Block-Parallel Programming for Real-Time Embedded Applications

Abstract

Abstract-Embedded media applications have traditionally used custom ASICs to meet their real-time performance requirements. However, the combination of increasing chip design cost and availability of commodity many-core processors is making programmable devices increasingly attractive alternatives. Yet for these processors to be successful in this role, programming systems are needed that can automate the task of mapping the applications to the tens-tohundreds of cores on current and future many-core processors, while simultaneously guaranteeing the real-time throughput constraints. This paper presents a block-parallel program description for embedded real-time media applications and automatic transformations including buffering and parallelization to ensure the program meets the throughput requirements. These transformations are enabled by starting with a high-level, yet intuitive, application description. The description builds on traditional stream programming structures by adding simple control and serialization constructs to enable a greater variety of applications. The result is an application description that provides a balance of flexibility and power to the programmer, while exposing the application structure to the compiler at a high enough level to enable useful transformations without heroic analysis.

A DSEL for high throughput and low latency software‐defined radio on multicore CPUs

“…Many languages dedicated to streaming applications have been introduced. [41][42][43][44][45][46][47][48] These languages are often variants of the cyclo-static dataflow model and propose automatic parallelization techniques such as pipelining and forks/joins.…”

DSL Stream Programming on Multicore Architectures

“…Block Parallel Block Parallel [6] also targets signal applications. The author argues that the multidimensional formulations proposed, for example, by Array-OL are difficult to optimize since each new dimension increases the number of possible data traversals.…”

“…StreamIt uses fusion, fission and reordering transformations to optimize the throughput and Brook leverages the optimizations offered by affine partitioning [23]. Array-OL or Block Parallel on the other hand propose a high-level description of data dependences [16] [6]. Nevertheless the high-level description comes at a price: optimizations in theses languages are harder to implement, in particular optimization regarding the routing of data through the application.…”

“…To accomplish this objective, two requirements must be satisfied: first, the representation must be simple enough to enable a wellunderstood set of optimizations ; second, the representation should capture all the possible static data reorganizations (we cannot optimize what we cannot model). A high-level multidimensional representation like Array-OL and Block Parallel does not satisfy the first requirement, since the optimization complexity grows with the number of dimensions [6]. A simple graph language like StreamIt is much easier to optimize.…”

Triple-C: Resource-usage prediction for semi-automatic parallelization of groups of dynamic image-processing tasks

“…Within performance computing, several techniques are reported in the literature for performance prediction of parallel applications [3][4] [5]. Most of them are based on blockparallel or dataflow-like approaches.…”