using 32bit nvidia opencl using 64bit system

Accelerated Computing CUDA CUDA Programming and Performance
1

if you like to use the 32bit version using a 64bit system, you can try this guide.

you can download the patch here: http://home.in.tum.de/~schreibm/opencl/patch_32on64.patch

this solution works using ubuntu jaunty using the driver NVIDIA-Linux-x86_64-185.18.08-pkg2.run

  1. first, install necessary 32bit libs (maybe i forgot some):

[font=“Courier New”]martin@vanadium:/usr/lib32$ sudo aptitude install lib32stdc++6 lib32gcc1 libc6-dev-i386[/font]

  1. on my ubuntu system, it was necessary to create a symlink for the 32bit version of libstdc++

[font=“Courier New”]martin@vanadium:/usr/lib32$ sudo ln libstdc++.so.6.0.10 ./libstdc++.so -s[/font]

  1. then you can install, patch and run the example programs:

[codebox]

[… install sdl using default installation options …]

martin@vanadium:~/NVIDIA_GPU_Computing_SDK$ patch -p1 < …/patch_32on64.patch

[…]

martin@vanadium:~/NVIDIA_GPU_Computing_SDK$ cd OpenCL/

martin@vanadium:~/NVIDIA_GPU_Computing_SDK/OpenCL$ make force32on64=1

[…]

martin@vanadium:~/NVIDIA_GPU_Computing_SDK/OpenCL$ make runall[/codebox]

good luck!