abrown commented on Issue #1762:
I am also thinking of doing the same for
ExtractLane
--it's a bit too specific--but that can be a separate PR.
github-actions[bot] commented on Issue #1762:
Subscribe to Label Action
cc @bnjbvr
<details>
This issue or pull request has been labeled: "cranelift", "cranelift:meta", "cranelift:wasm"Thus the following users have been cc'd because of the following labels:
- bnjbvr: cranelift
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
iximeow commented on Issue #1762:
I feel it's important to note here that this is a pretty x86-oriented change - do NEON instructions have similar "select-a-lane-of-the-dest-with-an-immediate" instructions? It seems like the alternative is to have different formats for different place
imm8
makes sense in clif IR, which is admittedly not great..
iximeow edited a comment on Issue #1762:
I feel it's important to note here that this is a pretty x86-oriented change - ~do NEON instructions have similar "select-a-lane-of-the-dest-with-an-immediate" instructions?~ It seems like the alternative is to have different formats for different place
imm8
makes sense in clif IR, which is admittedly not great..edit: NEON instructions would go through the new backend machinery anyway, so that question isn't so relevant.
I don't want to single-handedly encourage orienting the old backend machinery towards x86, so: this seems like a fair change in considering other
reg, reg, imm
-encoded SIMD instructions but I'd like a second :heavy_check_mark: on the broader idea of this code really just supporting x86 encoding at this point.
abrown commented on Issue #1762:
Closing this one in favor of #1770 after chatting offline with @iximeow.
Last updated: Dec 23 2024 at 13:07 UTC