================ @@ -3,6 +3,12 @@ ; RUN: -vplan-print-after=scalarizeMemOpsWithIrregularTypes \ ; RUN: -enable-mem-access-versioning=false 2>&1 | FileCheck %s +; Only interested in interactions with trip counts for the scalable +; vectorization, the rest has shared code path with fixed-width vectors. +; RUN: llvm-extract -S -rfunc=btc < %s | opt -p loop-vectorize -force-vector-width="vscale x 4" -force-target-supports-scalable-vectors -force-target-supports-masked-memory-ops -force-target-supports-gather-scatter-ops -disable-output \ ---------------- lukel97 wrote:
I don't think we need both the LLVM IR + VPlan tests for the scalable cases, either one is enough. My main concern is that we don't really use llvm-extract anywhere else in the llvm/test/transforms directory, and I'm not sure if this test case is special enough to warrant it. In any case splitting the files seems like the easiest way to avoid blocking #182595, we can always revisit the testing structure and format separately https://github.com/llvm/llvm-project/pull/214047 _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
