PolyMage

Abstract

This paper presents the design and implementation of PolyMage, a domain-specific language and compiler for image processing pipelines. An image processing pipeline can be viewed as a graph of interconnected stages which process images successively. Each stage typically performs one of point-wise, stencil, reduction or data-dependent operations on image pixels. Individual stages in a pipeline typically exhibit abundant data parallelism that can be exploited with relative ease. However, the stages also require high memory bandwidth preventing effective utilization of parallelism available on modern architectures. For applications that demand high performance, the traditional options are to use optimized libraries like OpenCV or to optimize manually. While using libraries precludes optimization across library routines, manual optimization accounting for both parallelism and locality is very tedious.The focus of our system, PolyMage, is on automatically generating high-performance implementations of image processing pipelines expressed in a high-level declarative language. Our optimization approach primarily relies on the transformation and code generation capabilities of the polyhedral compiler framework. To the best of our knowledge, this is the first model-driven compiler for image processing pipelines that performs complex fusion, tiling, and storage optimization automatically. Experimental results on a modern multicore system show that the performance achieved by our automatic approach is up to 1.81× better than that achieved through manual tuning in Halide, a state-of-the-art Permission to make digital

Reconciling Compiler Optimizations and WCET Estimation Using Iterative Compilation

“…Evaluation of the optimization selection is supported by a large selection of benchmarks from different domains: Mälardalen benchmark suite, that are reference benchmarks for WCET estimation [11]; Polybench benchmarks [12] for linear algebra, physics and statistics; MiBench [13] benchmarks for its applications in security and telecommunications; and PolyMage [14] for image processing. As this work focuses on optimization, we paid special attention to include complex kernels, which can be challenging for WCET estimation.…”

The Next 700 Accelerated Layers

“…Polyhedral techniques have also been tailored for domain-specific purposes. State-of-the-art examples include the PolyMage [46] DSL for image processing pipelines and the PENCIL approach to the construction of parallelizing and compilers for DSLs [5,9]. PolyMage is a clear illustration of the benefits of operating at a high level of abstraction, closer to the mathematics of the domain of interest: While GCC/Graphite and LLVM/Polly struggle to recover affine control and flow from low-level code, PolyMage natively captures patterns amenable to domain-specific optimization, such as stencil-specific overlapped tiling with or without recomputation, and cache-conscious fusion and tiling heuristics; it also offers a more productive programming experience for end-users.…”

“…The polyhedral framework of compilation emerged as a natural candidate to design a versatile optimization flow satisfying the needs of the domain and target hardware. It has demonstrated strong results in domain-specific optimization [5,9,20,46], expert-driven meta-programming [6,15,26], embedding of third-party library code [40], and automatic generation of efficient code for heterogeneous targets [5,7,43,51,70,77]. We attempt to take the best of both worlds, defining a domain-specific language rich enough to capture full sub-graphs of modern Machine Learning (ML) models while enabling aggressive compilation competitive to native libraries.…”

Flextended Tiles

“…The latter may be implemented by either splitting the shaded region of overlapped tiles or introducing more advanced scheduling constraints on the bounding faces of a tile. No implementations of overlapped tiling are known to exist in generalpurpose compilers, although the technique was implemented in domain-specific compilers for image processing pipelines [24,31] or stencil code generator for GPU [19,37]. There was also an implementation of split tiling for iterated stencils on GPU architectures [15].…”

“…PolyMage [24] is the state-of-the-art polyhedral compilation framework automatically generating high-performance schedules for such image processing pipelines, benefiting from the full inter-tile parallelism enabled by overlapped tiling [21]. It takes as input a domain-specific language (DSL) inspired by Halide [31], computes schedules competitive with manually written ones for image processing pipelines, and generates high-performance imperative code.…”