Add compile-time unrolled for loops #15456
Conversation
juancarlospaco
commented
Oct 2, 2020
•
juancarlospaco
commented
- Add compile-time unrolled for loops.
- Documentation, examples, links, runnableExamples, changelog, etc.
- Tiny diff.
| @@ -178,7 +180,7 @@ else: # not defined(nimNewRoof) | |||
| inc i | |||
|
|
|||
|
|
|||
| iterator `||`*[S, T](a: S, b: T, annotation: static string = "parallel for"): T {. | |||
| iterator `||`*[S, T](a: S, b: T, annotation: static string = "omp parallel for"): T {. | |||
There was a problem hiding this comment.
If you need a new iterator type, make one, don't abuse the parallel for iterator.
Removing the omp as a prefix of the iterator annotation is a breaking change for anyone using this iterator with a custom annotation.
There was a problem hiding this comment.
I dont understand what you mean, the || keeps working.
There was a problem hiding this comment.
Now anyone that use || with a custom annotation must prefix their annotation with omp, so this is a breaking change.
@mratsim's laser is a big user of this iterator, and this change will cause it not to compile.
There was a problem hiding this comment.
My advise would be to create a brand new loop type. And note that #pragma unroll can unroll any loop, so a solution that allows tagging a generated loop would be much more efficient. Basically implementing the {.unroll.} pragma, if you may.
…lang#16472) * continue nim-lang#15456 * follow the advice from juan_carlos
…lang#16472) * continue nim-lang#15456 * follow the advice from juan_carlos
