Hi Praveen, first, llvm is an interesting project, but it does not generate faster code in general. You might want to look at this benchmark: http://www.phoronix.com/scan.php?page=article&item=gcc_46_llvm29&num=1 The big advantage of llvm I see is that the compile times are often much lower.
Second, I tried to compile deal.II with llvm a few months ago. I couldn't get it to work due to bugs in llvm. I worked with the svn version of llvm and submitted several bug reports (I think they got fixed now), but in February I was still not able to compile deal.II. Some errors might be our mistakes, but some were clearly llvm bugs. I think it would make a lot of sense to make deal.II work with llvm. It makes deal.II more portable and we can spot undefined behavior or even find bugs in our code (the errors and warnings are much better than in g++). I still have the stuff laying around on my disk and was hoping to look at it with Wolfgang at some point. @Wolfgang: what do you think about all this? Best, Timo On Mon, Apr 4, 2011 at 12:50 PM, Praveen C <[email protected]> wrote: > Hello all > The llvm-g++ compiler is supposed to produce faster code than g++. Is this > true for deal.II ? Is there an appreciable difference in speed ? Also is it > possible to compile deal.II with llvm-g++ and link it to libraries (like > Trilinos) which have been compiled with g++ ? > Thanks > praveen -- Timo Heister http://num.math.uni-goettingen.de/~heister _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
