[Compiler/Runtime/External Libs] Add flash attention external library & pass through mechanism #99
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
zhekunz2
changed the title
zhekunz2
force-pushed
the
zhekunz/flash_attn
branch
from
b632130
to
4d7dc8d
Compare
zhekunz2
requested review from
liwenchangbdbz,
yaochengji,
qingyunqu and
xrzhang111
liwenchangbdbz
approved these changes
Jan 24, 2024
yaochengji
reviewed
Jan 24, 2024
yaochengji
approved these changes
Jan 24, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR moves runtime flash attention kernels to
external_libs/, supports flash attention kvcache, and uses it in a pass-through fashion by declaringByre::CustomOp.Byre::CustomOphas three main components:StrAttr:$lib_path, which specifies the path of the library file.StrAttr:$api_name, which specifies the symbol name of the library for this custom op.ArrayAttr:$extra_args, which specifies the additional arguments that needs to be passed to the api call.The following changes are made to support pass-through on flash attention op:
external_libs/(adds flash attention kvcache support).Byre::CustomOpsupport.runtime/lib/backends/cuda/providers/default/custom/custom.ccHloToByreCustomPassconversion pass.compiler/lib/Conversion/HloToByreTensor/HloToByreCustom.cpp