Archives Discussions

[darule0]

I have been going through the tutorials and applications that came with 1.4 Beta

I have been going through the tutorials and applications that came with 1.4 Beta and it seems they all read very small matrices into the GPU <40KB each matrix. If there is a cap on the matrix size, how can the 1GB of the AMD FireStream be filled up? Can this memory only be used to queue up many small matrices for ordered processing? Is there another way to fill up GPU memory? Also, how can I view the current available/free memory of the GPU? Is there some kind of monitor tool for this?

[darule0]

I am running this with:
   Catalyst 9.9
   Windows vista 32-bit
   Stream Computing SDK 1.4 (beta)

[darule0]

The GPU I am using is: FireStream 9250

[gaurav_garg]

CAL provides different methods to query this information.

Yopu can use calDeviceGetInfo, calDeviceGetAttribs, calDeviceGetStatus methods.