So far I am moving my c code to NSight Eclipse Edition Version: 5.5.0 on OS C 10.8.5 with Builtin Intel HD GPU VRAM = 288 MB
and NVIDIA GeForce GT 330M 512 MB
I am compiling the c files as is now and will convert them to cu and develop gpu kernels later.
Trying to debugging the legacy code, I get this message:
Error in final launch sequence Failed to execute MI command: -gdb-set cuda software_preemption off Error message from debugger back end: Undefined set cuda command: "software_preemption off". Try "help set cuda". Undefined set cuda command: "software_preemption off". Try "help set cuda".
I tried running from the command line
set cuda software_preemption on
and no errors, I wasn’t sure how to set CUDA_LAUNCH_BLOCKING , CUDA_MEMCHECK , and CUDBG_ENABLE_PREEMPTION_DEBUGGING to 1 as listed:
So I used them as environment variables, and it didn’t work.
I appreciate any help, thank you in advance,