3 ms·

My prediction: Numpy/SciPy will steal a large portion of the Matlab market for scientific computing in the not-too-distant future. A GPU-accelerated Numpy alon

by aklein 16y ago

My prediction: Numpy/SciPy will steal a large portion of the Matlab market for scientific computing in the not-too-distant future. A GPU-accelerated Numpy along the lines of Jacket would be supremely valuable. Side project, anyone?

nkassis 16y ago
I'll have no problem with that if it happens but, Matlab is so entrenched and they now have GPGPU computing in their products, I don't see them disappearing for a long long time. At least not in the field my lab is in. I don't think matlab is as prominent in many other sciences right now as it is in neuroscience. There are moves toward things like NumPy but the community seems reluctant. EDIT: I'd like to see some uptake of clojure too. Things like Incater look interesting for scientific computing.
tomrod 16y ago
I've actually found Numpy/SciPy to have a lot of trouble in basic maximization routines as compared with Matlab.
earl 16y ago
Do you actually use scipy? I tried briefly and was awfully disappointed. For starters, the documentation was crap. Followed closely by the fact that I wanted to take the norm of a vector and had to try 3 different norm functions before one worked. Further, I had a sparse matrix and all the functions -- of course, this was mentioned nowhere -- magically desparsified my vectors. Apparently if you want to operate on sparse matrices/vectors, you need to pull in an entirely separate import and potentially change the name of all functions. I was left completely underwhelmed. Not to mention that installing matlab was a pain, but installing scipy was an order of magnitude worse. ports failed to work, the binary distributions wanted much older python than I had, etc. It would be awesome to see them seriously take on matlab, but I have to conclude they have a very long way to go. If their gpu implementation is anything like their sparse vector/matrix implementation, matlab will remain the language of choice for a long time.
- tomrod 16y ago
  I have a friend in Apple's data mining group who told me they use Numpy/SciPy exclusively for their programming. I was surprised. My experience has not been all that great either. But it's fun to tinker.
- melonakos 16y ago
  The MATLAB GPU implementation is seriously flawed. For a better approach, see http://www.accelereyes.com/products/compare http://www.accelereyes.com/products/compare
  - earl 16y ago
    Dude, I'm not hating, but when you are plugging accelereyes, maybe you should mention you're the CEO and cofounder. That being said, it looks like interesting tech.
    - melonakos 16y ago
      I'm definitely partial :) I was relying on my profile to make it clear, but that's a bit buried. BTW, you're profile has me cracking up in how you share your email. I tweeted it here: http://twitter.com/melonakos/status/11101399047737344 http://twitter.com/melonakos/status/11101399047737344 Do you think GPUs in R would be useful? We're heading that way with LIBJACKET.

by aklein 16y ago

nkassis 16y ago

I'll have no problem with that if it happens but, Matlab is so entrenched and they now have GPGPU computing in their products, I don't see them disappearing for a long long time. At least not in the field my lab is in. I don't think matlab is as prominent in many other sciences right now as it is in neuroscience. There are moves toward things like NumPy but the community seems reluctant. EDIT: I'd like to see some uptake of clojure too. Things like Incater look interesting for scientific computing.

tomrod 16y ago

I've actually found Numpy/SciPy to have a lot of trouble in basic maximization routines as compared with Matlab.

earl 16y ago

Do you actually use scipy? I tried briefly and was awfully disappointed. For starters, the documentation was crap. Followed closely by the fact that I wanted to take the norm of a vector and had to try 3 different norm functions before one worked. Further, I had a sparse matrix and all the functions -- of course, this was mentioned nowhere -- magically desparsified my vectors. Apparently if you want to operate on sparse matrices/vectors, you need to pull in an entirely separate import and potentially change the name of all functions. I was left completely underwhelmed. Not to mention that installing matlab was a pain, but installing scipy was an order of magnitude worse. ports failed to work, the binary distributions wanted much older python than I had, etc. It would be awesome to see them seriously take on matlab, but I have to conclude they have a very long way to go. If their gpu implementation is anything like their sparse vector/matrix implementation, matlab will remain the language of choice for a long time.

tomrod 16y ago
I have a friend in Apple's data mining group who told me they use Numpy/SciPy exclusively for their programming. I was surprised. My experience has not been all that great either. But it's fun to tinker.
melonakos 16y ago
The MATLAB GPU implementation is seriously flawed. For a better approach, see http://www.accelereyes.com/products/compare http://www.accelereyes.com/products/compare
- earl 16y ago
  Dude, I'm not hating, but when you are plugging accelereyes, maybe you should mention you're the CEO and cofounder. That being said, it looks like interesting tech.
  - melonakos 16y ago
    I'm definitely partial :) I was relying on my profile to make it clear, but that's a bit buried. BTW, you're profile has me cracking up in how you share your email. I tweeted it here: http://twitter.com/melonakos/status/11101399047737344 http://twitter.com/melonakos/status/11101399047737344 Do you think GPUs in R would be useful? We're heading that way with LIBJACKET.