Hi. My apologies - I am pretty sure this is not a unique issue, but I couldn’t find something exactly like this in the forums.
-
I have run the installation, installed license file, etc… to test the system I am compiling and running openACC “hello_world” vector addition.
-
I have found that I have to manually start the license server in order to get the code to compile.
a) is this normal?; and
b) what would you suggest to automate it (if anything)
- After starting the license server by running “lmgrd start”, I get some unexpected verbosity from the license manager when compiling:
sblair@sblair-P650SE:~/work/projects/acc_saxpy$ pgc++ -o acc_saxpy acc_saxpy.cpp -acc -Minfo=acc
18:23:48 (pgroupd) DENIED: “pgc++” sblair@sblair-P650SE (INVALID error code (-9999,333:104 “Connection reset by peer”))
18:23:48 (pgroupd) OUT: “PGI2015-700257” sblair@sblair-P650SE
18:23:48 (pgroupd) OUT: “pgc++” sblair@sblair-P650SE
main:
37, Generating copyin(X[:N],Y[:N])
Generating copyout(Z[:N])
Accelerator kernel generated
Generating Tesla code
40, #pragma acc loop gang, vector(128) /* blockIdx.x threadIdx.x */
18:23:48 (pgroupd) IN: “PGI2015-700257” sblair@sblair-P650SE
18:23:48 (pgroupd) IN: “pgc++” sblair@sblair-P650SE
- Despite the unpromising “DENIED pgc++…” the code compiles and appears to run normally.
a) Is this a problem?; and
b) what could you suggest to make the extraneous info from the license manager go away?
My system is Ubuntu 14.04.2 LTS 64-bit laptop with CUDA 7.5 drivers/7.0 runtime.