Why is MATLAB crashing on medium-to-large matrices?

6 visualizaciones (últimos 30 días)
Richard Rast
Richard Rast el 7 de Mzo. de 2016
Comentada: Richard Rast el 7 de Mzo. de 2016
To keep this really short, the following code crashes MATLAB on my machine:
X = ones(100, 100);
A = X*X;
One could argue that, perhaps, MATLAB should not handle something like ones(1000000,1000000) that well; I think it should, but perhaps some reasonable size limit is acceptable. But the above is trivial; I can't imagine why it's crashing. Other similar examples result in a crash. Multiplication is what's causing the problem (e.g. addition is fine). It's related to size -- a 5x5 matrix works with no problems. And the instantiation of the matrix is also fine; I can even construct much larger matrices with no issues.
My MATLAB is version 2015b. I'm happy to supply whatever other information is relevant.
  1 comentario
Walter Roberson
Walter Roberson el 7 de Mzo. de 2016
Are you using Windows 32 or Windows 64? How much RAM do you have? When you say it crashes MATLAB, do you mean that it creates the crash dialog?

Iniciar sesión para comentar.

Respuesta aceptada

Steven Lord
Steven Lord el 7 de Mzo. de 2016
What type of processor are you using? If you're using an AMD Jaguar processor, see if the information in Bug Report 1043644 is applicable.
  1 comentario
Richard Rast
Richard Rast el 7 de Mzo. de 2016
I don't believe my processor was advertised as a Jaguar, but it is AMD. I have upgraded to 2016a, so whatever the problem turned out to be, this fixed it.

Iniciar sesión para comentar.

Más respuestas (2)

Image Analyst
Image Analyst el 7 de Mzo. de 2016
Look into memmapfile().

Chris Turnes
Chris Turnes el 7 de Mzo. de 2016
Does it create a crash log? If so, what's in the log? You can find it using the directions located here:

Categorías

Más información sobre Logical en Help Center y File Exchange.

Productos

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by