freyrs3

34 ms·

301.
by freyrs3 14y ago
It dumps the internal Core representation of expression. It's closer to the lambda calculus ( System F ) and a good way to understand internals.
302.
by freyrs3 14y ago
The author seems to have glazed over quite a bit of the rather mature Python ecosystem simply because it lives in a general purpose language.
303.
by freyrs3 14y ago
Begs the question whether doing meaningful work in a field will take more time given that it takes a decade or two to get up to speed with current research.
304.
by freyrs3 14y ago
In numeric Python packages computations are never done inside the CPython runtime, they're done in specialized kernels written in C, Fortran and sometimes fine-tuned assembly including a sampling of the various BLAS/LAPACK flavors... so yes
305.
by freyrs3 14y ago
mmap'ing /dev/zero is the fastest way I know.
306.
by freyrs3 14y ago
Your observations are pretty accurate. I'm fortunate enough to work with the maintainer of llvmpy and he's done some great work in making llvmpy pretty Pythonic. Should also check llvm-cbuilder the EDSL for building up higher level LLVM con
307.
by freyrs3 14y ago
You can do the same in Haskell with the flip function. In fact the bindWith2 is the same as the flip function for binary functions, but without the need for "functions objects" with bound methods. λ: (/) 2 4 0.5 λ: flip (/) 2 4 2.0
308.
by freyrs3 14y ago
There's a Python version here: http://www.llvmpy.org/llvmpy-doc/dev/doc/kaleidoscope/index....
309.
by freyrs3 14y ago
> Isn't your productivity gain coming from using a statically typed language rather than Haskell specifically I think it's a bit more than that, Haskell has a much more sophisticated type system than quite a few of the popular staticall
310.
by freyrs3 14y ago
That's probably more Coq.
311.
by freyrs3 14y ago
> how does this compare to theano? it seems like some of the ideas are similar? It's quite similar, we just take some of the ideas farther and try to generalize the data storage to include storage backends that data scientists use more
312.
by freyrs3 14y ago
Lua has some areas where it excels in performance, but using Python you can leverage 10 years of work on numeric libraries that unrivaled in any other general purpose language. NumPy and SciPy are extremely powerful.
313.
by freyrs3 14y ago
GPU support is definitely planned and already supported in NumbaPro[2]. Here's a video of Travis Oliphant's talk about targeting CUDA through Numba: [1] http://www.ustream.tv/recorded/26973799 [2] https://store.continuum.io/cshop/numbapr
314.
by freyrs3 14y ago
Yes, using Numba we can just-in-time compile numeric Python logic straight down to machine code, so naturally we can achieve some pretty impressive numbers on kernel execution. In case many people didn't reach the bottom here are the links
315.
by freyrs3 14y ago
> THis doesn't happen anymore. And if you, instead, grow a boy to a man the woman way, what you get is not a man but an overly feminine male who is at loss with his manhood and who has two balls that are barely working. This reasoning
316.
by freyrs3 14y ago
They'll think its old code. I work with plenty of old Fortran linear algebra libraries and I don't think they're crap, they're just old but they still get the job done more efficiently than any modern software.
317.
by freyrs3 14y ago
This article is loaded with implicit assumptions about what "business value" is and how it gets created. All software businesses are different and I dare say that most are not web startups. A company that develops SaaS, or compilers, or flu
318.
by freyrs3 14y ago
It's only falling behind if you consider MongoDB to be a forward technology. I don't.
319.
by freyrs3 14y ago
In non-homoiconic languages I think this aphorism is more apt: > Metaprogramming is the language feature that helps you write code that you won't be able to understand once the cocaine wears off.
320.
by freyrs3 14y ago
Most of the internet runs on a kernel that is GPL'd.
321.
by freyrs3 14y ago
Here's a bigger list: http://programming-motherfucker.com/become.html
322.
by freyrs3 14y ago
> i'm assuming this guy has done real world work before? Yes, very much. He's one of the leading voices in the parallel Haskell world and a very accomplished programmer: * http://hackage.haskell.org/package/stm * http://hackage.haske
323.
by freyrs3 14y ago
Replace the word "app" with "custom software" and you'll see how ludicrous this article sounds.
324.
by freyrs3 14y ago
There is a connection, we can view mappings between linear memory spaces to higher arrays constructs as a series of coordinate transformations in the same way we deal with transformations between charts on manifolds. Except in our case our
325.
Blaze: Next Generation NumPy(speakerdeck.com)
118 points by freyrs3 14y ago|28 comments
326.
by freyrs3 14y ago
libeio makes some strides in this direction. http://software.schmorp.de/pkg/libeio.html
327.
by freyrs3 14y ago
Numba and Cython are also worth mentioning if you're doing numeric programming.
328.
by freyrs3 14y ago
So we live in a zoo and eat gazelles. Ok, that makes sense now.
329.
by freyrs3 14y ago
> We will not celebrate. Every day will be awesome, or we stop what we’re doing and make it awesome. Cake is for gazelles, and we are lions. What does this even mean?
330.
by freyrs3 14y ago
Yes, very little of what he said applies to anything but the Django side of Python. Compared to say, the numeric side of Python where I'm constantly seeing rapid innovation these days (Numba, Scikits.*, Numba, Pandas, GPU Runtimes, etc).

More ›