I am interested in evaluating the performance of SciDB, but I don’t need to show the output for now. I tried to use “set no fetch” command and “set timer” command, but the timer will be masked by the “no fetch” mode.
How can I get the time statistics without showing the large output on the console? Or can I store the output in an (existing) array? I tried to use the AFL store function, but it couldn’t work for me.
Furthermore, a concrete example is the window aggregation. If I simply perform a window aggregation over 1 GB data, and if I set no fetch, then the query will be executed in one second. I really doubt if the query has actually been executed in this mode. That’s TOOOO fast, since if I use the same grid size for the grid aggregation (by using the regrid function), it will still take more than 25 seconds on my PC, and we know that a window aggregation should calculate way more grid aggregates than the grid aggregation with the same grid size. Is there anyone who can explain that?