dragandj

59 ms·

661.
by dragandj 9y ago
But all major Deep Learning frameworks use Nvidia-s cuDNN library. They provide lots of sugar, but the central thing is developed (or paid for if you prefer) by Nvidia.
662.
by dragandj 9y ago
Exactly that! Why the CEO of Intel or the CEO of AMD didn't do the same? It was obvious! Well, it is obvious now , but then it was far from obvious.
663.
by dragandj 9y ago
HPC techniques are exactly those that ML and deep learning use in implementation. I don't see how DL is much different from this point of view. It relies on fast linear algebra. HPC relies on fast linear algebra. Nvidia perfected the p
664.
by dragandj 9y ago
On top of that, he acknowledges that there is a much, much simpler solution, known from the dawn of Java, and this is to have a polymorphic children() method. But that would (gasp!) require adding those implementations. Which is tedious (it
665.
by dragandj 9y ago
Did you know that there are rich distributions that have done that for you? You can literally install those with a few simple steps and have fully functional super environment. Key words for Google search: Emacs Prelude, or Spacemacs
666.
by dragandj 9y ago
The difficult part is to stop hodling at the right moment though.
667.
by dragandj 9y ago
> The only thing that I notice about Lisp are the bracket jokes and memes. You have also noticed that lots of people use it to create cool things which reach the HN front page.
668.
by dragandj 9y ago
> looking at the past 5 years How much it extrapolates in 20 years then? 100 million?
669.
by dragandj 9y ago
I am not sure people around the world will start being crazy about all thing Kazakh now that those are written with a Latin alphabet. As far as popular culture goes, Borat is the only thing most people recognize Kazakhtan for (besides it be
670.
by dragandj 9y ago
I think it is also a historical coincidence. Cyrillic is an original Slavic thing and its early variant Glagolitic srcipt ( https://en.wikipedia.org/wiki/Glagolitic_script ) was actually created for West Slavic "cus
671.
by dragandj 9y ago
> The author is trying to sell a freemium product, its a gift insofar as Dropbox's 2 gigs are a gift. In other words, it is a gift. You get something useful for nothing. As for the other point, I agree that the author should use mor
672.
by dragandj 9y ago
Thanks. I got (and agree) that it was not written very skillfully, but, as you said, it was not hostile either.
673.
by dragandj 9y ago
I'm not a native speaker either, but having a quite high exposure to English language, I am still puzzled at why would anyone think those sentences were hostile . Can any native speaker translate what I quoted to a non-hostile version
674.
by dragandj 9y ago
OTOH, the license is optional. You can use it for $0. I'm not even a potential user of this library. I'm just amused with how people expect open source programmers to work for peanuts.
675.
by dragandj 9y ago
What are you talking about? The author literally says: "for help please go to the yue/help repo where users help themselves. To report a bug, you have to create a pull request with a test case of your problem, or you can choose to
676.
Reminiscences of a Stock Operator (1923)(en.wikipedia.org)
1 points by dragandj 9y ago|0 comments
677.
Prelude is an enhanced Emacs distribution(github.com)
2 points by dragandj 9y ago|0 comments
678.
by dragandj 9y ago
William Bolstad - Introduction to Bayesian Statistics Or E.T. Jaynes - Probability Theory, The Logic of Science (but this one is much more detailed and has lots of philosophy) Or Ian Hacking - An Introduction to Probability and Inductive Lo
679.
by dragandj 9y ago
Source: https://github.com/uncomplicate/neanderthal Also used for Bayesian data analysis on the GPU in: https://github.com/uncomplicate/bayadera
680.
More specialized matrix data structures in Clojure on CPU and GPU(dragan.rocks)
6 points by dragandj 9y ago|1 comments
681.
by dragandj 9y ago
The problem with these kind of extremely short texts is that they jump into explaining technical concepts without even mentioning basic things such as random variable for example. Then, readers learn that statistics is about applying a fe
682.
by dragandj 9y ago
Direct link to Getting Started: http://clojurecl.uncomplicate.org/articles/getting_started.h... source code: https://github.com/uncomplicate/clojurecl Heavily used in other Clojure libraries: htt
683.
Getting started with GPU programming in Clojure(clojurecl.uncomplicate.org)
4 points by dragandj 9y ago|0 comments
684.
ClojureCL – GPU programming in Clojure(clojurecl.uncomplicate.org)
18 points by dragandj 9y ago|1 comments
685.
OpenCL BLAS library CLBlast(github.com)
2 points by dragandj 9y ago|0 comments
686.
by dragandj 9y ago
Sure. My intention was not to correct your statement, but to clarify it a bit.
687.
by dragandj 9y ago
It is very easy to run unoptimized code despite sending everything to cuBLAS. How easy it is to make the wrong turn - that depends on the library.
688.
by dragandj 9y ago
No impact whatsoever! You get the full speed.
689.
by dragandj 9y ago
Hi Adam, Perhaps you'd be interested in writing a joint article with me where we explore both Neanderthal and ND4J side by side? Maybe you could also benefit of ClojureCUDA's ease of use vs using low-level Java bindings? This coul
690.
by dragandj 9y ago
You do not need to read any of those Intel MKL guides - they are for the developers who compile the dependencies (me in this case). You as a user only need to have Intel's .so files on your PATH (or dylib for Mac or dll for Windows),

More ›