I bought a Radeon 7970 with 3GB of memory. I wrote an OpenCL program which runs fine on the GPU.
But I have a problem with Memory allocation. I allocate 6 Buffers with the same size.
Every buffer has the size of ~240 MegaByte, if I increase all buffers a little bit, 5 buffers can be allocated
but the 6th buffer can not be allocated on the GPU.
It seems that I can only uses ~2 GB out of 3 GB.
Windows 64 Bit
Visual Studio 2013 64 Bit Project
APP SDK 2.9
Catalyst Version: 13.11
I have set both environment variables to these values:
GPU_MAX_ALLOC_PERCENT = 95
GPU_MAX_HEAP_SIZE = 95
ClInfo.exe returns the following information:
CL_DEVICE_MAX_MEM_ALLOC_SIZE: 2674 MByte
CL_DEVICE_GLOBAL_MEM_SIZE: 2918 MByte
Is there any solution to make usage of all memory?
This seems to be a common issue, but I did not see any solution for this.