MATLAB: Is it possible to access a memory-mapped file in a C++ program to the page file from MATLAB

mapMATLABmemorypage

I am trying to use the MEMMAPFILE function to open an anonymous C++ memory map in MATLAB, but am unable to do so. I am using 'CreateFileMapping' (from the Microsoft Windows API) to keep the memory map out of the file system and in the page file instead. There is a mapping object that has a name and lives in memory, but no actual disk file.

'CreateFileMapping' is an operating system function that is documented on the MSDN website at:

 <http://msdn.microsoft.com/library/default.asp?url=/library/en-us/memory/base/createfilemapping.asp>

At the bottom of that page, there is an example named 'Creating Named Shared Memory' that my code is closely modeled on. A direct link to that example is:

 <http://msdn.microsoft.com/library/default.asp?url=/library/en-us/memory/base/creating_named_shared_memory.asp>

I would like to access this memory map from MATLAB.

Best Answer

The ability to read or write to an anonymous memory map in the page file is not available in MATLAB.

MATLAB functions for memory mapping are limited to MEMMAPFILE. MEMMAPFILE maps a saved disk file and cannot be used to access an anonymous memory map in the page file.

Related Solutions

MATLAB: How to enhance the memory allocation in the MEX-file

The reason MATLAB crashes when you run your application is because there isn't enough memory on the stack. Since the stack memory is shared with MATLAB, which is already using a lot of the memory, there is not enough memory left on the stack for the memory allocation done in your code.

A solution to this problem is to use the dynamic allocation of memory using the MALLOC command in C. This will force the memory allocation to be done on the heap. However, there is still an upper limit of how much memory can be allocated using MALLOC which will be governed by the operating system. Also make sure you free the memory allocated with MALLOC to prevent memory leaks. You will have to use the FREE command in C to free up the memory.

MATLAB: Does using memory-mapped files appear to be slower than using standard I/O functions in MATLAB

At the operating system level, using a memory-mapped file is faster than using standard I/O functions. However, there is some additional overhead incurred in using the MEMMAPFILE object within MATLAB.

With MEMMAPFILE, there is an overhead penalty associated with every access of the Data member of this class. Therefore you will likely not see improved performance when reading or writing a small number of values in a single assignment to a memory mapped file when compared to FWRITE.

To work around this issue, you can create a reference to the "Data" property, which cuts down on some overhead when accessing data. The following code using memory-mapped files should run faster than using standard I/O functions:

m = memmapfile ('myfile.dat', 'Format', 'double' );
dataRef = m.Data;
for k = 1 : N
      y(k) = dataRef (k);
end

The general rules and techniques for vectorization of code apply even more so to MEMMAPFILE objects. You can find details regarding vectorization here:

Techniques to improve performance (MATLAB Programming):

https://www.mathworks.com/help/matlab/matlab_prog/techniques-for-improving-performance.html

More information on memory mapped files in MATLAB is available here:

Accessing Files with Memory-Mapping :: Data Import and Export (MATLAB Programming)

http://www.mathworks.com/help/matlab/import_export/overview-of-memory-mapping.html

Best Answer

Related Solutions

MATLAB: How to enhance the memory allocation in the MEX-file

MATLAB: Does using memory-mapped files appear to be slower than using standard I/O functions in MATLAB

Related Question