[gambit-list] Performance of different versions of gcc

Bradley Lucier lucier at math.purdue.edu
Sun Apr 29 13:21:26 EDT 2012


On Apr 29, 2012, at 12:51 PM, Marc Feeley wrote:

> On 2012-04-29, at 10:52 AM, Bradley Lucier wrote:
>> 
>> What I'm trying to say is that while this test gives a good study of gcc's default behavior, a broader study would be needed to see which optimizations are applied to Gambit-generated code, how long they take at compile time, and how much they benefit execution speed.
> 
> Are you suggesting this should be done for every version of gcc?

I'm not suggesting anything, except not to draw too broad conclusions from this broad study.

>  It is hard enough to do for a single version, and the effort will mostly have to be redone for the next release.

I've found that '-fschedule-insns' and '-frename-registers' help in x86_64 and powerpc code (not in i386 code, which doesn't have enough registers).  It would be hard, without actually adding the code to gcc itself to implement -Wdisabled-optimization at all the right spots, to see what optimizations are actually performed, say, on _io.c with -O2.

> If I were a gcc developer I might be motivated to do this,

I'm sorry, I've lost the thread here, what is "this"?

> Would you or someone else be interested in doing this?

Adding the warnings in all the right places is on my "bucket list".  (It's such a boring list, really.)

Brad


More information about the Gambit-list mailing list