Bottleneck on Mac OSX

113 views
Skip to first unread message

Keith Goodman

unread,
Jun 26, 2011, 11:57:46 AM6/26/11
to bottl...@googlegroups.com
Has anyone installed bottleneck 0.5.0 on Mac OSX?

According to activestate [1] bottleneck 0.5.0 builds on linux and
windows but not on mac osx [2]. It hangs while compiling the moving
window functions. I suspect move_median since activestate reports that
the previous version of bottleneck builds fine on osx.

[1] http://code.activestate.com/pypm/bottleneck
[2] http://pypm-free.activestate.com/2.6/macosx10.5-i386-x86_64/pool/b/bo/bottleneck-0.5.0_macosx10.5-i386-x86_64_2.6_1.pypm.d/log

Wes McKinney

unread,
Jun 26, 2011, 1:15:56 PM6/26/11
to bottl...@googlegroups.com

I was able to compile Bottleneck in universal mode on OS 10.6 at
least. There are actually some more serious problems in the install
process in that I can't install this version of Bottleneck into 64-bit
EPD (which lives in EPD64.framework and not Python.framework).

- Wes

Keith Goodman

unread,
Jun 26, 2011, 1:27:03 PM6/26/11
to bottl...@googlegroups.com

Thanks, Wes. That's good to know.

Well, at least there is partial support for OSX.

Dani Valevski

unread,
Oct 6, 2011, 4:40:13 PM10/6/11
to bottl...@googlegroups.com
Wes, Can you please elaborate on how you installed bottleneck on Mac OSX? 

Keith Goodman

unread,
Oct 6, 2011, 4:54:20 PM10/6/11
to bottl...@googlegroups.com
On Thu, Oct 6, 2011 at 1:40 PM, Dani Valevski <dan...@gmail.com> wrote:
> Wes, Can you please elaborate on how you installed bottleneck on Mac OSX?

I'm curious about happens on osx when you do:

$ python setup.py build

If you have cython installed then you can do:

$ make all

in the top-level directory. That will build in place.

Dani Valevski

unread,
Oct 6, 2011, 5:50:17 PM10/6/11
to bottl...@googlegroups.com
Looks like the normal install works perfectly, if you are patient enough:

python setup.py build
python setup.py install

Maybe it's worth adding a note to the README saying that setup might take around 30 minutes with no console output (except a compiler warning).

Keith Goodman

unread,
Oct 6, 2011, 6:06:56 PM10/6/11
to bottl...@googlegroups.com

Wow, that is slow. I hope the bottleneck functions are fast enough to
make up for the lost 30 minutes :)

I recommend running bn.test() and bn.bench().

Thanks for the suggestion. README updated:

https://github.com/kwgoodman/bottleneck/commit/9ee4215620664422d8e90e962a5f6e86041f874e

Keith Goodman

unread,
Oct 6, 2011, 6:22:23 PM10/6/11
to bottl...@googlegroups.com
On Thu, Oct 6, 2011 at 2:50 PM, Dani Valevski <dan...@gmail.com> wrote:

> Maybe it's worth adding a note to the README saying that setup might take
> around 30 minutes with no console output (except a compiler warning).

How old is your mac?

Looking at the link in the first email in this thread I see the reason
why activestate failed to build bottleneck on osx was that it timed
out after 900 seconds. Linux and windows did not have that problem.
Odd.

Dani Valevski

unread,
Oct 6, 2011, 6:28:24 PM10/6/11
to bottl...@googlegroups.com
My machine isn't that old -- 
2.53 Ghz Intel Core 2 Duo, 4GB Memory, OS x 10.6.8

All the tests pass, and here are some of the benchmarks:

Bottleneck performance benchmark
    Bottleneck  0.5.0
    Numpy (np)  1.6.1
    Scipy (sp)  0.9.0
    Speed is NumPy or SciPy time divided by Bottleneck time
    NaN means one-third NaNs; float64 and axis=0 are used
    High-level functions used (mode='fast')

                 no NaN     no NaN     no NaN      NaN        NaN        NaN    
                (10,10)   (100,100) (1000,1000)  (10,10)   (100,100) (1000,1000)
median            3.58       1.98       1.73       3.80       3.34       1.97
nanmedian       106.62      38.07       3.81     124.26      82.73       5.35
nansum            6.94       2.53       3.29       6.97       3.91       3.73
nanmax            1.63       1.10       0.97       1.71       1.87       1.10
nanmean          16.02       7.33       4.27      17.85      20.15       7.85
nanstd           22.19       5.84       5.08      23.72       8.14       6.84
nanargmax         6.40       4.42       5.32       6.97       6.63       5.62
ss                3.62       2.69       2.71       3.22       2.97       2.02
rankdata         27.71      24.53      14.24      27.66      23.40      15.44
partsort          1.25       2.02       2.09       1.27       4.71       2.23
argpartsort       0.61       1.85       1.32       0.56       2.73       1.16


Reply all
Reply to author
Forward
0 new messages