Program Failed in the increment test

Hi all,

I just installed the PGICE 17.4 and cuda 8.0 on my computer (Win 10, I7, GTX1080).

I was trying to run the tests in the CUDA-FORTRAN-BOOK. The deviceQuery could be compiled and run normally.
When I try to run the case increment in chapter one, the code can be compiled but the result is not right. It shows **** Program Failed ****.

So I print the value of ‘a’ out. It turns out all the elements of ‘a’ are still 1 and the program never goes into the kernal.

Anyone know the reason or how to resolve this?


Hi Lazycatyi,

What compilation flags did you use?

Since a GTX1080 uses the Pascal architecture, my guess is that you just need to add the flag “-Mcuda=cc60” to tell the compiler to target Pascal.

Hope this helps,

It works. Thank you very much!