Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!amdcad!ames!rutgers!iuvax!pur-ee!uiucdcs!uxc.cso.uiuc.edu!ccvaxa!aglew From: aglew@ccvaxa.UUCP Newsgroups: comp.arch Subject: Re: Horizontal Pipelining -- a pair Message-ID: <28200073@ccvaxa> Date: Fri, 27-Nov-87 16:11:00 EST Article-I.D.: ccvaxa.28200073 Posted: Fri Nov 27 16:11:00 1987 Date-Received: Mon, 30-Nov-87 02:45:34 EST References: <391@sdcjove.CAM.UNISYS.COM> Lines: 35 Nf-ID: #R:sdcjove.CAM.UNISYS.COM:391:ccvaxa:28200073:000:1671 Nf-From: ccvaxa.UUCP!aglew Nov 27 15:11:00 1987 ..> John Mashey talking about barrel architectures Just making sure - barrel architectures are systems that multiplex each pipeline stage between different threads of execution? I assume this in my response. >Assume we're using split I & D caches. Assume that the cache line >is N words long, filled 1 word/cycle after a latency of L cycles. >One would expect that efficient cache designs have L <= N. I expect this is quite basic, but how do you show this for barrel architectures (the demonstrations I've seen have required regular architectures). >When filling an I-cache miss, you can do L more barrel slots, >then you must stall for N slots (or equivalent), because it doesn't >make sense to have the I-cache run faster than the chip (if it did, >you would run the chip faster). Why not have the I-cache run faster than the chip? I-caches are more regular structures than the cpu, and are probably that much easier to make run faster. Also, why stall for N slots? There are several schemes to deliver data from a partially filled cache line as soon as it is available. Finally, why not continue on a completely separate thread while the cache is filling for the thread that caused the cache miss? Barrel need not imply round robin. I think that I've missed part of John's point here. Andy "Krazy" Glew. Gould CSD-Urbana. 1101 E. University, Urbana, IL 61801 aglew@mycroft.gould.com ihnp4!uiucdcs!ccvaxa!aglew aglew@gswd-vms.arpa My opinions are my own, and are not the opinions of my employer, or any other organisation. I indicate my company only so that the reader may account for any possible bias I may have towards our products.