similar to: [LLVMdev] Correctness of Optimization Phases

On Wed, Jul 19, 2006 at 03:09:44AM +0800, deveshti at comp.nus.edu.sg wrote: > I may sound stupid but can someone tell me that after applying > optimization sequences randomly whether we can surely say that the > final output is correct? Answering this question is equivalent to solving the Halting Problem. > I mean if we apply 12 options in random sequence provided by 'opt'

>> I may sound stupid but can someone tell me that after applying >> optimization sequences randomly whether we can surely say that the >> final output is correct? > > Answering this question is equivalent to solving the Halting Problem. Prooving correctness of a compiler is really an NP problem. This goes for any compiler backend. All you can do is have enough test

On Fri, Sep 24, 2004 at 09:38:44AM -0700, Jeff Cohen wrote: > Here's the patch to Signals.cpp. assuming that stdio.h is acceptable > (can't imagine it won't work). We prefer #include <cstdio>, since this is C++ after all. :) -- Misha Brukman :: http://misha.brukman.net :: http://llvm.cs.uiuc.edu

Here's the patch to Signals.cpp. assuming that stdio.h is acceptable (can't imagine it won't work). On Fri, 24 Sep 2004 09:29:05 -0700 Jeff Cohen <jeffc at jolt-lang.org> wrote: > Uh... this may be a silly question, but why can't you include <stdio.h>? > It'd be much better than <iostream>. > > Anyway, I think I'll try this weekend to come

> Prooving correctness of a compiler is really an NP problem. This goes for > any compiler backend. All you can do is have enough test cases. No. To be in NP it would be necessary for a Non-deterministic Polynomial algorithm to exist. There is no such algorithm. The problem is undecidable. It doesn't means that it is impossible to prove the correctness of a compiler. It means that no

Sigh... take it up with Microsoft. On Fri, 24 Sep 2004 12:06:36 -0500 Misha Brukman <brukman at uiuc.edu> wrote: > On Fri, Sep 24, 2004 at 09:38:44AM -0700, Jeff Cohen wrote: > > Here's the patch to Signals.cpp. assuming that stdio.h is acceptable > > (can't imagine it won't work). > > We prefer #include <cstdio>, since this is C++ after all. :) >

Actually, Misha, that won't work. The -C option is used directly in docs/Makefile. So the change will have to go into docs/Makefile. Either that or upgrade install to version 5. Reid. On Thu, 2005-02-17 at 11:33, Misha Brukman wrote: > On Thu, Feb 17, 2005 at 01:21:20PM -0600, Feng Chen wrote: > > llvm[1]: Installing HTML documentation > >

Constant *strcon==ConstantArray::get("Value : %d\n"); Sorry Typo. On Tue, 21 Dec 2004, Misha Brukman wrote: > On Tue, Dec 21, 2004 at 03:45:33PM -0700, Sriraman Tallam wrote: > > I have this call instruction to printf inserted which is causing > > an assertion failure. Any pointers to where I am wrong : > > > > Function

On Wed, Jun 23, 2004 at 03:50:09PM -0500, Patrick Meredith wrote: > MetaSplit is an anlysis I just finished writing. It doesn't alter > anything, all it does is build a set of "program instructions". For > some reason even though if I run it with any other combination of > passes I've found, anytime I run it with mem2reg I get a seg fault in > dyn_cast!

Thanks Misha. I tried the "gcc -dM -E /tmp/file.c | grep __sparcv9" but there was no output. Maybe this is the probelm, I do have CXX = g++ -mcpu=v9 CC := gcc -mcpu=v9 in the Makefile.config. Do I need to add -m64 as well? Thanks. Shukang On Thu, 7 Oct 2004, Misha Brukman wrote: > Is your compiler configured to define __sparcv9 when it compiles? > The following should tell

On Fri, 20 Aug 2004 08:52:28 -0700 Reid Spencer <reid at x10sys.com> wrote: > Hi Jeff, > > On Fri, 2004-08-20 at 08:07, Jeff Cohen wrote: > > OK. I've built the front end without any heartaches, but I did > > encountered the following glitches: > > > > The documentation of --with-llvmgccdir is a bit ambiguous. I had to > > try several paths

On Thu, Feb 17, 2005 at 11:39:27AM -0800, Reid Spencer wrote: > Actually, Misha, that won't work. Sorry, Reid, but I think it will, for the following reason: I don't really know what the -c (lowercase) option does, but install says it's ignored anyway, so that's irrelevant for GNU install. We're talking about the -C option (uppercase). > The -C option is used directly

On Wed, 23 Jun 2004, Patrick Meredith wrote: > What's different about code that's been mem2reg'd from straight front end > code, or anything that mem2reg hasn't been run on? PHINODES! Yup, front-ends generally don't produce SSA form. :) > It appears to be crashing when I try to cast a Value* that's really a > BB* (from the PHInode operands) to a User*,

On Fri, Sep 03, 2004 at 03:01:01PM -0500, Anshu Dasgupta wrote: > ><snip> > >for (BasicBlock::iterator I = Dest->begin(); PHINode *PN = > >dyn_cast<PHINode>(I); ++I) > > visitPHINode(*PN); > ><snip> > > > >build_vc71\lib\Transforms\Scalar\SCCP.cpp(202) : error C2275: > >'llvm::PHINode' : illegal use of this type as

I'm CC'ing the llvm-dev list because other people are more knowledgeable about the bytecode format/encoding than I am. Please follow-up the replies to the list. On Wed, Oct 20, 2004 at 11:27:53AM -0700, Yiping Fan wrote: > We also want to extend the llvm instructions/intrinsic > functions/types/passes to support our high-level synthesis for > hardware. First of all, we want to

On Wed, Aug 04, 2004 at 12:16:12PM -0700, Reid Spencer wrote: > So I propose: > [snip] > -O3 = aggressive optimization, regardless of computation time with the > effect of producing the fastest executable I would suggest splitting -O3 into 2 or more levels of optimization, because as written, -O3 sounds pretty scary: "regardless of computation time", and given some

On Sat, Aug 21, 2004 at 04:15:49PM -0700, Jeff Cohen wrote: > I don't know if it's under cvs. It's the "getting started" page > (http://llvm.cs.uiuc.edu/docs/GettingStarted.html) in section "Getting > Started Quickly (A Summary)". But careful reading of the remainder of > the page does give the correct path. http://llvm.cs.uiuc.edu/docs/* is a copy of

I'm the guy who is working on the LLVM bytecode documentation. The document I have at present just supports the bytecodes my code generator processes, though, which is far from all of them. As I get farther along with my code generator I expect I'll get to the point where everything kind of fits together for me and I can finish it up. In the meantime, people are welcome to what I have

Hi, I met some errors when I tried to build LLVM. The tar file is llvm-1.3.tar.gz. I am using a sparc machine with gcc 3.2.2. ------------- Compiling SparcV9CodeEmitter.cpp /uf24/zhou/research/llvm/src/lib/Target/SparcV9/SparcV9CodeEmitter.cpp: In static member function `static void llvm::<unnamed>::JITResolver::CompilationCallback()':

Qiuyu, The dynamic optimization project (an internal research project in our group) uses some way to map LLVM basic blocks to native code addresses. If this is what you want, perhaps you can ask Brian Gaeke (gaeke at uiuc.edu) to give you some information about how that is done. --Vikram http://www.cs.uiuc.edu/~vadve http://llvm.cs.uiuc.edu/ On Aug 19, 2004, at 5:11 PM, Misha Brukman