similar to: [LLVMdev] strange behaviour after extracting optimization pass code

Hi, I am expieriencing strange behaviour of llvm's optimization passes and I don't understand what I am doing wrong. Basically all I've done is extracting code for optimization of a llvm-function in a llvm-module and put it into a separate function for better readability. The original code looks like follows (and works as expected): ----------------------------- std::string

On 2008-12-19, at 06:59, Ralf Karrenberg wrote: > I am expieriencing strange behaviour of llvm's optimization passes > and I don't understand what I am doing wrong. > Basically all I've done is extracting code for optimization of a > llvm-function in a llvm-module and put it into a separate function > for better readability. The original code looks like follows

Hello We are migrating from 2.9 to 3.2 Here is some code that does not compile any more llvm::PassManager *pm; llvm::FunctionPassManager *fpm; module = llvm::ParseBitcodeFile(mb,context,&err_str); if (!module) { error() <<"Failed to load module from bitcode file: " <<err_str <<endl; exit(1); } pm = new PassManager();

Hi, Here is the result of optimization using the following passes: llvm::createBasicAliasAnalysisPass() llvm::createInstructionCombiningPass() llvm::createReassociatePass() llvm::createGVNPass() llvm::createCFGSimplificationPass() The optimized IR seems to contain what look like dead stores on %8 and %9 in basic blocks 7 and haveData. How can I get rid of them? Thanks, Vinayak define

Hi Vinayak, > Here is the result of optimization using the following passes: > > llvm::createBasicAliasAnalysisPass() > llvm::createInstructionCombiningPass() > llvm::createReassociatePass() > llvm::createGVNPass() > llvm::createCFGSimplificationPass() > you should run the mem2reg pass too, and first. > ; <label>:7 ; preds =

Hi Rob, > I'm using the gvn pass, not sure about basic-aa. if you are using LLVM from svn then you need to specify the basic-aa analysis, otherwise gvn won't unify calls to readonly/readnone functions. This is new behaviour introduced by Dan; probably the tutorial should be updated. Ciao, Duncan. > > I've copied the code as-is from

Still no luck after switching to: theFPM->add( new llvm::TargetData( *theExecutionEngine->getTargetData() ) ); theFPM->add( llvm::createBasicAliasAnalysisPass() ); theFPM->add( llvm::createInstructionCombiningPass() ); theFPM->add( llvm::createReassociatePass() ); theFPM->add( llvm::createGVNPass() ); theFPM->add( llvm::createCFGSimplificationPass() ); Based on output

Hi Brent, > The code in this setup will be optimized according to the optimizer pipeline one > sets up as in the code below. I find that if I only use the passes below the > quality of the code is not that good code quality will be dreadful if you don't promote memory accesses to registers right at the start, using eg mem2reg or scalarrepl. (for example inlining is not

Hi, If hope this is the right list to post a question like this. If not, my apologies -- please redirect me. Following the Kaledoscope example I am trying to write a simple JIT and compile my own small language using LLVM. Compilation happens using the C++ api by constructing a VM and emitting code using the api (just as the Kaledoscope example). The code in this setup will be optimized

I'm using the gvn pass, not sure about basic-aa. I've copied the code as-is from http://llvm.org/docs/tutorial/LangImpl4.html#code and added "F->addFnAttr( Attribute::ReadOnly )" after "Function *F = Function::Create(FT, Function::ExternalLinkage, Name, TheModule)". The passes it sets up are: // Set up the optimizer pipeline. Start with registering info about

Hi Duncan, The mem2reg pass did not help. Moving the alloca instructions into the entry block did the trick. One other thing I noticed is that if there was only one alloca, the dead store is eliminated. Having the second one (does not matter which of the two) prevents the elimination. Is this a bug? Thanks, Vinayak On 9/10/12 8:16 AM, Duncan Sands wrote: > Hi Vinayak, > >>

Rob, I can reproduce the behaviour you observe using llvm top-of-tree. I will try to look into it. Best regards, -- Arnaud de Grandmaison -----Original Message----- From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Rob Pieke Sent: Monday, November 15, 2010 4:27 PM To: Duncan Sands; llvmdev at cs.uiuc.edu Subject: Re: [LLVMdev] Optimization of calls to

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> Tilmann Scheller wrote: <blockquote

I'm writing a front end for an existing interpreted language with slightly odd semantics for primitive values. Similar to the values in a database table, any value could be null, even for non-pointer types. For example a boolean variable could be true, false, or null. To model this behaviour, I'm passing an {i1, [type]} around for every numeric type. And using insertvalue / extractvalue

SUCCESS! Sorry, I had removed the addFnAttr() call by accident. Using the basicAA pass and ReadOnly, as you suggested, works like a charm! :) > -----Original Message----- > From: Rob Pieke > Sent: Monday, November 15, 2010 4:40 PM > To: 'Duncan Sands' > Cc: 'llvmdev at cs.uiuc.edu'; 'Dan Gohman' > Subject: RE: [LLVMdev] Optimization of calls to

Hi, I'm attempting to port some code which uses the GlobalMappingLayer in the Orc JIT. This code worked fine in LLVM 4, but I'm getting a compile error with LLVM 5. I think the problem is that this layer hasn't been modified to account for some of the changes made in LLVM 5, but I wanted to make sure that I wasn't missing something. I have code which looks like this:

Hello, So the current JIT will be superseded by the MCJIT completely and no further development should be expected for the current JIT? In any case, I am definitely interested in submitting a patch if you could help me by sending me in the right direction, since I really want this working. I managed to reproduce this behavior in LLVM 3.0 by modifying llc to read my .bc file and try to JIT the

Daniel Dunbar wrote: > Ah, in this case llvm_shutdown isn't what you are looking for. I've > never looked at the Statistic implementation so I can't be any help > here, but if you poke around starting with "llvm/ADT/Statistic.h" you > may find a way to do what you want. I looked into that already, but Statistic.h only gives me the possibility to implement tracking

Hi Brian, Yes - I believe you're correct. I'm working on a fix and extra test coverage now. In the meantime, I believe you should be able to fix the signatures in your copy and everything should "just work". Cheers, Lang. On Fri, Sep 22, 2017 at 2:04 PM, Brian Kahne via llvm-dev < llvm-dev at lists.llvm.org> wrote: > Hi, > > > > I’m attempting to port

Hi, Sean: I'm sorry I lie. I didn't mean to lie. I did try to avoid making a *BIG* change to the IPO pass-ordering for now. However, when I make a minor change to populateLTOPassManager() by separating module-pass and non-module-passes, I saw quite a few performance difference, most of them are degradations. Attacking these degradations one by one in a piecemeal manner is wasting