Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.

USM offload issue

jimdempseyatthecove
Honored Contributor III
156 Views

Following the example listed in

jimdempseyatthecove_0-1761919290112.pngjimdempseyatthecove_1-1761919403705.png

My variation of the test program errors out with

"PluginInterface" error: Failure to run target region 0x000002067f104068 in device 0: "unknown or internal error" Error in launch Kernel __omp_offloading_b46132e0_9a052b74_MAIN___l51: -1
omptarget error: Executing target region abort target.
omptarget error: Run with
omptarget error: LIBOMPTARGET_DEBUG=1 to display basic debug information.
omptarget error: LIBOMPTARGET_DEBUG=2 to display calls to the compute runtime.
omptarget error: LIBOMPTARGET_INFO=4 to dump host-target pointer mappings.
GPUtest2.f90:51:58: omptarget fatal error 1: failure of target construct while offloading is mandatory

Using Arc Pro B50

 

Jim Dempsey

 

0 Kudos
1 Reply
jimdempseyatthecove
Honored Contributor III
137 Views

I reduce the program to that as was illustrated (but added one print statement at the end).

Still fails

!  GPUtest3.f90 
program GPUtest3
    use omp_lib
    implicit none
    integer, parameter :: N=16
    integer :: i, dev
    integer, allocatable :: x(:)
    dev = omp_get_default_device()
    !$omp allocate allocator(omp_target_shared_mem_alloc)
    allocate(x(N))
    
    do i=1,N
            x(i) = i
    end do
    
    !$omp target has_device_addr(x)
    !$omp teams distribute parallel do
    do i=1,N
        x(i) = x(i) * 2
    end do
    !$omp end target
    print *, x
    deallocate(x)
end program GPUtest3

 

Jim

0 Kudos
Reply