[Numpy-discussion] numpy.dot and ACML
Yves Frederix
yves.frederix at gmail.com
Mon Feb 19 12:11:20 EST 2007
Hi all,
I have managed to compile numpy using pathscale and ACML on a 64 bit AMD
system. Now I wanted to verify that numpy.dot indeed uses the ACML
libs. The example for dot()
(http://www.scipy.org/Numpy_Example_List?highlight=%28example%29#head-c7a573f030ff7cbaea62baf219599b3976136bac) suggest a way of doing this:
1 u0050015 at lo-03-02 .../core $ python -c "import numpy; print id(numpy.dot)==id(numpy.core.multiarray.dot);"
True
This indicates that I am not using the acml libraries.
When running a benchmark (see attach) and comparing to a non-ACML
installation though, the strange thing is that there is a clear
speed difference, suggesting again that the acml libraries are indeed
used.
Because this is not all that clear to me, I was wondering whether there
exists an alternative way of verifying what libraries are used.
Many thanks,
YVES
-------------- next part --------------
ACML:
dim x.T*y x*y.T A*x A*B A.T*x
-----------------------------------------------------------------
5000 0.002492 0.002417 0.002412 0.002399 0.002416
50000 0.020074 0.020024 0.020004 0.020003 0.020024
100000 0.092777 0.093690 0.100220 0.093787 0.094250
200000 0.184933 0.198623 0.196120 0.197089 0.197273
300000 0.276583 0.279177 0.280898 0.284016 0.276204
500000 0.476340 0.481987 0.471875 0.480868 0.481501
1000000.0 0.892623 0.895500 0.915173 0.894815 0.922501
5000000.0 4.450555 4.465748 4.467870 4.468188 4.469083
No ACML:
dim x.T*y x*y.T A*x A*B A.T*x
-----------------------------------------------------------------
5000 0.002523 0.002428 0.002410 0.002430 0.002419
50000 0.024756 0.061520 0.036575 0.036399 0.036450
100000 0.338576 0.353074 0.169472 0.302087 0.334633
200000 0.670803 0.735732 0.538166 0.649335 0.744496
300000 1.004381 1.269259 0.482542 2.194308 0.611997
500000 1.110656 1.504701 1.571736 1.656021 1.491146
1000000.0 2.182746 2.234478 2.254645 2.439508 2.537558
5000000.0 10.878910 16.578266 8.265109 8.905976 17.124400
More information about the NumPy-Discussion
mailing list