Open64 Compiler Tuning On AMD Bulldozer FX-8150

[Shining Arcanine]

Why test only single options? At least one -O2 -march=bdver1 run and one with all recommended options would give more relevant results I think.

It could be that Michael doesn't use Gentoo. People who use Gentoo know to do that. Users of other distributions do not unless they are kernel/systems programmers.

[AgY!]

The default Optimization level of open64 is -O2 .. if you specify nothing you get O2's.

In other words: The "-march=bdver1" are with O2

+ quote from the article:

The options tested included stock (not overriding any CFLAGS/CXXFLAGS and Open64 defaults to the -O2 optimization level)

[AnonymousCoward]

The default Optimization level of open64 is -O2 .. if you specify nothing you get O2's.

Oh, good, then at least that one is covered. Still it would be nice to know how all recommended flags together change the results.

[pszilard]

Speaking of which - I'm always annoyed by the lack of analysis in these articles. "We ran foo and it yielded this number X. Next."

Articles like this teach readers next to nothing, it offers pretty much zero enhancement to understanding.

I would look at results for, e.g. GraphicsMagick and ask myself "why aren't the BD-specific optimizations helping?"
And then I'd re-run the test using valgrind cachegrind and see what the code that the compiler generated is actually doing, in an instruction-level profile, and look at the cache hits and misses. (Of course, this assumes that you've built a new enough valgrind that has already been updated to support the new AVX etc. instructions....)

Exactly! A technology-oriented website should strive for giving insights through analysis. Also, fragmenting a single topic - in this case analysis of the Bulldozer architecture - into multiple articles that just present the reader with bare benchmark data is IMO pointless.

I have previously mentioned this in the forums and not surprisingly Michael didn't comment. However, a fanatic-sounding guy jumped at me that I shouldn't expect Michael to be my brain-prosthesis - or something like that. :P

I think this kind of articles are driving down the quality of Phoronix and make it similar content-mills.

[yoshi314]

somebody explain - what is the point of measuring time TO COMPILE a piece of software?

it's obvious that more aggressive optimizations will increase build time, and -O0 will be the fastest.

my idea of benchmarking a compiler would be to measure build time of a source package with DIFFERENT BUILDS of the same compiler under the same CFLAGS, not comparing different cflags against the same build of compiler.

[staalmannen]

somebody explain - what is the point of measuring time TO COMPILE a piece of software?

it's obvious that more aggressive optimizations will increase build time, and -O0 will be the fastest.

my idea of benchmarking a compiler would be to measure build time of a source package with DIFFERENT BUILDS of the same compiler under the same CFLAGS, not comparing different cflags against the same build of compiler.

Actually you have a point that compile time is a relatively "boring" measure since it probably has less impact on the choice of compiler (although you might want to have a fast compiler during development and then a really good optimizing one for when you want to distribute binaries). The interesting measures would be:

* size of the resulting binaries
* performance of the resulting binaries