ross@apollo.HP.COM (Mike Ross) (12/19/89)
I also don't want to read a lot of religious wars about whether the Next or the Mac is better. Probably each machine has its own good features, and if you bought one, you probably liked one or more of them. People have differing needs and reasons for buying the machines, and you're unlikely to have the same exact needs as the person on the net with whom you are arguing. The original subject line was " What I'd like to see on an Apple of the 90's". Now the answer to *THAT* question is interesting. I'd like to hear what other people would like to see on a machine at the price of a MAC or NEXT in the 90s. For myself, I'd like to see a truly distributed file system, totally transparent from one machine to another on a common network, and a good data base system that is capable of utilitizing the disk and processing power of several machines at once to handle queries. I'd also like to see machines with up to 4 processors at this price, with some tools to take advantage of parallelism. Getting parallel computing down in price so that it can be easily utilized is an important step for the 90's, I think.
rlp@beach.cis.ufl.edu (Bob Powell) (12/23/89)
In article <47808ed2.20b6d@apollo.HP.COM> ross@apollo.HP.COM (Mike Ross) writes: >The original subject line was " What I'd like to see on an Apple of the 90's". >Now the answer to *THAT* question is interesting. I'd like >to hear what other people would like to see on a machine >at the price of a MAC or NEXT in the 90s. Hmmm, well, it *is* Christmas time, so why not a wish list. What I'd like to see in a Mac or Next-class machine of the 90s: * This is based on the idea of a $10,000 machine (approx. the cost of a Next + printer, or Mac II*, in a reasonable configuration [i.e., one that won't run out of memory with four processes]). * Base machine would have at least two processors, running in parallel. Shouldn't be that expensive; the Jan. issue of Discover has an article on computing in 1989. In the article it states that Thinking Machines of Cambridge, MA released in 1989 a system known as the CM-2a, with 4,000 processors. The price on this baby is about half a million bucks. If my calculations are correct, that works out to $125 per processor. Throw in some extra for basic support hardware and design cost recovery, and you've still got very inexpensive parallel processing. No flames for my oversimplification, please; || processors aren't that expensive (though the software to run 'em is), so why bother with squeezing more MIPS out of CISC/RISC single-processor machines. * If we're going to stick with single-processor systems, or even if we go to || processors, let's go with a 64-bit system. * The machine should have a minimum of 128M RAM. If it has > 2 processors, each should bring an additional 64M. * Minimum storage, in the main system chassis, of 4G, with access time of around 15 mS. For the college kiddies and techno-nuts, a removable optical disk, CD-sized (and with no cartridge around it) and capacity (600M [vs. 256M for the current Next cartridge]). Also, a cheap (couple of bucks per disk) system for flinging a data file across the office or to take home. In other words, a 3.5" 1.4M disk, much like that used in the (surprise!) Mac. I don't think anyone's going to release a laptop, or even a AC-powered portable, with an optical disk system (they're not particularly small nor light), and I wouldn't want the hassle of plugging my laptop into the Ethernet just to load the latest revision to the Jones contract. * High-resolution display, 300 dpi, on a flat 21" screen. If the system is configured for color (user's option, and the thing would work in either mono, grayscale, or color, with the hardware handling color-to-grayscale mapping), lower resolution would be tolerable. * Keyboard with a looong cord, for lap use. Well, that covers most of the hardware. I don't want to think about the software yet; it's too early in the morning, and it's snowing outside (yes, it's Florida, and it's snowing). Bob
jacob@gore.com (Jacob Gore) (12/23/89)
/ comp.sys.next / rlp@beach.cis.ufl.edu (Bob Powell) / Dec 23, 1989 / > * Base machine would have at least two processors, running in parallel. > Shouldn't be that expensive; the Jan. issue of Discover has an article > on computing in 1989. In the article it states that Thinking Machines > of Cambridge, MA released in 1989 a system known as the CM-2a, with > 4,000 processors. The price on this baby is about half a million bucks. > If my calculations are correct, that works out to $125 per processor. Uhm, Connection Machines use 1-bit processors. I don't think I'd want just two of these things in my computer-of-the-90's :-) Jacob -- Jacob Gore Jacob@Gore.Com boulder!gore!jacob
olson@brutus.cs.uiuc.edu (Robert Olson) (12/28/89)
In article <21562@uflorida.cis.ufl.EDU> rlp@beach.cis.ufl.edu (Bob Powell) writes: >In the article it states that Thinking Machines >of Cambridge, MA released in 1989 a system known as the CM-2a, with >4,000 processors. The price on this baby is about half a million bucks. >If my calculations are correct, that works out to $125 per processor. Yes, but if the CM-2a (which I haven't read anything about) uses the same processors as the CM-2 (which I consider a reasonable assumption), a couple of these won't do you much good, as they are one-bit processors. (The machine does 32-bit arithmetic one bit at a time). A CM-2 processor just isn't comparable to any of the current RISC/CISC machines. Bob Bob Olson University of Illinois at Urbana/Champaign Internet: olson@cs.uiuc.edu UUCP: {uunet|convex|pur-ee}!uiucdcs!olson
ramsiri@blake.acs.washington.edu (Enartloc Nhoj) (12/28/89)
In article <1989Dec27.172152.1755@ux1.cso.uiuc.edu> olson@cs.uiuc.edu (Robert Olson) writes: >In article <21562@uflorida.cis.ufl.EDU> rlp@beach.cis.ufl.edu (Bob Powell) writes: >>In the article it states that Thinking Machines >>of Cambridge, MA released in 1989 a system known as the CM-2a, with >>4,000 processors. The price on this baby is about half a million bucks. >>If my calculations are correct, that works out to $125 per processor. > >Yes, but if the CM-2a (which I haven't read anything about) uses the same >processors as the CM-2 (which I consider a reasonable assumption), a >couple of these won't do you much good, as they are one-bit processors. >(The machine does 32-bit arithmetic one bit at a time). A CM-2 processor >just isn't comparable to any of the current RISC/CISC machines. > >Bob >Bob Olson University of Illinois at Urbana/Champaign >Internet: olson@cs.uiuc.edu UUCP: {uunet|convex|pur-ee}!uiucdcs!olson Has anybody heard anything about the ATW? I don't know much, but i know it uses the T-800 and uses a 68000 for I/O... has dedicated video RAM.( I think 32 bit video) runs at 10mips with a single T-800.. can hold 64 T-800's.. runs HELIOS parallel processing OS ... Comes with C-Shell and UNIX command set.. i think.... I think the base price with an 80MB drive and 4Mb RAM is around $7,500. The system is designed for multiprocessing... I'd like to know more about it .. if anyone in this group is conversant with the product. Thanks -kevin ramsiri@blake.acs.washington.edu BTW: i brought this into this group having just read 125 articles about "APPLE"...