Closed collinarnett closed 10 months ago
I cannot reproduce on the nightly version. I'm pretty sure this was fixed in 0.25.9.
I am seeing an error which could be related, in the mss test.
cpj@utopia-1:/scratch-fast/cpj/futfut/bin$ futhark --version
Futhark 0.26.0 (prerelease - include info below when reporting bugs)
git: HEAD @ 5234eb8 (Tue Dec 5 14:54:42 2023 +0100)
-- Parallel maximum segment sum
--
-- ==
-- random input { [1000]i32 } auto output
-- random input { [1000000]i32 } auto output
-- random input { [1000000000]i32 } auto output
def main(xs: []i32): i32 =
let max = i32.max
let redOp (mssx, misx, mcsx, tsx) (mssy, misy, mcsy, tsy) =
( max mssx (max mssy (mcsx + misy))
, max misx (tsx+misy)
, max mcsy (mcsx+tsy)
, tsx + tsy)
let mapOp x =
( max x 0
, max x 0
, max x 0
, x)
in (reduce redOp (0,0,0,0) (map mapOp xs)).0
I get this error:
cpj@utopia-1:/scratch-fast/cpj/futfut/bin$ futhark c ../benches/mss.fut
Internal compiler error (unhandled IO exception).
Please report this at https://github.com/diku-dk/futhark/issues
<<loop>>
No, that is a very different bug - an infinite loop (and simple enough that the Haskell RTS can recognise it).
But I also cannot reproduce that one.
I am going to close this because I think it is already fixed.
Hello,
I'm trying to build a simple linear regression example to benchmark against some jax code (for fun) and I came across this error when trying to compile my code as a python library.
Here's my code as well. I assume my usage of vjp here is wrong but I figured I'd see if I could use it regardless.
System Information
Futhark Version