Skip to content

Implement VNNI based flash attention that uses SumOfMulQuadAccumulate #957

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
copybara-service wants to merge 1 commit into
base: dev
Choose a base branch
from
Open

Implement VNNI based flash attention that uses SumOfMulQuadAccumulate #957

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 3 additions & 0 deletions gemma/configs.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -863,6 +863,7 @@ constexpr std::pair<const char*, AttentionImpl> kAttentionImplNameToEnum[] = {
{"flash_transposed_qs", AttentionImpl::kFlashTransposedQs},
{"flash_transposed_qs_bf16", AttentionImpl::kFlashTransposedQsBF16},
{"flash_transposed_qs_int16", AttentionImpl::kFlashTransposedQsInt16},
{"flash_transposed_qs_int8", AttentionImpl::kFlashTransposedQsInt8},
{"flash_matrix_accumulation", AttentionImpl::kFlashMatrixAccumulation},
{"int8_matrix_accumulation", AttentionImpl::kInt8MatrixAccumulation},
};
Expand Down Expand Up @@ -897,6 +898,8 @@ std::string KVEncodingToString(KVEncoding encoding) {
return "Int8";
case KVEncoding::kInt8TwoTranspositions:
return "Int8TwoTranspositions";
case KVEncoding::kInt8VNNITwoTranspositions:
return "Int8VNNITwoTranspositions";
case KVEncoding::kBF16MatrixAccumulation:
return "BF16MatrixAccumulation";
case KVEncoding::kInt8MatrixAccumulation:
Expand Down
2 changes: 2 additions & 0 deletions gemma/configs.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,6 +94,7 @@ enum class KVEncoding {
kInt8TwoTranspositions = 6,
kBF16MatrixAccumulation = 7,
kInt8MatrixAccumulation = 8,
kInt8VNNITwoTranspositions = 9,
};

// Returns a string representation of the KVEncoding.
Expand All @@ -110,6 +111,7 @@ enum class AttentionImpl {
kFlashTransposedQsInt16,
kFlashMatrixAccumulation,
kInt8MatrixAccumulation,
kFlashTransposedQsInt8,
kSentinel,
};

Expand Down
Loading
Loading