Jay Taylor's notes

back to listing index

Pattern-defeating quicksort | Hacker News

[web search]
Original source (news.ycombinator.com)
Tags: sorting news.ycombinator.com
Clipped on: 2017-06-29

Image (Asset 1/2) alt=


Image (Asset 2/2) alt=
Rust's stable sort is based on timsort (https://doc.rust-lang.org/std/vec/struct.Vec.html#method.sor...) and unstable sort is based on pattern-defeating quicksort (https://doc.rust-lang.org/std/vec/struct.Vec.html#method.sor...). The documentation says that 'It [unstable sorting] is generally faster than stable sorting, except in a few special cases, e.g. when the slice consists of several concatenated sorted sequences.'

how much faster?

Here are a few benchmarks results from a recent rayon pull:

https://github.com/nikomatsakis/rayon/pull/379

So, for a dual-core with HT 8.26s vs 4.55s


I think it's fair to say that pdqsort (pattern-defeating quicksort) is overall the best unstable sort and timsort is overall the best stable sort in 2017, at least if you're implementing one for a standard library.

The standard sort algorithm in Rust is timsort[1] (slice::sort), but soon we'll have pdqsort as well[2] (slice::sort_unstable), which shows great benchmark numbers.[3] Actually, I should mention that both implementations are not 100% equivalent to what is typically considered as timsort and pdqsort, but they're pretty close.

It is notable that Rust is the first programming language to adopt pdqsort, and I believe its adoption will only grow in the future.

Here's a fun fact: Typical quicksorts (and introsorts) in standard libraries spend most of the time doing literally nothing - just waiting for the next instruction because of failed branch prediction! If you manage to eliminate branch misprediction, you can easily make sorting twice as fast! At least that is the case if you're sorting items by an integer key, or a tuple of integers, or something primitive like that (i.e. when comparison is rather cheap).

Pdqsort efficiently eliminates branch mispredictions and brings some other improvements over introsort as well - for example, the complexity becomes O(nk) if the input array is of length n and consists of only k different values. Of course, worst-case complexity is always O(n log n).

Finally, last week I implemented parallel sorts for Rayon (Rust's data parallelism library) based on timsort and pdqsort[4].

Check out the links for more information and benchmarks. And before you start criticizing the benchmarks, please keep in mind that they're rather simplistic, so please take them with a grain of salt.

I'd be happy to elaborate further and answer any questions. :)

[1] https://github.com/rust-lang/rust/pull/38192

[2] https://github.com/rust-lang/rust/issues/40585

[3] https://github.com/rust-lang/rust/pull/40601

[4] https://github.com/nikomatsakis/rayon/pull/379


Hi stjepang,

If the following criteria is met, then perhaps the branch mis-predict penalty is less of a problem: 1. you are sorting a large amount of data, much bigger than the CPU LLC 2. you can effectively utilize all cores, i.e. your sort algorithm can parallelize Perhaps in this case you are memory bandwidth limited. If so, you are probably spending more time waiting on data than waiting on pipe flushes (i.e. consequence of mis-predicts).


Absolutely - there are cases when branch misprediction is not the bottleneck. It depends on a lot of factors.

Another such case is when sorting strings because every comparison causes a potential cache miss and introduces even more branching, and all that would dwarf that one misprediction.


Comparing sorting algo's often says more about your benchmark than the algo's themselves. Random and pathological are obvious, but often your dealing with something in between. Radix vs n log n is another issue.

So, what where your benchmarks like?


That is true - the benchmarks mostly focus on random cases, although there are a few benchmarks with "mostly sorted" arrays (sorted arrays with sqrt(n) random swaps).

If the input array consists of several concatenated ascending or descending sequences, then timsort is the best. After all, timsort was specifically designed to take advantage of that particular case. Pdqsort performs respectably, too, and if you have more than a dozen of these sequences or if the sequences are interspersed, then it starts winning over timsort.

Anyways, both pdqsort and timsort perform well when the input is not quite random. In particular, pdqsort blows introsort (e.g. typical C++ std::sort implementations) out of the water when the input is not random[1]. It's pretty much a strict improvement over introsort. Likewise, timsort (at least the variant implemented in Rust's standard library) is pretty much a strict improvement over merge sort (e.g. typical C++ std::stable_sort implementations).

Regarding radix sort, pdqsort can't quite match its performance (it's O(n log n) after all), but can perform fairly respectably. E.g. ska_sort[2] (a famous radix sort implementation) and Rust's pdqsort perform equally well on my machine when sorting 10 million random 64-bit integers. However, on larger arrays radix sort starts winning easily, which shouldn't be surprising.

I'm aware that benchmarks are tricky to get right, can be biased, and are always controversial. If you have any further questions, feel free to ask.

[1]: https://github.com/orlp/pdqsort

[2]: https://github.com/skarupke/ska_sort


I would love to see the benchmark results against Timsort, the Python sorting algorithm that also implements a bunch of pragmatic heuristics for pattern sorting. Timsort has a slight advantage over pdqsort in that Timsort is stable, whereas pdqsort is not.

I see that timsort.h is in the benchmark directory, so it seems odd to me that the README doesn't mention the benchmark results.


There are multiple reasons I don't include Timsort in my README benchmark graph:

1. There is no authoritative implementation of Timsort in C++. In the bench directory I included https://github.com/gfx/cpp-TimSort, but I don't know the quality of that implementation.

2. pdqsort intends to be the algorithm of choice of a system unstable sort. In other words, a direct replacement for introsort for std::sort. So std::sort is my main comparison vehicle, and anything else is more or less a distraction. The only reason I included std::stable_sort in the benchmark is to show that unstable sorting is an advantage for speed for those unaware.

But, since you're curious, here's the benchmark result with Timsort included on my machine: http://i.imgur.com/tSdS3Y0.png

This is for sorting integers however, I expect Timsort to become substantially better as the cost of a comparison increases.


Has HN ever discussed the possibilities when purposely crafting worst-case input to amplify a denial-of-service attack?

If whoever you're targeting uses libc++, I already did the analysis: https://bugs.llvm.org/show_bug.cgi?id=20837

To my knowledge it's still not fixed.


Question as a non low-level developer, and please forgive my ignorance:

How is it that we're essentially 50 years in to writing sorting algorithms, and we still find improvements? Shouldn't sorting items be a "solved" problem by now?


Basically all comparison-based sort algorithms we use today stem from two basic algorithms: mergesort (stable sort, from 1945) and quicksort (unstable sort, from 1959).

Mergesort was improved by Tim Peters in 2002 and that became timsort. He invented a way to take advantage of pre-sorted intervals in arrays to speed up sorting. It's basically an additional layer over mergesort with a few other low-level tricks to minimize the amount memcpying.

Quicksort was improved by David Musser in 1997 when he developed introsort. He set a strict worst-case bound of O(n log n) on the algorithm, as well as improved the pivot selection strategy. And people are inventing new ways of pivot selection all the time. E.g. Andrei Alexandrescu has published a new method in 2017[1].

In 2016 Edelkamp and Weiß found a way to eliminate branch mispredictions during the partitioning phase in quicksort/introsort. This is a vast improvement. The same year Orson Peters adopted this technique and developed pattern-defeating quicksort. He also figured out multiple ways to take advantage of partially sorted arrays.

Sorting is a mostly "solved" problem in theory, but as new hardware emerges different aspects of implementations become more or less important (cache, memory, branch prediction) and then we figure out new tricks to take advantage of modern hardware. And finally, multicore became a thing fairly recently so there's a push to explore sorting in yet another direction...

[1] http://erdani.com/research/sea2017.pdf


Thanks for the alexandrescu paper

What makes things tricky is that there are a couple of common cases that can be sorted in O(n), and that more complicated algorithms might have better asmyptotic behaviour, while being worse for small or even moderately large lists.

To make matters worse there are also more specific sorting algorithms like radix sort, which can be even faster in cases where they can be used.


One of the problems is that hardware changes. Long time ago memory was very limited and there was virtually no cost with branching. Now we have very complex pipelined architecture with branch prediction, many levels of cache, microcode, etc. And memory is plenty.

This is an improvment in practical cases, not theoretical ones.

As far as I understood, the base algorithm is the same which has an average case of n log(n). The new algorithms only try to improve the pivot selection to avoid the worst cases and try to be better than the average case for most practical cases. But at the end there are not new algorithms improving the limit of n log(n).

[Post-edit] I made several edits to the post below. First, to make an argument. Second, to add paragraphs. [/Post-edit]

Tl;dr version: It seems to me you should either use heapsort or plain quicksort; the latter with the sort of optimisations described in the linked article, but not including the fallback to heapsort.

Long version:

Here's my reasoning for the above:

You're either working with lists that are reasonably likely to trigger the worst case of randomised quicksort, or you're not working with such lists. By likely, I mean the probability is not extremely small.

Consider the case when the worst case is very unlikely: you're so unlikely to have a worst case that you're gaining almost nothing for accounting for it except extra complexity. So you might as well only use quicksort with optimisations that are likely to actually help.

Next is the case that a worst case might actually happen. Again, this is not by chance; it has to be because someone can predict your "random" pivot and screw with your algorithm; in that case, I propose just using heapsort. Why? This might be long, so I apologise. It's because usually when you design something, you design it to a high tolerance; a high tolerance in this case ought to be the worst case of your sorting algorithm. In which case, when designing and testing your system, you'll have to do extra work to tease out the worst case. To avoid doing that, you might as well use an algorithm that takes the same amount of time every time, which I think means heapsort.


The overhead of including the fallback to heapsort takes a negligible, non-measurable amount of processing time that guarantees a worst case runtime of O(n log n), and to be more precise, a worst case that is 2 - 4 times as slow as the best case.

Your logic also would mean that any sorting function that is publicly facing (which is basically any interface on the internet, like a sorted list of Facebook friends) would need to use heapsort (which is 2-4 times as slow), as otherwise DoS attacks are simply done by constructing worst case inputs.

There are no real disadvantages to the hybrid approach.


Thanks for your reply.

> Your logic also would mean that any sorting function that is publicly facing (which is basically any interface on the internet, like a sorted list of Facebook friends) would need to use heapsort (which is 2-4 times as slow), as otherwise DoS attacks are simply done by constructing worst case inputs.

Why is that a wrong conclusion? It might be, I'm not a dev. But if I found myself caring about that sort of minutiae, I would reach exactly that conclusion.

Reasons:

* the paranoid possibility that enough users can trigger enough DoS attacks that your system can fall over. If this is likely enough, maybe you should design for the 2-4x worst case, and make your testing and provisioning of resources easier.

* a desire for simplicity when predicting performance, which you're losing by going your route because you're adding the possibility of a 2-4x performance drop depending on the content of the list. Ideally, you want the performance to solely be a function of n, where n is the size of your list; not n and the time-varying distribution of evilness over your users.

Finally, adding a fallback doesn't seem free to me, because it might fool you into not addressing the points I just made. That O(n^2) for Quicksort might be a good way to get people to think; your O(n log n) is hiding factors which don't just depend on n.


Presuming your internet based users will enter malicious input is not paranoia, it is the only sane starting place.

Anyone knows how this compares to Timsort in practice?

A quick google turns out nothing



To summarize:

If comparison is cheap (e.g. when sorting integers), pdqsort wins because it copies less data around and the instructions are less data-dependency-heavy.

If comparison is expensive (e.g. when sorting strings), timsort is usually a tiny bit faster (around 5% or less) because it performs a slightly smaller total number of comparisons.


Where is a high level description of the algorithm? How is it different from quick sort, it seems quite similar based on a quick observation of the code.

The readme file actually contains a fairly thorough description of how it differs from quicksort. Start with the section titled "the best case".

> On average case data where no patterns are detected pdqsort is effectively a quicksort that uses median-of-3 pivot selection

So basically is quicksort with a bit more clever pivot selection, but only for some cases.


You're forgetting probably the most important optimization: block partitioning. This one alone makes it almost 2x faster (on random arrays) than typical introsort when sorting items by an integer key.

I always wondered if there would be a way to have quicksort run slower than O(n ln(n)).

Due to that possibility, when I code up a sort routine, I use heap sort. It is guaranteed O(n ln(n)) worst case and achieves the Gleason bound for sorting by comparing keys which means that on average and worst case, on the number of key comparisons, it is impossible to do better than heap sort's O(n ln(n)) forever.

For a stable sort, sure, just extend the sort keys with a sequence number, do the sort, and remove the key extensions.

Quicksort has good main memory locality of reference and a possibility of some use of multiple threads, and heap sort seems to have neither. But there is a version of heap sort modified for doing better on locality of reference when the array being sorted is really large.

But, if are not too concerned about memory space, then don't have to care about the sort routine being in place. In that case, get O(n ln(n)), a stable sort, no problems with locality of reference, and ability to sort huge arrays with just the old merge sort.

I long suspected that much of the interest in in-place, O(n ln(n)), stable sorting was due to some unspoken but strong goal of finding some fundamental conservation law of a trade off of processor time and memory space. Well, that didn't really happen. But heap sort is darned clever; I like it.


Is there a analysis of its complexity ? The algorithm looks very nice !

Hey, author of pdqsort here, the draft paper contains complexity proofs of the O(n log n) worst case and O(nk) best case with k distinct keys: https://drive.google.com/open?id=0B1-vl-dPgKm_T0Fxeno1a0lGT0...

Best case? Give worst and average case when describing complexities.

There's not much value in discussing the asymptotic worst-case or average-case performance of sorting algorithms, because just about every algorithm worth talking about is already optimal in that regard.

The interesting performance differences are all about constant factors.


I already did, you may want to re-read my comment.

Am I misinterpreting your usage of "best case"?

Without enlightening me on what your interpretation is, I have no way of telling.

pdqsort has a worst case of O(n log n). That means, no matter what, the algorithm never takes more than a constant factor times n log n time to complete.

Since pdqsort is strictly a comparison sort, and comparison sorts can do no better than O(n log n) in the average case, pdqsort is asymptotically optimal in the average case (because the average case can never be worse than the worst case).

On top of the above guarantees, if your input contains only k distinct keys, then pdqsort has a worst case complexity of O(nk). So when k gets small (say, 1-5 distinct elements), pdqsort approaches linear time. That is pdqsort's best case.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: