- author = "Jonathan Ragan-Kelley and Connelly Barnes and Andrew Adams and Sylvain Paris and Fredo Durand and Saman Amarasinghe",
- title = "Halide: A Language and Compiler for Optimizing Parallelism, Locality, and Recomputation in Image Processing Pipelines",
- booktitle = "Proceedings of the 34th ACM SIGPLAN Conference on Programming Language Design and Implementation",
- year = "2013",
- pages = "519--530",
- address = "Seattle, Washington, USA",
- month = jun # " 16-19",
- publisher = "ACM",
- keywords = "genetic algorithms, genetic programming, genetic improvement, autotuning, compiler, domain specific language, GPU, image processing, locality, optimization, parallelism, redundant computation, vectorization",
- isbn13 = "978-1-4503-2014-6",
- acmid = "2462176",
-
URL = "
http://people.csail.mit.edu/jrk/halide-pldi13.pdf",
-
DOI = "
doi:10.1145/2491956.2462176",
- size = "12 pages",
-
abstract = "Image processing pipelines combine the challenges of
stencil computations and stream programs. They are
composed of large graphs of different stencil stages,
as well as complex reductions, and stages with global
or data-dependent access patterns. Because of their
complex structure, the performance difference between a
naive implementation of a pipeline and an optimized one
is often an order of magnitude. Efficient
implementations require optimization of both
parallelism and locality, but due to the nature of
stencils, there is a fundamental tension between
parallelism, locality, and introducing redundant
recomputation of shared values.
We present a systematic model of the trade-off space fundamental to stencil pipelines, a schedule representation which describes concrete points in this space for each stage in an image processing pipeline, and an optimizing compiler for the Halide image processing language that synthesizes high performance implementations from a Halide algorithm and a schedule. Combining this compiler with stochastic search over the space of schedules enables terse, composable programs to achieve state-of-the-art performance on a wide range of real image processing pipelines, and across different hardware architectures, including multicores with SIMD, and heterogeneous CPU+GPU execution. From simple Halide programs written in a few hours, we demonstrate performance up to 5 times faster than hand-tuned C, intrinsics, and CUDA implementations optimized by experts over weeks or months, for image processing applications beyond the reach of past automatic compilers.",
-
notes = "and an optimizing compiler for the Halide image
processing language... Autotuning Pipeline
Schedules
\cite{Mullapudi:2015:ASPLOS} says GA tool no longer available
MIT CSAIL",
