Re: [PATCH] add statistics counting to postreload, copy-rename, and math-opts

[Richard Guenther <richard.guenther@gmail.com>]

On Tue, Apr 12, 2011 at 5:09 PM, Nathan Froyd <froydnj@codesourcery.com> wrote:
> On Tue, Apr 12, 2011 at 04:54:43PM +0200, Richard Guenther wrote:
>> On Tue, Apr 12, 2011 at 4:51 PM, Nathan Froyd <froydnj@codesourcery.com> wrote:
>> > True, but maybe those testcases should be adjusted--per-pass flags,
>> > rather than blindly assuming -O2 includes them.  And it's not clear to
>>
>> It's easier to add things to GCC than to argue removing things ...
>
> And sometimes not even easy to argue for adding things. :)
>
>> > me that the statistics_counter_event infrastructure really helps
>> > catching do-nothing passes, since it doesn't record stats that increment
>> > by zero...
>>
>> Well, if the overall count is zero then nothing was done.
>
> Granted, but that fact should still be recorded.  The situation we have
> today, for something like:
>
> func1: statistic for "statx" was 0
>  - nothing is recorded in the statistics table
> func2: statistic for "statx" was 0
>  - nothing is recorded in the statistics table
> func3: statistic for "statx" was 0
>  - nothing is recorded in the statistics table
> ...
>
> and so forth, is that at the end of the day, the dump file won't even
> include any information about "statx".  If you had some func7387 where
> "statx" was non-zero, you could infer that nothing else happened in the
> previous 7386 functions.  For the case where a pass is truly useless on
> a TU, it's hard to figure out from the statistics dump alone.  And I'd
> argue that it's useful to see explicitly that the pass only helped in 1
> out of 7387 functions, rather than trying to infer it from missing data.

I always use statistics-stats (thus, overall stats, not per function).  The
per function ones omit zero counts during dumping on purpose
(to make the dump smaller).

Richard.

> -Nathan
>