AIX

AIX

Connect with fellow AIX users and experts to gain knowledge, share insights, and solve problems.


#Power
 View Only
  • 1.  Processor Load Balancing

    Posted Thu February 04, 2010 04:27 AM

    Originally posted by: sck


    Hi,

    As show below in the sar output system has four processor but only two of them are high utilized. Two processor remains idle most of the time due to which system shows high CPU utilization.

    Is there any way by which I can share the system load to each of the running CPU, so that system will not reflect high CPU utilization.

    smt is enabled on the system.
    AIX bwprd1 3 5 0002BFB8D700 02/04/10

    System configuration: lcpu=4

    14:37:00 cpu %usr %sys %wio %idle physc
    14:39:10 0 12 11 65 12 0.57
    1 4 2 10 85 0.43
    2 12 11 64 13 0.57
    3 5 2 11 82 0.43
    - 8 7 42 43 2.00
    14:41:20 0 12 14 65 9 0.58
    1 3 3 10 83 0.42
    2 11 13 65 11 0.57
    3 4 3 13 80 0.43
    - 8 9 42 40 2.00
    14:43:30 0 11 10 69 10 0.57
    1 3 2 7 89 0.43
    2 11 10 69 10 0.57
    3 3 2 8 87 0.43
    - 7 7 42 44 2.00
    14:45:40 0 11 11 67 11 0.57
    1 3 2 7 88 0.43
    2 11 11 66 12 0.57
    3 4 2 10 83 0.43
    - 8 7 42 43 2.00
    14:47:50 0 10 10 68 12 0.57
    1 3 2 7 88 0.43
    2 10 10 68 12 0.57
    3 4 2 7 88 0.43
    - 7 6 42 45 2.00

    Average 0 11 11 67 11 0.57
    1 3 2 8 86 0.43
    2 11 11 66 12 0.57
    3 4 2 10 84 0.43
    - 8 7 42 43 2.00
    #AIX-Forum


  • 2.  Re: Processor Load Balancing

    Posted Thu February 04, 2010 06:07 AM

    Originally posted by: MarkTaylor


    You have 4 x Logical CPUs, I am assuming that you have 2 x Online Virtual and SMT turned on ? In which case it seems you application does not make use of SMT .. I would read up on SMT if I were you.

    Ref: https://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp?topic=/com.ibm.aix.genprogc/doc/genprogc/smt.htm

    Benefitting from Simultaneous Multi-Threading

    Simultaneous multi-threading is primarily beneficial in commercial environments where the speed of an individual transaction is not as important as the total number of transactions that are performed. Simultaneous multi-threading is expected to increase the throughput of workloads with large or frequently changing working sets, such as database servers and Web servers.
    Workloads that see the greatest simultaneous multi-threading benefit are those that have a high Cycles Per Instruction (CPI) count. These workloads tend to use processor and memory resources poorly. Large CPIs are usually caused by high cache-miss rates from a large working set. Large commercial workloads are somewhat dependent upon whether the two hardware threads share instructions or data, or the hardware threads are completely distinct. Large commercial workloads typically have this characteristic. Workloads that share instructions or data, including those that run extensively in the operating system or within a single application, might see increased benefits from simultaneous multi-threading.
    Workloads that do not benefit much from simultaneous multi-threading are those in which the majority of individual software threads use a large amount of any resource in the processor or memory. For example, workloads that are floating-point intensive are likely to gain little from simultaneous multi-threading and are the ones most likely to loose performance. These workloads heavily use either the floating-point units or the memory bandwidth. Workloads with low CPI and low cache miss rates might see a some small benefit.
    Measurements taken on a dedicated partition with commercial workloads indicated a 25%-40% increase in throughput. Simultaneous multi-threading is should help shared processor partition processing. The extra threads give the partition a boost after simultaneous multi-threading is dispatched because the partition recovers its working set more quickly. Subsequently, the threads perform like they would in a dedicated partition. Although it might be somewhat counterintuitive, simultaneous multi-threading performs best when the performance of the cache is at its worst.

    HTH
    Mark Taylor
    #AIX-Forum


  • 3.  Re: Processor Load Balancing

    Posted Thu February 04, 2010 07:51 AM

    Originally posted by: j.gann


    also, consider how much runnable threads you actually have. a single process/thread application will never make use of more than a single processor. a unix scheduler for good reason will try to schedule a thread onto the same processor during each turn it gets to run. thereby increasing the probability that the thread's code and data still be in the processor's caches.

    you are writing about high cpu utilization which maybe a simplistic monitoring application flags you as warning, right? monitoring system load should be more appropriate.

    j.gann
    #AIX-Forum