Compiling Generalised Histograms for GPU (extended abstract)
We present and evaluate an implementation technique for histogram-like computations on GPUs. We show the need for a specialised language construct, embodied as a higher-order function, in order to ensure both a work-efficient asymptotic cost, and an efficient implementation via hardware-supported atomic operations. We demonstrate an implementation in the Futhark compiler and demonstrate that it performs competitively with hand-written GPU code.
Sun 18 AugDisplayed time zone: Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna change
15:20 - 16:40
|Compiling Generalised Histograms for GPU (extended abstract)|
|Position-Dependent Arrays and Their Applicationfor High Performance Code Generation|
Federico Pizzuti University of Edinburgh, Michel Steuwer University of Glasgow, Christophe Dubach University of EdinburghLink to publication DOI Pre-print File Attached
|Safety at speed: In-place array algorithms from pure functional programs by safely re-using storage|