similar to: [LLVMdev] algorithm for GVN

Hello everyone I am Rakshit Singla, a third year undergrad at IIT Hyderabad, India. I finished a basic compilers course last semester and am currently doing a compiler optimizations course. I have been exploring LLVM for the past few months (wrote a front-end for the Classroom Object Oriented Language and have been studying pieces of code.) I would like to work with LLVM and contribute to the

The GVN algorithm used in LLVM currently (I'm rewriting it) is the basic hash based RPO algorithm. The new one i'm writing is based on http://dl.acm.org/citation.cfm?id=512536 (see https://github.com/dberlin/llvm-gvn-rewrite) LLVM has different algorithms for both scalar PRE and load PRE. They are basically variants of standard PRE algorithms transformed into SSA, but with some

Thank you David for prompt reply. I tried with SmallVector. I inserted elements with push_back(). But when I retrieve elements using pop_back_val the elements are returned in reverse order of insertion (I mean like LIFO order). I need this to be FIFO order. How to achieve that? Regards, Rekha On Sun, Nov 3, 2013 at 8:31 PM, David Blaikie <dblaikie at gmail.com> wrote: > If you

Hi, I am writing an analysis which requires creating worklist of basic blocks. The worklist should be in FIFO order. I checked SmallVector (and similar others) and found out this is working in LIFO order when I use the functions push_back and pop_back_val to insert and delete elements in the worklist. Can someone suggest an appropriate DS to implement my worklist. Note: I am not concerned about

Hi, I wrote a pass in LLVM and would like to test it on the programs in LLVM test-suite. To be specific, I would like to get the stats upon running my pass on these programs. What are the commands to run my pass on the test suite? -- Rekha -------------- next part -------------- An HTML attachment was scrubbed... URL:

On Wed, Jul 23, 2014 at 4:06 PM, Rekha R <rekharamapai at nitc.ac.in> wrote: > Ok. Got it. > > If *add nsw* overflows, this results in undefined value. > But then *add* on same arguments results in well-defined value. > > Hence treating first one as redundant based on the second is acceptable. > But vice versa is not. > If they are in different code paths, sure.

IMHO; On undefined behaviour we can do whatever we want. If the "add nsw" overflows this would lead to undefined behaviour. Therefore we can assume that "add", with the same arguments will not overflow. On Wed, Jul 23, 2014 at 3:32 PM, Tim Northover <t.p.northover at gmail.com> wrote: > On 23 July 2014 06:25, Rekha R <rekharamapai at nitc.ac.in> wrote: >

On Sun, Nov 3, 2013 at 7:38 AM, Rekha R <rekharamapai at nitc.ac.in> wrote: > Thank you David for prompt reply. > > I tried with SmallVector. I inserted elements with push_back(). > But when I retrieve elements using pop_back_val the elements are returned > in reverse order of insertion (I mean like LIFO order). > I need this to be FIFO order. How to achieve that? >

Hi, I was trying to write a sample optimization pass in LLVM, by following instructions in LLVM doc for Writing a New Pass. But then ran into *make*issues. The issue is with having separate *source *and *build* directories. The LLVM doc for 'Getting Started' suggests keeping the two separate, but the 'Writing a new pass' doc assumes they are merged. Hence the *make *issue. I did

Hi, Thank you David. My question was from a machine independent code analyzer/optimizer perspective where types are of less importance. Hence I felt difficult to understand the IR with the types. Rekha On Mon, Oct 27, 2014 at 6:44 PM, David Chisnall <David.Chisnall at cl.cam.ac.uk > wrote: > On 27 Oct 2014, at 12:46, Rekha R <rekharamapai at nitc.ac.in> wrote: > > > Why

Hi, I am curious to know the design decision on why operands of LLVM IR instructions are type annotated. Why wasn't the C style of explicit declaration of variables (or values in LLVM context) followed by their uses model not adopted here? or may be even annotate the value (result of the Instruction) with type rather than annotating each operands on the Instruction? Just felt that by

Hi, I am trying to understand the semantics of Instructions in llvm. Are the following instructions semantically same? * %add2 = add nsw i32 %add, %add1 %add3 = add i32 %add, %add1* Based on my understanding from the Language Reference Manual, I think they are different. But then why is the *gvn* pass detecting *%add3* as redundant and deleting it? Your views are appreciated. Rekha

Hi, I was studying the LLVM IR instruction set by analyzing how constructs in C language is getting converted to IR form. I found difficulty in finding C constructs which could generate Invoke, Resume, Unreachable, ExtractValue, InsertValue instructions. Can someone help me find some simple C code that generates these instructions in LLVM IR? -- Regards, Rekha -------------- next part

If you don't care about efficiency you can push (or at least insert) at the front of a(small or otherwise) vector. On Nov 3, 2013 3:32 AM, "Rekha R" <rekharamapai at nitc.ac.in> wrote: > Hi, > > I am writing an analysis which requires creating worklist of basic blocks. > The worklist should be in FIFO order. I checked SmallVector (and similar > others) and found

Hi, I am trying to implement Available Expressions data flow analysis. I created the following class (I am giving here code snippet.): namespace { typedef DenseMap<Expression, uint32_t> DMTy; //Expression is a class I defined. struct DataFlowValue { DMTy ExprMap; llvm::BitVector* DFV; // Functions operating on the data // bool operator==(const DataFlowValue V) const;

On Oct 3, 2013 11:47 PM, "Rekha R" <rekharamapai at nitc.ac.in> wrote: > > Hi, > > I was trying to write a sample optimization pass in LLVM, by following instructions in LLVM doc for Writing a New Pass. But then ran into make issues. The issue is with having separate source and build directories. The LLVM doc for 'Getting Started' suggests keeping the two

On Mon, Nov 4, 2013 at 10:35 AM, Rekha R <rekharamapai at nitc.ac.in> wrote: > Hi, > > I am trying to implement Available Expressions data flow analysis. I created > the following class (I am giving here code snippet.): > > namespace { > typedef DenseMap<Expression, uint32_t> DMTy; //Expression is a class I > defined. > struct DataFlowValue { >

Hi, I intend to write an optimization using LLVM. But before doing that I felt like understanding the tools. I wrote a hello world program to be compiled and executed. But then one of the tutorials suggested using Clang, an other said dragonegg, a third suggested using llvm-gcc. I am confused as to which one to use. Can some one help me in 1. What are the differences between these three? 2. Which

----- Original Message ----- > From: "Daniel Berlin via llvm-dev" <llvm-dev at lists.llvm.org> > To: "Amos Robinson" <amos.robinson at gmail.com> > Cc: "llvm-dev" <llvm-dev at lists.llvm.org> > Sent: Tuesday, May 3, 2016 7:39:54 PM > Subject: Re: [llvm-dev] GVN pass: does global value numbering remove > duplicate computations in

> The GVN on the newgvn branch i have will remove these, and is more complicated > The one i have implemented unifies AWZ and hash based and will also do predication/value inference. This is exciting news. It sounds like it will find a lot of the interesting cases. > Note that we don't do full-on polynomial time equivalence finding. While it would be fun to play with such