Add avx512 permute and pmaddbw

[folkertdev]

Turns out we need a couple more, but the implementation can be shared with existing versions.

[rustbot]

Thank you for contributing to Miri! A reviewer will take a look at your PR, typically within a week or two.
Please remember to not force-push to the PR branch except when you need to rebase due to a conflict or when the reviewer asks you for it.

[RalfJung]

Thanks! I can't say I can make sense of this maddubs test (it looks like lots of random numbers), but the code is the same as before so 🤷

I assume you ran the tests on real hardware?

View changes since this review

[folkertdev]

I can't say I can make sense of this maddubs test (it looks like lots of random numbers), but the code is the same as before so 🤷

The numbers attempt to exercise some of the interesting behavior with the saturation. But yeah it's tough.

I assume you ran the tests on real hardware?

Yes

@rustbot ready

[folkertdev]

Oh btw has your opinion on force pushes changed now that we have the improved diffs? Or should I just squash all changes into one commit anyway?

[RalfJung]

Yeah I guess we can be more relaxed about force-pushes now, though that does require some discipline to always submit a comment with the review so that we get the diff links.

This one is ready to have the fixup commit squashed in whatever way you think makes sense (the one-commit ./miri squash automation is just there to make it easier for people). Thanks!

@rustbot author

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[folkertdev]

I absorbed the changes into the earlier commits

@rustbot ready

resolved