As a first investigation I took Neil Schemenauer’s bpnn.py (a 200 line back-prop artificial neural network library with doc and comparison). The intention was to see how much faster the code might run using psyco and shedskin.

The results were really quite surprising, notes and src follow.

The run-time in minutes for the harder test case are:

1. Vanilla Python 153 minutes
2. Python + Psyco 1.6.0.final.0 57 minutes (2.6* faster)
3. Shedskin 0.0.29 0.78 minutes (196* faster)

The test machines uses Python 2.5.2 on Ubuntu 8.04. The box is an Intel Core Duo 2.4GHz running a single process.

The ‘hard’ problem trains the ANN using 508 patterns with 57 input neurons, 50 hidden and 62 output neurons over 1000 iterations. If you know ANNs then the configuration (0.1 learning rate, 0 momentum) might seem unusual, be assured that this is correct for my researcher’s problem.

There is a shorter version of this problem using just 2 patterns, this is useful if you want to replicate these results but don’t want to wait 3 hours on your first run.

My run times for the shorter problem are:

1. Vanilla Python 42 seconds
2. Python + Psyco 14 seconds
3. Shedskin 0.2 seconds (210* faster)
   

Shedskin has an issue with numerical stability - it seems that internally some truncation occurs with floating point math. Whilst the results for vanilla Python and Python+Psyco were identical, the results with Shedskin were similar but with fractional divergences in each result.

Whilst these divergences caused some very different results in the final weights for the ANN, my researcher confirms that all the results look equivalent.

Mark Dufour (Shedskin’s author) confirms that Python’s C double is used the same in Shedskin but notes that rounding (or a bug) may be the culprit. Shedskin is a young project, Mark will welcome extra eyes if you want to look into this.

Running the code with Shedskin was fairly easy. On Ubuntu I had to install libgc-dev and libpcre3-dev (detailed in the Shedskin docs) and g++, afterwards shedskin was ready. From download to first run was 15 minutes.

On my first attempt to compile bpnn.py with Shedskin I received an error as the ‘raise’ keyword isn’t yet supported. I replaced the ‘raise’ calls with ‘assert False’ for sanity, afterwards compilation was fine.

Edit - Mark notes that the basic form of ‘raise’ is supported but the version used in bpnn.py isn’t yet supported.  Something like ‘raise ValueError(’some msg’)’ works fine.

Mark notes that Shedskin currently works well up to 500 lines (maybe up to 1000), since bpnn.py is only 200 lines compilation is quick.

Note that if you can’t use Psyco because you aren’t on x86, Shedskin might be useful to you since it’ll work anywhere that Python and g++ compile.

Running this yourself

If you want to recreate my results, download bpnn_shedskin_src_20081117.zip. You’ll see bpnn_shedskin.py, this is the main code. bpnn_shedskin.py includes either ‘examples_short.py’ or ‘examples_full.py’, short is the easier 2 pattern problem and full has 508 patterns.

Note that these patterns are stored as lists of tuples (Shedskin doesn’t support the csv module so I hardcoded the input patterns to speed development), the full version is over 500 lines of Python and this slows Shedskin’s compilation somewhat.

By default the imports for Psyco are commented out and the short problem is configured. At the command line you’ll get an output like this:

python bpnn_shedskin.py
Using 2 examples
ANN uses 57 input, 50 hidden, 62 output, 1000 iterations, 0.100000 learning rate, 0.000000 momentum
error 65.454309      2008-11-17 15:22:58.318593
error 45.176110      2008-11-17 15:22:59.060787
error 44.616933      2008-11-17 15:23:00.246280
error 44.026883      2008-11-17 15:23:01.743821
error 44.049276      2008-11-17 15:23:02.815876
error 44.905183      2008-11-17 15:23:03.860352
error 44.674506      2008-11-17 15:23:05.270307
error 43.365627      2008-11-17 15:23:06.757126
error 43.299160      2008-11-17 15:23:08.244466
error 42.540076      2008-11-17 15:23:09.732035
Elapsed: 0:00:41.472192

If you uncomment the two Psyco lines your code will run about 2.6* faster.

Using Shedskin

To use shedskin, first run the Python through shedskin and then ‘make’ the result. The compiled binary will run much faster than the vanilla Python code, the result below shows the short problem taking 0.19 seconds compared to 41 seconds above.

shedskin bpnn_shedskin.py
*** SHED SKIN Python-to-C++ Compiler 0.0.29 ***
Copyright 2005-2008 Mark Dufour; License GNU GPL version 3 (See LICENSE)
[iterative type analysis..]
***
iterations: 3 templates: 519
[generating c++ code..]
*WARNING* bpnn_shedskin.py:178: function (class NN, ‘weights’) not called!
*WARNING* bpnn_shedskin.py:156: function (class NN, ‘test’) not called!

make
g++  -O2 -pipe -Wno-deprecated  -I. -I/usr/lib/shedskin/lib /usr/lib/shedskin/lib/string.cpp /usr/lib/shedskin/lib/random.cpp /usr/lib/shedskin/lib/datetime.cpp examples_short.cpp bpnn_shedskin.cpp /usr/lib/shedskin/lib/builtin.cpp /usr/lib/shedskin/lib/time.cpp /usr/lib/shedskin/lib/math.cpp -lgc  -o bpnn_shedskin

./bpnn_shedskin
Using 2 examples
ANN uses 57 input, 50 hidden, 62 output, 1000 iterations, 0.100000 learning rate, 0.000000 momentum
error 65.454309      2008-11-17 16:11:08.452087
error 44.970416      2008-11-17 16:11:08.476869
error 46.444249      2008-11-17 16:11:08.506324
error 44.209054      2008-11-17 16:11:08.519375
error 44.058518      2008-11-17 16:11:08.532430
error 45.655892      2008-11-17 16:11:08.545741
error 44.518816      2008-11-17 16:11:08.558520
error 43.643572      2008-11-17 16:11:08.571705
error 44.800429      2008-11-17 16:11:08.584241
error 43.710905      2008-11-17 16:11:08.597465
Elapsed: 0:00:00.198747

Why is the math different?

An open question remains as to why the evolution of the floating point arithmetic is different between Python and Shedskin. If anyone is interested in delving in to this, I’d be very interested in hearing from you.

Extension modules

Mark notes that the extension module support is perhaps a more useful way to use Shedskin for this sort of problem.

A single module can be compiled (e.g. ’shedskin -e module.py’) and with Python you just import it (e.g. ‘import module’) and use it…with a big speed-up.

This ties the code to your installed libs - not so great for easy distribution but great for lone researchers needing a speed boost.

Shedskin 0.1 in the works

Mark’s plan is to get 0.1 released over the coming months. One aim is to get the extension module to a similar level of functionality as SWIG and improve the core library support so that Shedskin comes with (some more) Batteries Included.

Mark is open to receiving code (up to 1000 lines) that doesn’t compile.  The project would always happily accept new contributors.

See the Shedskin homepage, blog and group.

This is the same location that we used for the last meet and again we’re combining this with the Brighton freelancer Farm meetup for a larger crowd.

Some more details are in this previous post, hope to see you there!

We threw one months back along with Paul Silver’s The Farm freelancers, we filled the The Hampton Arms with close to 60 geeks IIRC.  The pub is a 15 minute walk from Brighton station (note to GnuBlade and other Londoners, this pub is much closer to the station than the building we use for £5 App meets!).

Amongst others we had Jim of SecondLife UK, a Microsoftie who likes lightweight languages, several pyQT programmers from a medical company and a whole bunch of others.

As before it’ll just be a meet n’greet pub outing, I’m expecting 10-20 Pythonistas along with 20-40 Farmers (they’re PHP, Web, db, graphics etc local freelancers), lots of beer and conversation.

Please sign-up (Upcoming) so we’ve an idea of who will be along.

Yesterday we had Mark Shuttleworth’s key-note on how he wanted to see Python give better support to Transactional Memory and Cloud Computing.Both talks interesting and inspiring. It felt like Mark really buys into the Python community with lots of ‘our language’ references, he’s a great speaker to boot.  He dug at our GIL and said ‘please solve scaling above 1 core!’.

Raymond Hettinger gave some great talks including a look at Python 2.6 and 3.0, behind-the-scenes Python containers (cool src link) and A.I. with Python.  Did you know that a list’s growth pattern is 0, 4, 8, 16, 25, 35, 46, 58, 72, 88, … elements and for large lists you never waste more than 12.5% of memory?  Neato.

It looks like PyPy is coming up to a 1.0 release later this year.  James Gardner gave a good Birds of a Feather session on Pylons, this is relevant to me due to TurboGear’s use of Pylons - ShowMeDo is written in TurboGears.

Zeth tells me that last year there were about 150 attendees, I think we have over 200 this year, next year EuroPython meets PyConUK so we’ll have closer to 500.  I also met a bunch of regional Python usergroups including Python Ireland - there’s obvious growth in our userbase with a lot of smart people coding away.

At the end of last night’s dinner we had a talk on the Lunar Society, it was long (2 hours!) but told a great story.  The drinking, inevitably, went on late <ouch>.

I look forward to next year’s PyCon UK!  Pictures etc via the PyCon wiki which links to flickr.

As a side note it was great to talk to Pythonistas and hear that our ShowMeDo is fairly well known (given that we’ve never had a marketing budget, or time, or resources…) and well-respected.  I think I’ve recruited a few more authors along the way.

1. Chumby+Python hacking by John
2. ‘Groundbreaking movie production’ with ASwarmOfAngels by Matt
3. Arduino for physics display hacking by Seb
4. Locomatrix GPS Gaming by Richard
5. “How not to launch a web app” by Dan on the soon-to-really-be-launched Tails bug tracker

This event is a Demo Camp like last month’s. Usual rules - sign-up on Upcoming so we know how much beer to buy (we buy it out of our own pockets…unless you want to sponsor?) and cake to bake, UNattend if you can’t make it.

If you’re local then you’ll know John (£5 App co-founder) and the Chumby, Arduinos and Locomatrix.  You might not know ASwarmOfAngels:

“A groundbreaking project to create a £1 million film and give it away to over 1 million people using the Internet and a global community of members”

There is more about Matt Hanson (founder of ASOA) in this interview.

We’re in our usual location of the refurbished Regency Town House in Hove (thanks Danny!), if you’re coming down by train then you’re a 10 minute bus ride away.

From London - get to Brighton station, walk 5 minutes South to Churchill Square, get a number 1 bus heading West (actually, most busses will probably do if they go West from Churchill Square), get off by Brunswick Square (ask the driver, it’ll take 5 mins on the bus).  It is a 20 minute walk from Brighton Station if you’ve got a map.

I figured that a look at some of the recent series might be useful.  In total we have over 330 Python screencasts, with 8 series alone for Django.  Over 150 of the screencasts are aimed at Beginner Python coders.

We’ve had several series which cover test-driven development (TDD), unit testing and coverage, as well as discussing the whole development process.  Seeing these hard-core videos teaching ‘a good way of coding’ is really nice, these skills are hard to describe and easy for a beginner to ignore.  Three that spring to mind are:

• Agile Development Tools in Python inc. VirtualEnv, PasteScript
• Agile Python+Pylons inc. TDD, coverage, unit testing
• Developing emol! inc. use-cases and class diagrams in 24 beautiful episodes
  

It is nice to see some web-framework series sprouting up:

• Pylons in action (mentioned above)
• Quick-start guide to TurboGears 2
• Setting Up the Django Dev Environment

In the Club (for paying users) we’ve been busy, currently we’re simultaneously publishing several meaty series.  They’re aimed at beginner/intermediate Python programmers, mixing Python background with useful worked examples:

• Loops and Iteration
• Common Variables and Containers
• What does Python look like?
• Database Programming
• Introducing the Google App Engine

In total we have 11 strong series in the Club which reduce the learning time for a beginner.  We’ve also commissioned some of our open-source authors to join us in the Club, two new series on ‘batteries included’ and ‘C+Python’ are in the works.

We welcome new open-source authors - each author’s contributions are special to us, we don’t stick ads on the pages or require ownership of any rights.  We just do our best to bring an author’s knowledge to a wider audience.

I guess the biggest thing that we’ve achieved is ‘recognition’ - it is nice mailing a project author to discuss a relevant series that we’ve had submitted and to have them say ‘ShowMeDo - yes, I was watching some cool stuff there just the other day…’.

Each of the 6 talks lasted 15 minutes and covered a wide range of geek and tech-business topics:

• ‘Google Maps + GPS + Python + Series 60 + giant pencil == live biking location demo’, Tristan Roddis (CogApp)
• LiquidFold using Google’s App Engine, Simon Willison & Natalie Downe
• “An airport that you can turn-off” - evolvable flight-paths for crash-resistant landing vectors at a major airport, Matt Sarjent
• ‘Pistach.io: a cracking idea! [blog-based advertising platform, new start-up]’, Aral Balkan and Dave Stone
• AllSizes+ - hacking flickr with GreaseMonkey, Premasagar Rose (of Dharmafly)
• ‘Inuda’s first public app [howsociable.com - brand metrics company side-project]… after two years of ‘Innovating’ alongside client work’, Jon Markwell and Kev The Dev

Ribot both sponsored the event (great beer - thanks!) and made live Qik video streams for the evening.  I believe these segments are correct:

1. Video 1 - quick intro from John and myself
2. Video 2 - Tristan’s Python Series 60 demo starts
3. Video 3 - Tristan concludes, full talk for Simon and Matt (followed by break)