lkml.org 
[lkml]   [2004]   [Aug]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: RCU issue with SELinux (Re: SELINUX performance issues)
    On Tue, 24 Aug 2004, Kaigai Kohei wrote:

    > o UNIXbench
    >
    > * INDEX value comparison
    > 2.6.8.1 2.6.8.1 2.6.8.1 2.6.8.1
    > (Disable) (Enable) (rwlock) (RCU)
    > Dhrystone 2 using register variables 268.9 268.8 269.2 269.0
    > Double-Precision Whetstone 94.2 94.2 94.2 94.2
    > Execl Throughput 388.3 379.0 377.8 377.9 +
    > File Copy 1024 bufsize 2000 maxblocks 606.6 526.6 515.6 504.8 *
    > File Copy 256 bufsize 500 maxblocks 508.9 417.0 410.4 395.2 *
    > File Copy 4096 bufsize 8000 maxblocks 987.1 890.4 876.0 857.9 *
    > Pipe Throughput 525.1 406.4 404.5 408.8 +
    > Process Creation 321.2 317.8 315.9 316.3 +
    > Shell Scripts (8 concurrent) 1312.8 1276.2 1278.8 1282.8 +
    > System Call Overhead 467.1 468.7 464.1 467.2 +
    > ========================================
    > FINAL SCORE 445.8 413.2 410.1 407.7

    This benchmark somewhat characterizes 1P performance, and the ones I've
    marked with (*) get noticably worse with the RCU patch compared to the
    current locking scheme. Tests marked (+) show no or insignificant
    improvement.

    Might be useful to compare with the lmbench macrobenchmark, to see if it
    shows a similar pattern.

    > o dbench [ 4 processes run parallely on 4-CPUs / 10 times trials ]
    > ---- mean ---- - STD -
    > 2.6.8.1(disable) 860.249 [MB/s] 44.683
    > 2.6.8.1(enable) 714.254 [MB/s] 32.359
    > 2.6.8.1(+rwlock) 767.904 [MB/s] 27.968
    > 2.6.8.1(+RCU) 830.678 [MB/s] 16.352

    Can you show the figures for 1 and 2 clients?

    > In IA-32 or x86_64, can anybady implement atomic_inc_return()?
    > If it can not, I'll try to make alternative macros or inline functions.

    If you can get this done, it will be very useful, as I could allso run
    some benchmarks on my test systems.


    - James
    --
    James Morris
    <jmorris@redhat.com>





    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

    \
     
     \ /
      Last update: 2005-03-22 14:05    [W:0.024 / U:153.748 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site