Wasmtime GitHub notifications bot (Aug 20 2020 at 18:12): Wasmtime GitHub notifications bot (Aug 20 2020 at 18:12):jlb6740 commented on Issue #2151:
Still in draft form. Generating some unexpected moves and also haven't checked correctness yet.
Wasmtime GitHub notifications bot (Aug 20 2020 at 18:12):jlb6740 edited a comment on Issue #2151:
Still in draft form. Generating some unexpected moves and also haven't verified correctness yet.
Wasmtime GitHub notifications bot (Aug 20 2020 at 18:31):github-actions[bot] commented on Issue #2151:
Subscribe to Label Action
cc @bnjbvr
<details>
This issue or pull request has been labeled: "cranelift", "cranelift:area:x64"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>
Wasmtime GitHub notifications bot (Aug 21 2020 at 16:06):bnjbvr commented on Issue #2151:
Thanks for the patch! I'm not sure what you expect from the review, since the PR is marked as a draft and you're not sure about it working; the approach of using a sequence of lowered instructions to emulate the addition looks good, if that's what you wanted to validate. An alternative is to extract each vector's lane, do the multiply, and reconstruct the result vector from this (iirc we do this in Spidermonkey; this seems alright since the vectors only have 2 lanes). Otherwise, I'd rather have a look once this has been tested, just to spare some back-and-forth, if that's alright with you :-) Thanks!
Wasmtime GitHub notifications bot (Aug 21 2020 at 17:32):jlb6740 commented on Issue #2151:
@bnjbvr Hi, Yes thanks for the comment! Sorry shouldn't have added folks as reviewers yet, did it to simply let you know the patch was coming where any feedback was a bonus .. so thanks. There are some redundant moves that I need to get rid of and check correctness which I'll get too soon.
Wasmtime GitHub notifications bot (Aug 28 2020 at 20:58):jlb6740 commented on Issue #2151:
Hi ... sorry this took so long. I've confirmed the output is as expected and results seem consistent with the previous backend when running some clif examples through clif-util.
Wasmtime GitHub notifications bot (Sep 09 2020 at 05:36):jlb6740 commented on Issue #2151:
Not sure if this was ready for another review spin, so I did give a look. I have a slightly different suggestion that would avoid one instruction, please let me know if this makes sense!
Hi ... Yes sorry I was going to try to address all your comments to avoid you having to look at this too many times but none-the-less your comments are very helpful. Yes, factoring out the shift makes sense and will do this. Thanks!
Last updated: Nov 22 2024 at 17:03 UTC