Font Size: a A A

Transplanting And Optimizing Dalvik For Android Based CK-CPU

Posted on:2012-05-04Degree:MasterType:Thesis
Country:ChinaCandidate:J Y JiangFull Text:PDF
GTID:2178330332483350Subject:Circuits and Systems
Abstract/Summary:PDF Full Text Request
More and More people jump into the development of Android along with Android being flourishing. It's expected that Android will keep progressing at a high speed in the next few years. On the other hand, CK-CPU(CK-Core)——a series of 32bits, high perfromace, low power, CPU IPs applied in the field of Embedded System and SOC, developed by Hangzhou C-SKY Micro-system Ltd. and Zhejiang University, is growing up fleetly as a excellent competitor home during the last few years. This paper researches and analyzes Dalvik VM deeply, under the background of transplanting Android to architecture of CSKY(the instructions architecture of CK-CPU), for the objective of driving the developing of CK-CPU and Android. First, this paper expatiates the theory and differences between JVM and Dalvik VM. The advantages of Dalvik VM compared to JVM is emphasized, which show the reasons that Dalvik VM is more suitable for protable equipments. Register architecture and DEX file format are just the two measures that Dalvik VM optimizes for portable equipments. The main transplanting work of this paper are implementing a assemble interpretor——CSKY Mterp(include a set of bytecode implemented for float computing based on CK-CPU with a hardware float coprocessor) and JNI call bridge.We also completing the work supporting Dalvik VM running on CK-CPU, including supporting Linux 2.6 for Android, and transplanting Bionic that a exclusive C library of A ndroid and so on. We then test CSKY Dalvik VM we have transplanted. The results show that CSKY Mterp improves 200%-300% compared to the portable interpretor and fast interpretor of Dalvik source code. And the CSKY Mterp optimized for float computing improves about 300% for the performance of float computing. At the last of this paper, we bring forward several optimizing measures for Dalvik based on execute engine, memory managament and JNI, showing the work that should to do in the future. And optimizing execute engine is the core of optimizing CSKY Dalvik VM.
Keywords/Search Tags:Android, Dalvik, JVM, CK-CPU, Interpretor, Register Architecture, DEX
PDF Full Text Request
Related items