[SciPy-user] Benchmark data

Travis Oliphant oliphant.travis at ieee.org
Fri Dec 9 04:14:49 CST 2005


I'd like people to try out scipy core in SVN.  I made improvements to the
buffered ufunc section of code that I think will make a big difference
in the recently published benchmarks. 

Here are some results that I get now (I've got a debug build with no 
optimizations).

Optimization flags: -g -Wall -Wstrict-prototypes
CPU info: getNCPUs has_3dnow has_3dnowext has_mmx is_32bit is_AMD 
is_singleCPU
Numeric-24.2
numarray-1.5.1
scipy-core-0.8.1.1614
benchmark size = 4  (vectors of length 256)
label            Numeric       numarray     scipy.base
    1          0.0001121       0.002115      9.108e-05
    2          0.0001349      0.0008631       0.000133
    3          7.105e-05       0.000989      8.202e-05
    4          0.0001509       0.003712      0.0001481
    5           6.39e-05        0.04328      6.986e-05
    6          6.008e-05      5.412e-05      9.012e-05
    7          0.0001211      0.0003929       0.000129
    8          4.292e-05       0.000118      3.004e-05
    9          0.0005081        0.00541      0.0006039
   10          0.0004249      0.0006261        0.00056
   11           0.000329       0.000932       0.000422
TOTAL           0.002019         0.0585       0.002359

Optimization flags: -g -Wall -Wstrict-prototypes
CPU info: getNCPUs has_3dnow has_3dnowext has_mmx is_32bit is_AMD 
is_singleCPU
Numeric-24.2
numarray-1.5.1
scipy-core-0.8.1.1614
benchmark size = 6  (vectors of length 4096)
label            Numeric       numarray     scipy.base
    1           0.000401       0.002003      0.0003419
    2          0.0003641       0.001114      0.0004022
    3           0.000263       0.001245       0.000315
    4          0.0005591       0.004512       0.000577
    5           0.000298       0.001587       0.000283
    6           0.000268       0.000284      0.0002999
    7           0.000464       0.000725      0.0004301
    8          0.0002861       0.000536      0.0001659
    9           0.004739         0.0104       0.004864
   10            0.04717       0.004666       0.004523
   11           0.003594       0.004486       0.003582
TOTAL             0.0584        0.03156        0.01578


Optimization flags: -g -Wall -Wstrict-prototypes
CPU info: getNCPUs has_3dnow has_3dnowext has_mmx is_32bit is_AMD 
is_singleCPU
Numeric-24.2
numarray-1.5.1
scipy-core-0.8.1.1614
benchmark size = 11  (vectors of length 4194304)
label            Numeric       numarray     scipy.base
    1              0.414          0.231         0.3357
    2             0.3932          0.509         0.4095
    3             0.2833         0.3737         0.3759
    4              1.157          1.118         0.6301
    5             0.4359         0.4895         0.3693
    6             0.2868         0.3933         0.3205
    7              1.211          1.354         0.6187
    8             0.7242         0.8617         0.4587
    9              8.506          12.46          7.865
   10              9.726          13.13          6.891
   11              7.518          10.44          8.416
TOTAL              30.66          41.35          26.69


I think these benchmarks are showing that scipy base is at least doing 
better than it was

-Travis





More information about the SciPy-user mailing list