[DAG] SimplifyDemandedBits - ICMP_SLT(X,0) - only sign mask of X is required

[RKSimon]

Similar to what we did in #163981 for X86ISD::PCMPGT - when we are testing for a negative value, then we only need to demand the signbit

define <4 x i32> @select(<4 x i32> %x, <4 x i32> %y, <4 x i32> %a, <4 x i32> %b) {
  %umax = tail call <4 x i32> @llvm.umax.v4i32(<4 x i32> %x, <4 x i32> %y)
  %cmp = icmp slt <4 x i32> %umax, zeroinitializer
  %res = select <4 x i1> %cmp, <4 x i32> %a, <4 x i32> %b
  ret <4 x i32> %res
}

llvm -mcpu=x86-64-v4

select: # @select
  vpmaxud %xmm1, %xmm0, %xmm0
  vpmovd2m %xmm0, %k1
  vpblendmd %xmm2, %xmm3, %xmm0 {%k1}
  retq

the VPMAXUD is unnecessary - could be:

select: # @select
  vpor %xmm1, %xmm0, %xmm0
  vpmovd2m %xmm0, %k1
  vpblendmd %xmm2, %xmm3, %xmm0 {%k1}
  retq

We already have the ISD::UMAX -> ISD::OR simplification in place but we don't recursively call SimplifyDemandedBits from the ISD::SETCC node

[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)

Similar to what we did in #163981 for X86ISD::PCMPGT - when we are testing for a negative value, then we only need to demand the signbit

define &lt;4 x i32&gt; @<!-- -->select(&lt;4 x i32&gt; %x, &lt;4 x i32&gt; %y, &lt;4 x i32&gt; %a, &lt;4 x i32&gt; %b) {
  %umax = tail call &lt;4 x i32&gt; @<!-- -->llvm.umax.v4i32(&lt;4 x i32&gt; %x, &lt;4 x i32&gt; %y)
  %cmp = icmp sge &lt;4 x i32&gt; %umax, zeroinitializer
  %res = select &lt;4 x i1&gt; %cmp, &lt;4 x i32&gt; %a, &lt;4 x i32&gt; %b
  ret &lt;4 x i32&gt; %res
}

llvm -mcpu=x86-64-v4

select: # @<!-- -->select
  vpmaxud %xmm1, %xmm0, %xmm0
  vpxor %xmm1, %xmm1, %xmm1
  vpcmpnltd %xmm1, %xmm0, %k1
  vpblendmd %xmm2, %xmm3, %xmm0 {%k1}
  retq

the VPMAXUD is unnecessary - could be:

select: # @<!-- -->select
  vpor %xmm1, %xmm0, %xmm0
  vpxor %xmm1, %xmm1, %xmm1
  vpcmpnltd %xmm1, %xmm0, %k1
  vpblendmd %xmm2, %xmm3, %xmm0 {%k1}
  retq

We already have the ISD::UMAX -> ISD::OR simplification in place but we don't recursively call SimplifyDemandedBits from the ISD::SETCC node

[an1k3sh]

I'd like to work on this :)