Trevor L. McDonell

Results 25 issues of Trevor L. McDonell

The CPU backend is currently broken on windows because it lacks an object file loader ([here](https://github.com/AccelerateHS/accelerate-llvm/blob/master/accelerate-llvm-native/Data/Array/Accelerate/LLVM/Native/Link/COFF.hs)). This is the same as what GHCi does, but a bit simpler because we...

windows
llvm-cpu
help wanted

The standard accelerate test suite, used by all the backends, can be quite slow. Several of the tests are significantly slower than the others, for example segmented folds and scans,...

good first issue
help wanted

**Description** This bug contains some examples where the pretty printer currently generates incorrect or ambiguous results. Please feel free to add new examples as you find them! **Steps to reproduce**...

frontend
bug

Accelerate's internal AST has accrued some redundant terms in the AST over its lifetime, which might be good to consolidate. - `Replicate` and `Slice` are effectively moved to the scalar...

Now that travis has proper build stages, we should use this for the [accelerate-travis-buildbot](https://travis-ci.org/tmcdonell-bot/accelerate-travis-buildbot) so that it is only updated when all builds pass. https://blog.travis-ci.com/2017-05-11-introducing-build-stages

help wanted

Performance of multidimensional reductions is not good when the array is highly skewed. For example, a `fold` where the number of columns is (innermost dimension) is very small. See also...

good first issue
llvm-ptx
help wanted

We need to set up some kind of performance regression testing. @rrnewton and I started to set this up at http://codespeed.crest.iu.edu, following @nomeata's example for ghc at http://ghcspeed-nomeata.rhcloud.com, but so...

help wanted

Originally mentioned here: https://groups.google.com/forum/#!topic/accelerate-haskell/UUNYssdEykQ

new feature
language

@rrnewton notes in #48 that the current (driver default) behaviour is to spin when waiting for GPU operations to complete, which is not friendly towards other Haskell threads that want...

new feature

We don't have any form of loop-invariant code-motion implemented. At the scalar level we could assume that the backend compiler (CUDA / LLVM) will do that for us, but at...

frontend
new feature