Re: [Chicken-hackers] Why isn't OPTIMIZE_FOR_SPEED the default?

[Sven Hartrumpf]

Mon, 3 Nov 2014 09:40:09 +0100, Peter.Bex wrote:
> On Mon, Nov 03, 2014 at 09:38:37AM +0100, Sven Hartrumpf wrote:
>> Mon, 3 Nov 2014 09:04:27 +0100, Peter.Bex wrote:
>> > How about restoring the optimization option to the defaults and seeing
>> > what breaks?  For me it was highly unexpected that CHICKEN was producing
>> > completely unoptimized code, and I'm pretty sure I'm not the only one,
>> > so if this doesn't break on too many systems, it's a saner default.
>> > If it's too aggressive, we could try -O2 instead or even -O1.
>> 
>> Just my experience with a large application (chicken generates one C file
>> of 700 KLOC) and gcc 4.8.N:
>> Chicken built
>> - with "-O2    -fomit-frame-pointer -march=corei7": has been ok for 2 years 
>> or so
>> - with "-Ofast -fomit-frame-pointer -march=corei7": caused errors (last 
>> checked
>>   in late 2013; I can retry if someone is interested)
> 
> Cool, that's useful info!  What is -Ofast?  Is it -O3, or something higher?
> If retrying isn't too much of a hassle, I'd be interested to know the results
> with a recent GCC (and perhaps clang, too).

OK. I reran my tests with GCC 4.9.2. -Ofast works without any problems,
but is only slightly faster (1-2 %) than -O2 for my binaries 

> Using -O2 by default would be a big improvement already, I think.

Yes.

Sven