Title :
ePUMA: A novel embedded parallel DSP platform for predictable computing
Author :
Wang, Jian ; Sohl, Joar ; Kraigher, Olof ; Liu, Dake
Author_Institution :
Dept. of Electr. Eng., Linkoping Univ., Linkoping, Sweden
Abstract :
In this paper, a novel parallel DSP platform based on master-multi-SIMD architecture is introduced. The platform is named ePUMA [1]. The essential technology is to use separated data access kernels and algorithm kernels to minimize the communication overhead of parallel processing by running the two types of kernels in parallel. ePUMA platform is optimized for predictable computing. The memory subsystem design that relies on regular and predictable memory accesses can dramatically improve the performance according to benchmarking results. As a scalable parallel platform, the chip area is estimated for different number of co-processors. The aim of ePUMA parallel platform is to achieve low power high performance embedded parallel computing with low silicon cost for communications and similar signal processing applications.
Keywords :
coprocessors; embedded systems; information retrieval; parallel architectures; parallel memories; communication overhead; coprocessor; data access kernel; data algorithm kernel; ePUMA; embedded parallel DSP platform; embedded parallel computing; master-multiSIMD architecture; memory subsystem design; parallel processing; predictable computing; predictable memory access; scalable parallel platform; Computer architecture; Concurrent computing; Coprocessors; Costs; Digital signal processing; Embedded computing; Kernel; Parallel processing; Signal processing algorithms; Silicon; SIMD architecture; chip multi-processor; conflict-free memory access; data permutation; multi-bank memory;
Conference_Titel :
Education Technology and Computer (ICETC), 2010 2nd International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-6367-1
DOI :
10.1109/ICETC.2010.5529952
