very different compilation times [no optimisation]

[Ian Lance Taylor]

Dave Korn <dave.korn.cygwin@googlemail.com> writes:

> Basile STARYNKEVITCH wrote:
>
>>>> Notice that gcc 4.4 is running four times slower that gcc-4.3!
>
>> glinka ~/tmp 16:39 % gcc-4.3 -ftime-report -O0 -g -fPIC
>> warmelt-normal-1.i -c -o warmelt-normal-1-gcc-4.3.o
>> 
>> Execution times (seconds)
>>  TOTAL                 :  33.13             3.33           
>> 36.64             363506 kB
>
>> glinka ~/tmp 16:40 % gcc-4.4 -ftime-report -O0 -g -fPIC
>> warmelt-normal-1.i -c -o warmelt-normal-1-gcc-4.4.o
>>  TOTAL                 :  10.83             3.03           
>> 13.90             429820 kB
>> glinka ~/tmp 16:41 %
>
>   Wait, what? :-P
>
>
>   Assuming you got those results swapped round, I see it's mostly expand and
> global alloc.  Side effect of IRA on functions with loooong unbroken sequences
> of initialisations maybe?

He  didn't swap them--you can tell by the presence of "integrated RA"
rather than "global".

Basile, you posted the time for the wrong file, or you got something
else wrong.

Ian