Closed jramapuram closed 9 years ago
Are we trying to use double precision? Shouldn't be needed for nnets. The trajectory seems to be going to smaller and smaller types actually (i.e. nirvana's 16bit ultra fast impl)
It should not be using double precision. Which version of ArrayFire are you using ?
3.02
Can you try this with our devel branch ?
This seems to be a problem in the main arrayfire library. I pushed a fix to this library that works around the issue.
Nice, ty Ill try it with the tip
jason@tesla ~/p/a/build> ./test/Activations_opencl in[0] [5 1 1 1] -8.9262 32.2371 45.1800 -32.0635 -8.0176
libc++abi.dylib: terminating with uncaught exception of type af::exception: ArrayFire Exception(401): Double precision not supported for this device In /var/lib/jenkins-slave/workspace/arrayfire-osx-graphics-installer/src/api/cpp/data.cpp:27 fish: Job 1, './test/Activations_opencl ' terminated by signal SIGABRT (Abort)