32 bit executable fails with insufficient driver version on 64 bit Linux OS

Accelerated Computing CUDA CUDA Programming and Performance
1

I am able to build and execute the deviceQuery example and it works correctly for my Fedora 18, 64-bit computer. If I build the same example with a 32 bit target (i.e., by passing -m32 to the compiler and linker), the example terminates with the following error message:

cudaGetDeviceCount returned 35
→ CUDA driver version is insufficient for CUDA runtime version

The driver version is 319.32.

Is this a bug in the driver, CUDA runtime, or something I can fix?

2

I tried to compile deviceQuery as you did. and I got that error too.

I found out the reason is:
The system does not have 32bit gcc library.

You can install:

glibc.i386
libstdc++.i386

and try again.