dragandj

91 ms·

931.
by dragandj 10y ago
Thanks. I doubt that XeonPhi would be any faster than my old AMD R9 290X, and the 10x price tag is also not inviting.
932.
by dragandj 10y ago
But there ARE hierarchical models in the examples. One is 158-dimensional. With highly correlated parameters. Works like a charm in Bayadera.
933.
by dragandj 10y ago
Thanks! About the quil utils - you mean get rid of them, or you think that they would be useful on their own? Btw, I only use quil for the sketch setup, the plots are done in low-level Processing. Quil does some(un)boxing, so It would be a
934.
by dragandj 10y ago
Please point me to models that absolutely need HMC, and I can try to see how Bayadera fares.
935.
by dragandj 10y ago
I'll be glad to send you the paper once it gets accepted.
936.
by dragandj 10y ago
It uses different MCMC algorithm - affine invariant ensemble MCMC. The difference comes from the fact that this algorithm is parallelizable, while JAGS/Stan's isn't. So, many GPU cores are the main factor. But, the algorithm
937.
by dragandj 10y ago
I can not guarantee, but this should not be a problem for this particular algorithm (MCMC).
938.
by dragandj 10y ago
Recompile time is under a second for most models that I tried. Let's say 150 - 700 ms. Once you compile it, you can use it many times. The diference for the inference time is in the post below (but YMMV).
939.
by dragandj 10y ago
Nothing is universal and guaranteed, of course. YMMV, and all that. For example, robust linear regression from chapter 17, that fits 300 points over 4 parameters (easy, but far from trivial) runs in 180 seconds in JAGS and 485 in Stan, in p
940.
by dragandj 10y ago
Characteristic examples from the book Doing Bayesian Data Analysis 2nd edition [1] programmed in Clojure and OpenCL to run on the GPU. Much, much faster than Stan or JAGS! The library used (Bayadera) is still pre-release, so much polishing
941.
Doing Bayesian Data Analysis On the GPU: Examples(github.com)
112 points by dragandj 10y ago|28 comments
942.
by dragandj 10y ago
The problem is that, if you put mere reproduction as a goal, many scientists would see that as low hanging fruit to beef up the resume, so we'd get countless unnecessary "experiments". I'd say the goal that gets credited
943.
by dragandj 10y ago
And yet, not only all speakers of one can understand perfectly the other, everyone can speak it effortlessly, and even understand localized accents and subtleties.
944.
by dragandj 10y ago
You forget a very big factor: only US citizens can move to SV or NY without many hoops. If you are from EU, you can go to Berlin with no fuss, if you can get a job. Maybe your US counterpart can go to SV, but you need a visa rain dance, whi
945.
by dragandj 10y ago
The Gutenberg version is abridged, and the translation is also much worse than Trask. I do not consider $20 too much for a 800 page 2-volume book, very well typeset, with plenty of notes, and huge scholarly work put in it to make it accessi
946.
by dragandj 10y ago
I read Casanova's memoirs in English translation, and the translation to my native language (the original is in French). I cannot recommend it enough. If you are interested, do NOT take an abridged publication (of which there are many)
947.
by dragandj 10y ago
While I agree with the author about the main topic, the actual content does not illustrate the point well. The author wrote a naive loop in a bunch of languages, and, got a faster result in OCaml (0.3 sec) than in C(0.8 sec). Right, but the
948.
by dragandj 10y ago
On the other hand, for using it from a higher-level language (through various bindings) and writing my own kernels instead of just using Nvidia's libraries I find OpenCL much easier. Another reason is that I can run that on my AMD GPU
949.
by dragandj 10y ago
The only catch in that perspective is that the language specification is open source, while Microsoft's compiler implementation is not. So, yes, we can edit those TeX documents they put on GitHub to our heart's delight.
950.
by dragandj 10y ago
Only if we don't count Netscape and JavaScript. They put it in the browser, for better or for worse.
951.
by dragandj 10y ago
In reply to myself: The times are changing indeed. It seams that HN-ers started liking Microsoft. Whenever I post some concern, even if only mild, there is first a few upvotes, and then a bunch of downvotes. Microsoft became the darling of
952.
by dragandj 10y ago
Being pushed by Microsoft could be exactly the reason not to convert, for many people. Especially when there is no shortage of alternatives that are backed by a bit more benevolent entities (Rust, and many other less known variants).
953.
by dragandj 10y ago
In core libraries? If low-level functions checked the inputs for each odd case, they'd (often unnecessarily) spend 90% of time doing those checks.
954.
by dragandj 10y ago
Because checks to avoid some kinds of garbage can incure significant performance penalty. There is always a tradeoff in that sense, and I think Clojure is pretty well balanced in that way. So, read the docs, try the functions out to see how
955.
by dragandj 10y ago
What editor do you use? Emacs Cider does this, and does it much better than IDEs (IMO).
956.
by dragandj 10y ago
Maybe you did not offer a fair enough compensation for that service. She has lots of important work improving horrible state of scientific knowledge dissemination. I don't think she has much time for consulting private corporations, un
957.
by dragandj 10y ago
I think that his argument was that other minorities do much better, despite also being hugely disadvantaged. Not to mention immigrants, who often go to USA with noting else than a bag of clotes and a desire to work hard and get a better lif
958.
by dragandj 10y ago
And there is also Boot, which might be considered a better replacement for Leiningen.
959.
by dragandj 10y ago
That's their private concern. TPB is not a busybody.
960.
by dragandj 10y ago
All I can see is that people use it to advertise the TOC of their folders. The links do not even guarantee that the files in those folders match description. But, I don't see what TPB has with that? None of these is on TPB. The users d

More ›