Why are the readings in perfmon and zabbix perf_counter different?

Question

Zabbix

Good afternoon.
Please tell me how to correctly analyze the processor queue length counter data obtained via zabbix get and in the perfmon snap-in itself?
With an average queue in perfmon of 120, zabbix_get gives values ​​of about 0.5
With an average queue in perfmon of 90, zabbix_get gives values ​​of about 2

It seems that they still calculate the load in different ways, despite the fact that the data for calculation are taken the same ( perf_counter values).

As far as I understand, critical loads will be considered if the queue is higher than 2 * number of cores, and, judging by the server load, the value in perfmon is close to this option.

zabbix_get gets data via perf_counter, from the same counter.