HGT: An Open-Source Framework for Simulating Parallel Real-Time Tasks
Posted: Wed Jun 07, 2017
Title: HGT: An Open-Source Framework for Simulating Parallel Real-Time Tasks
Authors:
Ignacio Sañudo, Paolo Burgio and Marko Bertogna (HiPeRT Lab, University of Modena, Italy)
Abstract: With the increasing complexity of multi/many-core architectures, academy-industry research collaborations on this topic are today intensifying. In this sense, the characterization of industrial applications is still a big challenge, due to industry’s reluctance to share application code details. Luckily, this trend is partly changing, and today several industrial partners can disclose high-level details of their software suites, e.g., the timing constraints or even the memory footprint and access patterns, to perform a better application characterization. However, it is still extremely cumbersome to reproduce the behavior of real applications in their actual environment, due to IPR on source code.
In this paper, we introduce the HiPeRT Generator Tool that helps researchers creating synthetic yet realistic test cases, using a variety of techniques based on the model-driven development approach. The result is an open-source framework, that generates ready to use ANSI C code from high-level behavioral description of an application represented with a Directed acyclic graph (DAG).
Attached paper:
Authors:
Ignacio Sañudo, Paolo Burgio and Marko Bertogna (HiPeRT Lab, University of Modena, Italy)
Abstract: With the increasing complexity of multi/many-core architectures, academy-industry research collaborations on this topic are today intensifying. In this sense, the characterization of industrial applications is still a big challenge, due to industry’s reluctance to share application code details. Luckily, this trend is partly changing, and today several industrial partners can disclose high-level details of their software suites, e.g., the timing constraints or even the memory footprint and access patterns, to perform a better application characterization. However, it is still extremely cumbersome to reproduce the behavior of real applications in their actual environment, due to IPR on source code.
In this paper, we introduce the HiPeRT Generator Tool that helps researchers creating synthetic yet realistic test cases, using a variety of techniques based on the model-driven development approach. The result is an open-source framework, that generates ready to use ANSI C code from high-level behavioral description of an application represented with a Directed acyclic graph (DAG).
Attached paper: