[Skimmertalk] Skimmer in CQWW and CPU Utilization Problem

[W3OA]

I ran a Skimmer for 48 hours during CQWW and I think it was a great 
success.  My score and number of q's went up 15 percent over last year 
for the same number of hours worked.

The SDR was a Softrock Lite+Xtall version 8.3.  The antenna connection 
was from the Mark-V Rx Antenna Out jack.  I used a relay to cut the 
audio connection between the Softrock and the sound card to prevent a 
lot of false signals in the waterfall while I was transmitting.

I had Skimmer set for 96 kHz sampling rate, Do not send callsigns 
without "CQ", and Paranoid validation.  The Master.dta file was one I 
had modified to remove all the US calls.  Thus the only spots sent out 
were for "counters" that were participants in previous contests.  I used 
Wintelnetx to combine Skimmer spots with spots from a Cluster node.  
These were fed into a simple chat server program to allow other Carolina 
DX Association users to see the combined spot stream via a telnet 
connection.  My local display was the N1MM bandmap.

Skimmer was on a dedicated computer, a 3.2 Ghz Pentium 4 with 2 gigs of 
RAM.  I noticed a couple of times (like the bottom 96 kHz of 40 during 
the first hour and 20 on Saturday morning) the CPU utilization was 
bouncing up to 100 percent and "Decoders" showed something like 600 of 
1000 or 500 of 1300.  During these times the number of spots put out 
seemed to drop dramatically.  I wonder if this is normal operation?  I'm 
thinking a way to get around this is to use 48 kHz sampling rate and 
switch between band segments every so often, maybe ever 5 minutes.  I'd 
appreciate hearing comments from others on this.

73 - Dick, W3OA