Fusing index transformations is hard. Two days of work and nothing to show for it.
Generics for small, fixed-size vectors http://bit.ly/9ToHYB
3
2
Just when I thought we've finally beaten GHC's optimiser into submission it's being naughty again. Drowning in Core.
Back to a bit of vector and DPH hacking after a long break
@Paul_Bone Moving to London, will try to work on DPH in my spare time
@bos31337 London, gonna work for Standard Chartered
Bye-bye Sydney and thanks for nice 5 years
1
RT @TacticalGrace: New paper: An LLVM Backend For GHC http://post.ly/jYPF #codegen #compilers #ghc #haskell #llvm
1
Too much optimisation. Now those loops are slow because they don't allocate enough. Such a cruel world!
Wrote a multithreaded Haskell HTTP server in only 25 lines, I am very impressed (based on @simonmar's original paper) (via @ndm_haskell)
First release of Repa — a high-performance, parallel array library for Haskell http://post.ly/dlTc (via @TacticalGrace)
I wish there was some way to specify complexity and preconditions in haddock without making the docs look horrible
Sparking imperatives or how to fake lazy threads in Haskell http://bit.ly/9fvD3e
3
2
Package vector has a trac now! http://trac.haskell.org/vector/
1
1
Regular, shape-polymorphic, parallel arrays in Haskell http://post.ly/Y8qF #haskell #multicore #parallelism (via @TacticalGrace)
2
Meh, NoSlow segfaults with the LLVM backend
It turns out I'm not qualified enough for a Haskell programmer position. Maybe in my next life...
Finally released vector 0.5 http://bit.ly/dAtQkK
2
1
NoSlow: Microbenchmarks for Haskell array libraries http://bit.ly/66nnbm
1