[Numpy-discussion] Numpy/Scipy for EC2
Thu Oct 29 01:24:00 CDT 2009
I have recently created an AMI for running python processes.
I recommend using the ubuntu server ami's provided by http://alestic.com/.
Alestic is a well known provider of public AMI images. I think this is
exactly the place you want to start from; anything you need is an apt-get or
>From the moment you launch an instance, you are literally minutes away from
being able to run a computation with Python.
I also recommend the FireFox plugin called ElasticFox for interfacing the
AWS api. It is a lot easier than the command line api tools!
I left some rough notes on my AMI creation/setup process here:
The notes include the ami-id of my resulting image, which you should be able
to launch if you wish. If you are interested, I can dive into more detail on
how I set up the os/python environment, etc.
The image I created is used as the Codenode live public notebook backend:
You can create an account, login, start a Notebook, import Numpy and run any
code you want right now!
Hope this is useful,
I cross-posted this to codenode-devel, sympy, and sage-notebook; I think
this topic could be of interest to others on those lists.
On Wed, Oct 28, 2009 at 9:29 PM, Dan Yamins <email@example.com> wrote:
> Hi all:
> I'm gearing up to build an Amazon Machine Instance (AMI) for use in doing
> Numpy/Scipy computations on the Amazon EC2 cloud.
> I'm writing to ask if anyone has any advice for which (if any) publicly
> available AMI I should start with.
> If any one has any specific AMI's that they think are good bases from which
> to modify -- or really, any other advice about using numpy/scipy on EC2 --
> I'd love to know.
> Beyond that, even if you don't know which AMI to recommend (or even what an
> AMI is), I still would like advice about which Linux flavor to use. I've
> had some experience with Mac OSX (and, with David Cornapeau's help over this
> list, I was able to build 64-bit Scipy with Python 2.6!), but I really know
> nothing about what the build process is like on Linux (and most likely,
> unless someone recommends a good AMI with optimized BLAS/LAPACK already
> built, I'm going to have to built it from scratch). So, should I use
> Ubuntu or Debian or Fedora or Centos or ...?
> NumPy-Discussion mailing list
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the NumPy-Discussion