search for: tup1

...ig function. The O2 compilation is almost a decimal order of magnitude slower with the inline version, but twice as fast when the function is split up into 4 pieces (plus some other stuff). The actual C++ here: 1729 5498 93095 /Users/johnskaller/.felix/cache/text/Users/johnskaller/felix/tup1.cpp 753 2246 17518 /Users/johnskaller/.felix/cache/text/Users/johnskaller/felix/tup1.hpp plus library files. Felix optimises the use of headers so only required library headers are actually used. 100 seconds for 2000 lines is only 20 lines a second. At that speed a 2 million line progra...

Adding LLVMdev, since this is intimately related to the optimization passes. > I think this is roughly because some function level optimisations are > worse than O(N) in the number of instructions. Please profile this and mail llvmdev regarding passes with significantly superlinear behavior (e.g. O(n^2)). My understanding is that these kinds of algorithmic problems are generally considered

...on is almost a decimal order of magnitude slower > with the inline version, but twice as fast when the function is split up > into 4 pieces (plus some other stuff). > > The actual C++ here: > > 1729 5498 93095 /Users/johnskaller/.felix/cache/text/Users/johnskaller/felix/tup1.cpp > 753 2246 17518 /Users/johnskaller/.felix/cache/text/Users/johnskaller/felix/tup1.hpp How about providing a preprocessed version of those files as a test case? I don't think there are many people on this list that speak Felix fluently and/or are willing to install your compile...