This question already has answers here:
How do I declare a 2d array in C++ using new?
(29 answers)
Closed 7 years ago.
How do I define a dynamic multi-dimensional array in C++? For example, two-dimensional array? I tried using a pointer to pointer, but somehow it is failing.
The first thing one should realize that there is no multi-dimensional array support in C++, either as a language feature or standard library. So anything we can do within that is some emulation of it. How can we emulate, say, 2-dimensional array of integers? Here are different options, from the least suitable to the most suitable.
Improper attempt #1. Use pointer to pointer
If an array is emulated with pointer to the type, surely two-dimensional array should be emulated with a pointer to pointer to the type? Something like this?
int** dd_array = new int[x][y];
That's a compiler error right away. There is no new [][] operator, so compiler gladly refuses. Alright, how about that?
int** dd_array = new int*[x];
dd_array[0][0] = 42;
That compiles. When being executed, it crashes with unpleasant messages. Something went wrong, but what? Of course! We did allocate the memory for the first pointer - it now points to a memory block which holds x pointers to int. But we never initialized those pointers! Let's try it again.
int** dd_array = new int*[x];
for (std::size_t i = 0; i < x; ++i)
dd_array[i] = new int[y];
dd_array[0][0] = 42;
That doesn't give any compilation errors, and program doesn't crash when being executed. Mission accomplished? Not so fast. Remember, every time we did call a new, we must call a delete. So, here you go:
for (std::size_t i = 0; i < x; ++i)
delete dd_array[i];
delete dd_array;
Now, that's just terrible. Syntax is ugly, and manual management of all those pointers... Nah. Let's drop it all over and do something better.
Less improper attempt #2. Use std::vector of std::vector
Ok. We know that in C++ we should not really use manual memory management, and there is a handy std::vector lying around here. So, may be we can do this?
std::vector<std::vector<int> > dd_array;
That's not enough, obviously - we never specified the size of those arrays. So, we need something like that:
std::vector<std::vector<int> > dd_array(x);
for(auto&& inner : dd_array)
inner.resize(y);
dd_array[0][0] = 42;
So, is it good now? Not so much. Firstly, we still have this loop, and it is a sore to the eye. What is even more important, we are seriously hurting performance of our application. Since each individual inner vector is independently allocated, a loop like this:
int sum = 0;
for (auto&& inner : dd_array)
for (auto&& data : inner)
sum += data;
will cause iteration over many independently allocated inner vectors. And since CPU will only cache continuous memory, those small independent vectors cann't be cached altogether. It hurts performance when you can't cache!
So, how do we do it right?
Proper attempt #3 - single-dimensional!
We simply don't! When situation calls for 2-dimensional vector, we just programmatically use single-dimensional vector and access it's elements with offsets! This is how we do it:
vector<int> dd_array(x * y);
dd_array[k * x + j] = 42; // equilavent of 2d dd_array[k][j]
This gives us wonderful syntax, performance and all the glory. To make our life slightly better, we can even build an adaptor on top of a single-dimensional vector - but that's left for the homework.
Related
A classic, I'm looking for optimisation here : I have an array of things, and after some processing I know I'm only interested in elements i to j. How to trim my array in the fatset, lightest way, with complete deletions/freeing of memory of elements before i and after j ?
I'm doing mebedded C++, so I may not be able to compile all sorts of library let's say. But std or vector things welcome in a first phase !
I've tried, for array A to be trimmed between i and j, with variable numElms telling me the number of elements in A :
A = &A[i];
numElms = i-j+1;
As it is this yields an incompatibility error. Can that be fixed, and even when fixed, does that free the memory at all for now-unused elements?
A little context : This array is the central data set of my module, and it can be heavy. It will live as long as the module lives. And there's no need to carry dead weight all this time. This is the very first thing that is done - figuring which segment of the data set has to be at all analyzed, and trimming and dumping the rest forever, never to use it again (until the next cycle where we get a fresh array with possibily a compeltely different size).
When asking questions about speed your millage may very based on the size of the array you're working with, but:
Your fastest way will be to not trim the array, just use A[index + i] to find the elements you want.
The lightest way to do this would be to:
Allocate a dynamic array with malloc
Once i and j are found copy that range to the head of the dynamic array
Use realloc to resize the dynamic array to the size j - i + 1
However you have this tagged as C++ not C, so I believe that you're also interested in readability and the required programming investment, not raw speed or weight. If this is true then I would suggest use of a vector or deque.
Given vector<thing> A or a deque<thing> A you could do:
A.erase(cbegin(A), next(cbegin(A), i));
A.resize(j - i + 1);
There is no way to change aloocated memory block size in standard C++ (unless you have POD data — in this case C facilities like realloc could be used). The only way to trim an array is to allocate new array. copy/move needed elements and destroy old array.
You can do it manually, or using vectors:
int* array = new int[10]{0,1,2,3,4,5,6,7,8,9};
std::vector<int> vec {0,1,2,3,4,5,6,7,8,9};
//We want only elements 3-5
{
int* new_array = new int[3];
std::copy(array + 3, array + 6, new_array);
delete[] array;
array = new_array;
}
vec = std::vector<int>(vec.begin()+3, vec.begin()+6);
If you are using C++11, both approaches should have same perfomance.
If you only want to remove extra elements and do not really want to release memory (for example you might want to add more elements later) you can follow NathanOliver link
However, you should consider: do you really need that memory freed immideately? Do you need to move elements right now? Will you array live for such long time that this memory would be lost for your program completely? Maybe you need a range or perharps a view to the array content? In many cases you can store two pointers (or pointer and size) to denote your "new" array, while keeping old one to be released all at once.
I have below code
PositionSummary** psArray_= new PositionSummary*[instSize]
for (int k = 0; k < instSize_; k++)
{
PositionSummary* ps = new PositionSummary();
psArray_[k] = ps;
}
when I try to do a copy as below, it doesn't work out
std::copy(begin(psArray_),end(psArray_),std::back_inserter(psContainer_));
Compiler reports "no matching function for call to ‘begin(PositionSummary*&)", how to get around this?
That doesn't work because psArray_ is not an array at all, it is a pointer. You may know that it points to an array of a specific size, but the compiler has no (general) way of knowing this. If you were to make it an array it would work fine. Try to declare psArray_ like this:
PositionSummary* psArray_ [instSize];
This will work, but presumably not for you because I'm guessing instSize is not a compile time constant. You can fix that by using a vector instead of doing your own memory allocation:
std::vector<PositionSummary*> psArray_;
for (int k = 0; k < instSize; k++)
psArray_.push_back (new PositionSummary());
std::copy(psArray_.begin()),psArray_.end(),std::back_inserter(psContainer_));
For further improvement, you could consider using a smartpointer in the vector (so you don't have to worry about freeing later). And while I have no idea of the greater context of your code, I'm still wondering why the copy is necessary at all - why not just construct your vector directly in the data structure that is ultimately supposed to hold it?
psArray is just a pointer. std::begin and std::end cannot possibly know the length of the array the pointer points to. They work for plain arrays, not for pointers to arrays.
You can do this instead:
std::copy(psArray_, psArray_ + instSize, std::back_inserter(psContainer_));
assuming psContainer_ is a container holding pointers to PositionSummary.
Depending on the details of your application, you may be better off using an std::vector, either of smart pointers or values:
std::vector<some_smart_pointer<PositionSummary>> psArray_;
or
std::vector<PositionSummary> psArray_;
where the second option may be the better one if you don't need referential semantics.
I recently found this won't work in my global CUDA C++ code that I plan to compile and later to be called in Matlab:
int M = 10; float V[M];
or if I were to import M value from the matlab host code.
But this works:
float V[10];
I was told there exists a function called new that I can use to avoid this problem, but I read online and am still quite confused how to use this new function, and it seems only to apply to host code, is that right? If so, it won't apply to my case then, since my host code is in matlab. Is this a way to get around this, so that I don't have to change vector lengths one by one? Thank you!
I don't know anything about MATLAB or CUDA, but your problem is in C++. Arrays declared like that must have sizes fixed at compile-time.
Solution 1: Fix the size
Declare your variable M const. These are equivalent:
int const M = 10;
const int M = 10;
The compiler would then know that it can assume these variables will always have the same value no matter how you run the program.
Solution 2: C-style dynamic allocation
Dynamic allocation with new and delete. Arrays allocated on the abstract section of memory called the "free-store" (rather than on the "stack", like those arrays you have) can determine their sizes on the fly. You use it like this:
float * V = new V[M]; //V is a pointer to some freestore memory
//You use it and pass it like you would a normal array:
V[2] = 5.5;
int x = some_func(V);
//But after you're done, you should manually free the memory
delete [] V; //don't forget the [], since you used [] in the allocation
I don't recommend this, because of the possiblity of forgetting to delete the memory.
Solution 3: Automatic memory management with C++'s vector
In C++, the work of memory management can be hidden behind structures called classes.
#include<vector>
using std::vector;
vector<float> V(M); //V is a vector of floats, with initial size M
//You use it like a normal array
V[2] = 5.5;
//But to pass it as an array, you need to pass a pointer to the first element
int x = some_func(&V[0]); //read as &(V[0]): pass in the address of V[0]
Solution 3b: CUDA-compatible vector
Thrust is a C++ template library for CUDA based on the Standard Template Library (STL). Thrust allows you to implement high performance parallel applications with minimal programming effort through a high-level interface that is fully interoperable with CUDA C.
http://docs.nvidia.com/cuda/thrust/#vectors
Conclusion
If you're using fixed sizes, I recommend solution 1. If you're using sizes determined during runtime, I recommend vector.
(By the way, when you pass an ordinary array to a function, you are actually passing a pointer to the first element, NOT the array. The name of the array is automatically converted to a pointer type.)
int * a;
a = new int[10];
cout << sizeof(a)/sizeof(int);
if i would use a normal array the answer would be 10,
alas, the lucky number printed was 1, because sizeof(int) is 4 and iszeof(*int) is 4 too. How do i owercome this? In my case keeping size in memory is a complicated option. How do i get size using code?
My best guess would be to iterate through an array and search for it's end, and the end is 0, right? Any suggestions?
--edit
well, what i fear about vectors is that it will reallocate while pushing back, well you got the point, i can jus allocate the memory. Hoever i cant change the stucture, the whole code is releevant. Thanks for the answers, i see there's no way around, so ill just look for a way to store the size in memory.
what i asked whas not what kind of structure to use.
Simple.
Use std::vector<int> Or std::array<int, N> (where N is a compile-time constant).
If you know the size of your array at compile time, and it doens't need to grow at runtime, then use std::array. Else use std::vector.
These are called sequence-container classes which define a member function called size() which returns the number of elements in the container. You can use that whenever you need to know the size. :-)
Read the documentation:
std::array with example
std::vector with example
When you use std::vector, you should consider using reserve() if you've some vague idea of the number of elements the container is going to hold. That will give you performance benefit.
If you worry about performance of std::vector vs raw-arrays, then read the accepted answer here:
Is std::vector so much slower than plain arrays?
It explains why the code in the question is slow, which has nothing to do with std::vector itself, rather its incorrect usage.
If you cannot use either of them, and are forced to use int*, then I would suggest these two alternatives. Choose whatever suits your need.
struct array
{
int *elements; //elements
size_t size; //number of elements
};
That is self-explanatory.
The second one is this: allocate memory for one more element and store the size in the first element as:
int N = howManyElements();
int *array = int new[N+1]; //allocate memory for size storage also!
array[0] = N; //store N in the first element!
//your code : iterate i=1 to i<=N
//must delete it once done
delete []array;
sizeof(a) is going to be the size of the pointer, not the size of the allocated array.
There is no way to get the size of the array after you've allocated it. The sizeof operator has to be able to be evaluated at compile time.
How would the compiler know how big the array was in this function?
void foo(int size)
{
int * a;
a = new int[size];
cout << sizeof(a)/sizeof(int);
delete[] a;
}
It couldn't. So it's not possible for the sizeof operator to return the size of an allocated array. And, in fact, there is no reliable way to get the size of an array you've allocated with new. Let me repeat this there is no reliable way to get the size of an array you've allocated with new. You have to store the size someplace.
Luckily, this problem has already been solved for you, and it's guaranteed to be there in any implementation of C++. If you want a nice array that stores the size along with the array, use ::std::vector. Particularly if you're using new to allocate your array.
#include <vector>
void foo(int size)
{
::std::vector<int> a(size);
cout << a.size();
}
There you go. Notice how you no longer have to remember to delete it. As a further note, using ::std::vector in this way has no performance penalty over using new in the way you were using it.
If you are unable to use std::vector and std::array as you have stated, than your only remaning option is to keep track of the size of the array yourself.
I still suspect that your reasons for avoiding std::vector are misguided. Even for performance monitoring software, intelligent uses of vector are reasonable. If you are concerned about resizing you can preallocate the vector to be reasonably large.
I'm trying to think of a method demonstrating a kind of memory error using Arrays and C++, that is hard to detect. The purpose is to motivate the usage of STL vector<> in combination with iterators.
Edit: The accepted answer is the answer i used to explain the advantages / disadvantages. I also used: this
Improperly pairing new/delete and new[]/delete[].
For example, using:
int *array = new int[5];
delete array;
instead of:
int *array = new int[5];
delete [] array;
And while the c++ standard doesn't allow for it, some compilers support stack allocating an array:
int stack_allocated_buffer[size_at_runtime];
This could be the unintended side effect of scoping rules (e.g constant shadowed by a member variable)... and it works until someone passes 'size_at_runtime' too large and blows out the stack. Then lame errors ensue.
A memory leak? IMO, vector in combination with iterators doesn't particularly protect you from errors, such as going out of bounds or generally using an invalidated iterator (unless you have VC++ with iterator debugging); rather it is convenient because it implements a dynamically resizable array for you and takes care of memory management (NB! helps make your code more exception-safe).
void foo(const char* zzz)
{
int* arr = new int[size];
std::string s = zzz;
//...
delete[] arr;
}
Above can leak if an exception occurs (e.g when creating the string). Not with a vector.
Vector also makes it easier to reason about code because of its value semantics.
int* arr = new int[size];
int* second_ref = arr;
//...
delete [] arr;
arr = 0; //play it safe :)
//...
second_ref[x] = y;
//...
delete [] second_ref;
But perhaps a vector doesn't automatically satisfy 100% of dynamic array use cases. (For example, there's also boost::shared_array and the to-be std::unique_ptr<T[]>)
I think the utility of std::vector really shows when you need dynamic arrays.
Make one example using std::vector. Then one example using an array to realloc. I think it speaks for itself.
One obvious:
for (i = 0; i < NUMBER_OF_ELEMENTS; ++i)
destination_array[i] = whatever(i);
versus
for (i = 0; i < NUMBER_OF_ELEMENTS; ++i)
destination_vector.push_back(whatever(i));
pointing out that you know the second works, but whether the first works depends on how destination_array was defined.
void Fn()
{
int *p = new int[256];
if ( p != NULL )
{
if ( !InitIntArray( p, 256 ) )
{
// Log error
return;
}
delete[] p;
}
}
You wouldn't BELIEVE how often I see that. A classic example of where any form of RAII is useful ...
I would think the basic simplicity of using vectors instead of dynamic arrays is already convincing.
You don't have to remember to delete your memory...which is not so simple since attempts to delete it might be bypassed by exceptions and whatnot.
If you want to do dynamic arrays yourself, the safest way to do it in C++ is to wrap them in a class and use RAII. But vectors do that for you. That's kind of the point, actually.
Resizing is done for you.
If you need to support arbitrary types, you don't have to do any extra work.
A lot of algorithms are provided which are designed to handle containers, both included and by other users.
You can still use functions that need arrays by passing the vector's underlying array if necessary; The memory is guaranteed to be contiguous by the standard, except with vector<bool> (google says as of 2003, see 23.2.4./1 of the spec).
Using an array yourself is probably bad practice in general, since you will be re-inventing the wheel...and your implementation will almost definitely be much worse than the existing one...and harder for other people to use, since they know about vector but not your weird thing.
With a dynamic array, you need to keep track of the size yourself, grow it when you want to insert new elements, delete it when it is no longer needed...this is extra work.
Oh, and a warning: vector<bool> is a dirty rotten hack and a classic example of premature optimization.
Why don't you motivate it based on the algorithms that the STL provides?
In raw array, operator[] (if I may call so) is susceptible to index-out-of-bound problem. With vector it is not (There is at least a run time exception).
Sorry, I did not read the question carefully enough. index-out-of-bound is a problem, but not a memory error.