Tried to compile an accelerated program using the following flag $ pgcc -ta=nvidia,cc11 program.c as well as trying $ pgcc -ta=nvidia(cc11) program.c as shown in the User’s Guide for 9.0 Are these suboptions (cc10, cc11, cc13) available for use yet, and if so what is the format for compiling using them?
The first option you used is correct (i.e. “pgcc -ta=nvidia,cc11 program.c”) if your NVIDIA card uses compute capability 1.1. By default the compiler generates code targeting compute capability 1.3. In other words, “-ta=nvidia” implies “cc13” unless “cc10” or “cc11” are used.
Hope this helps,
When using $ pgcc -ta=nvidia,cc11 program.c I receive the following error:
pgcc-Error-Switch -ta with unknown keyword cc11
Choose target accelerator
nvidia Select NVIDIA accelerator target
analysis Analysis only, no code generation
nofma Don’t generate fused mul-add instructions
none Select NO accelerator target
time Collect simple timing information
Are the cc suboptions just not implemented with the 9.0 version yet or is there something else I’m missing?
Something else is happening since cc11 is implemented in 9.0.
% pgcc -V pgcc 9.0-1 64-bit target on x86-64 Linux -tp k8-64e Copyright 1989-2000, The Portland Group, Inc. All Rights Reserved. Copyright 2000-2009, STMicroelectronics, Inc. All Rights Reserved. % pgcc -ta=nvidia,cc11 program.c -o prog.out % prog.out 100000 iterations completed
What is the output of “pgcc -V”?
Got it to work. Path was pointing to the previous version.