Large matrices and vectors
Customers frequently ask us how large their matrices can be.
NMath matrices (and vectors) are stored internally as an array of floats or doubles.
32-bit .NET processes are limited to 1.2GB of memory. Theoretically a matrix could take up most of that space. Let’s say it’s feasible to have a 1 GB matrix. That matrix would contain 134,217,728 doubles or 268,435,456 floats—for example, a 11,585 x 11,585 square DoubleMatrix or a 16,384 x 16,384 square FloatMatrix.
There’s a workaround in 32-bit .NET to increase the memory to 2.4GB…
1. Add the /3GB switch to boot.ini.
2. After building the application, run the
linker as follows:
link -edit -LARGEADDRESSAWARE myapp.exe
where myapp.exe is your application.
Increasingly, our customers are switching to 64-bit computing in part to get around these memory limitations. Although a 64-bit .NET process’s memory is only limited by the available RAM, the .NET runtime limits any one object to 2GB. For that reason, our matrices are limited to a theoretical maximum of 402,653,184 doubles or 805,306,368 floats —for example, a 20,066 x 20,066 square DoubleMatrix or a 28,377 x 28,377 square FloatMatrix.
- Trevor
Update
We have spoken with Microsoft. They are well aware of this issue and are working on a solution. That solution could involve a blob object that could be created of arbitrary size. We have no word on when this will be available.
December 7th, 2011 at 7:37 am
Trevor,
Any furhter update on this? 2GB limit on 64-bit systems is restricting. Large engineering simulations can easily have 100,000+ equations. Thanks.
Kurt
December 12th, 2011 at 7:46 pm
It’s certainly a very frustrating .NET limitation. Microsoft, in general, still seems unaware of the seriousness of the issue. I believe that they feel that no one would ever create such a large object. The technical computing guys get it but they can’t promise a fix. It could be years, unfortunately.
Internally, we have gotten around this problem by creating memory on the native heap inside our data structures. We have created a custom LU factorization for a customer and underlying matrix, vector. All were long indices with native data arrays. It all worked flawlessly.
We hope to have a version of the product in the future that uses the native heap. Stay tuned.
- Trevor