Paul Osmialowski
Paul Osmialowski
I don't know about gfortran, yet I know that there is plenty of code omitting those `::`'s that can be successfully (and correctly) compiled by some of the commercial compilers....
@abrahamtovarmob the problem here is very deeply hidden and even if you compile and run this example, it may or may not work properly. If it works, you may think...
@abrahamtovarmob finally I've managed to spot missing `call` in the example above. The complete program that reproduces the issue looks like this: ``` module test type t(k) integer, len::k real...
I'm skeptical on our ability of replicating gfortran logic behind this. As using `-Hx,4,0x100000` flag prevents flang from generating calls to those heap allocation functions (resulting in gfortran-like performance on...
I'm starting to suspect that the decision this compiler would have to make cannot be made at compile time. Maybe all we need is a different approach: the compiler should...
Michal, what I wanted to say is that we may need to redefine the problem to make finding a solution more realistic.
Hi @kiranchandramohan, The code snippet you quoted above comes from the 'Background' comment you've kindly wrote above: > There is no temporary needed here since even if val1 aliases with...
> In Polly, the only promising pass I found was polly-dependences, but this is actually addressing reduction dependencies, so again - we'd have to create something from scratch. Polly won't...
Hi @kiranchandramohan, This was purely conceptual work with no documentation ever written. This PR consists of three commits: 1. `machreg: AArch64 has 64-bit registers, no need to pack`, this commit...
The test case I was using is following: ``` ! ********************************************************* program main implicit none ! --------------------------------------------------- character(len=500) :: cart real(kind=8) :: t1,t2 ! --------------------------------------------------- open(unit=9,status='old',file='my_file.txt') open(unit=10,file='my_new_file.txt') call cpu_time(t1) do...