Skip to content

codegen: better support for half precision floats#324

Merged
sgt0 merged 1 commit into
mainfrom
f16
Jul 1, 2026
Merged

codegen: better support for half precision floats#324
sgt0 merged 1 commit into
mainfrom
f16

Conversation

@sgt0

@sgt0 sgt0 commented Jul 1, 2026

Copy link
Copy Markdown
Owner

I didn't see any native f16<->f32 conversion functions in Cranelift, so the vectorized paths decode/encode the bit patterns inline based on an implementation described here: https://gist.github.com/rygorous/2156668.

I didn't see any native f16<->f32 conversion functions in Cranelift, so
the vectorized paths decode/encode the bit patterns inline based on an
implementation described here: <https://gist.github.com/rygorous/2156668>.
@sgt0 sgt0 enabled auto-merge (rebase) July 1, 2026 02:41
@sgt0 sgt0 merged commit b65ff68 into main Jul 1, 2026
20 checks passed
@sgt0 sgt0 deleted the f16 branch July 1, 2026 02:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant