Athas

43 ms·

421.
Futhark 0.3.0 released(futhark-lang.org)
2 points by Athas 9y ago|0 comments
422.
How Futhark manages GPU memory(futhark-lang.org)
50 points by Athas 9y ago|1 comments
423.
by Athas 9y ago
It is. I encourage everyone to buy a subscription to LWN (it's cheap) to support this kind of high-quality technical reporting.
424.
by Athas 9y ago
The main assumption is that the specification (the proposition you want to prove) is (much) simpler than the proof. This means that the user/customer can read the specification, and understand what it means, even if the proof is not c
425.
Reflections on a PhD accidentally spent on language design(futhark-lang.org)
2 points by Athas 9y ago|0 comments
426.
by Athas 9y ago
That looks very interesting. I'm OK with it only supporting recent AMD GPUs, but I hope the other changes will be upstreamed soon.
427.
by Athas 9y ago
Is there also a useful open-source OpenCL implementation for AMD GPUs? I would find that even more useful.
428.
by Athas 9y ago
This raises a question that I'm still wondering about. Matrix operations are useful for high-performance computing because a lot of useful operations can be transformed (with varying effort) into matrix operations. Hence, with access
429.
by Athas 9y ago
Does this solve some of the core problems with GPU programming, such the difficulty of writing reusable code without significant performance overhead (via nested parallelism or fusion), or the need to follow some potentially awkward rules f
430.
by Athas 9y ago
But it might not be undefined behaviour at compile-time. Consider if instead of just 'f(nullptr)', the call in the main() function had been 'f(getchar() == 'x' ? nullptr : &x)' (where x is some variable in
431.
by Athas 9y ago
These algorithms are written in NESL, which is perhaps the cleanest example of the oft-mentioned theory that parallel programming becomes simple in a functional setting. It really is remarkable how flexible it is. Unfortunately, I'm
432.
by Athas 9y ago
This is great! I'm really inspired by the work CMU does on parallelism. They have a parallel programming course[0] that results in some really impressive final projects [1]. I also hear that CMU takes a "parallel-first" ap
433.
by Athas 9y ago
OpenCL is a very sound design on a technical level. Unfortunately, it's mostly designed to be a good target for code generation/libraries, and is awkward to use directly. The problems are that everything is very explicit and spe
434.
by Athas 9y ago
I just hope it won't be all too focused on deep learning. That's boring. We need massively parallel machines that are more general than that. Modern NVIDIA-style GPUs are actually doing quite well, although they've started
435.
by Athas 9y ago
Piracy makes it easier for me to obtain papers that are otherwise behind paywalls, for example because I am at home. As an academic, my main interest is that others are able to read my work. If piracy is how they obtain access, then that&#
436.
by Athas 9y ago
Not all fields suffer from this to the same degree. For example, I am a fledgling computing scientist. I mostly publish in ACM-related conferences and workshops, where ACM is an organisation run by and for academics and professionals in t
437.
by Athas 9y ago
Yes, but based on an subjective judgment. I am the author of the blog post, and I tried to do my utmost to emphasize that my judgment here is highly subjective, using phrases such as "for my own pedantic notion of correctness", a
438.
by Athas 9y ago
Does this mean that the following comment would be a syntax error in an OCaml program? (* Strings are enclosed by '"'. *)
439.
by Athas 9y ago
Line comments are no hindrance to documentation generation. Extracting the contents from a paragraph line comment is a rather simple regular expression[0]. Haskell almost exclusively uses line comments for documentation, and it works fine
440.
by Athas 9y ago
Yes, heredocs (or "here-comments", since you want them ignored, not just stored in a string variable) would likely be a solution for all but the most pedantic. The final recourse of the pedant would be to point out that no single
441.
Block Comments Are a Bad Idea(futhark-lang.org)
48 points by Athas 9y ago|82 comments
442.
by Athas 9y ago
I use OpenBSD on my personal server. It does not do anything advanced, but runs email, a few websites, IRC bots, and such. I am not a professional systems administrator, nor do I consider myself a particularly competent one (I'm a co
443.
by Athas 9y ago
It's about picking a CPU for workstations, though, not normal desktop machines. It's assumed you will need serious computational power, and that's not free. That's also why they highlight the presence of so many PCIe l
444.
by Athas 9y ago
We already have both a radix sort[0] and a bitonic merge sort[1]. They perform alright - nothing world-shaking, but fine. I was more interested in whether string sorting has some particularly clever algorithms. [0]: https://git
445.
by Athas 9y ago
It's a little tricky to implement irregular algorithms in Futhark, but I'd like to give it a try. Do you know of a nice data-parallel algorithm for string sorting?
446.
by Athas 9y ago
Yes, both futhark-c and futhark-opencl use the exact same infrastructure.
447.
by Athas 9y ago
That's good to hear! You will probably encounter some limitations due to the missing support for error recovery. When you do, we'd very much like to hear from you, to get an understanding of what a proper solution should look li
448.
by Athas 9y ago
The biggest issue with compiling APL and J is the presence of dynamic types (and other dynamic behaviour). For the subset of APL can that can be resolved statically, I don't see why they should be particularly difficult (Co-dfns[0] se
449.
Calling Futhark from C and Haskell(futhark-lang.org)
88 points by Athas 9y ago|15 comments
450.
by Athas 9y ago
I am very fascinated by this notion of teaching compilers by starting at code generation and working backwards from there. I've read about nanopass compilers before, but do they also work well if the compiler is written in a staticall

More ›