orlp

35 ms·

361.
by orlp 3y ago
> but if AIs invent a new type of math, they would be able to teach it to us There are already math proofs made by humans on this very day that are hundreds upon hundreds of pages of lemmas that are highly advanced and building on other
362.
by orlp 3y ago
I mean, it addresses the pain point but with the wrong conclusion.
363.
by orlp 3y ago
Yes, using slotmap you can get "use after free"-style bugs that you would not encounter if you strictly stayed with the borrow checker. So if the borrow checker fits your purpose, by all means, go ahead. But the borrow checker can
364.
by orlp 3y ago
> Slotmap uses unsafe everywhere, it's a memory usage pattern not supported by the borrow checker. Author of slotmap here. This is patently false. Yes, the slotmap crate uses a lot of unsafe to squeeze out maximum performance. But i
365.
by orlp 3y ago
Author of slotmap here. There is absolutely no need for unsafe in slotmap. I chose to use unsafe (wrapped in a safe API) to reduce memory usage using intrusive linked freelists. If done using safe Rust this would involve `enum`s that woul
366.
Show HN: PolymurHash, a fast hash with a mathematically proven collision rate(github.com)
8 points by orlp 3y ago|0 comments
367.
by orlp 3y ago
Please check my edit right as you commented :)
368.
by orlp 3y ago
I am not cynical about the research itself, I am critical of claims such as "new sorting algorithm uncovered", "up to 70% faster", or "first change in a decade". The research is good. The achieved results are m
369.
by orlp 3y ago
You'll love this list of examples where the AI successfully accomplished reward hacking: https://docs.google.com/spreadsheets/d/e/2PACX-1vRPiprOaC3Hs...
370.
by orlp 3y ago
It's not even a cmov. Look at Figure 3 in the paper: https://www.nature.com/articles/s41586-023-06004-9 They eliminated a register-register mov.
371.
by orlp 3y ago
> AlphaDev uncovered new sorting algorithms that led to improvements in the LLVM libc++ sorting library that were up to 70% faster for shorter sequences and about 1.7% faster for sequences exceeding 250,000 elements. As someone that know
372.
by orlp 3y ago
All of quantum computing is done with strictly reversible logic. But this is not a reduction of power. The Toffoli gate is an example of a reversible 3-bit logic gate that is universal. You can transform any classical circuit into a reversi
373.
by orlp 3y ago
> But is there a way to somehow construct this modulus automatically for some functions? In a sense. You would end up with https://en.wikipedia.org/wiki/Automatic_differentiation . The amount of functions you can con
374.
by orlp 3y ago
https://en.wikipedia.org/wiki/Modulus_of_convergence
375.
by orlp 3y ago
a) All functions from computable reals to computable reals are continuous. This makes sense, because an arbitrarily small input difference leading to an arbitrarily large output difference is not a good basis for computation. You can still
376.
by orlp 3y ago
I've made the argument many times before that the reals are certainly an interesting set of objects, but a poor abstraction for the concept of a number. We should work towards replacing them with a more well-behaved set without eldritc
377.
by orlp 3y ago
This isn't true for two different objects in the same orbit with a small phase difference. To be fair that arrangement is rather unstable, especially with planet-sized objects.
378.
by orlp 3y ago
> OpenAI has no leverage (e.g. terminating service). Uhh... the United States Justice system? Prior art is morally nice but ultimately meaningless unless a judge agrees with you, which can cost a lot of money before you get there, if you
379.
by orlp 3y ago
> To begin with, when n is exactly 2^k-1, I would expect all the functions to show k comparisons, including Skarupke's. In the plot, all the curves overlap at 64, 127 and 255 except Skarupke's which is off by 1. I'm afraid
380.
by orlp 3y ago
That is very kind of you :)
381.
by orlp 3y ago
The benchmarking code is available, I did not: https://github.com/orlp/bitwise-binary-search . The compiler also did not appear to have autovectorized it. That said, I would expect the overhead of vectorization to domin
382.
by orlp 3y ago
> This is surprising, shouldn't the number of comparisons be identical to the standard implementation? No, to make it branchless (in an efficient way, it can be done regardless of course), you always take ceil(n / 2) as your re
383.
by orlp 3y ago
> In this post, I don't think there is a comparison with a classical binary search implemented in a branch-free fashion? Searching along dyadic intervals is a cute trick, but it's not clear whether the improvement comes from th
384.
by orlp 3y ago
I wasn't aware of __builtin_unpredictable until yesterday. I was aware of __builtin_expect_with_probability. Unfortunately I've found both of them to be essentially useless with today's compilers.
385.
Bitwise Binary Search: Elegant and Fast(orlp.net)
3 points by orlp 3y ago|3 comments
386.
by orlp 3y ago
> If they generate verbatim copies of any non-trivial-enough part that would be the subject of a copyright license, yes, it would be copyright infringement. Yet, could anyone give such practical examples? Yes, there are many such example
387.
by orlp 3y ago
And my algorithm pdqsort uses Hoare-style partitioning, is branchless, and properly solves the Dutch National Flag problem by being 0(n log k) for k unique values on average using only a single extra comparison per partitioning. https:&#x2
388.
by orlp 3y ago
I... don't know what the hell happened. I was doing the calculation on mobile while on the move and I must've screwed up the parenthesis in the denominator. Embarrassing... Your calculation is correct. My point still stands (2 mil
389.
by orlp 3y ago
Adding four randomly generated characters a-z to your password adds a factor of 456976x to the bruteforce time required. A password that is derived in 1 millisecond with these characters appended takes longer to crack than a password that i
390.
by orlp 3y ago
> His encryption password was supposedly greater than 20 characters and included a mixture of cases, numbers, and punctuation, so in the absence of any sort of opsec failures this implies that even relatively complex passwords can now be

More ›