Skip to content
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

Packed QKV and Rotary Embedding Support for sm<80 GQA #20012

Merged
merged 4 commits into from
Mar 23, 2024
Merged

Packed QKV and Rotary Embedding Support for sm<80 GQA #20012

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
b14651d
rotary memory efficient gqa
aciddelgado Mar 21, 2024
4f5a21e
formatting and naming
aciddelgado Mar 21, 2024
7908046
remove unused print
aciddelgado Mar 22, 2024
284030a
lint fix
aciddelgado Mar 22, 2024
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
23 changes: 20 additions & 3 deletions onnxruntime/contrib_ops/cuda/bert/group_query_attention.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,8 +159,6 @@
!use_flash_attention &&
!disable_memory_efficient_attention_ &&
local_window_size_ == -1 &&
do_rotary_ == false &&
key != nullptr &&
(parameters.head_size & 7) == 0 &&
parameters.sequence_length <= parameters.seqlen_past_kv_cache + parameters.sequence_length &&
(sizeof(T) == 2 || parameters.sequence_length >= attention::kMinSeqLenForMemoryEfficientAttentionFp32) &&
Expand All @@ -172,18 +170,31 @@
if (use_memory_efficient_attention && needs_buff) {
kv_buffer_bytes = (sizeof(T) * parameters.batch_size * parameters.num_heads * parameters.seqlen_present_kv_cache * parameters.head_size);
}
size_t rotary_buffer_bytes = 0;
if (use_memory_efficient_attention && do_rotary_) {
tianleiwu marked this conversation as resolved.
Show resolved Hide resolved
rotary_buffer_bytes = 2 * sizeof(T) * parameters.batch_size * parameters.num_heads * parameters.sequence_length * parameters.head_size;

Check warning on line 175 in onnxruntime/contrib_ops/cuda/bert/group_query_attention.cc

View workflow job for this annotation

GitHub Actions / Lint C++ 

[cpplint] reported by reviewdog 🐶
 Lines should be <= 120 characters long  [whitespace/line_length] [2]

Raw Output:
onnxruntime/contrib_ops/cuda/bert/group_query_attention.cc:175:  Lines should be <= 120 characters long  [whitespace/line_length] [2]
rotary_buffer_bytes += sizeof(int64_t) * parameters.batch_size * parameters.sequence_length;
}
size_t fmha_buffer_bytes = 0;
if (use_memory_efficient_attention && MemoryEfficientAttentionParams::need_workspace(parameters.head_size, sizeof(T) == sizeof(float))) {
fmha_buffer_bytes = (parameters.batch_size * parameters.sequence_length * parameters.num_heads * parameters.head_size * sizeof(float));
}
size_t unpacked_qkv_bytes = 0;
if (use_memory_efficient_attention && parameters.is_packed_qkv) {
unpacked_qkv_bytes = (parameters.batch_size * parameters.sequence_length * (parameters.num_heads + 2 * parameters.kv_num_heads) * parameters.head_size * sizeof(T));

Check warning on line 184 in onnxruntime/contrib_ops/cuda/bert/group_query_attention.cc

View workflow job for this annotation

GitHub Actions / Lint C++ 

[cpplint] reported by reviewdog 🐶
 Lines should be <= 120 characters long  [whitespace/line_length] [2]

Raw Output:
onnxruntime/contrib_ops/cuda/bert/group_query_attention.cc:184:  Lines should be <= 120 characters long  [whitespace/line_length] [2]
}
auto k_buffer = GetScratchBuffer<void>(kv_buffer_bytes, context->GetComputeStream());
auto v_buffer = GetScratchBuffer<void>(kv_buffer_bytes, context->GetComputeStream());
auto rotary_buffer = GetScratchBuffer<void>(rotary_buffer_bytes, context->GetComputeStream());
auto fmha_buffer = GetScratchBuffer<void>(fmha_buffer_bytes, context->GetComputeStream());
auto unpacked_qkv_buffer = GetScratchBuffer<void>(unpacked_qkv_bytes, context->GetComputeStream());
#else
constexpr bool use_memory_efficient_attention = false;
auto k_buffer = GetScratchBuffer<void>(0, context->GetComputeStream());
auto v_buffer = GetScratchBuffer<void>(0, context->GetComputeStream());
auto rotary_buffer = GetScratchBuffer<void>(0, context->GetComputeStream());
auto fmha_buffer = GetScratchBuffer<void>(0, context->GetComputeStream());
auto unpacked_qkv_buffer = GetScratchBuffer<void>(0, context->GetComputeStream());
#endif

// seqlens_k buffer
Expand Down Expand Up @@ -251,7 +262,13 @@
if (fmha_buffer != nullptr) {
data.fmha_buffer = reinterpret_cast<CudaT*>(fmha_buffer.get());
}
// Rotary
if (unpacked_qkv_buffer != nullptr) {
data.unpacked_qkv_buffer = reinterpret_cast<CudaT*>(unpacked_qkv_buffer.get());
}
if (rotary_buffer != nullptr) {
data.rotary_buffer = reinterpret_cast<CudaT*>(rotary_buffer.get());
}
// Rotary Embedding
if (parameters.do_rotary) {
data.cos_cache = reinterpret_cast<const CudaT*>(cos_cache->Data<T>());
data.sin_cache = reinterpret_cast<const CudaT*>(sin_cache->Data<T>());
Expand Down
Loading
Loading