LGTM.

LGTM.

LGTM.

In D104247#2827116, @jsji wrote:

In D104247#2827089, @spatel wrote:

In code that I've looked at (mostly C compiled with -ffast-math), we always have contract when we have reassoc, so I don't see much practical difference.
Can you explain more how we could benefit from this change - in the example in the description, we would have 4 instructions rather than 3 if we use FMA - is that better?

Yes, you are right, the performance of having FMA should be better. However, we have quite some scenarios that users care about precision more than performance, they want to precise control of when FMA can be generated. So The major motivation of this is to ensure that we respect the IR semantics. For users that care about performance, we still can get them through default global option or emitting respect flag in IR.

LG

Maybe we can use an MIR test?

LGTM but let's wait one or two days to see opinions from others.

Seems it still causes crash https://bugs.llvm.org/show_bug.cgi?id=50609

We found another crush: https://godbolt.org/z/vcbvddEKr

In D103274#2789960, @lebedev.ri wrote:

@pengfei @craig.topper @RKSimon does anyone feel strongly regarding the avx512 question and the question about not having fast-variable-crosslane-shuffle?

In D103274#2788519, @RKSimon wrote:

I'm happy with this but I do think we should always enable variable shuffles (both types) on all AVX512 targets as well (and maybe land that first to simplify this patch) - @pengfei @craig.topper what do you think?

In D103128#2785257, @davezarzycki wrote:

Sorry for the noise. I reran the bisect. The test in question is still failing, but it turns out to be failing unreliably so git-blame is fairly useless without a reliable test.

LGTM. But let's wait one or more days to see if others have more comments.

LGTM. Thanks!

In D102588#2774515, @MaskRay wrote:

In D102588#2772686, @pengfei wrote:

Updated LCPI regex to include Darwin style LCPI_[0-9]+_[0-9]+ (no
leading dot).

Do we have a test for it?

llvm/test/CodeGen/X86/WidenArith.ll

Updated LCPI regex to include Darwin style LCPI_[0-9]+_[0-9]+ (no
leading dot).

LG. Thanks for fixing it.
(do you want me to land this?)

Thanks @RKSimon , I will investigate it.

LGTM. Thanks for improving it.

I think you can add a NFC in the title.

I see. LGTM, but let's wait one or more days to see if others object it.

I don't find any problem in the tests. So I think it looks good. But I'm not familiar with call lowering, I'd like others to sign off.

In D101338#2726759, @rnk wrote:

Seems reasonable, thanks.

We really ought to solve the general problem one day. It is kind of pig-headed that the backend just picks two arbitrary CSRs, the base and frame pointers, and doesn't bother checking if any instructions clobber those registers.

LGTM.

Yes, it is a pretty implementation. Thanks Craig.

Update for test.

Ping~

Reapply the minor refactor.
Fix MSan build failures.

In D101067#2712921, @hctim wrote:

Looks like this patch broke the MSan buildbots: https://lab.llvm.org/buildbot/#/builders/5/builds/6967/steps/9/logs/stdio

Reproing MSan is a little tricky as it can require a multistage compilation, so you might find the instructions at https://github.com/google/sanitizers/wiki/SanitizerBotReproduceBuild helpful.

I happened to commit this patch with another one. Since Xiang had +1 for it, @LuoYuanke , do you think if I need to revert it or not?

Minor refactor.

Always insert before AMX instruction.

This patch LGTM, but I'd like other sign off.

Reset ShapeBBs[MBB] once we hoist any shapes in MBB.

In D101039#2707914, @LuoYuanke wrote:

Is there any test case for it?

The implementation is good to me in general. Let's wait a few days to see opinions from community and other reviewers.