[FP6-LLM] Port splitK map from DeepSpeed #283
Merged
+145
−2
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/283
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit b89ec10 with merge base 42c2376 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
msaroufim
reviewed
May 27, 2024
| @@ -111,6 +112,143 @@ def from_tc_float6_e3m2(tensor: Tensor, M: int, N: int, dtype: torch.dtype = tor | |||
| return from_float6_e3m2(tensor_fp6, no_bit_packing=True, dtype=dtype) | |||
|
|
|||
|
|
|||
| # https://github.com/microsoft/DeepSpeed/blob/3a3a6db3332e339cc9fd94efd4982f6d60635a3d/deepspeed/inference/v2/kernels/core_ops/cuda_linear/cuda_linear.py | |||
| _SPLIT_K_MAP = [ | |||
| { # tokens: [1, 64] | |||
There was a problem hiding this comment.
n00b q: what is meant by token counts here?
Also can these values be autotuned? I don't necesarily wanna merge in something that's only fast on A100
There was a problem hiding this comment.
I think it's the batch size. Yea, it will be great if we can autotune CUDA kernel also.
msaroufim
approved these changes
May 29, 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Outdated suggestions cannot be applied.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.
https://github.com/microsoft/DeepSpeed/blob/3a3a6db3332e339cc9fd94efd4982f6d60635a3d/deepspeed/inference/v2/kernels/core_ops/cuda_linear/cuda_linear.py
Optimal splitK for a given
(batch_size, out_dim)pair on A100-80G.