Smalltalk › Gemtalk › GLASS

[Glass] Where is the rest of the CPU time spent when using ProfMonit?

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

2 messages Options

GLASS mailing list

[Glass] Where is the rest of the CPU time spent when using ProfMonit?

Hi guys,

I am profiling something that is slow, using the ProfMonit. What I don't understand is the following:

================

STACK SAMPLING TREE RESULTS

elapsed CPU time: 12585 ms

monitoring interval: 10.0 ms

reporting down to: 1 hits

100.0% (1097) JadeServer >> doIt:in:

100.0% (1097) JadeServer >> evaluate:inContext:

100.0% (1097) String >> evaluateInContext:symbolList:

100.0% (1097) executed code [UndefinedObject]

So the profiler says the total CPU elapsed time is 12585. And indeed, if I do a #timeToRun I also see a similar value. However....the 100% of the stack is only 1097.

So... where are the 12585-1097 spent? How can I see that? I need to improve the performance of this scenario and I don't know where to start because I have no clue what is taking the time....

Thanks in advance,

--
Mariano
http://marianopeck.wordpress.com

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

GLASS mailing list

Re: [Glass] Where is the rest of the CPU time spent when using ProfMonit?

On Mon, Oct 27, 2014 at 10:55 AM, Mariano Martinez Peck via Glass <[hidden email]> wrote:

Hi guys,

I am profiling something that is slow, using the ProfMonit. What I don't understand is the following:

================
STACK SAMPLING TREE RESULTS
elapsed CPU time: 12585 ms
monitoring interval: 10.0 ms
reporting down to: 1 hits

100.0% (1097) JadeServer >> doIt:in:
100.0% (1097) JadeServer >> evaluate:inContext:
100.0% (1097) String >> evaluateInContext:symbolList:
100.0% (1097) executed code [UndefinedObject]

I don't often use this format (It's created from ProfMonitorTree) to look at results (I have a Profile Monitory in tODE that I prefer to use since I can view method source, callers and senders breakdown), but reading through the code it looks like the value in parens is the tally or the count of samples in which the method showed up on the stack ... so presumably there were 1097 samples (and at 10ms per sample we're in the neighborhood)...

Dale

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass