GCC 4.8 To GCC 6.1 Benchmarks For A Complex Program

[phoronix]

Phoronix: GCC 4.8 To GCC 6.1 Benchmarks For A Complex Program

Here are some more compiler performance metrics to share of the GNU Compiler Collection (GCC) for a complex program...

http://www.phoronix.com/scan.php?page=news_item&px=OPM-Performance-GCC-6.1

[Dea1993]

gcc 6 has HSA support right?
can you test also on APU AMD?

[Michael]

gcc 6 has HSA support right?
can you test also on APU AMD?

Will need to look again to see if it's still a pain in the ass to setup or not. And to see how many of my OpenMP benchmarks really can take advantage of it. bridgman can hopefully comment.

[SaucyJack]

Looks like stock optimizations, would have been nice to see if optimized options improved over previous versions.

[SystemCrasher]

I'm feeling like a nut, but damn, which compiler flags were used during this benchmark? It seems I can't easily find this information anywhere. Has it been running with no flags at all? If yes, it makes benchmark rather pointless (who builds programs with no flags, anyway?). If no, where can I see flags used?

[Del_]

I believe the default flags were used, and hence the compile switches are set by the build system. You will find the set-up here: https://github.com/OPM/opm-common/bl...mization.cmake
From what I gather this means -O2 -DNDEBUG -mtune=native