[llvm] [LoopVectorize] Generate wide active lane masks (PR #147535)
Kerry McLaughlin via llvm-commits
llvm-commits at lists.llvm.org
Thu Jul 10 07:29:08 PDT 2025
================
@@ -0,0 +1,121 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mattr=+sve < %s | FileCheck %s -check-prefix CHECK-SVE
+; RUN: llc -mattr=+sve2p1 < %s | FileCheck %s -check-prefix CHECK-SVE2p1
+
+target triple = "aarch64-unknown-linux"
+
+define void @scalable_wide_active_lane_mask(ptr %dst, ptr %src, i64 %n) #0 {
----------------
kmclaughlin-arm wrote:
I added this using the output of one of the loops in `test/Transforms/LoopVectorize/AArch64/sve-wide-lane-mask.ll`, just to test that it is actually able make use of the SVE2.1 whilelo pair instruction.
I've removed this test from the PR now though. The reason is that I don't think this is adding anything new compared to the existing tests in `test/CodeGen/AArch64/get-active-lane-mask-extract.ll`, which contains tests of the get.active.lane.mask intrinsic followed by a sequence of extracts in the same pattern.
https://github.com/llvm/llvm-project/pull/147535
More information about the llvm-commits
mailing list