en Copyright 2015 Java http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#95f76bff [LV] Create & use VPScalarIVSteps for all scalar users.This patch is a follow-up to D115953. It updates optimizeInductionsto also introduce new VPScalarIVStepsRecipes if an IV has both vectorand scalar uses.It updates all uses that only need scalar values to use the newlycreated recipe for the scalar steps.This completes untangling of VPWidenIntOrFpInductionRecipecode-generation. Now the recipe *only* creates the widened vectorvalues, as it says on the tin.The code to genereate IR has been moved directly toVPWidenIntOrFpInductionRecipe::execute.Note that the recipe has been updated to hold a reference toScalarEvolution, which is needed to expand the step, until we can placethe corresponding SCEV expansion in the pre-header.Depends on D120827.Reviewed By: AyalDifferential Revision: https://reviews.llvm.org/D120828 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Sun, 13 Mar 2022 17:15:24 +0000 Florian Hahn <flo@fhahn.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#e6ad9ef4 [instcombine] Canonicalize constant index type to i64 for extractelement/insertelementThe basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order.I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions.We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause.Differential Revision: https://reviews.llvm.org/D115387 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Tue, 14 Dec 2021 00:29:19 +0000 Philip Reames <listmail@philipreames.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#1a18de3d Autogen a bunch of instcombine and vectorizer testsDone in advance of D115387. These are all the ones which my local script could handle, there's a couple more which need manual updates. List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Mon, 13 Dec 2021 18:07:28 +0000 Philip Reames <listmail@philipreames.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#2a48b699 [IR] In ConstantFoldShuffleVectorInstruction use zeroinitializer for splats of 0When creating a splat of 0 for scalable vectors we tend to create themwith using a combination of shufflevector and insertelement, i.e.shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 0, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)However, for the case of a zero splat we can actually just replace theabove with zeroinitializer instead. This makes the IR a lot simpler andeasier to read. I have changed ConstantFoldShuffleVectorInstruction touse zeroinitializer when creating a splat of integer 0 or FP +0.0 values.Differential Revision: https://reviews.llvm.org/D113394 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Mon, 08 Nov 2021 12:02:21 +0000 David Sherwood <david.sherwood@arm.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#87a294d5 [LoopVectorize] Change getRuntimeVFAsFloat to use unsigned int->FP conversionWe never expect the runtime VF to be negative so we should usethe uitofp instruction instead of sitofp.Differential revision: https://reviews.llvm.org/D112610 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Wed, 27 Oct 2021 10:40:01 +0000 David Sherwood <david.sherwood@arm.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#84394153 [IR] Let ConstantVector::getSplat use poison instead of undefThis patch updates ConstantVector::getSplat to use poison insteadof undef when using insertelement/shufflevector to splat.This follows on from D93793.Differential Revision: https://reviews.llvm.org/D107751 List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Mon, 09 Aug 2021 10:36:41 +0000 David Sherwood <david.sherwood@arm.com> http://172.16.0.5:8080/history/llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll#e7452770 [AArch64] NFC: Make some AArch64-SVE LoopVectorize tests generic.This change moves most of `sve-inductions.ll` to non-AArch64 specificLV tests using the `-target-supports-scalable-vectors` flag, because they'renot explicitly AArch64-specific. One test builds on AArch64-specificknowledge regarding masked loads/stores, and remains in sve-inductions.ll. List of files: /llvm-project-15.0.7/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll Mon, 26 Jul 2021 19:17:22 +0000 Sander de Smalen <sander.desmalen@arm.com>