunks * num_heads));
} else {
if (ENABLE_PREFILL) {
tmp_workspace =
allocator->Allocate(phi::SizeOf(qkv.dtype()) *
static_cast<size_t>(token_num * num_chunks *
num_heads * HEAD_DIM));
tmp_m = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(token_num * num_chunks * num_heads));
tmp_d = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(token_num * num_chunks * num_heads));
} else {
tmp_workspace = allocator->Allocate(
phi::SizeOf(qkv.dtype()) *
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads * HEAD_DIM));
tmp_m = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads));
tmp_d = allocator->Allocate(
phi::SizeOf(paddle::DataType::FLOAT32) *
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads));
}
}
split_kv_kernel<<<grids, blocks, smem_size, stream>>>(
reinterpret_cast<NV_TYPE *>(const_cast<T *>(qkv.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_k.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_v.data<T>())),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(smooth_weight.get().data<T>()))
: nullptr,
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
batch_ids.data<int>(),
tile_ids_per_batch.data<int>(),
cu_seqlens_q.data<int>(),
block_table.data<int>(),
max_seq_len,
max_dec_len,
max_block_num_per_seq,
scale,
quant_max_bound,
quant_min_bound,
in_scale,
chunk_size,
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
speculate_max_draft_token_num);

// merge
constexpr int vec_size = num_elems_per_128b<NV_TYPE>();
if (is_decoder) {
constexpr int blockx = HEAD_DIM / vec_size;
constexpr int blocky = (128 + blockx - 1) / blockx;
dim3 grids_merge(bsz, num_heads);
dim3 blocks_merge(blockx, blocky);
merge_multi_chunks_decoder_kernel<NV_TYPE,
vec_size,
blocky,
HEAD_DIM,
OUT_NV_TYPE,
ENABLE_PREFILL>
<<<grids_merge, blocks_merge, 0, stream>>>(
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
seq_lens_encoder.data<int>(),
cu_seqlens_q.data<int>(),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(const_cast<T *>(
smooth_weight.get().data<T>()))
: nullptr,
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
quant_max_bound,
quant_min_bound,
in_scale,
max_seq_len,
num_chunks,
num_heads,
chunk_size,
HEAD_DIM);
} else {
constexpr int blockx = HEAD_DIM / vec_size;
constexpr int blocky = (128 + blockx - 1) / blockx;
dim3 grids_merge(min(sm_count * 4, token_num),
num_heads);
dim3 blocks_merge(blockx, blocky);
merge_multi_chunks_v2_kernel<NV_TYPE,
vec_size,
blocky,
HEAD_DIM,
OUT_NV_TYPE,
ENABLE_PREFILL>
<<<grids_merge, blocks_merge, 0, stream>>>(
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
seq_lens_encoder.data<int>(),
batch_id_per_token.data<int>(),
cu_seqlens_q.data<int>(),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(const_cast<T *>(
smooth_weight.get().data<T>()))
: nullptr,
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
quant_max_bound,
quant_min_bound,
in_scale,
max_seq_len,
num_chunks,
num_heads,
chunk_size,
HEAD_DIM,
token_num,
speculate_max_draft_token_num);
static_cast<size_t>(speculate_max_draft_token_num * bsz *
num_chunks * num_heads));
}
}
split_kv_kernel<<<grids, blocks, smem_size, stream>>>(
reinterpret_cast<NV_TYPE *>(const_cast<T *>(qkv.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_k.data<T>())),
reinterpret_cast<NV_TYPE *>(const_cast<T *>(cache_v.data<T>())),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(smooth_weight.get().data<T>()))
: nullptr,
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
batch_ids.data<int>(),
tile_ids_per_batch.data<int>(),
cu_seqlens_q.data<int>(),
block_table.data<int>(),
max_seq_len,
max_dec_len,
max_block_num_per_seq,
scale,
quant_max_bound,
quant_min_bound,
in_scale,
chunk_size,
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
speculate_max_draft_token_num);

// merge
constexpr int vec_size = num_elems_per_128b<NV_TYPE>();
if (is_decoder) {
constexpr int blockx = HEAD_DIM / vec_size;
constexpr int blocky = (128 + blockx - 1) / blockx;
dim3 grids_merge(bsz, num_heads);
dim3 blocks_merge(blockx, blocky);
merge_multi_chunks_decoder_kernel<NV_TYPE,
vec_size,
blocky,
HEAD_DIM,
OUT_NV_TYPE,
ENABLE_PREFILL>
<<<grids_merge, blocks_merge, 0, stream>>>(
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
seq_lens_encoder.data<int>(),
cu_seqlens_q.data<int>(),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(const_cast<T *>(
smooth_weight.get().data<T>()))
: nullptr,
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
quant_max_bound,
quant_min_bound,
in_scale,
max_seq_len,
num_chunks,
num_heads,
chunk_size,
HEAD_DIM);
} else {
constexpr int blockx = HEAD_DIM / vec_size;
constexpr int blocky = (128 + blockx - 1) / blockx;
dim3 grids_merge(min(sm_count * 4, token_num),
num_heads);
dim3 blocks_merge(blockx, blocky);
merge_multi_chunks_v2_kernel<NV_TYPE,
vec_size,
blocky,
HEAD_DIM,
OUT_NV_TYPE,
ENABLE_PREFILL>
<<<grids_merge, blocks_merge, 0, stream>>>(
reinterpret_cast<NV_TYPE *>(tmp_workspace->ptr()),
static_cast<float *>(tmp_m->ptr()),
static_cast<float *>(tmp_d->ptr()),
seq_lens_q.data<int>(),
seq_lens_kv.data<int>(),
seq_lens_encoder.data<int>(),
batch_id_per_token.data<int>(),
cu_seqlens_q.data<int>(),
shift_bias ? reinterpret_cast<NV_TYPE *>(
const_cast<T *>(shift_bias.get().data<T>()))
: nullptr,
smooth_weight ? reinterpret_cast<NV_TYPE *>(const_cast<T *>(
smooth_weight.get().data<T>()))
: nullptr,
reinterpret_cast<OUT_NV_TYPE *>(out->data<OutT>()),
quant_max_bound,
quant_min_bound,
in_scale,
max_seq_len,
num_chunks,
num_heads,
chunk_size,
HEAD_DIM,
token_num,
speculate_max_draft_token_num);
}
}
}

Expand Down
Loading
Loading