This site will undergo a maintenance outage on Saturday, 13 September beginning at 2:30 AM Central/1 PM IST for a platform upgrade. The downtime will be ~three hours.

Note on using normalized zAAP/zIIP Utilization

On some zSeries models (2086, 2096, 2094-4xx, -5xx and -6xx), the speed of General Purpose Central Processor (GCP) is lowered artificially.

IBM sometime refers to these models as knee-capped processors.

However, the speed of specialized processors (zAAP, zIIP, ICF, and IFL) on these models has not changed.

Therefore, the same number of instructions will execute faster on specialized processors than on the GCP. The ratio of GCP and specialized processors speed is reported in RMF type 72 record fields, R723NFFI and R723NFFS, for zAAP and zIIP respectively. This ratio for different zSeries models ranges from 1.25 to 15.

Related topic

Visualizer-graphs

In Visualizer graphs, the zAAP/zIIP utilization is treated differently:

  • In all Partition graphs zAAP/zIIP utilization is displayed as actual % Dispatch time, that is, as ratio of actual time the processor(s) of a particular type was dispatched to the interval length:

    %Dispatch time = (Dispatch time in sec) / (Interval length in sec) x 100

    This means that in the Partition with N zAAP processors, zAAP Dispatch time can in theory go up to N x 100%.

    Important

    The following times include Dispatch Time for both GCP and all specialized processors:

    • Partition Total Dispatch Time
    • Effective Dispatch Time
    • Management time and Physical System Total Dispatch Time
    • Effective Dispatch Time and Management Time

  • Normalized zAAP/zIIP Utilization is displayed in all of the following:

    • Logical System zAAP/zIIP Utilization graphs
    • Workload zAAP/zIIP Utilization graphs
    • Suite zAAP/zIIP Utilization graphs
    • Subsystem Address Space zAAP/zIIP Utilization graphs
    (Normalized zAAP/zIIP Utilization) = (Actual zAAP/zIIP busy time) / (Interval length) x (Ratio of zAAP/zIIP speed to the ratio of GCP in the same Physical system)

    The Physical meaning of normalized Utilization is the Utilization of GCP, if all work executed on the specialized processor is executed on a GCP instead.

    Example

    An LPAR on 2086-110 has 1 GCP and 1 zAAP.
    The Partition Dispatch Time by Engine type shows 50% GCP dispatch and 80% zAAP dispatch. But zAAP, zIIP, and CP Utilization Analysis shows 50% CP utilization and 1120% zAAP normalized utilization (as the speed of zAAP is about 14 times higher than the speed of GCP on 2086-110).

    Note that on all Logical System CPU Utilization, Workload Utilization, Suite Utilization, Application Utilization, and so on, the Utilization metric does not include the utilization of specialized processors.

 

Tip: For faster searching, add an asterisk to the end of your partial query. Example: cert*