FHPNC 2023
who it's for
array programmers, functional programmers, compiler engineers, gpu programmers, parallelism researchers, performance engineers
topics
rank-polymorphic arrays; shape-guided cache blocking; size-dependent array types; dependently-typed array safety; Haskell-to-GPU pipeline; reverse-mode AD via Dual category; ConCat to Accelerate lowering; BMMC and BPC index permutations; tiled shared-memory GPU kernels; multicore fork-join refactoring; CPS and defunctionalization for parallelism; automated existential size bookkeeping in Futhark
| title | author | length |
|---|---|---|
| Rank-Polymorphism for Shape-Guided Blocking | Artjoms Šinkarovs, Thomas Koopman, Sven-Bodo Scholz | 36m |
Claims a method to express arbitrary levels of cache blocking by encoding block depth and sizes in rank-polymorphic arrays, with reshaping switching between… | ||
| Fast Deep Learning with Categories | Michael Sperber | 42m |
Builds a Haskell-to-GPU pipeline: plain Haskell neural nets get transformed by ConCat into categorical form, compiled to a reverse AD category using the Dual… | ||
| Efficient GPU Implementation of Affine Index Permutations on Arrays | Mathis Bouverot-Dupuis, Mary Sheeran | 42m |
Presents GPU kernels that implement BMMC (bit-matrix-multiply-complement) index permutations with performance comparable to as fast as a simple array copy… | ||
| The best multicore-parallelization refactoring you've never heard of | Mike Rainey | 39m |
Refactors a recursive tree-sum into two paths—a parallel fork‑join version and an optimized serial iterator—then merges them so task creation is triggered only… | ||
| Shape-Constrained Array Programming with Size-Dependent Types | Lubin Bailly, Troels Henriksen, Martin Elsman | 43m |
A dependent-type extension enforces array-size consistency in an ML-style data-parallel language and is implemented in the Futhark compiler; sizes are… | ||
| nothing matches | ||