GCC 5.3 Optimization Level Tests From -O0 To -Ofast

Here are some fresh tests of Fedora 23 with the GCC 5.3.1 compiler when running a series of benchmarks after the binaries were compiled each time with an assortment of optimization levels.

A new Phoronix Premium subscriber had inquired this morning about having more GCC optimization benchmarks. Besides Phoronix Premium getting you an ad-free experience and the ability to view multi-page articles on a single page while the funds go to ensure more content and Linux hardware tests, etc, on Phoronix, I always try my best to honor Phoronix Premium subscriber requests for tests and future articles. Thus I managed to find some time today to start running some fresh GCC optimization benchmarks.

In this first article are GCC tests with the new Intel Xeon E3 1245 v5 Skylake system while running Fedora 23 64-bit with its GCC 5.3.1 compiler. The CFLAGS/CXXFLAGS tested with a variety of open-source tests via the Phoronix Test Suite were -O0, -O1, -O2, -O3, -Ofast, and -Ofast -march=native. If you don't know what those compiler switches mean, see the GCC documentation.
The results come largely in as expected.
I'll be working on some other GCC/Clang optimization level tests on other hardware shortly, these were just spur of the moment tests to satisfy a new premium member's request.
To see many more of these GCC compiler optimization benchmarks, visit this OpenBenchmarking.org result file. If you would like to support all of the Linux hardware/testing work done at Phoronix and/or see your own requests fulfilled, consider joining Phoronix Premium this holiday season.