compiler: Fix min/max reductions to be backend-portable #2315
Conversation
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## master #2315 +/- ##
=======================================
Coverage 86.69% 86.69%
=======================================
Files 229 229
Lines 42983 43043 +60
Branches 7967 7983 +16
=======================================
+ Hits 37262 37318 +56
- Misses 5033 5034 +1
- Partials 688 691 +3 ☔ View full report in Codecov by Sentry. |
| ispce = cluster.ispace.project(lambda i: i not in dims) | ||
| init.append(cluster.rebuild(exprs=expr, ispace=ispce)) | ||
|
|
||
| processed.append(e.func(lhs, Max(lhs, rhs))) |
There was a problem hiding this comment.
bit of redundance but it's fine,
There was a problem hiding this comment.
thought about factoring some code out but honestly it doesn't get better imho
| limits = FindApplications(ValueLimit).visit(iet) | ||
| includes = set() | ||
| if limits & (set(limits_mapper[np.int32]) | set(limits_mapper[np.int64])): | ||
| includes.add('limits.h') |
There was a problem hiding this comment.
Note: I remember this one being a pain because it's in two different places in the nvidia SDK and one of them is broken
There was a problem hiding this comment.
interesting. For me CUDA and OpenACC are compiling fine, let's see if/when an issue pops up
There was a problem hiding this comment.
I think it was popping up when I was trying to install cupy that was finding weird headers can't remember exactly but I know I had issues with it
No description provided.