Changes between Initial Version and Version 1 of PJMEDIA-MIPS


Ignore:
Timestamp:
Jul 4, 2008 11:10:50 PM (16 years ago)
Author:
bennylp
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • PJMEDIA-MIPS

    v1 v1  
     1= PJMEDIA MIPS Measurement = 
     2 
     3This page shows the CPU requirements/MIPS (Million Instructions Per Second) measurements of various PJMEDIA components which would be useful to evaluate PJMEDIA performance. Please do not interpret these numbers as the official or definite performance number, as there are many compilation flags in PJMEDIA that can be set, as well as compiler switches, to improve the performance. 
     4 
     5All the results below are done with the stock settings that come with PJSIP distribution. The test source is in {{{pjmedia/src/test}}} directory. 
     6 
     7== Interpreting the Results == 
     8 
     9=== Columns === 
     10 
     11There are four columns in the result table: 
     12 
     13 '''Clock Rate:''' :: 
     14  This shows the sampling rate of the component being tested, in Hz. We test both 8KHz and 16Hz. Most components can work in both 8KHz and 16KHz hence there will be two test results row for the same component, each with different clock rate. Some components (mostly related codec) can only work in one of the clock rate (e.g. GSM is only shown in 8KHz, while G.722 is only shown in 16KHz). 
     15 
     16 '''Time (usec):''' :: 
     17 This shows the time elapsed to process 1 second worth of audio samples, in microseconds. 
     18 
     19 '''CPU (%):''' :: 
     20 From the elapsed time above, we can measure how much CPU usage needed to run this component in real-time. For example, if the time elapsed is 1 second (one million microseconds) then this component will take 100% of CPU time when run in real-time. Or if the time elapsed is 0.5 second (500 thousands microseconds) then this component will take 50% of the CPU time when run in real-time. 
     21 
     22 '''MIPS:''' :: 
     23 Also from the elapsed time above, we can measure the MIPS needed to run this component in real-time.  
     24 
     25 
     26=== Rows === 
     27 
     28The rows shows the measurement result of a particular components. The components tested are described below. 
     29 
     30 '''get from memplayer:''' :: 
     31  The memory/buffer based player port supplies the audio samples for almost all of the tests, so its time adds as overhead for all tests. 
     32 
     33 '''conference bridge with N call(s):''' :: 
     34  This measures the performance of the conference bridge with N calls. Note that we don't use actual call for the test since we only want to measure the conference bridge performance and not codec performance (this will be measured in separate tests). So for this test we use memplayer for each ''call'' to supply audio to the bridge. 
     35 
     36 '''upsample+downsample:''' :: 
     37  This measures the performance of the resampling algorithm used. The test gets the audio from the memplayer, upsample it twice the clock rate, then downsample it half the clock rate again so that the clock rate now is the same as originally. This test measures linear resampling and polyphase resampling using small filter and large filter. 
     38 
     39 '''WSOLA PLC - N% loss:''' :: 
     40  This measures the performance of Waveform Similarity based Overlap and Add (WSOLA) algorithm when it is used to generate/emulate lost packet (a.k.a Packet Lost Concealment/PLC). Timing for various loss percentage is shown. 
     41 
     42 '''WSOLA discard N% excess:''' :: 
     43  This measures the performance of Waveform Similarity based Overlap and Add (WSOLA) algorithm when it is used to discard excess audio samples (e.g. caused by clock drifts). Timing for various excess percentage is shown. 
     44 
     45 '''echo canceller Nms tail len:''' :: 
     46  This measures the performance of the acoustic echo canceller (AEC) for various echo tail settings. 
     47 
     48 '''tone generator with single/dual freq:''' :: 
     49  This measures the performance of the tone generator to continuously generate single or dual frequency tone. 
     50 
     51 '''codec encode/decode:''' :: 
     52  This measures the time to encode and then decode 1 second worth of audio samples using the specified codec. 
     53 
     54 '''stream TX/RX:''' :: 
     55  This test is intended to measure the performance/overhead of the stream, which consist of RTP/RTCP processing and de-jitter buffering. In addition it also tests the performance of Secure RTP (SRTP) for various settings combination and codec bandwidth. Since the test here also consists of codec processing (actual encoding and decoding), you need to subtract the result with the result of the corresponding codec to measure the overhead of the stream and SRTP only. 
     56 
     57 
     58== Results == 
     59 
     60=== PJSIP-0.9.0, ARM9 (ARM926EJ-S) Linux === 
     61 
     62||Hardware:||Olimex SAM9-L9260 board|| 
     63||Platform:||Linux 2.6.23|| 
     64||Processor:||ARM926EJ-S rev 5 (v5l)|| 
     65||Speed:||180 MHz|| 
     66||Assumed MIPS:||198 MIPS|| 
     67||BogoMIPS:||98.91|| 
     68||Compilation:||arm-926-linux-gnu-gcc -O2 -msoft-float -DNDEBUG -DPJ_HAS_FLOATING_POINT=0|| 
     69 
     70Result: 
     71{{{ 
     7200:59:38.531 os_core_unix.c pjlib 0.9.0-trunk for POSIX initialized 
     73MIPS test, with CPU=180Mhz,  198.0 MIPS 
     74Clock  Item                                      Time     CPU    MIPS 
     75 Rate                                           (usec)    (%) 
     76---------------------------------------------------------------------- 
     77 8KHz get from memplayer                          181    0.018    0.04 
     78 8KHz conference bridge with 1 call              6682    0.668    1.32 
     79 8KHz conference bridge with 2 calls            11943    1.194    2.36 
     80 8KHz conference bridge with 4 calls            22402    2.240    4.44 
     81 8KHz conference bridge with 8 calls            42969    4.297    8.51 
     82 8KHz conference bridge with 16 calls           83328    8.333   16.50 
     83 8KHz upsample+downsample - linear               5815    0.581    1.15 
     84 8KHz upsample+downsample - small filter        66786    6.679   13.22 
     85 8KHz upsample+downsample - large filter       870754   87.075  172.41 
     86 8KHz WSOLA PLC - 0% loss                         605    0.060    0.12 
     87 8KHz WSOLA PLC - 2% loss                        1004    0.100    0.20 
     88 8KHz WSOLA PLC - 5% loss                        1541    0.154    0.31 
     89 8KHz WSOLA PLC - 10% loss                       1803    0.180    0.36 
     90 8KHz WSOLA PLC - 20% loss                       3102    0.310    0.61 
     91 8KHz WSOLA PLC - 50% loss                       8431    0.843    1.67 
     92 8KHz WSOLA discard 2% excess                     214    0.021    0.04 
     93 8KHz WSOLA discard 5% excess                     488    0.049    0.10 
     94 8KHz WSOLA discard 10% excess                   1178    0.118    0.23 
     95 8KHz WSOLA discard 20% excess                   2009    0.201    0.40 
     96 8KHz WSOLA discard 50% excess                   6432    0.643    1.27 
     97 8KHz echo canceller 100ms tail len            335870   33.587   66.50 
     98 8KHz echo canceller 128ms tail len            336225   33.623   66.57 
     99 8KHz echo canceller 200ms tail len            349240   34.924   69.15 
     100 8KHz echo canceller 256ms tail len            363206   36.321   71.91 
     101 8KHz echo canceller 400ms tail len            400026   40.003   79.21 
     102 8KHz echo canceller 500ms tail len            426646   42.665   84.48 
     103 8KHz echo canceller 512ms tail len            432291   43.229   85.59 
     104 8KHz echo canceller 600ms tail len            454965   45.496   90.08 
     105 8KHz echo canceller 800ms tail len            516487   51.649  102.26 
     106 8KHz tone generator with single freq             920    0.092    0.18 
     107 8KHz tone generator with dual freq              1428    0.143    0.28 
     108 8KHz codec encode/decode - G.711                2701    0.270    0.53 
     109 8KHz codec encode/decode - GSM                 75750    7.575   15.00 
     110 8KHz codec encode/decode - iLBC              2856203  285.620  565.53 
     111 8KHz codec encode/decode - Speex 8Khz         436162   43.616   86.36 
     112 8KHz codec encode/decode - L16/8000/1           1704    0.170    0.34 
     113 8KHz stream TX/RX - G.711                       6786    0.679    1.34 
     114 8KHz stream TX/RX - G.711 SRTP 32bit           21688    2.169    4.29 
     115 8KHz stream TX/RX - G.711 SRTP 32bit +auth     33501    3.350    6.63 
     116 8KHz stream TX/RX - G.711 SRTP 80bit           21725    2.172    4.30 
     117 8KHz stream TX/RX - G.711 SRTP 80bit +auth     33551    3.355    6.64 
     118 8KHz stream TX/RX - GSM                        82035    8.203   16.24 
     119 8KHz stream TX/RX - GSM SRTP 32bit             90890    9.089   18.00 
     120 8KHz stream TX/RX - GSM SRTP 32bit + auth      99334    9.933   19.67 
     121 8KHz stream TX/RX - GSM SRTP 80bit             90893    9.089   18.00 
     122 8KHz stream TX/RX - GSM SRTP 80bit + auth      99356    9.936   19.67 
     12316KHz get from memplayer                          239    0.024    0.05 
     12416KHz conference bridge with 1 call             12780    1.278    2.53 
     12516KHz conference bridge with 2 calls            23052    2.305    4.56 
     12616KHz conference bridge with 4 calls            43174    4.317    8.55 
     12716KHz conference bridge with 8 calls            82096    8.210   16.26 
     12816KHz conference bridge with 16 calls          158565   15.856   31.40 
     12916KHz upsample+downsample - linear              11469    1.147    2.27 
     13016KHz upsample+downsample - small filter       133088   13.309   26.35 
     13116KHz upsample+downsample - large filter      1739742  173.974  344.47 
     13216KHz WSOLA PLC - 0% loss                         980    0.098    0.19 
     13316KHz WSOLA PLC - 2% loss                        1910    0.191    0.38 
     13416KHz WSOLA PLC - 5% loss                        3734    0.373    0.74 
     13516KHz WSOLA PLC - 10% loss                       7867    0.787    1.56 
     13616KHz WSOLA PLC - 20% loss                      13007    1.301    2.58 
     13716KHz WSOLA PLC - 50% loss                      29022    2.902    5.75 
     13816KHz WSOLA discard 2% excess                     551    0.055    0.11 
     13916KHz WSOLA discard 5% excess                    1027    0.103    0.20 
     14016KHz WSOLA discard 10% excess                   1973    0.197    0.39 
     14116KHz WSOLA discard 20% excess                  10454    1.045    2.07 
     14216KHz WSOLA discard 50% excess                  22276    2.228    4.41 
     14316KHz echo canceller 100ms tail len            664649   66.465  131.60 
     14416KHz echo canceller 128ms tail len            682686   68.269  135.17 
     14516KHz echo canceller 200ms tail len            720924   72.092  142.74 
     14616KHz echo canceller 256ms tail len            752928   75.293  149.08 
     14716KHz echo canceller 400ms tail len            877528   87.753  173.75 
     14816KHz echo canceller 500ms tail len            970559   97.056  192.17 
     14916KHz echo canceller 512ms tail len            989839   98.984  195.99 
     15016KHz echo canceller 600ms tail len           1065465  106.547  210.96 
     15116KHz echo canceller 800ms tail len           1285075  128.508  254.44 
     15216KHz tone generator with single freq            1617    0.162    0.32 
     15316KHz tone generator with dual freq              2632    0.263    0.52 
     15416KHz codec encode/decode - G.722              148080   14.808   29.32 
     15516KHz codec encode/decode - Speex 16Khz        979202   97.920  193.88 
     15616KHz codec encode/decode - L16/16000/1          3244    0.324    0.64 
     15716KHz stream TX/RX - G.722                     155685   15.568   30.83 
     158}}}