- User Since
- Oct 3 2018, 7:41 AM (195 w, 2 d)
Wed, Jun 29
Rework patch to use VLA lowering for the VLS types.
Mon, Jun 27
Fix expansion for VLS on SVE2
Mon, Jun 13
Thu, Jun 9
Remove redundant code
Wed, Jun 8
Tue, Jun 7
Mon, Jun 6
Clean up code based on review comments
Move additional tests to existing file
add tests for vector non-scalar diagnostic
May 26 2022
May 25 2022
I've resubmitted this split from the diagnostic changes as D126380
May 23 2022
Add additional tests and fix bugs associated with new tests
May 12 2022
I think this needs more tests to ensure every diagnostic is being triggered correctly so I will be adding those in another patch, but I wanted to put up the fix for the original issue as soon as possible!
Fix assertion failures
May 10 2022
May 3 2022
Apr 27 2022
Fix crash where condition and splatted result don't have the same width
Apr 21 2022
Add additional tests for vector splats
Apr 20 2022
Apr 14 2022
Apr 7 2022
Apr 5 2022
Mar 31 2022
Not sure what you mean by this; LLVM supports extractelement on <vscale x 16 x i1> vectors. I guess the fact that it's a "vscale x 16" element vector might not be intuitive?
Mar 30 2022
Mar 24 2022
Mar 23 2022
Mar 21 2022
Mar 18 2022
Fix non-valid operation diagnostics
Add correct float tests
Add negative tests for invalid types
Mar 17 2022
When updating the float tests it seems this is causing some issues with fp16 support, which I am investigating.
Mar 16 2022
Fix mistake in unsigned tests
Mar 14 2022
Use ArithConvKind enum instead of boolean
Mar 11 2022
Mar 7 2022
Add tests for in-place operators
Feb 25 2022
Feb 22 2022
Feb 7 2022
Feb 2 2022
Add back special f64 handling for NEON to avoid going through integer register
Feb 1 2022
Rebase on main
Remove special f64 handling which was erroneous and no longer necessary after previous change.
Reorder the operands and flip bit mask.
This changes the mask to select all the bits other than the sign from the first
operand and then only the sign from the second, making this match the operand order for std::copysign which should be more beneficial.
Jan 31 2022
Fix SVE2 lowering. BSL argument ordering is different for SVE2 and NEON, my mistake!
Apologies to reviewers for the confusion above, I believe this patch should now be correct in all the given cases.
Fix operand ordering in all cases
There's currently a bug in the output for scalars here that I am investigating
Fix operand ordering mistake
Jan 28 2022
Add missing SVE2 tests
Jan 27 2022
Jan 26 2022
Use isSignedIntSetCC to check which extend to perform.
Add tests for signed and unsigned comparisons.
Jan 25 2022
When we come around to doing this, it should be done sufficiently differently to this anyway, so abandoning for now.
Fix warnings related to switch statement.