PIPELINE INTERLEAVED PROGRAMMABLE DSPS - ARCHITECTURE

被引:25
作者
LEE, EA
MESSERSCHMITT, DG
机构
[1] Univ of California, Dep of, Electrical Engineering & Comput, Sci, Berke, Univ of California, Dep of Electrical Engineering & Comput Sci, Berke
来源
IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING | 1987年 / 35卷 / 09期
关键词
COMPUTERS; MICROCOMPUTER - SIGNAL PROCESSING - Digital Techniques;
D O I
10.1109/TASSP.1987.1165274
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
The authors apply an old but rarely used architectural approach to the design of single-chip signal processors so that the potential benefits of extensive pipelining can be fully realized. Instead of a single program, multiple programs are interleaved in a pipeline. Each program experiences none of the difficulties of pipelining and consequently are easier to write and optimize, regardless of the amount of pipelining used. The main advantage is that much more pipelining can be used without disturbing the programming. A specific experimental architecture is outlined.
引用
收藏
页码:1320 / 1333
页数:14
相关论文
共 39 条
[1]  
ACKERMAN WB, 1982, IEEE COMPUT, V15
[2]  
ALLEN J, 1975, P IEEE, V63
[3]  
[Anonymous], DIGITAL SIGNAL PROCE
[4]   OPTIMAL CHOICE OF INTERMEDIATE LATCHING TO MAXIMIZE THROUGHPUT IN VLSI CIRCUITS [J].
CAPPELLO, PR ;
LAPAUGH, A ;
STEIGLITZ, K .
IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING, 1984, 32 (01) :28-33
[5]  
CAPPELLO PR, 1983, IEEE T ACOUST SPEECH, V31
[6]  
CHAPMAN RC, 1981, BELL SYST TECH J 2, V60
[7]  
COHN LA, 1983, THESIS COLUMBIA U NE
[8]  
DAVIDSON ES, 1973, PIPELINING PARALLELI
[9]  
Flynn M. J., 1970, Parallel processor systems, technologies and applications, P251
[10]  
FLYNN MJ, 1972, IEEE T COMPUT, V21