similar to: [LLVMdev] Dynamic optimalization passes in LLVM based compiler

On Sat, Nov 17, 2012 at 4:44 AM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > Hi! > I'm new to LLVM but I've read tons of articles, I want to implement my own > compiler and I came across a big problem. > I have several questions, that I cannot answer myself: > > 1) If I'm writing custom compiler do I have to "hardcode" passes that it >

Thank you for yours response :) I know that LLVM Pass was designed to transform IR, but lets focus on an example - LLVM Pass is a function that transform some set of input into output. It can transform IR into graph of lets say strongly connected components and then other passes can use it (that data - not IR) to generate other data OR to manipulate the IR. So why I can not create passes, that

> > I know that LLVM Pass was designed to transform IR, but lets focus on an > > example - LLVM Pass is a function that transform some set of input into > > output. It can transform IR into graph of lets say strongly connected > > components and then other passes can use it (that data - not IR) to > generate > > other data OR to manipulate the IR. > > >

I would love to ask you one more question. You have written: It's not really going to give you a lot of value compared to just building your own graph transformation pipeline & then producing IR at the end of that. Could you please tell more about this topic? Why my custom solution (custom pass manager etc) would be better than making LLVM (non IR) passes? 2012/11/17 Wojciech Daniło

On Sat, Nov 17, 2012 at 1:09 PM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > Thank you for yours response :) > I know that LLVM Pass was designed to transform IR, but lets focus on an > example - LLVM Pass is a function that transform some set of input into > output. It can transform IR into graph of lets say strongly connected > components and then other passes

On Nov 17, 2012, at 1:56 PM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > > > I know that LLVM Pass was designed to transform IR, but lets focus on an > > example - LLVM Pass is a function that transform some set of input into > > output. It can transform IR into graph of lets say strongly connected > > components and then other passes can use it

Hi! I'm new to LLVM and I'm planning to write a compiler, which would behave slighty different than all current tools. As an input it will not get text but it will read a file (lets assume for a moment, that in this file there will be a graph structure). Before generating IR code I want to be able to run optimalization and analysis passes on this graph and after these passes I want to run

I have noticed that I was installing LLVM according to: http://llvm.org/docs/GettingStarted.html#getting-started so it was no the 3.1 but the 3.2 version from svn (trunk 167573). Maybe in the svn, not stable version, there is a bug? 2012/11/9 Wojciech Daniło <wojtek.danilo.ml at gmail.com> > Does anybody have Idea why I get such error? The LLVM is compiled from > source - is is the

Thank you, I have seen this page, but before I start developing my tool and digging deeply into LLVM I would love to know if I can generate LLVM IR in one of passes (as a result of one pass, before running other). Please answer this simple question :) Thank you :) 2012/11/6 Eli Friedman <eli.friedman at gmail.com> > On Mon, Nov 5, 2012 at 5:17 PM, Wojciech Daniło >

If anyone is interested in this topic, I filled a bug here: http://llvm.org/bugs/post_bug.cgi 2012/11/9 Wojciech Daniło <wojtek.danilo.ml at gmail.com> > I have noticed that I was installing LLVM according to: > http://llvm.org/docs/GettingStarted.html#getting-started > so it was no the 3.1 but the 3.2 version from svn (trunk 167573). > Maybe in the svn, not stable version,

Hi! I'm trying to learn LLVM and I want to compile the Kaleidoscope compiler. according to: http://llvm.org/docs/tutorial/LangImpl3.html I'm compiling it with: clang++ -g -O3 test.cpp `llvm-config --cppflags --ldflags --libs core` -o toy and it indeed works. The problem is that when using eclipse it forces me to divide the compilation into two steps: g++ -O0 -g3 -Wall -c

Thank you, now it is more clear whats going on. There is a line: /bin/sh: /home/wdanilo/dev/llvm/src/projects/Debug+Asserts/bin/llvm-config: No such file or directory but why it is searching for it in that directory? (all makefile configs, are default configs) Below, there is full output of make: >> make VERBOSE=1 for dir in lib tools; do \ if ([ ! -f $dir/Makefile ] || \ command

On Mon, Nov 5, 2012 at 5:17 PM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > Hi! > I'm new to LLVM and I'm planning to write a compiler, which would behave > slighty different than all current tools. > As an input it will not get text but it will read a file (lets assume for a > moment, that in this file there will be a graph structure). > Before

I will take a look into NVPTX. I don't want to deconstruct C++ classes to generate Python code - I want to generate Python backend :) Its purpose is to be able to generate Python code from LLVM IR code (not C++ code! - C++ code is needed for me only to get a sample IR code). I want to write my custom compiler in the future and I want to be able to output Python code from it (of course not

Does anybody have Idea why I get such error? The LLVM is compiled from source - is is the newest version (3.1) from http://llvm.org/releases/ (from 22 may 2012) I created project in llvmsrc/projects/test (as copy of sample project) and I get these errors while attaching LLVM lbraries :( 2012/11/8 Wojciech Daniło <wojtek.danilo.ml at gmail.com> > Thank you, now it is more clear whats

Hi! I'm trying to use the LLVMProject toolchain (according to this page: http://llvm.org/docs/Projects.html). I created a project "test" with kaleidoscope example source code from: http://llvm.org/docs/tutorial/LangImpl3.html It compiles successfully with: clang++ -g -O3 toy.cpp `llvm-config --cppflags --ldflags --libs core` -o toy according to the project documentation, my Makefile

On Tue, Nov 6, 2012 at 6:38 AM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > Hi! > I'm trying to learn LLVM and I want to compile the Kaleidoscope compiler. > according to: > http://llvm.org/docs/tutorial/LangImpl3.html > > I'm compiling it with: > clang++ -g -O3 test.cpp `llvm-config --cppflags --ldflags --libs core` -o > toy > > and it

You are of course right. I wanted to output "good looking" Python code though. (if its possible) I see that NVPTX backend is really big project. Is there any hello world "common codegen infrastructure"? 2012/11/21 Eli Bendersky <eliben at google.com> > On Tue, Nov 20, 2012 at 4:10 PM, Wojciech Daniło > <wojtek.danilo.ml at gmail.com> wrote: > > I will

On Tue, Nov 6, 2012 at 1:10 AM, Wojciech Daniło <wojtek.danilo.ml at gmail.com> wrote: > Thank you, I have seen this page, but before I start developing my tool and > digging deeply into LLVM I would love to know if I can generate LLVM IR in > one of passes (as a result of one pass, before running other). > Please answer this simple question :) > > Thank you :) Passes can

Hi, I have a version of the old CBackend which is working with the current trunk. I am in the progress of rewriting it into a new version but that project is not going as fast as I would hope as I have been distracted by other tasks with higher priority lately. Anyway, for those interested, I can provide patches, optionally separated into several parts (core, testing, misc). I am afraid that