lcw@halley.UUCP (Larry Wolfe) (03/14/90)
In article <36880@mips.mips.COM> mash@mips.COM (John Mashey) writes: >One more time: calling load-interlocks scoreboards is just marketing >silliness, confusing, and contradictory to terminology long-used >in computer architecture. >Jonathan Smith recently posted a reference to one of the original >360/91 papers that describe REAL scoreboards, and maybe somebody >will post a reference to one of the descriptions of the CDC machines >that have them. > see: Design of a Computer, The Control Data 6600 J.E. Thornton Scott, Foresman and Company Copyright 1970 Library of Congress Catalog Number 74-96462 pp 125 - 134 deal with the scoreboard. Thornton is identified as VP of the Advanced Design Laboratory, CDC "In spite of the large number of computing systems which have been de- signed and are in use today there is no clear-cut optimum approach to a gen eral purpose computing system. Rather, it would seem, we are just begin- ning to explore the really basic variations from the one address sequential machines that launched the digital computing industry. . . . System design then began to diverge into parallel structures. This book describes one of the early machines attempting to explore parallelism in electrical structure without abandoning the serial structure of the com- puter programs. Yet to be explored are parallel machines with wholly new programming philosophies in which serial execution of a single program is abandoned. . . . ...from the foreword by Seymore Cray ...much deleted from JM's posting > >-- >-john mashey DISCLAIMER: <generic disclaimer, I speak for me only, etc> >UUCP: {ames,decwrl,prls,pyramid}!mips!mash OR mash@mips.com >DDD: 408-991-0253 or 408-720-1700, x253 >USPS: MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086 Larry Wolfe (speaking soley for himself) ...!cs.utexas.edu!timex!lcw 512-244-8027
aglew@oberon.csg.uiuc.edu (Andy Glew) (03/15/90)
In article <36880@mips.mips.COM> mash@mips.COM (John Mashey) writes: >One more time: calling load-interlocks scoreboards is just marketing >silliness, confusing, and contradictory to terminology long-used >in computer architecture. >Jonathan Smith recently posted a reference to one of the original >360/91 papers that describe REAL scoreboards, and maybe somebody >will post a reference to one of the descriptions of the CDC machines >that have them. Umm, usually the term "scoreboard" is reserved for Thornton's work at CDC, with the less aggressive approach used by Cray, etc. Tomasulo's work on the 360/91 is usually considered to go beyond scoreboarding. IE. scoreboarding is not the ultimate in aggressive scheduling. At least not in the terminology that I have seen applied in this field for 8 years. (Personally, I tend to use the term "Tomasulo" to describe the most aggressive approaches, that let you skip O->I, I->O, and O-> dependencies, as well as functional unit busy conditions, etc. Many researchers, however, harp on the limitations in Tomasulo's implementation (the Common Data Bus) and invent their own terms for the ultimate...) -- Andy Glew, aglew@uiuc.edu