GCC 4.6 Leaves Stage 1 With New Features

[phoronix]

Phoronix: GCC 4.6 Leaves Stage 1 With New Features

Novell's Richard Guenther has just announced that GCC 4.6.0 has now left stage one of development and has immediately entered the third stage. This means no new features or other major work aside from bug-fixes will be accepted into this next major release of the GNU Compiler Collection...

http://www.phoronix.com/vr.php?view=ODc1MQ

[droidhacker]

Cool stuff... but there is a minor glitch... "it doesn't look like Intel's Core 2 and Core i7 optimizations didn't make it".... unintentional (I suspect) double negative.

[yoshi314]

Sadly, it doesn't look like Intel's Core 2 and Core i7 optimizations didn't make it into GCC 4.6.

yeah, it's sad that they don't look like they didn't make it, because we all hate intel cpus in here.

[yoshi314]

also, how come this wasn't linked in the article?

http://gcc.gnu.org/gcc-4.6/changes.html


also, explanation for -Ofast :

A new general optimization level, -Ofast has been introduced. It combines the existing optimization level -O3 with options that can affect standards compliance but result in better optimized code. For example -Ofast enables -ffast-math.

now that is interesting.

[smitty3268]

The compilation time improvements could place 4.6 around the same speed as LLVM, i think.

Michael, can you include 1 compilation speed test in your upcoming compiler roundup and not just test the runtime speed of the binaries they produce?

[XorEaxEax]

The compilation time improvements could place 4.6 around the same speed as LLVM, i think.

Michael, can you include 1 compilation speed test in your upcoming compiler roundup and not just test the runtime speed of the binaries they produce?

Well I don't know, even though llvm 2.8 is slightly slower than llvm 2.7 in my tests, I think it will still hold it's own against gcc 4.6 in terms of compilation speed. But it's definately a good thing that gcc is getting faster in this area. A bit bummed though that the core i7 optimizations won't make it into 4.6 though, gcc is already produces the fastest code of the compilers I use (gcc, clang/llvm, visual studio express) and it would be nice to see how much these machine specific optimizations could improve performance on my core i5 and i7. Also hoping that LTO will get some polishing since I find it to be a bit rough around the edges, like having to manually add the optimization options to the linking stage. I did read that the normal lto mode will be deprecated and whopr will be default which makes sense. Maybe it's time to start trying some gcc 4.6 snapshots.

Also, Phoronix has been slashdotted, damn it's slow!