I've got a piece of code that takes into account a given amount of features, where each feature is Boolean. I'm looking for the most efficient way to store a set of such features. My initial thought was to try and store these as a BitSet. But then, I realized that this implementation is meant to be used to store numbers in bit format rather than manipulate each bit, which is something I'd like to do (see the effect of switching any feature on and off). I then thought of using a Boolean array, but apparently the JVM uses much more memory for each Boolean element than the one bit it actually needs.
I'm therefore left with the question: What is the most efficient way to store a set of bits that I'd like to treat as independent bits rather than the building blocks of some number?
Please refer to this question: boolean[] vs. BitSet: Which is more efficient?
According to the answer of Peter Lawrey, boolean[] (not Boolean[]) is your way to go since its values can be manipulated and it takes only one byte of memory per bit to store. Consider that there is no way for a JVM application to store one bit in only one bit of memory and let it be directly (array-like) manipulated because it needs a pointer to find the address of the bit and the smallest addressable unit is a byte.
The site you referenced already states that the mutable BitSet is the same as the java.util.BitSet. There is nothing you can do in Java that you can't do in Scala. But since you are using Scala, you probably want a safe implementation which is probably meant to be even multithreaded. Mutable datatypes are not suitable for that. Therefore, I would simply use an immutable BitSet and accept the memory cost.
However, BitSets have their limits (deriving from the maximum number of int). If you need larger data sizes, you may use LongBitSets, which are basically Map<Long, BitSet>. If you need even more space, you may nest them in another map Map<Long, LongBitSet>, but in that case you need to use two or more identifiers (longs).
Related
It is a common fact that it is better in a certain circumstances to pass a parameter by reference to avoid costly copying. But recently I watched a Handmade Hero series where Casey said that if the object is not too complex sometimes it's better to pass it by value. I'm not too familiar with low-level details, but I assume it's connected with a cache. Could someone give more solid explanation of what's going on?
If you pass by value you are likely passing via registers (assuming not too many arguments and each one is not too large). That means the callee doesn't need to do anything to use the values, they are already in registers. If passing by reference, the address of each value may be in a register, but that requires a dereference which needs to look in the CPU cache (if not main memory), which is slower.
On many popular systems you can pass-by-value roughly 5-10 values which are each as wide as an address.
I know that Immutable objects offer several advantages over mutable objects like they are easier to reason about than mutable ones, they do not have complex state spaces that change over time, we can pass them around freely, they make safe hash table keys etc etc.So my question is what are the disadvantages of immutable objects??
Quoting from Effective Java:
The only real disadvantage of immutable classes is that they require a
separate object for each distinct value. Creating these objects can be
costly, especially if they are large. For example, suppose that you
have a million-bit BigInteger and you want to change its low-order
bit:
BigInteger moby = ...;
moby = moby.flipBit(0);
The flipBit method
creates a new BigInteger instance, also a million bits long, that
differs from the original in only one bit. The operation requires time
and space proportional to the size of the BigInteger. Contrast this to
java.util.BitSet. Like BigInteger, BitSet represents an arbitrarily
long sequence of bits, but unlike BigInteger, BitSet is mutable. The
BitSet class provides a method that allows you to change the state of
a single bit of a millionbit instance in constant time.
Read the full item on Item 15: Minimize mutability
Apart from possible performance drawbacks (possible! because with the complexity of GC and HotSpot optimisations, immutable structures are not necessarily slower) - one drawback can be that state must now be threaded through your whole application. For simple applications or tiny scripts the effort to maintain state this way might be too high to buy you concurrency safety.
For example think of a GUI framework like Swing. It would be definitely possible to write a GUI framework entirely using immutable structures and one main "unsafe" outer loop, and I guess this has been done in Haskell. Some of the problems of maintaining nested immutable state can be addressed for example with lenses. But managing all the interactions (registering listeners etc.) may get quite involved, so you might instead want to introduce new abstractions such as functional-reactive or hybrid-reactive GUIs.
Basically you lose some of OO's encapsulation by going all immutable, and when this becomes a problem there are alternative approaches such as actors or STM.
I work with Scala on a daily basis. Immutability has certain key advantages as we all know. However sometimes it's just plain easier to allow mutable content in some situations. Here's a contrived example:
var counter = 0
something.map {e =>
...
counter += 1
}
Of course I could just have the map return a tuple with the payload and count, or use a collection.size if available. But in this case the mutable counter is arguably more clear. In general I prefer immutability but also allow myself to make exceptions.
To answer this question I would quote Programming in Scala, second Edition, chapter "Next Steps in Scala", item 11, by Lex Spoon, Bill Venners and Martin Odersky :
The Scala perspective, however, is that val and var are just two different tools in your toolbox, both useful, neither inherently evil. Scala encourages you to lean towards vals, but ultimately reach for the best tool given the job at hand.
So I would say that just as for programming languages, val and var solves different problems : there is no "disavantage / avantage" without context, there is just a problem to solve, and both of val / var address differently the problem.
Hope it helps, even if it does not provide a concrete list of pros / cons !
We're encouraged to use struct over class in Swift.
This is because
The compiler can do a lot of optimizations
Instances are created on the stack which is a lot more performant than malloc/free calls
The downside to struct variables is that they are copied each time when returning from or assigned to a function. Obviously, this can become a bottleneck too.
E.g. imagine a 4x4 matrix. 16 Float values would have to be copied on every assign/return which would be 1'024 bits on a 64 bit system.
One way you can avoid this is using inout when passing variables to functions, which is basically Swifts way of creating a pointer. But then we're also discouraged from using inout.
So to my question:
How should I handle large, immutable data structures in Swift?
Do I have to worry creating a large struct with many members?
If yes, when am I crossing the line?
This accepted answer is not entirely answering the question you had: Swift always copies structs. The trick that Array/Dictionary/String/etc do is that they are just wrappers around classes (which contain the actual stored properties). That way sizeof(Array) is just the size of the pointer to that class (MemoryLayout<Array<String>>.stride == MemoryLayout<UnsafeRawPointer>.stride)
If you have a really big struct, you might want to consider wrapping its stored properties in a class for efficient passing around as arguments, and checking isUniquelyReferenced before mutating to give COW semantics.
Structs have other efficiency benefits: they don't need reference-counting and can be decomposed by the optimiser.
In Swift, values keep a unique copy of their data. There are several
advantages to using value-types, like ensuring that values have
independent state. When we copy values (the effect of assignment,
initialization, and argument passing) the program will create a new
copy of the value. For some large values these copies could be time
consuming and hurt the performance of the program.
https://github.com/apple/swift/blob/master/docs/OptimizationTips.rst#the-cost-of-large-swift-values
Also the section on container types:
Keep in mind that there is a trade-off between using large value types
and using reference types. In certain cases, the overhead of copying
and moving around large value types will outweigh the cost of removing
the bridging and retain/release overhead.
From the very bottom of this page from the Swift Reference:
NOTE
The description above refers to the “copying” of strings, arrays, and dictionaries. The behavior you see in your code will always be as if a copy took place. However, Swift only performs an actual copy behind the scenes when it is absolutely necessary to do so. Swift manages all value copying to ensure optimal performance, and you should not avoid assignment to try to preempt this optimization.
I hope this answers your question, also if you want to be sure that an array doesn't get copied, you can always declare the parameter as inout, and pass it with &array into the function.
Also classes add a lot of overhead and should only be used if you really must have a reference to the same object.
Examples for structs:
Timezone
Latitude/Longitude
Size/Weight
Examples for classes:
Person
A View
This paper presents a technique for the implementation of functional languages with fast equality, sets and maps, using hash-consing under the hoods. As far as my understanding goes, it uses the address of a hash-consed value as its key when inserting it on a map. This has the advantage that figuring the hashed key of essentially any value is O(1), as opposed to the O(N) standard. What I don't understand, though, is: what happens with a map after a garbage collection? Since the GC process will cause the address of every value to change, then the configuration of the map will be incorrect. In other words, there is no guarantee that addr(value) will be the same for the lifetime of the program.
Since the GC process will cause the address of every value to change
Only moving garbage collectors do that. When using non-moving algorithms like mark-and-sweep, all that happens is that unused objects are freed during the GC cycle - used objects stay exactly where they are.
Moving garbage collectors are generally seen as preferable to mark-and-sweep, but according to the abstract of the paper "mark-and-sweep becomes fast in a maximal sharing environment", which is further expanded on in section 2.4.4.
The paper also describes a way to make moving garbage collectors work (by assigning each object a unique id and using that instead of its address), but deems that impractical (section 2.4.2).
First off, I've seen this, but it doesn't quite seem to suit my needs.
I've got a situation where I need a sparse array. Some situations where I could have, say 3000 potential entries with only 20 allocated, other situations where I could have most or all of the 3000 allocated. Using an NSMutableDictionary (with NSString representations of the integer index values) would appear to work well for the first case, but would seemingly be inefficient for the second, both in storage and lookup speed. Using an NSMutableArray with NSNull objects for the empty entries would work fairly well for the second case, but it seems a bit wasteful (and it could produce an annoying delay at the UI) to insert most of 3000 NSNull entries for the first case.
The referenced article mentions using an NSMapTable, since it supposedly allows integer keys, but apparently that class is not available on iPhone (and I'm not sure I like having an object that doesn't retain, either).
So, is there another option?
Added 9/22
I've been looking at a custom class that embeds an NSMutableSet, with set entries consisting of a custom class with integer (ie, element#) and element pointer, and written to mimic an NSMutableArray in terms of adds/updates/finds (but not inserts/removals). This seems to be the most reasonable approach.
A NSMutableDictionary probably will not be slow, dictionaries generally use hashing and are rather fast, bench mark.
Another option is a C array of pointers. Allocation a large array only allocates virtual memory until the real memory is accessed (cure calloc, not malloc, memset). The downside is that memory is allocated in 4KB pages which can be wasteful for small numbers of entries, for large numbers of entries many may fall in the same page.
What about CFDictionary (or actually CFMutableDictionary)? In the documentation, it says that you can use any C data type as a key, so perhaps that would be closer to what you need?
I've got the custom class going and it works pretty well so far. It's 322 lines of code in the h+m files, including the inner class stuff, a lot of blank lines, comments, description formatter (currently giving me more trouble than anything else) and some LRU management code unrelated to the basic concept. Performance-wise it seems to be working faster than another scheme I had that only allowed "sparseness" on the tail end, presumably because I was able to eliminate a lot of special-case logic.
One nice thing about the approach was that I could make much of the API identical to NSMutableArray, so I only needed to change maybe 25% of the lines that somehow reference the class.
I also needed a sparse array and have put mine on git hub.
If you need a sparse array feel free to grab https://github.com/LavaSlider/DSSparseArray