In the PICO noise simulations, I am running with just 1 pixel and 1 MPI process per node and I can fit 3 days worth of data per node. I am using a configuration very similar to the tiny satellite example for everything else.
PICO is 1 KHz, so I expect 1 day of 1 pixel to occupy 1000*3600*24*8*2/1e9 = 1.4 GB
so 3 days would be 4.2 GB
I see madam allocating 6.1 GB for TOD + 14 GB pointing.
TiMemory reports 34 GB current and 53 GB peak.
I am just checking if you think all these numbers look reasonable, so I can use the metric of fitting between 4 GB and 5 GB of data per node to guess the initial size of jobs.
In the PICO noise simulations, I am running with just 1 pixel and 1 MPI process per node and I can fit 3 days worth of data per node. I am using a configuration very similar to the tiny satellite example for everything else. PICO is 1 KHz, so I expect 1 day of 1 pixel to occupy
1000*3600*24*8*2/1e9 = 1.4 GB
so 3 days would be
4.2 GB
I see madam allocating
6.1 GB
for TOD +14 GB
pointing.TiMemory reports
34 GB
current and53 GB
peak.I am just checking if you think all these numbers look reasonable, so I can use the metric of fitting between
4 GB
and5 GB
of data per node to guess the initial size of jobs.