[Sugar-devel] OT: determining memory usage of short-lived processes

[Sascha Silbe]

On Mon, Jun 08, 2009 at 05:31:59AM -0400, Albert Cahalan wrote:

>> I've examined memory usage of long running processes (i.e. daemons
>> and applications) in the past, no problems. But for my VCS comparison
>> I need to determine the peak memory usage of all child processes
>> (combined), which are rather short-lived. What's the best way to do
>> that?
> atop, vmstat, sar,
These are system-level tools and don't tell me anything about a 
particular process.

> pidstat,
Doesn't support memory statistics if used with child processes and 
probably won't work with short-lived processes.

> sa, accton
Up to now closest to what I'm looking for, thanks! I gives average 
memory size instead of peak, though.

> Never rule out valgrind. Find a way.
Well, the way is to either let the benchmark run for several days or 
reduce the sample size (and thus accuracy). I had hoped there's some way 
to avoid that.

> Valgrind supports tools that you can use to discover where you suffer 
> from cache line misses. Valgrind finds any use of uninitialized 
> memory.
While quite useful during development, these won't help me with 
comparing memory usage of existing VCSs (the "massif" valgrind tool does 
help, though).

CU Sascha

-- 
http://sascha.silbe.org/
http://www.infra-silbe.de/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: Digital signature
Url : http://lists.sugarlabs.org/archive/sugar-devel/attachments/20090609/b684fb48/attachment.pgp