GTX Titan Z Global Memory

I have a GTX Titan Z graphics card. It has two GPUs with 12GB total memory (6GB + 6GB). When I use the DeviceQuery app in the Cuda Sample (V6.5) folder to view the spec, it shows two devices, each with 4GB total memory. Also, in my C ++ code, I can only access 4GB of memory. On the other hand, when I run the GPU-Z software, it shows two Titan Zs, each with 6GB of memory. Can anyone please explain what caused this problem and how can I fix it?


source to share

1 answer

The problem was that the program was being compiled as a 32-bit application. With 32 bits, the program can only handle 4GB of memory. It seems that the CUDA call to check device characteristics ( cudaGetDeviceProperties

) recognizes this fact and only reports 4GB that you can actually use.

Compiling as a 64-bit application should solve this problem.



All Articles