[Skimmertalk] How did your Skimmer Server computer do in CQWW

[Pete Smith]

I agree - this is very useful data.  Makes me wonder how many spots we 
could have delivered on the RBN if Skimmers were not throttled by CPU 
capacity.

I was away over the weekend - drat! - but my Skimmer continued to run 
throughout.  I second Dave's idea that a log of CPU usage and number of 
decoders vs time would be useful.

For the record, I am using a puny E2200 dual core Pentium with 1.75 GB 
of RAM (.25 used by the onboard video).  I typically run 96 KHz 
bandwidth on 7 bands, and doubt that this machine would support even 
1000 decoders.  For the contest weekend I was running on the contest 
bands only, 160-10.

73, Pete N4ZR

The World Contest Station Database, updated daily at www.conteststations.com
The Reverse Beacon Network at http://reversebeacon.net, blog at reversebeacon.blogspot.com,
spots at telnet.reversebeacon.net, port 7000 AND now
at arcluster.reversebeacon.net port 7000



On 11/28/2011 6:19 PM, Dave Pascoe wrote:
> Wes-
>
> Thanks for starting this thread.  My Skimmer Server has only been in
> production mode for a short while - previously it was on and off over
> the last year during experiments.  I believe it may be the only
> QS1R-based server in New England at this time.
>
> I was at the KC1XX Multi-Multi operation and not closely monitoring my
> Skimmer until I saw a couple of emails over the weekend hinting at other
> Skimmer Server operators having load issues.  I then remotely logged in
> to my machine to poke around.  I did need to limit the number of bands
> covered at various times of the day but probably ran the bulk of the
> first 20 hours or so at reduced capacity.  I felt that it was better to
> get 100% decoding than to run more bands at reduced capacity - wonder
> what others feel about that.
>
> Machine/CPU: Dual-core Pentium 4 - 3.4 GHz/1GB RAM
> Receiver: QS1R
> OS: Windows XP SP 3 32-bit
>
> Notes:
> - On this hardware CPU would become overloaded approaching 2,000
> decoders (usually around 1,800 or so)
> - Generally I was forced to run no more than 3 bands at a time
>        - I could not cover 14,21, and all of 28 MHz (had to limit to
> 28.000-28.091)
>       - At night I generally covered 160-40m (which seemed to do OK)
>
> I should have kept closer watch on the number of decoders running at
> various times but that would have taken a lot of effort.  I was busy
> contesting.  :-)  It would be nice to have an option in Skimmer to log
> the number of decoders on each band every X minutes to a log file or to
> a remote syslog server.
>
> I am pretty convinced that it may be time to get an i7/2600 or something
> more powerful, but it only seems to be needed in periods of very high
> activity (contests).  The other option is to run a pair of QS1Rs and two
> separate machines (but more $$$).
>
> Based on Wes' data, I think there may be a somewhat nonlinear relation
> with number of cores and number of decoders that can safely run.  That
> is, increasing cores may not linearly increase the number of possible
> decoders.  But we need more detailed data (and data from other stations)
> to understand this better.
>
> 73,
> Dave KM3T
> Amherst, NH
>
> On 11/28/11 9:14 AM, Wes Cosand wrote:
>> --How did your machine perform?
>> --How many decoders were you seeing?
>> --Are you using 32 or 64 bit OS?
>> --What is your CPU
>>
>> My Core 2 Quad Q8300 could not handle more than 2800 decoders and I
>> was seeing many more than that on 5 bands.  I am running XP Pro (32
>> bit)
>>
>> Wes, WZ7I
> _______________________________________________
> Skimmertalk mailing list
> Skimmertalk at contesting.com
> http://dayton.contesting.com/mailman/listinfo/skimmertalk
>
>