Early GCC 7 Compiler Benchmarks On Linux: Some Performance Improvements

[phoronix]

Phoronix: Early GCC 7 Compiler Benchmarks On Linux: Some Performance Improvements

While GCC 7 is still under heavy development and the GCC 7.1 stable release will not come until a few months into 2017, here are some early benchmarks of GCC 7.0 compared to GCC 6.2 and GCC 5.4 on an Ubuntu Linux x86_64 system.

http://www.phoronix.com/vr.php?view=23512

[davidbepo]

it seems that gcc is getting better than clang
i hope they share the improvements

[SaucyJack]

Gcc has been better than clang since clang was created. I'm hoping llvm/clang catches up though, they've been making great progress.

[Adarion]

I hope besides all the performance that this will result in completed compilation of code and binaries that do not exhibit strange runtime problems. It might have a reason that Gentoo dev team takes some time to declare new compilers as "stable".

[davidbepo]

Gcc has been better than clang since clang was created. I'm hoping llvm/clang catches up though, they've been making great progress.

nope. around clang 3.7 (if i remember correctly) clang showed better build time with equal binary performance

[davidbepo]

link:

http://www.phoronix.com/scan.php?page=article&item=clang-37-gcc52&num=2

[SaucyJack]

nope. around clang 3.7 (if i remember correctly) clang showed better build time with equal binary performance

In one test? Who are you trying to fool here? Go fanboy somewhere else. I'm excited about clang as much as anyone, but the binary speed just isn't there yet to swap from gcc for general use. It works well enough of you need to use it though.

[s_j_newbury]

link:
http://www.phoronix.com/scan.php?pag...37-gcc52&num=2

Nice cherries there! :-) Clang with default flags is beating out GCC with default flags when running SciMark. This was also seen with the recent BSD benchmarks. Now this GCC7 benchmark is showing improvement with the SciMark performance from GCC. That's good news, quite possibly the GCC devs saw Clang was producing better code in that case and did something about it. There are of course a lot of optimisation passes that are not enabled at default (on both compilers), or are arch specific, ... it's very likely that different optimisation strategies will improve some tests and regress others.

[davidbepo]

In one test? Who are you trying to fool here? Go fanboy somewhere else. I'm excited about clang as much as anyone, but the binary speed just isn't there yet to swap from gcc for general use. It works well enough of you need to use it though.

im no fanboy and i dont win anything fooling anyone, i posted a link with the only tests that i have available, if you have a test showing another thing post it and stop trolling

[emblemparade]

Wow, weird politics going on here. You do know that gcc and clang developers are pretty much all friends? clang enjoyed the benefit of being able to start from scratch with a new architecture, so the internal design is much better. The gcc folks want to get there, but they have the much bigger problem of requiring huge legacy compatibility, so they move more slowly with refactoring. It's impossible to share code because of the license incompatibility, but everyone is talking and sharing ideas.