I am trying to understand the defintion of CPU, CPU USE, CPU SKEW, IMPACT CPU. I have referred viewpoint manual for this but still not able to grab it. Can anybody explain how these parametrs are calculated.
I have a related question to calculate number CPU cycles a particular system can support. For ex: I have a 6700 production box. If i want to calculate the capacity of that box in terms of N number of cycles, how can i do that.
I heard there is a formula to do that. Any comments will be appreciable.
Thanks in advance.
The number of CPU seconds for the entire system, which is used to calculate metrics such as CPU Use as mentioned above, can be calculated quite easily. Assuming a homogenous system, the formula is:
Number of nodes * Number of CPUs per node * Number of cores per CPU
In the case of coexistent systems Viewpoint accounts for that as it gets the number of CPU cores on each individual node in the system.
Viewpoint always assumes that 100% of the CPU cycles are available when calculating these types of metrics. In reality, there's some system overhead that will consume a small percentage of the CPU cycles and therefore those won't be available for user work.
Thank you for the quick answer.
If i wanted to know exactly how many cpu cycles available for 1hr on PROD box with 6650C and 5555H cooexistence system, I think my below formula is correct. Please correct me if i am wrong. I multiplied by 2 for hyperthreading.
NumberOfNode * NumberOfCPUsPerNode*NumberOfCoresPerCPU*2forHyperT
And while counting the nodes, i think i need to consider hot standby nodes too.
I don't think you should multiply by 2 for hyperthreading. Viewpoint doesn't specifically account for hyperthreading. It relies upon the number of CPUs returned from PM/API, which is also available in the ResUsageSPMA table.
I see, thank you.
I agree for Viewpoint server while calculating the CPU seconds no need to multiply by 2, but what about other systems like 6650, 5550 etc boxes.?