Unable to compile pgfortran on GPU

I have followed M. Wolfe youtube videos on using pgfortran. I was able to compile with -ta option for “host” and “multicore”. When I try to compile for GPU with -ta:tesla:cc30 option it tells me
“pgfortran-Error-Switch -ta with unknown keyword tesla:cc30
-ta=host|multicore Choose target accelerator
host Compile for serial execution on the host CPU
multicore Compile for parallel execution on the host CPU”
and proceeds to offer “host” and “multicore” options.

I assume that it is an installation problem. I have:
Macbook 15 with Geforce 750M
OSX High Sierra 10.13.3
xcode and command line tool 8.3.3
Nvidia web driver
CUDA Driver Version: 387.128
GPU Driver Version: 387.
Cuda toolkit 9.1 (I can compile samples on GPU for these)

I execute the code with the command:
“pgfortran -o jsolvef.exe jsolvef.F90 -fast -Minfo=opt -ta:tesla:cc30,managed -Minfo=accel”

OS X versions of PGI compilers DO NOT support GPUs.

You can compile with -acc, which is meant as -ta=multicore,
and can develop OpenACC code. But you cannot run it on a GPU
(Nvidia or AMD) under OS X.

But -ta=tesla is not supported.

Mr. Wolfe, interestingly, does most of his writing on a MAC, but does
almost all of his coding on Linux.


Dave thanks for reply. I hope PGI has a change of mind in the future.