stream 内存带宽测试Docker性能损失

逃离我推掉我的手 2022-12-18 10:51 238阅读 0赞

# 1。 宿主机测试 #

下载源码

[yeqiang@harbor STREAM]$ git clone https://github.com/jeffhammond/STREAM

单线程测试

[yeqiang@harbor STREAM]$ gcc -O  -DSTREAM_ARRAY_SIZE=1000000000 -DNTIME=20 -mcmodel=large stream.c -o stream
    [yeqiang@harbor STREAM]$ ./stream 
    -------------------------------------------------------------
    STREAM version $Revision: 5.10 $
    -------------------------------------------------------------
    This system uses 8 bytes per array element.
    -------------------------------------------------------------
    Array size = 1000000000 (elements), Offset = 0 (elements)
    Memory per array = 7629.4 MiB (= 7.5 GiB).
    Total memory required = 22888.2 MiB (= 22.4 GiB).
    Each kernel will be executed 10 times.
     The *best* time for each kernel (excluding the first iteration)
     will be used to compute the reported bandwidth.
    -------------------------------------------------------------
    Your clock granularity/precision appears to be 1 microseconds.
    Each test below will take on the order of 566061 microseconds.
       (= 566061 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    -------------------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    -------------------------------------------------------------
    Function    Best Rate MB/s  Avg time     Min time     Max time
    Copy:           20774.8     0.774718     0.770165     0.783492
    Scale:          20689.9     0.778666     0.773325     0.790584
    Add:            23589.8     1.023929     1.017391     1.042266
    Triad:          23338.4     1.032000     1.028349     1.039814
    -------------------------------------------------------------
    Solution Validates: avg error less than 1.000000e-13 on all three arrays

多线程测试

[yeqiang@harbor STREAM]$ gcc -O -fopenmp  -DSTREAM_ARRAY_SIZE=1000000000 -DNTIME=20 -mcmodel=large stream.c -o stream
    [yeqiang@harbor STREAM]$ ./stream 
    -------------------------------------------------------------
    STREAM version $Revision: 5.10 $
    -------------------------------------------------------------
    This system uses 8 bytes per array element.
    -------------------------------------------------------------
    Array size = 1000000000 (elements), Offset = 0 (elements)
    Memory per array = 7629.4 MiB (= 7.5 GiB).
    Total memory required = 22888.2 MiB (= 22.4 GiB).
    Each kernel will be executed 10 times.
     The *best* time for each kernel (excluding the first iteration)
     will be used to compute the reported bandwidth.
    -------------------------------------------------------------
    Number of Threads requested = 6
    Number of Threads counted = 6
    -------------------------------------------------------------
    Your clock granularity/precision appears to be 1 microseconds.
    Each test below will take on the order of 498106 microseconds.
       (= 498106 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    -------------------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    -------------------------------------------------------------
    Function    Best Rate MB/s  Avg time     Min time     Max time
    Copy:           21930.7     0.748195     0.729571     0.782713
    Scale:          22101.5     0.738321     0.723932     0.757996
    Add:            24832.2     0.978239     0.966488     0.999065
    Triad:          24901.5     0.975178     0.963797     0.987780
    -------------------------------------------------------------
    Solution Validates: avg error less than 1.000000e-13 on all three arrays
    -------------------------------------------------------------

# 2. Docker内STREAM测试 #

单线程

[root@c6f0d2296598 STREAM]# gcc -O  -DSTREAM_ARRAY_SIZE=1000000000 -DNTIME=20 -mcmodel=large stream.c -o stream
    [root@c6f0d2296598 STREAM]# ./stream 
    -------------------------------------------------------------
    STREAM version $Revision: 5.10 $
    -------------------------------------------------------------
    This system uses 8 bytes per array element.
    -------------------------------------------------------------
    Array size = 1000000000 (elements), Offset = 0 (elements)
    Memory per array = 7629.4 MiB (= 7.5 GiB).
    Total memory required = 22888.2 MiB (= 22.4 GiB).
    Each kernel will be executed 10 times.
     The *best* time for each kernel (excluding the first iteration)
     will be used to compute the reported bandwidth.
    -------------------------------------------------------------
    Your clock granularity/precision appears to be 1 microseconds.
    Each test below will take on the order of 555674 microseconds.
       (= 555674 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    -------------------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    -------------------------------------------------------------
    Function    Best Rate MB/s  Avg time     Min time     Max time
    Copy:           21290.9     0.756061     0.751496     0.763196
    Scale:          20938.2     0.772817     0.764154     0.785505
    Add:            24033.4     1.004074     0.998611     1.014896
    Triad:          23819.8     1.013162     1.007565     1.021040
    -------------------------------------------------------------
    Solution Validates: avg error less than 1.000000e-13 on all three arrays
    -------------------------------------------------------------

多线程

[root@c6f0d2296598 STREAM]# gcc -O -fopenmp  -DSTREAM_ARRAY_SIZE=1000000000 -DNTIME=20 -mcmodel=large stream.c -o stream
    [root@c6f0d2296598 STREAM]#./stream 
    -------------------------------------------------------------
    STREAM version $Revision: 5.10 $
    -------------------------------------------------------------
    This system uses 8 bytes per array element.
    -------------------------------------------------------------
    Array size = 1000000000 (elements), Offset = 0 (elements)
    Memory per array = 7629.4 MiB (= 7.5 GiB).
    Total memory required = 22888.2 MiB (= 22.4 GiB).
    Each kernel will be executed 10 times.
     The *best* time for each kernel (excluding the first iteration)
     will be used to compute the reported bandwidth.
    -------------------------------------------------------------
    Number of Threads requested = 6
    Number of Threads counted = 6
    -------------------------------------------------------------
    Your clock granularity/precision appears to be 1 microseconds.
    Each test below will take on the order of 494348 microseconds.
       (= 494348 clock ticks)
    Increase the size of the arrays if this shows that
    you are not getting at least 20 clock ticks per test.
    -------------------------------------------------------------
    WARNING -- The above is only a rough guideline.
    For best results, please be sure you know the
    precision of your system timer.
    -------------------------------------------------------------
    Function    Best Rate MB/s  Avg time     Min time     Max time
    Copy:           22041.1     0.728465     0.725915     0.733348
    Scale:          22164.6     0.727715     0.721871     0.744801
    Add:            25010.1     0.967849     0.959611     0.979415
    Triad:          25066.3     0.964457     0.957460     0.972716
    -------------------------------------------------------------
    Solution Validates: avg error less than 1.000000e-13 on all three arrays
    -------------------------------------------------------------

# 总结 #

Docker对内存性能无明星影响