GHC takes huge amounts of memory and compile time when compiling ZipWith from accelerate
In #15488 (closed), we found that GHC takes huge amounts of memory by building up excessively large amounts of Core. Looking a the Data.Array.Accelerate.Analysis.Hash module there, the problem seems to be due to excessive (and unproductive) inlining.
However, the Data.Array.Accelerate.Test.NoFib.Prelude.ZipWith module *also* takes a very large amount of time and memory to compile, and while it also builds up huge amounts of Core in unproductive ways, the blowup happens in a different step.
https://github.com/AccelerateHS/accelerate/issues/428#issuecomment-425019263 has a build log that points at the first round of Specialise as the culprit.
Just like #15488 (closed), this problem seems to trace back to GHC 8.0.2 at least, possibly further.
Trac metadata
| Trac field | Value |
|---|---|
| Version | 8.0.2 |
| Type | Bug |
| TypeOfFailure | OtherFailure |
| Priority | normal |
| Resolution | Unresolved |
| Component | Compiler |
| Test case | |
| Differential revisions | |
| BlockedBy | |
| Related | |
| Blocking | |
| CC | |
| Operating system | |
| Architecture |