similar to: [LLVMdev] Instruction Cost

CostModule::getInstructionCost also consults TTI ( http://llvm.org/docs/doxygen/html/CostModel_8cpp_source.html#l00380). No? Jingyue On Wed, Jan 14, 2015 at 4:05 PM, Chandler Carruth <chandlerc at google.com> wrote: > > On Wed, Jan 14, 2015 at 3:54 PM, Jingyue Wu <jingyue at google.com> wrote: > >> I'm looking for APIs that compute instruction costs, and noticed

Hi, I am a little unsure about the semantics of the ShuffleKind SK_ExtractSubvector. It seems a subvector is to be extracted, starting from a given index of a given subtype. First of all, if index 0 is passed, I suppose this would mean a noop? But what about calls like the one made of LoopVectorizer for Instruction::PHI in getInstructionCost(): return

Thanks all for replying! I'll try the CostModel class first. Jingyue On Thu, Jan 15, 2015 at 3:47 AM, Jonas Wagner <jonas.wagner at epfl.ch> wrote: > Hi, > > 3. CostModel::getInstructionCost::getInstructionCost in >> lib/Analysis/CostModel.cpp >> > > I've been using the CostModel class in a project, and it has worked quite > well. I don't have

Dear LLVM contributors, I work on the "Improvement of vectorization process in Polly". At the moment I'm trying to implement tiling, interchanging and unrolling of specific loops based on the following algorithm for the analytical modeling [1]. It requires information about the following parameters of a target architecture: 1. Size of double-precision floating-point number. 2.

Hi, I'm trying to implement the TTI hooks for AMDGPU to avoid unrolling loops for operations with huge expansions (i.e. integer division). The values that are ultimately reported by opt -cost-model -analyze (the actual cost model tests) seem to not matter for this. The huge cost I've assigned division doesn't prevent the loop from being unrolled, because it isn't actually

Hi Nadav, I'm interested in knowing how you'll work up the ARM cost model and how easy it'd be to split the work. As far as I can see, LoopVectorizationCostModel is the class that does all the work, with assistance from the target transform info. Do you think that updating ARMTTI would be the best course of action now, and inspect the differences in the CostModel later? I also

Hi Renato, > I'm interested in knowing how you'll work up the ARM cost model and how easy it'd be to split the work. Yes, I am starting to work on the ARM cost model and I would appreciate any help in the form of: advice, performance measurements, patches, etc. I tune the cost model by running the cost model analysis pass and I compare the output of the analysis to the output

There is CostModel.cpp since LLVM3, I am wondering if anyone can give me an concrete example on how to use this pass to estimate cycles used in a given IR file. Thank you very much. Don

On 9 January 2013 17:10, Nadav Rotem <nrotem at apple.com> wrote: > For example: > "opt -cost-model -analyze dumper.ll -mtriple=thumbv7 > -mcpu=cortex-a15" > > I also run the vectorizer with -debug-only=loop-vectorize because it dumps > the costs of all of the instructions with different vectorization factors, > and it also detects the different kinds

Hi Ryan, I think that it would be difficult to back port the CostModel back to LLVM3.0 because it uses the new TargetTransformInfo analysis. I also wanted to mention that only _lowering_ passes (target-specific optimization passes) may use TTI and the cost model. Higher-level canonicalization passes should not use it. Thanks, Nadav On Feb 6, 2013, at 1:26 AM, ryan <stdstack at

Thanks Nadav for the info. It clears my query :) Yes its an integer ADD, and since AVX2 supports 256 bits integer arithmetic, so its cost is less than AVX1. One query though - shouldn't then the cost of integer ADD/SUB/MUL (which would be 1) be explicitly specified in AVX2 cost table? Because right now this entry is missing and cost of these operations are taken from BaseTTI (which is

----- Original Message ----- > From: "Bjarke Roune" <broune at google.com> > To: "Hal Finkel" <hfinkel at anl.gov> > Cc: llvmdev at cs.uiuc.edu, "Jingyue Wu" <jingyue at google.com> > Sent: Wednesday, July 1, 2015 2:27:59 PM > Subject: Re: [LLVMdev] Deriving undefined behavior from nsw/inbounds/poison for scalar evolution > >

Hi All, I wanted to do some basic cost estimation of Instruction/BB. There're CostModelAnalysis and CodeMetrics available in 3.1 and 3.2 releases. I've been using 3.0 for a while. I'm wondering whether similar analysis can be done in the old 3.0 release, e.g. back porting the implementation. Thank you! Best, Ryan -------------- next part -------------- An HTML attachment was

On Tue, Jun 17, 2014 at 2:33 PM, Jingyue Wu <jingyue at google.com> wrote: > Hi Eric, > > In the IR, besides "target datalayout" and "target triple", we have a > special "target cpu" string which is set by the Clang front-end according to > its -target-cpu flag. We also write a Module::getTargetCPU() method to > retrieve this string from the

Hi, I'd like to propose a change to our cost interfaces so that instead of returning an unsigned value from functions like getInstructionCost, getUserCost, etc., we instead return a wrapper class that encodes an integer cost along with extra state. The extra state can be used to express: 1. A cost as infinitely expensive in order to prevent certain optimisations taking place. For example,

I know, I said a bad word -- analysis group. But it works pretty much the way I think we want here. We *always* want a TargetTransformInfo, and we have reasonable (conservative) stubs in place. We would just like the option of providing one from the target that has very clever implementations. I would propose that we make TargetTransformInfo be an analysis group, and provide

On 17 June 2014 06:41, Eli Bendersky <eliben at google.com> wrote: > On Tue, Jun 17, 2014 at 1:38 AM, Nick Lewycky <nicholas at mxc.ca> wrote: > >> Chandler Carruth wrote: >> >>> This seems fine to me, but I'd like to make sure it looks OK to Nick as >>> well. >>> >> >> I strongly prefer baking in knowledge about the

Chandler Carruth wrote: > This seems fine to me, but I'd like to make sure it looks OK to Nick as > well. I strongly prefer baking in knowledge about the intrinsics themselves into the passes if possible. Metadata will always be secondary. Separately, should value tracking look use range metadata when it's available? Absolutely. I think it should apply to all CallInst not just

Eh? How do you envision this? -eric On Tue, Jun 17, 2014 at 2:09 PM, Jingyue Wu <jingyue at google.com> wrote: > Hi Nick, > > That makes sense. I think a main issue here is that the ranges of these PTX > special registers (e.g., threadIdx.x) depend on -target-cpu which is only > visible to clang and llc. Would you mind we specify "target cpu" in the IR > similar

Hi, I noticed a significant performance regression (up to 40%) on some internal CUDA benchmarks (a reduced example presented below). The root cause of this regression seems that IndVarSimpilfy widens induction variables assuming arithmetics on wider integer types are as cheap as those on narrower ones. However, this assumption is wrong at least for the NVPTX64 target. Although the NVPTX64 target