Low perfomance at Group By in Script

Wed, 23 Jul 2025 14:48:52 GMT

Who may be watching the level of CPU utilization during the execution of Grouping?

While creating a model, you must count the number of entries for each day.

Perform this simple but CPU usage does not exceed 10-15% if you do the Group By in the script.

In the case where data are loaded ungrouped and creates a graph with dimension date and the expression of the counter, the CPU 100%.

This situation would not cause problems if the data volume is insignificant, but in my case it is tens of gigabytes on a server with 12 physical cores and load the script takes about 30 minutes, but if you calculate in the interface table, the execution time is less than 20 seconds.

Is there a solution to help calculate the grouping of the data in the script with maximum efficiency?

I know that if calculated on the SQL server, it will be faster, but my source is QVD.

Example to try:

'detailed':
LOAD
          Floor(RowNo()/10000) as dimention,
          RowNo() as value
AutoGenerate 10000000;
'grouped':
LOAD
          dimention,
          Sum(value) as sum_value
Resident detailed
Group By dimention;

topic Low perfomance at Group By in Script in Archived Groups

Low perfomance at Group By in Script