gnu@hoptoad.uucp (John Gilmore) (01/04/87)
I was thinking the other day about how people claim that their Atari ST has the power of a Vax 750 and why this is bull because the power is not harnessed. The Vax can be configured in hundreds of different ways to meet peoples' needs -- e.g. local choice of disks, ports, ram, networks, software, etc. With the micros you are stuck waiting while somebody figures out how to hook things up. (Now they are all trying to figure out how to retrofit multitasking and hook up read/write devices to the cartridge ports and turn the joystick socket into a network and such.) Garbage! It's a lot easier if you just do it right in the first place. A while ago people were speculating about the new Apple Unix machine; some people claim it would have a Nu bus. I could see Apple going with the Nu bus only so they could claim to be following standards while in actuality not doing so. (I don't know whether they would want to do that or not.) The VMEbus is not absolutely pristine, but hey, if you're building a machine with custom chips or gate arrays (I don't believe a new apple box would be TTL and PALs), saving a few gates or ns in your I/O bus interface is the least of your worries. What would be a win for their customers would be a VMEbus or Multibus I (or even a Q-bus or Unibus), since it gives some flexibility in finding vendors of boards. Like, you could buy SMD disk controllers, Pertec tape drives, *more serial ports*, networking, etc all off the shelf, rather than waiting for the gaggle of little companies, like ducklings following their mama, to bring out little boards for the new machine. The reason a Sun is as powerful as a Vax, while an Amiga or Atari is not, is because Sun interfaced the 68000 to just about anything you could get on a Vax, all the software and hardware options (except VMS). They worked hard to make sure that any HLL program that ran on a Vax would run on a Sun (modulo byte order and page zero problems), rather than defining Yet Another new programming environment. Then it *really* was as powerful as a 750, and the 68020 versions do much better. Just by doing that right, Suns are worth 10x what Ataris or Macs are -- people *queue up* with 90-day leadtimes to pay that much for them. [Please don't cross-post any replies to the mac/atari/amiga groups. I just named them as examples, not so a crowd of admirers could leap to their defense. Really, they're nice machines...........(urp)] I/O is the great ghetto of the micro world. PS: I think it's a great milestone that, now that there is a standard electrical bus interface for IBM PC compatible peripheral cards, all the manufacturers (including IBM) are altering the physical specs so the cards don't fit any more. It shows that they don't want standardization no matter what they say. -- John Gilmore {sun,ptsfa,lll-crg,ihnp4}!hoptoad!gnu gnu@ingres.berkeley.edu I forsee a day when there are two kinds of C compilers: standard ones and useful ones ... just like Pascal and Fortran. Are we making progress yet? -- ASC:GUTHERY%slb-test.csnet
ravi@mcnc.UUCP (Ravi Subrahmanyan) (01/05/87)
In article <1611@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: >I was thinking the other day about how people claim that their Atari ST >has the power of a Vax 750 and why this is bull because the power is >not harnessed. The Vax can be configured in hundreds of different ways >to meet peoples' needs -- e.g. local choice of disks, ports, ram, >networks, software, etc. I think John's way off the mark here as far as certain users (like me) are concerned. I really am not concerned with how many ways my ST can be configured, but rather with ensuring that I do have >>at least one<< configuration I can use. And, it happens to be something I like very much.. a choice of disks, RAM is available, extra ports aren't wanted, I have a fair approximation to my BSD environment, and networking isn't that big a deal as long as I can use kermit or something to move files. >With the micros you are stuck waiting while >somebody figures out how to hook things up. (Now they are all trying >to figure out how to retrofit multitasking and hook up read/write >devices to the cartridge ports and turn the joystick socket into a >network and such.) Garbage! It's a lot easier if you just do it right >in the first place. Who's waiting? And for what? People will hack, given anything. It doesn't mean I am waiting with bated breath for their success. I have *no need* for adding a read/write device to the cart port, and multitasking is available if I want it, but I really don't care.. >The reason a Sun is as powerful as a Vax, while an Amiga or Atari is >not, is because Sun interfaced the 68000 to just about anything you >could get on a Vax, all the software and hardware options (except >VMS). Agreed, but if I'd wanted to run stuff from our Vaxen on a micro, I'd have bought a Sun. As it is, I find it infinitely preferable to do edits or small compiles (for generic C only, of course), or make drawings, or just tool around, on a Mac or ST rather than on the (usually loaded) 780. When I'm writing a 20 page document (not quite out of purgatory yet!), it's so much faster to get it shipshape (and yes, I check my spellings there too) on a micro before sending it off to the VAX/Laser combo. >I/O is the great ghetto of the micro world. I think the point is that a <$1000 machine will never be a Sun precisely because of the corners that had to be cut to get the price down. But that doesn't mean it isn't preferable in it's own way.. -ravi
bill@voodoo.UUCP (Bill Sears) (01/05/87)
In article <1611@hoptoad.uucp> gnu@hoptoad.uucp (John Gilmore) writes: >I was thinking the other day about how people claim that their Atari ST >has the power of a Vax 750 and why this is bull because the power is >not harnessed. The Vax can be configured in hundreds of different ways >to meet peoples' needs -- e.g. local choice of disks, ports, ram, >networks, software, etc. With the micros you are stuck waiting while >somebody figures out how to hook things up. (Now they are all trying >to figure out how to retrofit multitasking and hook up read/write >devices to the cartridge ports and turn the joystick socket into a >network and such.) Garbage! It's a lot easier if you just do it right >in the first place. Here here!!! I get tired of people saying that they can configure their computer to outperform a vax. It sort of reminds me of the BMW commercial where all of these "inexpensive" cars are "just as good as a BMW" or "BMW would make one like this" or "You think BMW has high tech...", etc. There is no substitute for the "real thing". Just blowin' off some steam. I had to go back to work today. 8-( These are my opinions if you want 'em you can have 'em. -- Bill Sears ....uw-beaver!ssc-vax!voodoo!bill Masochist's Battle Cry - Stop it again!!! Quit it some more!!!
rb@cci632.UUCP (Rex Ballard) (01/06/87)
This debate is hardly new. One of the key points in determining which is "better" for a specific set of needs is to determine what is actually needed. First, a mini or mainframe has access to, and needs, more storage. If you wanted to access 4 gigabytes of data, application code, and tools, a mini or mainframe is probably a good idea. Even with CPU speed being divided among 100 or more users, it is likely, especially with "text only" processing, that the CPU won't be that heavily loaded, but rather that the drives will be "crunching away". On the other hand, if you want to do bit-mapped manipulations of graphics, windows, what-you-see-is-what-you-get editing, and similar loads that require a great deal of CPU overhead dedicated to one user, it's probably a good idea to incorporate a micro into the user level interface. In spite of the CPU benchmarks, the figures are very misleading. A VAX for example runs the main CPU at about the same speed as the Atari ST, yet when connected to 100 or so VT-100's or Techtronix terminals, and additional effective 10 to 100 mips is being used in a "distributed functionality" mode. In many cases the "VT-100" isn't a terminal at all, but rather a "terminal emulator", often running at speeds of up to 1 MIPs. Mainframe people like to think of micros/terminals/emulators/... as "dumb tubes" and attempt to do as much of the "intelligent work" in the host. Micro people tend to think of servers, telecommunications services, videotex,... as "dumb disks" and attempt to to the "intelligent work" in the micro. Slowly, the interconnections between host and micro are becoming more sophisticated. Interfaces like X-windows, and various "remote file systems" are causing a closer blend and a tighter, more efficient interface between the two. As this occurs, both "micro" and "mainframe" become more productive, with the mainframe handling more users and more storage, and the micros handling more complex presentation. Perhaps in a few years, we'll start seeing integration of Host and Micro become so tight that systems such as a VAX 8600 cluster, or a 6/32-FT will be running as many as 1000 users, 10 or 20 intelligent disk drives (built in caching, i-node searching, directory traversal...) and developing performance numbers measured in BIPS (billion instructions/second). A good example of such integration would be a simple editor. The host would appear to be running effectively a "line editor" like ed, the "disk drive" would be inserting and deleting blocks from the file, and the micro would be handling font presentation and converting the visual information to "ed" and/or NROFF type commands to the host. With proper load balancing, it would be possible to reach speeds well into the 2 BIPS region. Anybody wanna buy a used crystal ball? Rex B.
roberts@icst-ecf.arpa (ROBERTS, JOHN) (01/12/87)
>> I was thinking the other day about how people claim that their Atari ST >> has the power of a Vax 750 and why this is bull because the power is >> not harnessed. > Here here!!! I get tired of people saying that they can configure their > computer to outperform a vax. I fail to see how this controversy ever got started in the first place. Our Vax, along with its collection of disk drives, consumes thousands of watts of electrical power, and continually billows hot air out the back. I have yet to see a micro that comes anywhere near this awesome display of raw power. (On the other hand, the supercomputer at our site is so powerful that it requires a large supply of chilled water for cooling.) In any event, this debate draws attention away from a much more critical issue of our times: the increasing use of "colorization" in our computer video displays. Surely the use of color VDTs is an affront to the primal esthetic purity of the old reliable monochrome displays! A panel of the artistically sensitive should be formed to investigate this problem immediately! <Standard disclaimers. These are not even my opinions, let alone anyone else's.> John Roberts roberts@icst-ecf.ARPA ------