Merged
Conversation
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 19, 2024
I found the stride calculation for the cublas batched matmul is incorrect when we enable the parallel k optimization. And the wrong strides lead to the mismatching results in Llama inference. It's a bit complex to convert the matmuls in parallel k optimzation to a canonicalized batched matmul in cublas, so I just disabled it. Specifically, after splitting the K dimension, the extent in K dimension for each partition might be unequal, and we have to check if the K dimension is out-of-bound or not. But, actually, we didn't now. In addition, this complicates the conversion to a normal batched matmul in cublas. closes #419 --------- Co-authored-by: Ubuntu <ubuntu@ip-172-31-43-134.us-east-2.compute.internal> Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 20, 2024
I found the stride calculation for the cublas batched matmul is incorrect when we enable the parallel k optimization. And the wrong strides lead to the mismatching results in Llama inference. It's a bit complex to convert the matmuls in parallel k optimzation to a canonicalized batched matmul in cublas, so I just disabled it. Specifically, after splitting the K dimension, the extent in K dimension for each partition might be unequal, and we have to check if the K dimension is out-of-bound or not. But, actually, we didn't now. In addition, this complicates the conversion to a normal batched matmul in cublas. closes #419 --------- Co-authored-by: Ubuntu <ubuntu@ip-172-31-43-134.us-east-2.compute.internal> Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
vadiklyutiy
pushed a commit
that referenced
this pull request
Dec 26, 2024
I found the stride calculation for the cublas batched matmul is incorrect when we enable the parallel k optimization. And the wrong strides lead to the mismatching results in Llama inference. It's a bit complex to convert the matmuls in parallel k optimzation to a canonicalized batched matmul in cublas, so I just disabled it. Specifically, after splitting the K dimension, the extent in K dimension for each partition might be unequal, and we have to check if the K dimension is out-of-bound or not. But, actually, we didn't now. In addition, this complicates the conversion to a normal batched matmul in cublas. closes #419 --------- Co-authored-by: Ubuntu <ubuntu@ip-172-31-43-134.us-east-2.compute.internal> Co-authored-by: xiaocenxiaocen <xiao.zhang@centml.ai>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
No description provided.