Prints information about wall-clock time taken by various sub-processes. Running this command causes all delayed commands to be executed. The output includes the following time classes.
General overhead, e.g. parsing inputs, finding correct commands to run etc. This includes the total overhead time spent in the main process, and in possible cluster job processes.
Time spent in I/O-bound processing. This is the time when the disk I/O is the bottleneck. This includes the total I/O time spent in the main process, and in possible cluster job processes. Time spent in output data compression is counted to this time class.
Time spent in CPU/GPU-bound processing. This is the time when the CPU/GPU is the bottleneck. This includes the total computation time spent in the main process, and in possible cluster job processes. This is the default mode for all commands.
Total distributed job execution time. This value includes Overhead+IO+Computation of all jobs, plus workload manager node reservation, process starting, etc. overhead. This value does not include time spent in workload manager queue.
Total distributed job queuing time. This is the total time all jobs have spent in the workload manager queue, waiting to be executed.
Total job waiting
Total time from submitting the first distributed job until all of them are found to be finished. This is the total time spent in the job execution process, from submission to jobs until all of them are done.
Time spent in preparing for writing output images (e.g. NN5 write preparation). This is the total time spent in the submitting process while preparing writing of output images.
Total write finalization waiting
Time spent in write finalization jobs, including queuing (e.g. NN5 write finalization jobs). This is the total time spent in the submitting process, from submission of write finalization jobs until all of them are done.
This command can be used in the distributed processing mode, but it does not participate in distributed processing.