[flac-dev] 1.21 vs 1.3 encoding speed

[lvqcl]

Erik de Castro Lopo wrote:

>> I just built FLAC and noticed that the size of flac.exe is noticeably bigger,
>> so I compared the generated Makefiles before ang after this change.
>>
>> The difference is: "-g -O2" options were added to CFLAGS.
>>
>> before:
>> CFLAGS = -O3 -funroll-loops -Wall -W -Winline -Wall -Wextra -Wstrict-prototypes
>> -Wmissing-prototypes -Waggregate-return -Wcast-align -Wnested-externs -Wshadow
>> -Wundef -Wmissing-declarations -Wunreachable-code  -Wdeclaration-after-statement
>> -D_FORTIFY_SOURCE=2 -msse2 -Wextra
>>
>> after:
>> CFLAGS = -O3 -funroll-loops -Wall -W -Winline -g -O2 -Wall -Wextra
>> -Wstrict-prototypes -Wmissing-prototypes -Waggregate-return -Wcast-align
>> -Wnested-externs -Wshadow -Wundef -Wmissing-declarations -Wunreachable-code
>> -Wdeclaration-after-statement -D_FORTIFY_SOURCE=2 -msse2 -Wextra
>
> Well -g asks the compiler to add debug symbols. What hapens if you strip
> the binary?


After stripping: the binary is smaller than previously (520kB vs 652kB).

Probably that's because the compiler now uses -O2: according to
<https://gcc.gnu.org/onlinedocs/gcc-4.9.1/gcc/Optimize-Options.html>
"If you use multiple -O options, with or without level numbers,
the last such option is the one that is effective" which means that
GCC doesn't try to use SIMD to vectorize the code, etc.