I’m trying to using openacc in pvf with a GTX 750Ti, however -ta only has the Tesla option. When I complie it and run, the error comes out: No Nvidia/Cuda version of this construct available for the current device.
There is no problem when I run pgaccelinfo.
The version is 15.3. when I compile the code, only “Generating Tesla code”, never appear something like
“Generating compute capability 1.0 binary
Generating compute capability 2.0 binary
Generating compute capability 3.0 binary”.
I don’t know why.
PGI officially supports the Telsa line of Nvidia accelerators. In some cases when the Geforce product line shares the same architecture as a Tesla product, then PGI compiled code will work. However the Maxwell architecture which is used in a GTX 750ti does not currently have a Tesla equivalent product.