Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watmath!clyde!rutgers!mcnc!rti!xyzzy!throopw From: throopw@xyzzy.UUCP Newsgroups: comp.arch Subject: Re: Single tasking the wave of the future? Message-ID: <430@xyzzy.UUCP> Date: Mon, 7-Dec-87 12:40:16 EST Article-I.D.: xyzzy.430 Posted: Mon Dec 7 12:40:16 1987 Date-Received: Sat, 12-Dec-87 10:38:28 EST References: <201@PT.CS.CMU.EDU> <388@sdcjove.CAM.UNISYS.COM> <988@edge.UUCP> <147@sdeggo.UUCP> <1227@sugar.UUCP> Organization: Data General, RTP NC. Lines: 27 > peter@sugar.UUCP (Peter da Silva) > Until you have something incredible like 64000 processors in your > box, you're going to find yourself running out of the suckers. Note that argument that "we don't need to multiplex a single processor because we can simply throw more processors at the problem" is the same argument as "we don't need to have demand paging because we can simply buy more and more memory". I think that both are incorrect arguments, for exactly the reasons that Peter points out: > What happens when you want to start up that 5th or 9th program and you've > only got 4 or 8 CPUs? Or, what happens when you want to keep N processes handy at a convenient point in mid-execution, when they total to M pages of address space and you've only got (M-1) pages of physical memory? I think virtual processors are just as important as virtual memory. And I think virtual memory is *very* important. The fact that many PCs throw these things away is, I think, a Bad Thing. -- "Solve THIS!" "'Mighty Mouse is a wimp.' But what does it mean?" "It means you've made me MAD!" --- Mighty Mouse to alien brain. -- Wayne Throop!mcnc!rti!xyzzy!throopw