[KVCache] INT8 KV cache implementation and related changes #320
Conversation
This file contains hidden or 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
…feature/model_kvcache_dt
…feature/model_kvcache_dt
…feature/model_kvcache_dt
abenmao
reviewed
Apr 19, 2024
src/layers/attention.h
Outdated
|
|
||
| xft::copy(dstK, srcK, headSize); | ||
| xft::copy(dstV, srcV, headSize); | ||
| // Suppose dstK and dstV are the same data type |
Contributor
There was a problem hiding this comment.
Since there are type conversions in storeKVCache. The old comments can be removed.
Contributor
|
If it is convenient, maybe you can temporarily enable int8_t in type_selector.h to run ci test |
Contributor
Author
currently no interface. should be more convenient after adding interface. |
abenmao
approved these changes
Apr 22, 2024
changqi1
reviewed
Apr 22, 2024
| // ldq: leading dimension of query; lds: LD of score | ||
| template <typename T1, typename T2, typename T3> | ||
| void gemm1(T1 *query, T2 *key, T3 *score, int M, int N, int headSize, int ldq, int ldk, int lds) { | ||
| void gemm1(T1 *query, const std::tuple<T2 *, int, float *> &keyMat, T3 *score, int M, int N, int headSize, int ldq, |
| // score: M * K(keyLen), value: K * headSize, output: M * headSize | ||
| template <typename T1, typename T2, typename T3> | ||
| void gemm2(T1 *score, T2 *value, T3 *output, int M, int headSize, int K, int lds, int ldv, int ldo) { | ||
| void gemm2(T1 *score, const std::tuple<T2 *, int, float *> &valueMat, T3 *output, int M, int headSize, int K, |
Contributor
Author
There was a problem hiding this comment.
will move to kernel in the progress, will get a better name then.
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Outdated suggestions cannot be applied.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.
Main changes: