This is the mail archive of the
mailing list for the GCC project.
Re: inconsistent gcc performance on this code
At 12:55 PM 7/18/04 -0400, you wrote:
>On Sun, Jul 18, 2004 at 12:02:43PM -0400, Jason Papadopoulos wrote:
>> Compiling fgt5b with '-O3 -fomit-frame-pointer' generates
>> code that runs ~20% faster and is ~25% smaller than fgt5a.
>> The only difference between the two files is that 5a
>> writes each result to a different variable, while 5b
>> sometimes reuses the same set of 8 variables for common
>> (temporary) operations.
>You may want to compare the output of GCC HEAD (pre-3.5) on both files.
>SSA form excels at this sort of splitting.
Latest pre-3.5 snapshot still shows a timing difference; the
difference has narrowed because the fast runtime got slower
(by half the difference) and the slow time was unchanged.
Does this version of gcc use SSA form automatically? Also, using
-fnew-ra causes my code generator to segfault.