an idea how to get more out of latency measurements:

when nailing down latency issues, often there's a correlation between observed 
latency, and some other command being run in parallel [1]

the current latency-test doesnt help much except for 'reset statistics, observe 
again, take sceenshot'

what would really help is a rolling history display, like in a spectrum 
analyzer waterfall [2], and a way to tag it with timestamps correlating to 
command execution.

this would mean a 2D plot, scrolling up with time (Y=time)
on the X axis you'd have horizontal colored bars, like for min/avg/max latency 
in green/yellow/red

ideally one would be able to tag a line with some shell command, maybe just a 
text mark.

I'm out of my depth how to do this, but I think it'd be a great help to 
visualize and communicate obvserved behavior

- Michael

------

[1] yesterday we had a long 'track down these latency spikes' session on IRC; 
log is here: 
http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-02-11.html#16:43:23

if we had had such a tool, the impact could have been show by a screenshot

[2] example for a live spectrum waterfall: http://websdr.ewi.utwente.nl:8901/ 
(read here before: http://sdrrocks.blogspot.co.at/2008/04/web-based-sdr.html; 
stunning tool btw!)
------------------------------------------------------------------------------
Free Next-Gen Firewall Hardware Offer
Buy your Sophos next-gen firewall before the end March 2013 
and get the hardware for free! Learn more.
http://p.sf.net/sfu/sophos-d2d-feb
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to