Techniques in Optimizing Memory-Intensive Applications
Date Issued
2009
Date
2009
Author(s)
Tseng, Shian-Shuen
Abstract
Memory management is the act of governing the memory sub-system inomputers. Generally, memory management software involves the ways tollocate portions of memory requests by programs and recycle the memoryhen it is no longer needed. Due to the Memory Wall problem, memoryanagement has been an important research topic to boost the performancef computer systems. In this thesis, the performance of different memoryanagement strategies at different levels is evaluated. Several practical methodsor memory management are provided to boost the system performance.irst, the experiments were done to evaluate the performance of memoryopy operations with different configurations. From empirical experiments,e found that with proper configuration, the performance of memory copyperation can achieve about 8% speedup as compared with the default configuration.n the other hand, if the configuration is not chosen judiciously,he performance delivered by the best configuration, which is not obviouso find, can be 3.94 times faster than the worst configuration. Second, wevaluated the performance of Garbage Collection mechanism implementedn Android’s Dalvik virtual machine. The speed of garbage collecting operations improved up to 67% compared to original design. Finally, the memoryllocation mechanism in Dalvik is studied as well. The speed of memory allocation operation is accelerated by up to 43%.
Subjects
Memory Management
Memory Copy
Configurations
Garbage Collection
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-98-R96922102-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):3d2bbf394c3788876c91fb6992bea5a3