ATA_PuMa_FlyEye - david-macmahon/wiki_convert_test GitHub Wiki
Joeri's FlyEye ATA PuMa node log
SIGPROC
Installed sigproc 3.7, killed a bug (cf. install page)
filterbank-ata-02 runs fast (
30-min crab data dedispersed to 20 dm's 50-70, fg'd over 6 procs: 1m23
tee'd over 20 processes: 1m22 (75% use, over 8 procs), so appears CPU bound, not IO.
from ddplan below there are 44 x 250 trials per obs = 10.000/obs
if 20=1m22 then 10.000 = ~1hr = ~2x realtime
DDPLAN
(oak/FlyEye:)
/export/home/local/src/presto/bin/DDplan.py -l 100 -d 2000 -n 128 -t 0.000625 -f 1430 -b 208 -o ddplan.out.dm100-2000.128chan.625us.1430MHz.208MHz.eps
Minimum total smearing : 0.997 ms
--------------------------------------------
Minimum channel smearing : 0.461 ms
Minimum smearing across BW : 0.00295 ms
Minimum sample time : 0.625 ms
Setting the new 'best' resolution to : 0.625 ms
Best guess for optimal initial dDM is 2.118
Low DM High DM dDM DownSamp #DMs WorkFract
100.000 462.000 2.00 1 181 0.6925
462.000 720.000 3.00 2 86 0.1645
720.000 1290.000 5.00 4 114 0.109
1290.000 2000.000 10.00 8 71 0.03396
NOTES
currently no web access from machine so more notes kept local at
joeri@pulsar-1:~/notes.txt