Abstract
Pipelined computer architecture has received considerable attention since the 1960s when the need for faster and more cost-effective systems became critical. The merit of pipelining is that it can help to match the speeds of various subsystems without duplicating the cost of the entire system involved. As technology evolves, faster and cheaper LSI circuits become available, and the future of pipelining, either in a simple or complex form, becomes more promising.
This paper reviews the many theoretical considerations and problems behind pipelining, surveying and comparing various representative pipeline machines that operate in either sequential or vector pipeline mode, the practical solutions adopted, and the tradeoffs involved. The performance of a simple pipe, the physical speed limitation, and the control structures for penalty-incurring events are analyzed separately. The problems faced by the system designers are tackled, including buffering, busing structur, branching, and interrupt handling. Aspects of sequential and vector processing are studied. Fundamental advantages of vector processing are unveiled, and additional requirements (costs) are discussed to establish a criterion for the tradeoff between sequential and vector pipeline processing. Finally, two recent machines (the CRAY-1 and the Amdahl 470 V/6 systems) are presented to demonstrate how complex pipeline techniques can be used and how simple but advantageous pipeline concepts can be exploited.
- 1 CHEN, T. C. "Unconventional superspeed computer systems," in AFIPS 1971 Spring Jr. Computer Conf., AFIPS Press, Montvale, N.J., 1971, pp. 365-371.Google Scholar
- 2 MCINTYRE, D. "An introduction to the ILLIAC IV computer," Datamation (April 1970), 60-67.Google Scholar
- 3 EVENSEN, A. J.; AND TROY, J.L. "Introduction to the architecture of a 288-element PEPE," in Proc. 1973 Sagamore Conf. on Parallel Processing, Springer-Verlag, N.Y. 1973, pp. 162-169.Google Scholar
- 4 RUDOLPH, J. A. "A production implementation of an associattve array processor-- STARAN," in AFIPS 197~ Fall Jt. Computer Conf., AFIPS Press, Montvale, N.J., 1972, pp. 229-241.Google Scholar
- 5 MARVEL, O. E. "HAPPE--Honeywell associative parallel processing ensemble," in Proc. Symp. on Computer Architecture, Univ. of Florida, 1973, pp. 261-268. Google Scholar
- 6 STANGA, D.C. "Univac 110 multiprocessor system," in AFIPS 1967 Spring Jr. Computer Conf., Thompson Book Co., Washington, D.C., 1967, pp. 67-74.Google Scholar
- 7 CHEN, T. C. "Parallelism, pipelining and computer efficiency," Computer Design (Jan. 1971), 69-74.Google Scholar
- 8 RAMAMOORTHY, C. V.; AND LI, H.F. "Efficiency in generalized pipeline networks," in AFIPS 1974 National Computer Conf., AFIPS Press, Montvale, N.J., 1974, pp. 625-635.Google Scholar
- 9 ANDERSON, D. W.; SPARACIO, F. J.; AND TOMASULO, R.M. "IBM System 360 Model 91, machine philosophy and instruction handling," IBM J. R. and D. (Jan. 1967), 8-24.Google ScholarDigital Library
- 10 WATSON, W. J. "The TIASC--a highly modular and flexible super computer arehipp. 221-228. Google Scholar
- 11 HINTz, R. G.; ASD TATE, D.P. Control Data STAR-100 processor design," COMP- CON 7~,, IEEE, N.Y,,, 1972.Google Scholar
- 12 TOMASULO, R.M. An efficient algorimm for exploiting multiple arithmetic units," IBM J. R. and D. (Jan. 1967), 25-33.Google Scholar
- 13 TExAs INSTRUMENTS INC., A description ofA the advanced scientific computer system, ustin, Texas, April 1973.Google Scholar
- 14 CONTROL DA~X CORP., Control Data STAR- 100 computer hardware reference manual, 1974.Google Scholar
- 15 T~xAs INSTRUMENTS INC., The ASC system-central processor," Austin, Texas, Dec. 1971.Google Scholar
- 16 D.~.wvsoN, E. The design and control of pipeline function generator, Stanford Report Stanford University, 1972.Google Scholar
- 17 DAVIDSON, E. S.; SHAR, L. E.; THOMAS, A. T.; ANy PATEL, J.H. "Effective control for pipelined computers," COMPCON 75 IEEE, N.Y., 1975, pp. 181-184.Google Scholar
- 18 RAMAMOOR~HY, C. V.; AND LI, H.F. "Sequencing control in multifunctional pipelined systems," in Proc. 1975 Sagamore Conf. on Parallel Processing, Springer- Verlag, N.Y., 1975.Google Scholar
- 19 Kism, T.; Am) RUBY, T. "STAR TREK," in COMPCON 75, IEEE, N.Y., 1975, pp. 185-188.Google Scholar
- 20 WALI, ACV., C. S. "A suggestion for a fast multiplier," IEEE Trans. Computers EC-13 (1964), 14-17.Google ScholarCross Ref
- 21 A~DERSOS, S. F.; EARL~, J. G.; GOLD- SCHMIDT, R. E.; ASp PoweRs, I). M. "The IBM System/360 Model 91: floating-point execution unit," IBM J. R. and D. (Jan. 1967), 34-53.Google Scholar
- 22 CH.EN, T.C. "Overlap and pipeline proc.-Google Scholar
- 23 RAMAMOOBTHY, C. V.; AND KIM, K. H "Pipelining--the generalized concept an( sequencing strategies," in AFIPS 197, National Computer Conf., AFIPS Press Montvale, N.J., 1974, pp. 289-297.Google Scholar
- 24 K~M, K. H. "Optimizing architecture i~ parallel processing," Memorandum RLE M482, Electronics Research Lab., Univ. o Calif. Berkeley, Nov. 1974.Google Scholar
- 25 REDDI, S. S. "Sequencing strategies i~ pipeline computer systems," PhD Thesis Univ. of Texas, Austin, August 1972. Google Scholar
- 26 CRAY R~.S~aCH CORP., "Description o CR~Y-I", Minn., 1975.Google Scholar
- 27 COTTEN, L.W. "Circuit implementation o high speed pipeline systems," in AFIP2 1965, Fall Jr. Computer Conf., Thompso~ Book Co., Washington, D.C., 1967, pp 489-504.Google Scholar
- 28 H~LL~S T. G. ; ~D FLVS~, M. J. "Pipe lining of arithmetic functions," IEEE Trans Computers, C-21 (August 1972), 880-886Google Scholar
- 29 LARSEN, A. G.;AND DAVIDSON E.S. "Cost effective design of special purpose proces sors: a fast Fourier transform case study,' in 11th Allerton Conf., Univ. Illinois, 1973 pp. 547-554.Google Scholar
- 30 LI, H. F. "A structural study of paralle pipelined systems," PhD Dissertation Univ. of Cah'f., Berkeley, 1975.Google Scholar
- 31 AMDAHL CORP. Amdahl 470 V/6 machin~ reference, Sunnyvale, Calif., 1975.Google Scholar
- 32 B~A~, R.J. "Packaging for a super corn purer," Amdahl Corp., 1975.Google Scholar
- 33 MARTIN, J. T.; ZWAKENBERG, ~. G.; ANI SOLBECK, S. V. "LRLTRAN language use( with the CHAT and STAR compilers,' LTSS Chapter 207, Lawrence Livermor~ Lab., Lawrence, Calif.Google Scholar
- 34 W~D~, I). "FORTRAN for the Texa~, Instruments ASC System," SIGPLAN 1( (1975), 119-132. Google Scholar
- 35 BASILI, V. R.; AND KNIGHT, J. C. "A language design for vector machines,' SIGPLAN 10 (1973), 39-53. Google Scholar
Index Terms
- Pipeline Architecture
Recommendations
Split-Path Enhanced Pipeline Scheduling
Software pipelining increases the loop execution throughput by overlapping the execution of successive iterations in a pipelined fashion. For loops with control flows, however, software pipelining is not straightforward because we need to consider the ...
On-the-fly pipeline parallelism
SPAA '13: Proceedings of the twenty-fifth annual ACM symposium on Parallelism in algorithms and architecturesPipeline parallelism organizes a parallel program as a linear sequence of s stages. Each stage processes elements of a data stream, passing each processed data element to the next stage, and then taking on a new element before the subsequent stages have ...
Comments