Path: utzoo!utgpu!water!watmath!clyde!rutgers!mcnc!rti!sas!jcz From: jcz@sas.UUCP (John Carl Zeigler) Newsgroups: comp.arch Subject: Re: Single tasking the wave of the future? Summary: Multi - tasking . . . Message-ID: <310@sas.UUCP> Date: 10 Dec 87 03:39:30 GMT References: <201@PT.CS.CMU.EDU> <388@sdcjove.CAM.UNISYS.COM> <988@edge.UUCP> <151@sdeggo.UUCP> Organization: SAS Institute Inc.,Cary NC,25712 Lines: 38 It might be informative to remember why multi-tasking was invented in the first place. Back in the 60's Mr Warbucks wandered down to the Data Processing Department to see what kind of trouble that particular budgetary sink-hole was getting into. He wandered into the Computer Room and spied a Computer Programmer with a couple cases of cards turning dials and flipling switches. "Hi, what are you doing," he asked. "Oh, Hello Mr. Warbucks. I am testing our new Social Security Accounting and Collating System. Here, watch this." With that he flipped one of the switches and a machine began spewing cards into a box. After a few hundred cards had dropped, one popped out into a slot. "See, here is your card," explained the Programmer. "That's nice, but what does that light mean?" he said, pointing to a warm red light on the front panel. "That's the WAIT light, It goes on whenever the CPU is idle." "But, it NEVER went off!" "Right! I have made this program so fast that the light doesn't have time to stop glowing enough for you to see it." "You mean for most of the time the machine wasn't doing anything!" "Well, that is one way to look at it." "Why couldn't it have been doing something else???? I am not going pay a hundred times your salary a year just to have the damn thing WAITing!!!! Get that fixed!" Seriously. I do not mind having my computer doing something else while I am thinking, or while the disk drive is thinking, either. Most applications are NOT CPU bound. Most are I/O bound. The overhead of task switching will always be paid for by the increased utility of the machine. (not forgetting that L = yW, (y is upside down)) Until we get a better architectural paradigm than the current 'Von Neuman Bottle-Neck', that is. Any discussion on that?? -- --jcz John Carl Zeigler SAS Institute Inc. Cary, NC 27511 (919) 467-8000 ...!mcnc!rti!sas!jcz