ted@imsvax.UUCP (Ted Holden) (01/03/87)
There are several points to consider concerning the mainframe/micro question as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. A Novell server can fully use big disks such as the 300 meg Cores. Such a server can use several meg of ram as a cache area and adequately service 30 or so users. The neat thing is that the entire compute load has been put on the individual PCs; the server is only being required to fetch and store info to and from disk. As servers get more powerful, using 386 and future chips, and storage medea gets better, presumably using laser technologies, it is easy to anticipate such systems replacing mainframes for most if not all database work. The little ISI glass disks are available now for something like $2500, 256 meg disks going for $100, 1 ms track to track seek time. That's fast! I've seen 750 and 780 VAXs destroyed by fewer than 30 people doing database work (or attempting same) at the same time. The PCs are faster and better. The PC revolution, like our free-market system as opposed to communism, takes human nature into account, at least as it applies to bosses and managers. I've never yet seen one of these who, upon hiring three new people to work on a multi-user computer system, didn't simply add three new terminals to the system. With PC networks, they are forced to add three new computers to the system. And the beauty is that they do so with smiles on their faces, since PCs have gotten to the point of being cheaper than most mainframe terminals anyhow. Ted Holden IMS
tj@utcs.UUCP (01/03/87)
I think a lot of you might be missing some of the VERY important driving forces that are pushing people to micros. I have talked with a lot of micro users and I have seen our mainframes grow idle. People percieve micros as a fixed cost for computing. They buy it and no matter how much they use it it doesn't cost them more. Easy for them to budget for. Micro's are totally under their control. The software versions don't change unless THEY change it. On mainframes RARELY can a user bring in their program that used to work a year ago and have it still work. A version of software has changed, hardware has changed, files have been moved. Sure they may have written their program badly to make it device or version dependant, but that is not the point. Last year it worked. Micro users find text processing easier on micro's than on mainframes. So do I! There may be amazing programs on mainframes with amazing capabilities, but the speed of the video display on a micro and the almost WYSIWYG is quite nice. We are seeing a "returning" to mainframes by some micro users that are realizing that micro's can't do it all or at least not in a reasonable amount of time, but that above reasons are very good ones to avoid mainframes! (Save the flames, I am both a mainframe and a micro user. I use which ever is best suited for the purpose. Of course I work for the computer center so I don't have to pay which makes the decision of which to use much easier!) tj
dragheb@isis.UUCP (Darius "OPRDRT" Ragheb) (01/03/87)
Summary:lets please stop comparing Datsuns and Lamborghinis In article <657@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: > >There are several points to consider concerning the mainframe/micro question >as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. >A Novell server can fully use big disks such as the 300 meg Cores. Such a And IBM has a particular disk drive for its mainframes that stores 4 GigaBytes. (it spins around a horizontal axis rather than a vertical one) How many AT's can access one or more of these? >on the individual PCs; the server is only being required to fetch and store >info to and from disk. As servers get more powerful, using 386 and future >chips, and storage medea gets better, presumably using laser technologies, >it is easy to anticipate such systems replacing mainframes for most if not >all database work. The little ISI glass disks are available now for It is even easier to imagine and anticipate the growth of larger computers. >I've seen 750 and 780 VAXs destroyed by fewer than 30 people doing database These machines that you mention are not mainframes. I have seen an Amdahl machine with 200 users run faster than any PC or AT (and i think it was able to support an additional 300, i am not sure). >work (or attempting same) at the same time. The PCs are faster and better. Faster and better. very interesting statement. My car is faster and better than the space shuttle in the same way you are comparing PC's and mainframes. I would NEVER use the space shuttle to go to the corner grocery, but i would use my car, and I would never use my car to go to the moon, but I might consider the shuttle. In the same light: I would never use a PC to predict tomorrow's weather, but i would use a cray, and I would use my PC to handle my store's accounts, but not a cyber. >The PC revolution, like our free-market system as opposed to communism, takes interesting analogy -- Functionality, Efficiency, Luxury. isis!dragheb | dragheb@isis.cs.du.edu
news@cit-vax.Caltech.Edu (Usenet netnews) (01/04/87)
Organization : California Institute of Technology Keywords: From: tim@tomcat.Caltech.Edu (Tim Kay) Path: tomcat!tim In article <657@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: > >There are several points to consider concerning the mainframe/micro question >as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. You are totally in the wrong ball park. PCs aren't going to encroach on mainfames just because the 32M limit was broken. A _typical_ drive that IBM sells (3380) has about 600 Mbytes on it. A _typical_ machine they sell will have anywhere from 10 to 500 of these drives on it. A PC can handle only two drives. One of the main concerns in designing the XA architecture at IBM was that installations were running out of address space for numbering devices (disks, tape drives, terminals, CTCAs, etc.). The old 370 architecture can handle 4,096 devices! The new XA architecture can handle significantly more. Typically, a single machine configured as above is not adequate, so IBM offers (and has offered for about a thousand years now) the ability to network these large machines. There are installations that have dozens of the machines (of the size I just mentioned) networked together. Consider that an IBM mainframe handles all of its I/O through I/O channels. These are really separate computers that have direct access to memory and run simple programs that the main CPU constructs for them. Each channel on an IBM mainframe at least as powerful as an 80386. A mainframe has many channels for each CPU. >I've seen 750 and 780 VAXs destroyed by fewer than 30 people doing database >work (or attempting same) at the same time. The PCs are faster and better. A Vax 780 is a wimpy machine that was not designed for large tasks. Clearly PCs are encroaching upon DEC mainframe type sales. Actually, it is incorrect to call DEC machines mainframes. The term they coined for them is, I believe, "supermini." The present discussion started when a claim was put forward that PCs are cutting into IBM's mainframe market. DEC offers no machine that is anywhere near the power of IBM's mainframes. IBM can _routinely_ put hundreds of terminals on a single machine, and it will perform well. Just try to network more than a few tens of PCs. >The PC revolution, like our free-market system as opposed to communism, takes >human nature into account, at least as it applies to bosses and managers. >I've never yet seen one of these who, upon hiring three new people to work >on a multi-user computer system, didn't simply add three new terminals to >the system. With PC networks, they are forced to add three new computers >to the system. And the beauty is that they do so with smiles on their faces, >since PCs have gotten to the point of being cheaper than most mainframe >terminals anyhow. This argument doesn't fly. Quite often, PCs are used as terminals for IBM (and DEC) mainframes. It requires the simple addition of an IRMA or PCOX type card, which costs about the same as an Ethernet card. The only disadvantage in using PCs is that the screens are two small. IBM sells a terminal that can have four windows on it, each of which is (I believe) 132 characters by 25 lines. Alternatively, the same terminal can be used for one huge window. I will rephrase a statement that I made earlier. Most people have absolutely no idea what the bulk of the computing power in this country is used for. They see what is being done on PCs and Vaxes, and they see lots of them. They assume that this type of computing is typical. It is not. You could probably take all the Vaxes in the world, multiple by 100, and you would not have enough computing power to replace all the IBM mainframes in the world. Every year, IBM's profits are roughly equal to DEC's SALES. None of this discussion starts to consider machines like Crays or IBM 3090s with vector processors. They can compute thousands of times faster than a PC. An 8087, for example is about 5 kiloflops. A 68881 is about 30 kiloflops. A Cray XMP 48 (by now an old machine) was about 1 gigaflop. This is a factor of 1,000,000,000 / 30,000 , = 33,000 difference in performance. In conclusion, I'd like to suggest that a PC is to a mainframe like a rubber raft is to a yacht. They both have their uses, and many people that have yachts have uses for rubber rafts. Likewise, many people that can't afford or justify yachts can afford and do enjoy rubber rafts. On the other hand, it is ludicrous to think that, because rubber rafts are becoming less expensive, the sales of yachts are falling off. Comments? Tim Caltech Timothy L. Kay tim@csvax.caltech.edu Department of Computer Science Caltech, 256-80 Pasadena, CA 91125
ted@imsvax.UUCP (Ted Holden) (01/05/87)
Tim Kay (Caltech) writes: >In article <657@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: >>There are several points to consider concerning the mainframe/micro question >>as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. >You are totally in the wrong ball park. PCs aren't going to encroach on >mainfames just because the 32M limit was broken. A _typical_ drive that >IBM sells (3380) has about 600 Mbytes on it. A _typical_ machine they sell >will have anywhere from 10 to 500 of these drives on it. A PC can handle >only two drives. One of the main concerns in designing the XA architecture >at IBM was that installations were running out of address space for numbering >devices (disks, tape drives, terminals, CTCAs, etc.). The old 370 >architecture can handle 4,096 devices! The new XA architecture can handle >significantly more. I'm not the one who's in the wrong ballpark. The comparison between individual PCs and individual mainframes is no more meaningful than that between an individual army ant and one of the creatures which ARMIES of such small army ants EAT. What is important is the servers reach a certain level of power and sophistication, which they have, and the state of the control soft- ware for the army (e.g. Novell's Netware). These are good enough for such PC armies to eat VAX class machines today and I can't believe they won't be eating IBM mainframes in another year or two. >Typically, a single machine configured as above is not adequate, so IBM >offers (and has offered for about a thousand years now) the ability to >network these large machines. There are installations that have dozens >of the machines (of the size I just mentioned) networked together. And cost many times what equivalent compute power in the form of PCs would. As long as such systems could do things which small computers physically couldn't do, such costs could have been justified. The day when all such bets are off is rapidly approaching. >The present discussion started when a claim was put forward that PCs are cutting >into IBM's mainframe market. DEC offers no machine that is anywhere near the >power of IBM's mainframes. IBM can _routinely_ put hundreds of terminals >on a single machine, and it will perform well. ...... Assuming all but a few of these terminals are turned off. At every mainframe installation I've ever worked at or with, there was money to be made by some one selling IBM or Univac PUNCHING BAGS, just ordinary 70 lb heavy bags with "IBM" or "UNIVAC" or maybe "CDC" written on them. The people would have lined up. I mean, if any of these statements about mainframes serving 200 or 700 users at minimal (one guy quoted .05 for a CDC) load were true, PCs would never have seen the light of day. Get serious. >I will rephrase a statement that I made earlier. Most people have absolutely >no idea what the bulk of the computing power in this country is used for. I've been involved with porting mainframe applications to small computers for some time now, mostly because people finally couldn't stand dealing with the mainframes. Aside from the obvious database kinds of things, these have included large Fortran programs, including the Census X-11 time series routine and a couple of risk-analysis types of programs from Social Security in Baltimore, all of which ran way the hell faster on generic 68000 machines than they ever did on these organizations' Univac mainframes. Such applications are probably pretty typical of how the "bulk of the computing power in this country" is used. >None of this discussion starts to consider machines like Crays or IBM >3090s with vector processors. They can compute thousands of times >faster than a PC. An 8087, for example is about 5 kiloflops. A >68881 is about 30 kiloflops. A Cray XMP 48 (by now an old machine) >was about 1 gigaflop. This is a factor of 1,000,000,000 / 30,000 , >= 33,000 difference in performance. No argument here. The Crays will always be with us. Good coding on a 68020 device with Witech boards can approach a reasonable fraction of Cray power FOR SPECIFIC APPLICATIONS, as witnessed by the Silicon Graphics devices for example, but not for general usage. The Intel world doesn't come close here. Note, however, that supercomputing is a really small fraction of computer usage in the world, and that PC clone vendors will probably be satisfied with the other 95 percent. Ted Holden IMS
ma168a@sdcc3.ucsd.EDU (John Wavrik) (01/05/87)
Perhaps a better analogy is that the PC is to the mainframe as a yatch is to a battleship. As a UNIX user I can find out what other system users are doing at any time. I have no reason to believe the situation here is atypical: most users are editing files. I agree that a PC and a mainframe are two different types of computers -- but I argue that most of the needs of current users of mainframes could be met my PCs. Our overburdened VAXen provide an inferior environment (and slower response time) than an IBM-AT. I'd like to see tasks which could just as well be performed on a PC farmed out to these machines. I don't see mainframes as a dying breed -- I do see more and more users clogging them. PCs represent a viable alternative to the increased demand on computing services posed by the increase in the user population.
news@cit-vax.Caltech.Edu (Usenet netnews) (01/05/87)
Organization : California Institute of Technology Keywords: From: tim@tomcat.Caltech.Edu (Tim Kay) Path: tomcat!tim Ted Holden (IMS) writes: >Tim Kay (Caltech) writes: >>Ted Holden (IMS) writes: >>>There are several points to consider concerning the mainframe/micro question >>>as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. >>You are totally in the wrong ball park. PCs aren't going to encroach on >>mainfames just because the 32M limit was broken. A _typical_ drive that >>IBM sells (3380) has about 600 Mbytes on it. A _typical_ machine they sell >>will have anywhere from 10 to 500 of these drives on it. A PC can handle >>only two drives. > >I'm not the one who's in the wrong ballpark. The comparison between individual >PCs and individual mainframes is no more meaningful than that between an >individual army ant and one of the creatures which ARMIES of such small >army ants EAT. What is important is the servers reach a certain level of >power and sophistication, which they have, and the state of the control soft- >ware for the army (e.g. Novell's Netware). These are good enough for such >PC armies to eat VAX class machines today and I can't believe they won't be >eating IBM mainframes in another year or two. A single 3090 with 100 1 gigabyte disks has 100 gigabytes of storage of mass storage on it. How many PCs would it take, so that, at two disks per PC, you could come anywhere near that figure? Suppose you put two 400 megabyte Eagles on each PC. You would then have to network over 100 PCs, each of which must make 800 megabytes of data available to the network. Would this be workable? How much degradation would each user feel from network requests for data? What is the largest network of PCs that anybody out there has heard of? (Please respond to me, and I will summarize.) So, let's build a version of the IBM PC with I/O processors, high bandwidth point-to-point communications, additional, higher bandwidth busses, the ability to handle many disk drives, etc., to overcome the difficulties that I have just mentioned. For the CPU, we'll use an 80386, which I agree could be adequate. What you now have is a machine somewhere between today's minicomputers and a mainframes. You have lost your cost advantage, and the machine I just described is no longer a personal computer. But I'll agree that this machine could do what you are claiming. I have already conceded that smaller computers might eventually replace lots of mainframes. But I think it will take a bit longer than you are seeming to imply. I certainly disagree that IBM has lost past sales due to current PCs, and I don't think they will for a few years yet. And, don't get me wrong. I love PCs, and I am fascinated by what they can do. I can't imagine why anybody would buy something like a System 36, when you could have an AT or three that are just as powerful (more powerful?) and costs a fraction as much. Timothy L. Kay tim@csvax.caltech.edu Department of Computer Science Caltech, 256-80 Pasadena, CA 91125
sam@lanl.ARPA (Sam A Matthews) (01/07/87)
In article <657@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: > >There are several points to consider concerning the mainframe/micro question >as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. >A Novell server can fully use big disks such as the 300 meg Cores. Such a >server can use several meg of ram as a cache area and adequately service 30 >or so users. The neat thing is that the entire compute load has been put >on the individual PCs; the server is only being required to fetch and store >info to and from disk. As servers get more powerful, using 386 and future >chips, and storage medea gets better, presumably using laser technologies, >it is easy to anticipate such systems replacing mainframes for most if not >all database work. The little ISI glass disks are available now for >something like $2500, 256 meg disks going for $100, 1 ms track to track seek ^^^^^^^ ^^^^ I'll take 6! :-) >time. That's fast! This sounds great, but for every improvement in micro technology there is also an improvement in "mainframe" technology. They all use electronics after all. Current level of super-computer tech. is not big enough! Sure I concede that many differing applications will continue to migrate to micros, however there will *always* be jobs to big to be done on any computer! (NSA, NOAA, NASA, NORAD, Pan Am, IRS, DOE, DOD, etc) I still contend thousands of reservation desks and travel agents all accessing the same data base will not be done with micros. Sure, the terminal ends may be micros but the hub will be a monster machine. > >I've seen 750 and 780 VAXs destroyed by fewer than 30 people doing database >work (or attempting same) at the same time. The PCs are faster and better. VAXen are mini-computers and are also very old. I agree the 7xx series has lived long enough... The distinction between minis and super- micros and between super-micros and micros becomes muddier every day. >The PC revolution, like our free-market system as opposed to communism, takes >human nature into account, at least as it applies to bosses and managers. Tools are tools, each is used for a different job. Managers and bosses who refuse to recongnize this fact deserve every thing the free-market system can take away from them. :-) PCs are good at what they do and are getting better, but big machines are getting better too, you just don't read about it in PC-World. >I've never yet seen one of these who, upon hiring three new people to work >on a multi-user computer system, didn't simply add three new terminals to >the system. With PC networks, they are forced to add three new computers >to the system. And the beauty is that they do so with smiles on their faces, >since PCs have gotten to the point of being cheaper than most mainframe >terminals anyhow. It still depends on what you expect these 3 people to do... > >Ted Holden >IMS Sam Matthews /\|/\ "We put a star sam@lanl.ARPA --> * <-- in a box." (ihnp4 or cmcl2)!lanl!sam \/|\/
nelsons@psu-cs.UUCP (Shannon Nelson) (01/07/87)
In article <3690@sdcc3.ucsd.EDU> ma168a@sdcc3.ucsd.EDU (John Wavrik) writes: > >Perhaps a better analogy is that the PC is to the mainframe as a yatch is >to a battleship. > [...] > I agree that a PC and a mainframe are two different types >of computers -- but I argue that most of the needs of current users of >mainframes could be met my PCs. Our overburdened VAXen provide an inferior >environment (and slower response time) than an IBM-AT. I'd like to see tasks >which could just as well be performed on a PC farmed out to these machines. Yes, a lot (*not* all) of the functions that VAXen are used for can easily be replaces by PC's, but *please* remember that a VAX is *not* a mainframe: it is simply a mini-computer, and not designed to do the major batch number crunching that most mainframes do. In continuing the modification of an analogy, PC [XT, AT] = rubber raft (or maybe a rowboat?) VAX 11/780 = yacht IBM 4341 = oil tanker. Many larger sea-going vessels (oil tankers) include smaller vessels (life boats) to perform other functions. You don't use a lifeboat for hauling oil, and you don't use a PC for solving systems of 100 differential equations. As for the yacht, it's great for small cruises, going from one tanker to another, and the VAX is well suited for being a front-end to CRAYs and other "maxi"computers. Each tool has its use. It's possible to substitute if needed, but there's nothing like using the right tool. sn -- /======\ Shannon Nelson // \\ \\ ...tektronix!psu-cs!nelsons (( go\\to )) \\ \\ // "Keyboard. How quaint." \======/
sewilco@mecc.MECC.COM (Scot E. Wilcoxon) (01/07/87)
In article <1427@cit-vax.Caltech.Edu> tim@tomcat.UUCP (Tim Kay) writes: >Ted Holden (IMS) writes: >>Tim Kay (Caltech) writes: ... >>army ants EAT. What is important is the servers reach a certain level of >>power and sophistication, which they have, and the state of the control soft- >>ware for the army (e.g. Novell's Netware). These are good enough for such >>PC armies to eat VAX class machines today and I can't believe they won't be >>eating IBM mainframes in another year or two. "Speed" is relative. A PC is fast for balancing a checkbook. For drawing a complex image you'll want a faster processor if you don't want that PC on your desk busy for sixteen hours. Ditto for data crunching, as PC disk data paths are not fast. ... >So, let's build a version of the IBM PC with I/O processors, high bandwidth >point-to-point communications, additional, higher bandwidth busses, the >ability to handle many disk drives, etc., to overcome the difficulties that >I have just mentioned. For the CPU, we'll use an 80386, which I agree could be >adequate. What you now have is a machine somewhere between today's >minicomputers and a mainframes. You have lost your cost advantage, >and the machine I just described is no longer a personal computer. >But I'll agree that this machine could do what you are claiming. The above is currently called a "supermicro". It's a multiuser machine because for many applications it's cheaper to put dumb terminals on it than for each user to have a networked PC. As long as networking costs more than dumb terminals ($400 ASCII) that break-even point will be important. Micro *processors* are replacing larger processors, by being used to build larger computers, and will continue to push upward the lower limit of the "mainframe" and "supercomputer" ranges. A 30-processor 32032 or 680x0 machine with I/O processors packs a hefty punch, pulls a lot of data, and is only horrendously expensive when comparing it to a computer for your home. More power in a single processor is only needed for single-thread compute-intensive jobs (there will always be supercomputers). >I have already conceded that smaller computers >might eventually replace lots of mainframes. But I think ... >And, don't get me wrong. I love PCs, and I am fascinated by what they >can do. I can't imagine why anybody would buy something like a >System 36, when you could have an AT or three that are just as powerful >(more powerful?) and costs a fraction as much. Why is a System/36 not called a micro? Because the box is too big. -- Scot E. Wilcoxon Minn Ed Comp Corp {quest,dayton,meccts}!mecc!sewilco (612)481-3507 sewilco@MECC.COM ihnp4!meccts!mecc!sewilco National Enquirer seers: 4 Reality: 360
brandon@tdi2.UUCP (Brandon Allbery) (01/12/87)
Quoted from <658@imsvax.UUCP> ["Re: Mainframe vs Micro"], by ted@imsvax.UUCP (Ted Holden)... +--------------- | Tim Kay (Caltech) writes: | >In article <657@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: | >>There are several points to consider concerning the mainframe/micro question | >>as it relates to databases. The 32 meg limit doesn't apply to Novell Netware. | | >You are totally in the wrong ball park. PCs aren't going to encroach on | >mainfames just because the 32M limit was broken. A _typical_ drive that | >IBM sells (3380) has about 600 Mbytes on it. A _typical_ machine they sell | >will have anywhere from 10 to 500 of these drives on it. A PC can handle | >only two drives. One of the main concerns in designing the XA architecture | >at IBM was that installations were running out of address space for numbering | >devices (disks, tape drives, terminals, CTCAs, etc.). The old 370 | >architecture can handle 4,096 devices! The new XA architecture can handle | >significantly more. | | I'm not the one who's in the wrong ballpark. The comparison between individual | PCs and individual mainframes is no more meaningful than that between an | individual army ant and one of the creatures which ARMIES of such small | army ants EAT. What is important is the servers reach a certain level of | power and sophistication, which they have, and the state of the control soft- | ware for the army (e.g. Novell's Netware). These are good enough for such | PC armies to eat VAX class machines today and I can't believe they won't be | eating IBM mainframes in another year or two. +--------------- Show me a *network* on a PC whose data transfer rate is as fast as our ancient 14" Fujitsu drives (B-something-or-other, not Eagles), much less the current breed of fast hard disks. I would hesitate to use networked PCs for our database application -- and I am currently considering machines to replace our Plexus P/60 for that, as it isn't capable of handling 18 users running what is currently two applications, one in Unify and one in RM/COBOL. (I am in the midst of evaluating 4GLs to remedy this; looks like Progress will be the one we use. However, my speed tests don't show too much of an improvement from our current system to any of the 4GLs I have tested.) Any suggestions? +--------------- | >Typically, a single machine configured as above is not adequate, so IBM | >offers (and has offered for about a thousand years now) the ability to | >network these large machines. There are installations that have dozens | >of the machines (of the size I just mentioned) networked together. | | And cost many times what equivalent compute power in the form of PCs would. | As long as such systems could do things which small computers physically | couldn't do, such costs could have been justified. The day when all such bets | are off is rapidly approaching. +--------------- Nevertheless, the speed issue remains. You might network together 500 PCs to make the user-equivalent of an Amdahl, but network accesses will make it a lot slower than the Amdahl... +--------------- | >The present discussion started when a claim was put forward that PCs are cutting | >into IBM's mainframe market. DEC offers no machine that is anywhere near the | >power of IBM's mainframes. IBM can _routinely_ put hundreds of terminals | >on a single machine, and it will perform well. ...... | | Assuming all but a few of these terminals are turned off. At every mainframe | installation I've ever worked at or with, there was money to be made by some | one selling IBM or Univac PUNCHING BAGS, just ordinary 70 lb heavy bags with | "IBM" or "UNIVAC" or maybe "CDC" written on them. The people would have lined | up. I mean, if any of these statements about mainframes serving 200 or 700 | users at minimal (one guy quoted .05 for a CDC) load were true, PCs would | never have seen the light of day. Get serious. +--------------- Even as slow as it was, I preferred using CSU's IBM 370 to using a PC. And networked PCs I used at the same time made the bogged-down 43xx look fast by comparison. Get serious. +--------------- | >I will rephrase a statement that I made earlier. Most people have absolutely | >no idea what the bulk of the computing power in this country is used for. | | I've been involved with porting mainframe applications to small computers | for some time now, mostly because people finally couldn't stand dealing | with the mainframes. Aside from the obvious database kinds of things, | these have included large Fortran programs, including the Census X-11 | time series routine and a couple of risk-analysis types of programs from | Social Security in Baltimore, all of which ran way the hell faster on | generic 68000 machines than they ever did on these organizations' Univac | mainframes. Such applications are probably pretty typical of how the | "bulk of the computing power in this country" is used. +--------------- No doubt. And how many other things were running on the mainframe, as opposed to how many other things were running on the 68000? Many 68020 machines run slower than 68000 machines: the 68020s sold by Plexus have 80 user limits, whereas the 68000s (until recently) were limited to 40 users. Is this a reason to tout 68000s as faster than 68020s? ++Brandon -- ``for is he not of the Children of Luthien? Never shall that line fail, though the years may lengthen beyond count.'' --J. R. R. Tolkien Brandon S. Allbery UUCP: cbatt!cwruecmp!ncoast!tdi2!brandon Tridelta Industries, Inc. CSNET: ncoast!allbery@Case 7350 Corporate Blvd. INTERNET: ncoast!allbery%Case.CSNET@relay.CS.NET Mentor, Ohio 44060 PHONE: +1 216 255 1080 (home) +1 216 974 9210
li@tybalt.caltech.edu (James C. Li) (01/14/87)
In article <658@imsvax.UUCP> ted@imsvax.UUCP (Ted Holden) writes: >No argument here. The Crays will always be with us. Good coding on a >68020 device with Witech boards can approach a reasonable fraction of >Cray power FOR SPECIFIC APPLICATIONS, as witnessed by the Silicon Graphics > >Ted Holden >IMS Would someone DEFINE what a Micro is and what a Mainframe is? This argument seems to be going in circles because as Micro's increase in power, so do Mainframes. It isn't fair to compare Micro's of todays to Mainframes of the 60's. Also, what is a Mini? And what about a Supercomputer? What about the Hypercube(claim is 100 68020's in parallel can out compute a Cray 1)? (Yes, the right size/type for the right application, but what is similar about these systems that CAN be compared, or can't they?) li@tybalt.caltech.edu (James Li) __ __ (also li@citromeo.caltech.bitnet) | | "Official KANK symbol" ------> | |
ken@argus.UUCP (Kenneth Ng) (01/22/87)
In article <1490@cit-vax.Caltech.Edu>, li@tybalt.caltech.edu (James C. Li) writes: > Would someone DEFINE what a Micro is and what a Mainframe is? This argument > seems to be going in circles because as Micro's increase in power, so do > Mainframes. It isn't fair to compare Micro's of todays to Mainframes of the > 60's. Also, what is a Mini? And what about a Supercomputer? What about the > Hypercube(claim is 100 68020's in parallel can out compute a Cray 1)? > (Yes, the right size/type for the right application, but what is similar about > these systems that CAN be compared, or can't they?) > li@tybalt.caltech.edu (James Li) __ __ > (also li@citromeo.caltech.bitnet) | | > "Official KANK symbol" ------> | | We could use the old definition: a micro you can lift with two hands, a mainframe you can't. The Cray supercomputer line has supposedly been outpaced by several computers in recent years, especially by parallel and vector processing machines. But one must realize that no matter how much one parrellels stuff, there will always be stuff done in serial fashion. And that is where the Cray really excells, with one of the fastest (is it the fastest?) scaler processors around. By the way, here is an actual benchmark done here at NJIT on various computers. The benchmark is a compute bound matrix solution to find the capicatence of a square plate via point charge analysis. The speed killer is to solve a 4x4, 9x9, 16x16, up to a 400x400 floating point matrix. An IBM PC (without 8087) was stopped after a week, it didn't complete. An AT&T 3b5 with an average load of 5 was stopped after 3 days (system crash). An IBM 4361 with an average load of 3 users took about 9 hours. A Univac 90/80-4 with about 80 users took 10 minutes. Need I say more about the power of mainframes? And the Univac is 1970's technology to boot. Anyone have a Cray I can borrow to find out how long it will take there? Now, there are some differences that must be taken into account. First, was the quality of the compilers. The Univac compiler was NICE and generated VERY tight code. The 4361 compiler I have strong reservations as to how good it is. Perhaps one year when I get bored I'll code the routines all in assembly and then run the tests. -- Kenneth Ng: Post office: NJIT - CCCC, Newark New Jersey 07102 uucp !ihnp4!allegra!bellcore!argus!ken *** WARNING: NOT ken@bellcore.uucp *** bitnet(prefered) ken@orion.bitnet Gillian: "Are you sure you won't change your mind?" Spock: "Is there something wrong with the one I have?"
npollack@polyslo.UUCP (01/28/87)
In article <722@argus.UUCP> ken@argus.UUCP (Kenneth Ng) writes: >In article <1490@cit-vax.Caltech.Edu>, li@tybalt.caltech.edu (James C. Li) writes: >> Would someone DEFINE what a Micro is and what a Mainframe is? Around here, we generally make the following distinction; 1. A micro-computer can be thrown fairly far, and you have to squint to look at it very long. You generally need a large truck with new batteries close by. It usually has more than a 4-bit processor. 2. A Personal Computer can be thrown a few feet, and can not run on batteries. (But has been known to run all over your accounting disk.) It has an operating system with numerous dead ends (disk full, fatal error, hung, printer off-line lock-up, etc etc etc). 3. A Mini computer may have handles, but can not be thrown at all. Two or three people might use it to warm their lunch, as a punching bag when its down, or maybe shove it across the room. It has a "real" operating system that is capable of recovery from errors. 4. A Mainframe computer has no handles, but has replaced them with crane hooks. It can not be thrown or shoved. You can climb on it, use it to heat lunch (and a large building), and can go broke feeding it electricity, freon, and cooling water (cyber). They make nice artificial reefs off of the Pacific Ocean. They are also very very fast, depending on relative number of users. They have been known to swallow operaters past midnight. > >The Cray supercomputer line has supposedly been outpaced by several computers >in recent years, especially by parallel and vector processing machines. But >one must realize that no matter how much one parrellels stuff, there will >always be stuff done in serial fashion. And that is where the Cray really >excells, with one of the fastest (is it the fastest?) scaler processors around. ****************** Seriously people, I am enjoying the ongoing discussion about micros and mainframes, but please stop trying to argue which is better. The proper computer for the job depends 100.00000% on the type of job. I have a private jet and a road grader. I need to build a new runway. which is better, the jet or the grader???? I now need to fly to Hawaii. Which is better, the tractor or the jet???? There is no one computer on planet earth that will suit all apllications and problems equally well. A Personal Computer is the most cost effective tool for word-processing by an individual, and a Cray-1 is effective for massive mathematical problems such as simulation or "predicting" the weather for next week. You shall not attempt to run Chase Manhatten Bank on a network of IBM-PC's. You Equally shall not buy a Cray-X/MP for a home computer to do word-processing. I firmly believe that all the people arguing over which box is better should go back to a university and study advanced computer architecture, systems engineering, and basic seminars on how to match a computer to your specific task, problem, goal, or job. You all seem to be saying; "People should buy this computer because it is better, and then go see if anyone wrote software so I can do my data-processing-stuff." This is the case with everyone getting so HOT to buy the new intel 80386 machines, when very few people have even begun to use the power of their 80286 machines. There is very little software available yet for 80386 machines. Are we all looking for the 1987 Hood-ornament status symbol, or are we trying to do a professional job of information and data processing? At the same time, the distinction between micro and mainframe is becomming blurred. People are connecting mainframe disk drives (i.e. very fast) to micro computers. Some are using micro-processors in parallel to outperform minicomputers, and neither can be thrown very far. At the same time, current mainframes under development make the multi-processor micro look sick for number crunching speed. And the multi-processor 68020 unix boxes make a vax 750 look sick. I believe that there are too many directions of progress for any one person to be knowledgeable about all of them. The distinction between micro,mini, and maxi frame is not important. The distinction of how well a given machine solves YOUR data processing or computing problem is ALL important. Have a nice day, and try not to abuse your terminal after reading this. Neal Pollack Chief Engineer Computer Science Department California Polytechnic State University San Luis Obispo, CA 93407 Please direct all hate mail to bugs@Polyslo Newsgroups: comp.sys.ibm.pc Keywords: micro mainframe flames funnies