[X86] X86FixupInstTunings - attempt to convert VPERMQri to VINSERTI128rri

[RKSimon]

AVX2 target shuffle combining/lowering often performs a 128-bit subvector splat as VPERMQ as this allows memory folding if the subvector had been spilled.

But for most targets if we don't use the VPERMQmi and instead lower to VPERMQri %dst, %src, $0x44 we would have been much better off using VINSERTI128rri %dst, %src, xmm_sub(%src), 0x1.

X86FixupInstTunings should check the scheduler model agrees VINSERTI128rri is faster and adjusts the instruction encoding (not just the opcode but the registers and immediate need checking/adjusting as well).

[llvmbot]

@llvm/issue-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

AVX2 target shuffle combining/lowering often performs a 128-bit subvector splat as VPERMQ as this allows memory folding if the subvector had been spilled.

But for most targets if we don't use the VPERMQmi and instead lower to VPERMQri %dst, %src, $0x44 we would have been much better off using VINSERTI128rri %dst, %src, xmm_sub(%src), 0x1.

X86FixupInstTunings should check the scheduler model agrees VINSERTI128rri is faster and adjusts the instruction encoding (not just the opcode but the registers and immediate need checking/adjusting as well).

[RKSimon]

VPERMPDri as well of course

[llvmbot]

Hi!

This issue may be a good introductory issue for people new to working on LLVM. If you would like to work on this issue, your first steps are:

1. Check that no other contributor is working on this issue. If someone is assigned to the issue or claimed to be working on it, ping the person. After one week without a response, the assignee may be changed.
2. Leave a comment indicating that you are working on the issue, or just create a pull request after following the steps below. Mention this issue in the description of the pull request.
3. Fix the issue locally.
4. Run the test suite locally. Remember that the subdirectories under test/ create fine-grained testing targets, so you can e.g. use make check-clang-ast to only run Clang's AST tests.
5. Create a Git commit.
6. Run git clang-format HEAD~1 to format your changes.
7. Open a pull request to the upstream repository on GitHub. Detailed instructions can be found in GitHub's documentation. Mention this issue in the description of the pull request.

If you have any further questions about this issue, don't hesitate to ask via a comment in the thread below.

[llvmbot]

@llvm/issue-subscribers-good-first-issue

Author: Simon Pilgrim (RKSimon)

AVX2 target shuffle combining/lowering often performs a 128-bit subvector splat as VPERMQ as this allows memory folding if the subvector had been spilled.

But for most targets if we don't use the VPERMQmi and instead lower to VPERMQri %dst, %src, $0x44 we would have been much better off using VINSERTI128rri %dst, %src, xmm_sub(%src), 0x1.

X86FixupInstTunings should check the scheduler model agrees VINSERTI128rri is faster and adjusts the instruction encoding (not just the opcode but the registers and immediate need checking/adjusting as well).

[ckoparkar]

I'd like to give it a go.

[ckoparkar]

But for most targets if we don't use the VPERMQmi and instead lower to VPERMQri %dst, %src, $0x44 we would have been much better off using VINSERTI128rri %dst, %src, xmm_sub(%src), 0x1.

The basic idea is to match on VPERMQri/VPERMPDri and replace it with VINSERTI128rri, but VPERMQ and VINSERT seem like they do slightly different things? What am I missing. Is the 0x44 relevant here?

[RKSimon]

The 0x44 is very relevant as its the only case that we can perform this conversion!