Rupert Ford
Rupert Ford
If I remember correctly, some of the existing kernel transformations (that replace intrinsics) have logic similar to the above. Perhaps the splitting out of an intrinsic/function should have its own...
@christophermaynard has just found that his branch does not work with PSyclone due to him declaring a reduction scalar with precision `r_solver`.
Here is an example of generated code to help make it clear what the issues are ... ``` REAL(KIND=r_def), intent(out) :: asum ... TYPE(scalar_type) global_sum ... asum = 0.0_r_def !...
If the scalar reduction variable `asum` was declared as `r_solver` then the global sum (implemented in `global_sum`) would still be `r_def` as `scalar_type` uses `r_def` so any global sums would...
I think the reason we do not need colouring is not to do with accessing the halos, it is to do with the fact that the writes to shared dofs...
This also includes the interface name for mixed precision kernels as the interface name is essentially metadata information.
Created branch [1721_unsupported_rhs](https://github.com/stfc/PSyclone/tree/1721_unsupported_rhs)
The Fortran intrinsics used in the tracer advection example are `MAX`, `SIGN`, `MIN` and `ABS`.
I can't find any examples of using intrinsics in the dawn repo or in any previously supplied code so I've contacted Matthias requesting an example.
Created branch [1826_sir_intrinsics](https://github.com/stfc/PSyclone/tree/1826_sir_intrinsics)