OpenACC + Pinned Memory + Segmenation fault

[b]Used pinned memory like -ta=nvidia:pinned . But getting below error while execution.
[size=24][color=red][color=red]
[b]1 0.000000 0.000000 0.000000
Now time step = 1 (/ 300 )
[tng01:19053] *** Process received signal ***
[tng01:19053] Signal: Segmentation fault (11)
[tng01:19053] Signal code: Invalid permissions (2)
[tng01:19053] Failing at address: 0x3a2ebe17c

mpirun noticed that process rank 0 with PID 19053 on node tng01 exited on signal 11 (Segmentation fault).[/b]
[/color[/color]][/size]
Kindly suggest over it.
Note: Used GPGPU was K40

Hi Manoj_YADAV,

Unfortunately there’s not enough information here to give you good guidance. The “pinned” sub-option simply puts allocated host memory into physical memory rather than use a double buffering system to transfer data from virtual memory to the device. It’s unclear why this cause a seg fault. Though given that it’s a seg fault this means that the error is on the host rather than the device.

Does the same effort occur without the “pinned” sub-option?
Can you try running the program through a debugger, such as pgdbg, to see where the seg fault is occurring?
If you can, send a reproducing example to PGI Customer Service (trs@pgroup.com) and ask them to forward it to me so I can investigate.

Thanks,
Mat