[WIP][Feature] Support KV Partition for BatchPrefill kernel for Paged & Ragged KV-Cache. #75
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
yzh119
force-pushed
the
batch-prefill-partition-kv
branch
from
551858a to
bf6e4dc
Compare
3 tasks
|
@yzh119 is this PR good to use? This would be extremely useful for some of my work. |
|
@AgrawalAmey We did a huge amount of code refactor since the last commit of this PR, so I need to rebase and add some new commits, please stay tuned :) |
|
@yzh119 looking forward to it! I would be happy to help accelerate this, please let me know if I can help in any way. |
|
Looking forward to it!! |
|
@yzh119 Typing to ask if this is ready for use? I just find |
|
Moved to #310 |
|
@chenzhuofu @ZSL98 @AgrawalAmey |
|
Amazing, thanks a lot for the awesome work! 🙏 |
yzh119
added a commit
that referenced
this pull request
Jun 20, 2024
Duplicate of #75, but re-based on the main branch. Note that to support CUDAGraph, we cannot make `kv_chunk_size` a function argument, which will be passed by value, and cannot change once captured by CUDAGraph. Instead, we pass `kv_chunk_size` through a `kv_chunk_size_ptr` which is a pointer to a global memory address that stores the `kv_chunk_size`, its value can be set in `BeginForward` fuctions.
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Outdated suggestions cannot be applied.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.
Before this PR, FlashInfer supports KV sequence parallelism for single decode/prefill and batch decode, but not batch prefill, however, this feature is also important for batch prefill kernel. This PR implements KV partition for batch prefill kernels (on both Paged & Ragged KV-Cache).