Architecture-aware automatic computation offload for native applications

Abstract

Although mobile devices have been evolved enough to support complex mobile programs, performance of the mobile devices is lagging behind performance of servers. To bridge the performance gap, computation offloading allows a mobile device to remotely execute heavy tasks at servers. However, due to architectural differences between mobile devices and servers, most existing computation offloading systems rely on virtual machines, so they cannot offload native applications. Some offloading systems can offload native mobile applications, but their applicability is limited to wellanalyzable simple applications. This work presents automatic cross-architecture computation offloading for generalpurpose native applications with a prototype framework that is called Native Offloader. At compile-time, Native Offloader automatically finds heavy tasks without any annotation, and generates offloading-enabled native binaries with memory unification for a mobile device and a server. At run-time, Native Offloader efficiently supports seamless migration between the mobile device and the server with a unified virtual address space and communication optimization. Native Offloader automatically offloads 17 native C applications from SPEC CPU2000 and CPU2006 benchmark suites without a virtual machine, and achieves a geomean program speedup of 6.42× and battery saving of 82.0%.

SIMDOM: A framework for SIMD instruction translation and offloading in heterogeneous mobile architectures

“…Lee et al proposed an architecture‐aware native code‐offloading framework for mobile devices. The framework does not rely on code annotation and virtualization for offloading.…”

“…The RTT of existing high speed is still high for real-time migration of VM workloads. [22][23][24] Lee et al 25 proposed an architecture-aware native code-offloading framework for mobile devices. The framework does not rely on code annotation and virtualization for offloading.…”

Enabling Cross-ISA Offloading for COTS Binaries

“…MobiSys ' power constraints) and limited battery life compared to their desktop and server counterparts. A recent study [29] shows that a mobile device is more than 5 times slower than a desktop computer when used to play the same chess game. Therefore, users have to put up with longer waiting time while playing the same game on mobile devices.…”

“…Previous research work [22,29,41,31,38,18,43,20,30,42,27,19,35,26] has shown that work offloading is an effective way to improve performance and to prolong battery life of mobile devices. In a work offloading system, the computation-intensive tasks on a mobile device, i.e., the client, are sent to a more powerful server and executed remotely.…”

“…With more native binaries integrated into such VM systems, their portability across ISAs is significantly hampered. A recent study [29] shows that one third of the top 20 open source Android applications contain more than 50% native code and spend more than 20% of the execution time in native mode.…”