Software portability

About: Software portability is a research topic. Over the lifetime, 8987 publications have been published within this topic receiving 164922 citations. The topic is also known as: portability.

Investigating applications portability with the Uintah DAG-based runtime system on PetaScale supercomputers

Abstract: Present trends in high performance computing present formidable challenges for applications code using multicore nodes possibly with accelerators and/or co-processors and reduced memory while still attaining scalability. Software frameworks that execute machine-independent applications code using a runtime system that shields users from architectural complexities offer a possible solution. The Uintah framework for example, solves a broad class of large-scale problems on structured adaptive grids using fluid-flow solvers coupled with particle-based solids methods. Uintah executes directed acyclic graphs of computational tasks with a scalable asynchronous and dynamic runtime system for CPU cores and/or accelerators/co-processors on a node. Uintah's clear separation between application and runtime code has led to scalability increases of 1000x without significant changes to application code. This methodology is tested on three leading Top500 machines; OLCF Titan, TACC Stampede and ALCF Mira using three diverse and challenging applications problems. This investigation of scalability with regard to the different processors and communications performance leads to the overall conclusion that the adaptive DAG-based approach provides a very powerful abstraction for solving challenging multi-scale multi-physics engineering problems on some of the largest and most powerful computers available today.

Incremental partial evaluation: the key to high performance, modularity and portability in operating systems

Abstract: The most prominent recent trend in operating system (OS) design has been the move towards micro-kernel based OS's [2, 8, 9, 12, 14]. Micro-kernel based OS's allow highlevel OS code to be structured as a collection of modules above a minimal kernel. Despite the many advantages of this modular approach, the performance overhead associated with existing modular implementations has proven to be a major liability in the commercial acceptance of these systems [3]. Finding a solution to the con ict between performance and modularity remains a critical research issue of practical importance. Contrary to popular belief, the micro-kernel approach to OS structuring does not lead to major improvements in portability. The modularity exhibited by most micro-kernel designs is coarse grained and orthogonal to the issue of localizing machine-dependent code. Some micro-kernel OS's do de ne various machine-independent interfaces within their micro-kernels, but these are unrelated to the system structuring mechanisms used in the higher layers of their OS code, and make a coarse-grained distinction between machinedependent and portable code [1, 11]. This coarse-grained approach limits portability by limiting the amount of code that can be reused. The state of the art in OS design can be summarized as follows. The vast majority of OS's in active use are monolithic, having traded portability and modularity for performance. There has been considerable research investment in micro-kernel OS's which o er some coarse-grained modularity at the expense of performance, and hence have not received high acceptance commercially. Neither monolithic kernels nor micro-kernels have made major progress towards true portability since they do not possess ne-grain modularity.

An open distributed measurement system based on an abstract client-server architecture

Abstract: This paper describes in detail a Java-based, client-server architecture specifically designed to allow a flexible management of remote instruments. The main attributes of the proposed solution are portability and extensibility. The former feature is assured by the employment of the TCP/IP protocol suite and by the Java language properties. The latter is due to the high level of abstraction of the system implementation. This approach addresses a wide range of possible applications with high code reusability. In fact, the proposed architecture permits to drive many kinds of different devices and can be easily upgraded simply by adding a limited amount of code on the server computer whenever a new instrument is connected to the system.

GPU-accelerated molecular dynamics: State-of-art software performance and porting from Nvidia CUDA to AMD HIP:

Abstract: Classical molecular dynamics (MD) calculations represent a significant part of the utilization time of high-performance computing systems. As usual, the efficiency of such calculations is based on ...