simondanisch

9 ms·

61.
by simondanisch 8y ago
>Actually, your sample "solution" is (was?) also a much used pattern in Julia - especially before we had such a powerful broadcast implementation. To add some more historical facts to this: before broadcast, we mostly used maps
62.
by simondanisch 8y ago
Well, it imports from Base - Anyways, too many people got hung up on this rule, so I deleted it... It wasn't very important anyways, as long as the core implementation is somewhat from scratch.
63.
by simondanisch 8y ago
>There wouldn’t need to be a library providing this “broadcast” functionality Actually, your sample "solution" is (was?) also a much used pattern in Julia - especially before we had such a powerful broadcast implementation. So
64.
by simondanisch 8y ago
I think a lot of people had prepared articles, packages etc for the 1.0 release - and now they're slowly getting released ;) If you're a Julia developer, the release definitely motivates you to put in the extra work and try to rel
65.
by simondanisch 8y ago
Well, that'd be comparing apples and oranges... But fun fact: Julia actually has excellent GPU support! I will soon update the challenge with a GPU accelerated version. Spoiler: it won't need much more than 5 lines of code to get
66.
by simondanisch 8y ago
The rule is about the implementation itself, but the other one is part of the benchmarking code. You'll need to be able to use the implementation, that doesn't rely much on external libraries, with types from other packages. If yo
67.
by simondanisch 8y ago
Makes sense. I will try to update the article! Feel free to use whatever library that helps you to compose your program nicely. It would be nice though, if those libraries were in a set of standard libraries ;) If you need to implement all
68.
by simondanisch 8y ago
Ok fair enough - maybe this is just my bad english, let me know if you have a better way to phrase it. I do think, that it is quite showing, that google + tensorflow don't believe in the mix of python + c/c++. I've been worki
69.
by simondanisch 8y ago
as i said, if it makes sense, i'd totally allow using external libraries! maybe i should make that clearer since quite a few people get hung up on that
70.
by simondanisch 8y ago
What is the point?
71.
by simondanisch 8y ago
I must admit, that I find your sentiment very understandable! It is an egoistic challenge, after all. For me the problems I describe are highly relevant, and I'm very curious about any solution that isn't written in Julia. And I
72.
by simondanisch 8y ago
>If depicting usage for the amateur was the main point of the article You understood that right, it is not :) That's why I wrote, that I'd probably need years to create the same implementation in any other language. I do think
73.
by simondanisch 8y ago
that's exactly why I made that point debatable ;) Feel free to use Numpy, I'd consider that as the Python base array library anyways;) Note though, that this challenge is aimed at the people, that need to implement libraries like
74.
The Julia Language Challenge(nextjournal.com)
116 points by simondanisch 8y ago|52 comments
75.
by simondanisch 8y ago
>The purpose of multiple dispatch is extensibility, but the purpose of pattern matching is destructuring. Agreed! ;) Just wanted to point out, that I was able to use multiple dispatch elegantly in the unroll example, where you are using
76.
by simondanisch 8y ago
>You can in fact get loop unrolling with just standard functions in Spiral. We can do this too actually - and spirals approach sounds very similar to what you'd do in julia :) https://gist.github.com/SimonDanisch&#x2
77.
by simondanisch 8y ago
>GPU programming without macros looks like then take a peek at this: I'm not sure if you missed my examples, or if there is some other missunderstanding going on ;) GPU code in Julia works 100% without macros, but you are free to us
78.
by simondanisch 8y ago
On Julia 0.7, the broadcast kernel will look much more like this (and actually the version used in CuArrays/CLArrays currently also doesn't use those macros): https://github.com/jrevels/MixedModeBroadcastAD.jl
79.
by simondanisch 8y ago
The problem is still with apple forcing them to invest resources, without any reason, but to advance their vendor lock in. And if you're a developper of a small high performance 3d graphics and gpu computing library like me, its just
80.
by simondanisch 8y ago
I wouldn't say so. Cython is quite literally C for python, basically making all your python code look like C with most disadvantages, while not always reaching C performance. Julia in comparison is a fully featured language, with a lot
81.
A new detachable notebook design(medium.com)
1 points by simondanisch 9y ago|0 comments
82.
by simondanisch 9y ago
>type that explicitly recides on GPU Yes, that's what we do! More domain specific types will come, for e.g. views, lazy arrays and sparse arrays. > or does it manage the device state? It aims to manage most of the device state! S
83.
by simondanisch 9y ago
First of all, Julia is a proper multi purpose programming language! I consider that to be important, since I want to learn one language and then do most of my very different projects in it. I haven't used Futhark or Halide, but I know
84.
Writing extendable and hardware-agnostic GPU libraries(techburst.io)
45 points by simondanisch 9y ago|4 comments
85.
Compiling Julia binaries(medium.com)
1 points by simondanisch 9y ago|0 comments
86.
Drawing 2.7B points in 10s(medium.com)
2 points by simondanisch 9y ago|0 comments
87.
Writing extendable and hardware agnostic GPU libraries(medium.com)
1 points by simondanisch 9y ago|0 comments
88.
Julia's new GPU Array/Tensor library(github.com)
2 points by simondanisch 9y ago|0 comments
89.
by simondanisch 10y ago
Yeah and there is already an improvement to that: https://github.com/JuliaLang/julia/pull/12541 . Well, anyways, I think you can easily find a lot of unsatisfactory constructs in any language, especially in yo
90.
by simondanisch 10y ago
deployment + load times: obviously things aren't perfect, Julia is at version 0.4, after all ;) There are a lot of low hanging fruits for compile time and deployment, so I'm sure that the situation will look a lot different in a y

More ›