Originally posted by: JohnnyCloud

I'm trying to calculate the CPU utilization of an LPAR but I get wrong results. 
I can see the utilization in the HMC Web Interface, but Id like to calculate it myself for own purposes.

My test: I take an LPAR that has 0.2 EC and current utilization of 0.6 (0.6/0.2 = 3 -> 300%). So I am to expect 300% from this particular LPAR.

I login to the HMC, and run 

lslparutil -m <machine> -r lpar -n 2 --filter lpar_names=<lpar> -F time,capped_cycles,uncapped_cycles,entitled_cycles

I get the values from the last two minutes
04/25/2017 11:21:13,294282846772027,683494280098951,1050755263230847
04/25/2017 11:19:43,294273753737790,683455429211045,1050746050402243

I then use the formula to calculate processor utilization from IBMs manual on lslparutil (https://www.ibm.com/support/knowledgecenter/en/8408-E8E/p8edm/lslparutil.html)

Processor utilization % = ((capped_cycles + uncapped_cycles) / entitled_cycles) * 100 
Processor utilization % = (((294282846772027 - 294273753737790) + (683494280098951 - 683455429211045)) /  (1050755263230847 - 1050746050402243)) * 100 
Processor utilization % = 520.40%

I get a processor utilization of 520%, not 300%. Something seems way off here. What could it be?

Thanks!

Originally posted by: AncientAIXer

I don't know for sure, but I think it may be from comparing 2 different measurements that have differing time periods.  One is known, two minutes.  I don't know what algorithm the HMC uses to present the current utilization; if its a point-in-time or average over a time period.  IBM support might be able to answer that.