PSN-L Email List Message

Subject: Re: Drop Outs
From: Brett Nordgren brett3nt@.............
Date: Tue, 25 Aug 2009 11:12:43 -0400


Hi Larry and Dick,

Adding more memory did help significantly after I went from 500MB to 
1500MB.  My own problem seems to relate to the fact that the screen 
represented 24 hours of data at 200 samples per second.  That is 17.28 
million samples x the number of bytes stored per sample.  The dropouts 
still happen, but not nearly so often now.

Since then I have been doing some more checking with the Performance 
Monitor (XP) and have a couple of observations.  For me, dropouts mostly 
happen when redisplaying the screen (F8), not so often when the screen 
redraws itself at the end of the line, nor when generating the 
gif's.  Although virtual memory reads and writes go up a lot during a 
redisplay, so far I haven't found any indication that memory is causing a 
big problem.  At times the swapping rates to virtual memory get quite high, 
but as far as I can tell the machine appears to be handling it.

However.....  During redisplay, the processor usage goes solidly to 
100%.  I am wondering how much processor effort it takes to filter 17 
million data points?  I had the feeling that the problem was in some way 
related to filtering, but I never could really see what the relationship 
was.  I am pretty sure that when filtering is turned off the redisplay 
dropouts disappear.  I'll try to confirm that.

Hope there is something useful here.

Brett

At 06:02 PM 8/24/2009 -0700, you wrote:
>Hi Dick,
>
>Adding more RAM memory may help. It fixed Brett Nordgren's problem he 
>report the other day.
>
>Regards,
>Larry Cochrane
>Webtronics

Watch our wiggles
http://bnordgren.org/seismo/gif_images.htm

or watch some very very good wiggles
http://aslwww.cr.usgs.gov/Seismic_Data/telemetry_data/ANMO_24hr.html


__________________________________________________________

Public Seismic Network Mailing List (PSN-L)


[ Top ] [ Back ] [ Home Page ]