similar to: [RFC] I'd like to move RecurrenceDescriptor and LoopDescriptor in LoopUitls.h/.cpp from Transform/Utils tree to Analysis/Utils tree

Adding support for FP inductions through isInductionPHI() is certainly possible, I have a relatively small local patch that does exactly that for simple fp add-recurrence cases, along with changes to the vectorizer to make it aware of FP inductions. It won't get give you the powerful reasoning capabilities of SCEV, but for the B-like cases it should work. Amara On 20 May 2016 at 19:31, Adam

Hi Hideki, I like this summary overall, thanks. More below. > On May 20, 2016, at 10:04 AM, Saito, Hideki <hideki.saito at intel.com> wrote: > > > To the best of my experience, handling case B (secondary induction) is must-have, and if I’m not mistaken, > people aren’t opposed to that. > > For me, handling case A (primary induction) is “why not?”, but I certainly

What LoopVectorize.cpp has are the following. Each function may have to have a separate consolidation discussion. I'm bringing up getpointerOperand() since I actually found multiple instances defined/used. DependenceAnalysis.cpp has isLoadOrStore(). LoopAccessAnalysis.cpp has getAddressSpaceOperand(). I'm sure there are others that might be worth discussing within this thread or a follow

To the best of my experience, handling case B (secondary induction) is must-have, and if I’m not mistaken, people aren’t opposed to that. For me, handling case A (primary induction) is “why not?”, but I certainly admit that that can be very naïve thinking coming from lack of good understanding on SCEV and their proper usages. Now, let’s assume we can postpone discussion about case A. What is the

Hi all, During the Nov 2016 dev meeting, we had a hackers’ lab session where we discussed some issues about loop idiom recognition, IR representation and cost modelling. I took an action to write up an RFC about introducing reduction intrinsics to LLVM to represent horizontal operations across vectors. Vector reductions have been discussed in the past before, notably here:

Hi, TL;DR: A loop doesn't get vectorized due to the interaction of loop- rotate, licm and instcombine. What to do about it? Full story: In the benchmarks for our out-of-tree target we have a case that we would like to get vectorized, but currently it isn't. I've done some digging to see why and have some kind of idea what prevents it, but I don't know what the best way to fix

Thanks, Hal. I plan to post a patch w/o HW Legality early bailout first. That should enable further discussion on where the initial very high cost for "illegal masked load/store/gather/scatter" should be coming from --- like should LoopVectorize provide it? Or should it be provided by TTI? I prefer the latter (TTI) but the first revision of the patch will intentionally do the former

> On 5 Jan 2018, at 21:01, Saito, Hideki via llvm-dev <llvm-dev at lists.llvm.org> wrote: > > > All, > > I'm trying to refactor LoopVectorize such that it has better conformance to VPlan vision going forward > (http://www.llvm.org/docs/Proposals/VectorizationPlan.html). All VP*Recipe class definitions are now > moved to VPlan.h, and I have a patch under review

On 01/05/2018 06:28 PM, Saito, Hideki wrote: > Amara, > >> I support this direction > Thanks for the support. > >> but are there actually any real world workloads where gather/scatter scalarisation would be worth it, on any micro-architecture? If we don’t have examples and the compile time cost is non-negligible then I think we’d still like to keep the early >bailouts in

> One option would be to extend InductionDescriptor::isInductionPHI in the vectorizer to directly analyze the PHIs without SCEV support as Sanjoy suggested. I *think* that that could be sufficient to handle case B. I implemented this with FP SCEV and the code looks very structured, including SCEVExpander. Extending the existing structures without implementing FP SCEV will be problematic. And

All, I'm trying to refactor LoopVectorize such that it has better conformance to VPlan vision going forward (http://www.llvm.org/docs/Proposals/VectorizationPlan.html). All VP*Recipe class definitions are now moved to VPlan.h, and I have a patch under review to move LoopVectorizationPlanner class out of LoopVectorize.cpp (https://reviews.llvm.org/D41420). Next thing I'm working on is

Amara, >I support this direction Thanks for the support. >but are there actually any real world workloads where gather/scatter scalarisation would be worth it, on any micro-architecture? If we don’t have examples and the compile time cost is non-negligible then I think we’d still like to keep the early >bailouts in some form.’ It's not like I have specific application code in

One way you could go is to expose the interface in include/llvm/Transforms/Utils/LoopUtils.h. There's a similar approach in the LCSSA and LoopSimplify passes, both define functions used by other passes (e.g LoopUnroll and LICM). On Fri, Dec 19, 2014 at 10:58 PM, Philip Reames <listmail at philipreames.com> wrote: > I've come across similar use cases recently. In particular,

Thank You. i used -polly-ast-detect-parallel but there is no coincident info generated; my c code is simple vec-sum as follows; #include <stdio.h> int a[2048], b[2048], c[2048]; foo () { int i; for (i=0; i<2048; i++) { a[i]=b[5] + c[i]; } } i executed following commands; $clang -S -emit-llvm vec-sum.cpp -march=native -O3 -mllvm -disable-llvm-optzns -o vec-sum.s $opt -S

Hello, Michael, I'd like to ask if we can enhance the LoopVectorize LLVM module (I am currently using a version from Jul 2016). More exactly: - do you envision to support in the near future LLVM IR gather and scatter intrinsics (as described at http://llvm.org/docs/LangRef.html#llvm-masked-gather-intrinsics and scatter)? I see you have defined some methods that should

Hello I am developing a backend that generates OpenCL from llvm bitcode. I start with a revived fork of the original LLVM C-Backend and have been modifying it. One thing that the backend lacked was generating proper loop structures; instead it relied on labels and goto statements. Therefore, I am trying to find a way to identify the loop structure and print it out in the backend. So far, the main

Hi Michael, +llvmdev,Hal,Nadav For testing, I was currently thinking of a two pronged approach. Lit tests as you suggest with a dummy pass, probably with command line options to define what transform to do, and unit tests to test the delegate behaviour and return values. I'll try and produce a mega patch with at least the loop vectoriser moved over, then split it up again after review.

On 2/18/2019 4:15 AM, Michael Platings via llvm-dev wrote: > Taking my previous example [1]: > > InnerLoopVectorizer LB(L, PSE, LI, DT, TLI, TTI, AC, ORE, VF.Width, IC, > &LVL, &CM); > > If we imagine that over time it evolves such that 50% of the variables have been renamed to camelBack versions of the type names, then it will look like this: > >

Hi all, One of the reasons the Livermore Loops couldn't be vectorized is that it was using global structures to hold the arrays. Today, I'm investigating why is that so and how to fix it. My investigation brought me to LoopVectorizationLegality::canVectorizeMemory(): if (WriteObjects.count(*it)) { DEBUG(dbgs() << "LV: Found a possible read/write reorder:"

If I understand you correctly, conceptually you want two different objects to be returned for Foo.bl and Foo.al? Here is my take on this (take this with a grain of salt, Dan is the expert on this): http://llvm.org/docs/GetElementPtr.html#what-happens-if-an-array-index-is-out-of-bounds LLVM's semantic allows for arrays to be accessed out of bounds - this allows you to walk from the first