On 2010/03/26 13:57:33, zundel wrote:
http://codereview.chromium.org/1105009/diff/1/2
File include/v8-profiler.h (right):
http://codereview.chromium.org/1105009/diff/1/2#newcode88
include/v8-profiler.h:88: */
On 2010/03/23 09:35:37, Michail Naganov wrote:
> On 2010/03/23 08:25:11, Søren Gjesse wrote:
> > Should we consider using uint64_t and change to a higher resolution
than
> > milliseconds?
>
> I propose doubles for easier interconnection with JavaScript, where
Numbers
are
> doubles.
>
> I think there is no need to try to achieve a better resolution, than
> milliseconds with a software-only (that is, not using CPU h/w counters)
> profiler.
I concur on the double change, plus it would allow you to specify
sub-millisecond if you needed to.
I've been playing with profiling activity over short intervals. I did an
experimental change to the profiler on Mac which uses usleep() with lower
granularity profiles and it was somewhat useful (if not entirely
accurate).
But
the best profile timer (linux itimer TIMER_PROFILE) only gave about 2-4ms
granularity when set to 1ms anyway.
Yes. Higher sampling frequencies are possible on Mac and Windows, but not on
Linux.
http://codereview.chromium.org/1105009
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
To unsubscribe from this group, send email to v8-dev+unsubscribegooglegroups.com or reply
to this email with the words "REMOVE ME" as the subject.
