Lapedo: Hybrid Skeletons for Programming Heterogeneous Multicore Machines in Erlang
نویسندگان
چکیده
We describe Lapedo, a novel library of hybrid parallel skeletons for programming heterogeneous multi-core/many-core CPU/GPU systems in Erlang. Lapedo’s hybrid skeletons comprise a mixture of CPU and GPU components, allowing skeletons to be flexibly and dynamically mapped to available resources. We also describe a model for deriving near-optimal division of work between CPUs and GPUs, ensuring load balancing between resources. Finally, we evaluate the effectiveness of Lapedo on three realistic use cases from different domains, demonstrating significant speedups compared to executing the same application on only CPU cores or a GPU.
منابع مشابه
Toward optimised skeletons for heterogeneous parallel architecture with performance cost model
High performance architectures are increasingly heterogeneous with shared and distributed memory components, and accelerators like GPUs. Programming such architectures is complicated and performance portability is a major issue as the architectures evolve. This thesis explores the potential for algorithmic skeletons integrating a dynamically parametrised static cost model, to deliver portable p...
متن کاملRELEASE: A High-level Paradigm for Reliable Large-scale Server Software
Erlang is a functional language with a much-emulated model for building reliable distributed systems. This paper outlines the RELEASE project, and describes the progress in the first six months. The project aim is to scale the Erlang’s radical concurrency-oriented programming paradigm to build reliable general-purpose software, such as server-based systems, on massively parallel machines. Curre...
متن کاملThe Design of Scalable Distributed Erlang
The multicore revolution means that the number of cores in commodity machines is growing exponentially. Many expect 100,000 core clouds/platforms to become commonplace, and the best predictions are that core failures on such an architecture will become relatively common, perhaps one hour mean time between core failures. The RELEASE project aims to scale Erlang to build reliable general-purpose ...
متن کاملMulticore OSes: Looking Forward from 1991, er, 2011
Upcoming multicore processors, with hundreds of cores or more in a single chip, require a degree of parallel scalability that is not currently available in today’s system software. Based on prior experience in the supercomputing sector, the likely trend for multicore processors is away from shared memory and toward sharednothing architectures based on message passing. In light of this, the ligh...
متن کاملVacation model for Markov machine repair problem with two heterogeneous unreliable servers and threshold recovery
Markov model of multi-component machining system comprising two unreliable heterogeneous servers and mixed type of standby support has been studied. The repair job of broken down machines is done on the basis of bi-level threshold policy for the activation of the servers. The server returns back to render repair job when the pre-specified workload of failed machines is build up. The first (seco...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2015