This repository has been archived by the owner on Feb 2, 2024. It is now read-only.
Added bf16 convert proposal #1
Open
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
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
[Note: since this repository is not widely watched, I will be sending a similar message to the TG mailing list.]
A common theme in this task group from the outset has been distinguishing different degrees of BF16 support:
(1) Conversions between BF16 and "native" types (like FP32);
(2) Basic support for BF16 linear algebra, namely FMAs, dot products, and matrix multiplications more generally; and
(3) Comprehensive BF16 arithmetic, either matching existing FP support, or with AI-specific bells and whistles.
At SiFive we've seen a similar classification in interest/requests from our customers, and think it is best to separate the concerns of these use-cases.
This PR shares SiFive's proposal for use-case (1), i.e., conversions. (It does not address the other two use-cases.)
Your feedback is welcome.
Best,
Nick Knight (acting co-chair),
SiFive, Inc.