brooks@maddog.llnl.gov (Eugene Brooks) (03/01/90)
Well boys? The top IBM 6000 series Killer Micro turns in
about one third of the performance of a YMP for the Geometric
Mean on the Livermore Fortran Kernels. I won't mention what
it does on my "really crufty scalar code." It would appear that
more than the likes of Mips and Sun that are going to see some
carnage.
After seeing IBM's THREE PAGE add in the Wall Street Journal
I guess that Motorola decided that it could not sit quiet any
longer. Motorola featured a full page add preparing us for an
announcement (presumably in the same paper) on the 5'th.
Any comments or speculations? Will it be an IBM killer?
Will this monthly rabbit hop in performance between the vendors
ever end? Will Killer Micros reach a saturation point in performance
any time soon?
NO ONE WILL SURVIVE THE ATTACK OF THE KILLER MICROS!
brooks@maddog.llnl.gov, brooks@maddog.uucpdbradley@gibson.ncsa.uiuc.edu (David Bradley) (03/01/90)
In article <50437@lll-winken.LLNL.GOV> brooks@maddog.llnl.gov (Eugene Brooks) writes: >Well boys? The top IBM 6000 series Killer Micro turns in >about one third of the performance of a YMP for the Geometric >Mean on the Livermore Fortran Kernels. This was on a *single processor* of the YMP, right? -- David Bradley University of Illinois at Urbana Champaign
amir@smsc.sony.com (Amir ) (03/02/90)
In article <50437@lll-winken.LLNL.GOV> brooks@maddog.llnl.gov (Eugene Brooks) writes: >After seeing IBM's THREE PAGE add in the Wall Street Journal >I guess that Motorola decided that it could not sit quiet any >longer. Motorola featured a full page add preparing us for an >announcement (presumably in the same paper) on the 5'th. Here is the info that I have heard Motorola has leaked in UK. The workstations are supposed to be based on 88000. They are also supposed to have "higher resolution" than IBM systems (whatever that means). The killer is that they claim the price would be 35% lower and would have comparable performance! I can't imagine how the current 88K would have this level of performance. Even the 33 Mhz version would not be able to match the IBM box in floating point. Talking about the price, I noticed that the hard disk is standard but ethernet board is not in the IBM systems. This doesn't make much sense. Is this really true? If so, how much is the ethernet card? Also, if it is a microchannel card, what are the chances of it being high performance? Hopefully, it is not a PS/2 card.... >Any comments or speculations? Will it be an IBM killer? >Will this monthly rabbit hop in performance between the vendors >ever end? Will Killer Micros reach a saturation point in performance >any time soon? I don't see how it is going to be an IBM killer. Motorola still needs to prove that it can be a computer company and not just a semiconductor company. It has failed everytime in the past. Also, it is not clear how it is going keep its OEMs happy if it starts to compete with them. Also, as I mentioned before, unless it uses the next generation 88K (superscaler?), it is not going to win as far as absolute performance is concerned. > NO ONE WILL SURVIVE THE ATTACK OF THE KILLER MICROS! Agreed. Specially if they are standing still or taking a long time to come out with a product.... >brooks@maddog.llnl.gov, brooks@maddog.uucp -- Amir H. Majidimehr Operating Systems Group Sony Microsystems amir@smsc.sony.com | ...!{uunet,mips}!sonyusa!amir
gillies@m.cs.uiuc.edu (03/03/90)
/* Written 11:58 am Mar 1, 1990 by amir@smsc.sony.com in m.cs.uiuc.edu:comp.arch */ > I don't see how it is going to be an IBM killer. Motorola still needs to > prove that it can be a computer company and not just a semiconductor company. > It has failed everytime in the past. HP proved it can be a workstation company by purchasing Apollo. Well folks, Motorola purchased Gould's Urbana software division recently, so now it has a lot more UNIX expertise than in the past. I would look for much more sophisticated Motorola system software in the near future. Don Gillies, Dept. of Computer Science, University of Illinois 1304 W. Springfield, Urbana, Ill 61801 ARPA: gillies@cs.uiuc.edu UUCP: {uunet,harvard}!uiucdcs!gillies
sysmgr@KING.ENG.UMD.EDU (Doug Mohney) (03/03/90)
>I don't see how it is going to be an IBM killer. Motorola still needs to >prove that it can be a computer company and not just a semiconductor company. >It has failed everytime in the past. Also, it is not clear how it is going >keep its OEMs happy if it starts to compete with them. It isn't; it's going to make OEM boxes for other companies to slap their sticker on (Intel does this too, I understand). >Also, as I mentioned before, unless it uses the next generation 88K >(superscaler?), it is not going to win as far as absolute performance is >concerned. > Absolute performance? I think the name of the game is to sell boxes. If you can provide a faster box which has a good base of software (with the 88K series, sorta wishy-washy at this point), people will purchase them. >> NO ONE WILL SURVIVE THE ATTACK OF THE KILLER MICROS! > >Agreed. Specially if they are standing still or taking a long time to come >out with a product.... Uhhh, it isn't a microcomputer unless I can buy three of them for under $10,000, and still have money to put gas in the car.
amir@smsc.sony.com (Amir ) (03/03/90)
In article <00933152.FA563760@KING.ENG.UMD.EDU> sysmgr@KING.ENG.UMD.EDU (Doug Mohney) writes: >>I don't see how it is going to be an IBM killer. Motorola still needs to >>prove that it can be a computer company and not just a semiconductor company. >>It has failed everytime in the past. Also, it is not clear how it is going >>keep its OEMs happy if it starts to compete with them. > >It isn't; it's going to make OEM boxes for other companies to slap their >sticker on (Intel does this too, I understand). If this is the case, then we can assume that it is still not serious about computer business despite their $22M advertising budget for the new machine. >>Also, as I mentioned before, unless it uses the next generation 88K >>(superscaler?), it is not going to win as far as absolute performance is >>concerned. >> >Absolute performance? I think the name of the game is to sell boxes. If >you can provide a faster box which has a good base of software (with the >88K series, sorta wishy-washy at this point), people will purchase them. The name of the game *is* selling boxes. But the info that I got on the box quoted Motorola people as saying that the machines would be "comparable" in performance with IBM boxes. This was their word not mine. Also, this is net.arch. A higher frequency 88K is not too exciting. But if it is superscaler or some other nice architecture, then we have something to talk about. >>> NO ONE WILL SURVIVE THE ATTACK OF THE KILLER MICROS! >> >>Agreed. Specially if they are standing still or taking a long time to come >>out with a product.... > >Uhhh, it isn't a microcomputer unless I can buy three of them for under >$10,000, and still have money to put gas in the car. Do you mean microcomputer or personal computer? The two are not the same. Apple's new Mac IIfx (40Mhz, 68030) is going to be close to $10K for a single machine and is still considered a microcomputer(and a PC). I sure would love to get three of these for $10K :-). -- Amir H. Majidimehr Operating Systems Group Sony Microsystems amir@smsc.sony.com | ...!{uunet,mips}!sonyusa!amir
drake@sd2.almaden.ibm.com (Sam Drake) (03/04/90)
In article <1990Mar1.175823.10643@smsc.sony.com> amir@Sony.COM (Amir ) writes: > >Talking about the price, I noticed that the hard disk is standard but >ethernet board is not in the IBM systems. This doesn't make much sense. >Is this really true? Sure it makes sense; mebbee you want Ethernet, mebbee you want Token Ring, mebbee you want X.25, mebbee you don't want anything. Why pay for what you don't want? Options are good.... Note that the widely discussed $12,995 configuration DOES include an Ethernet card, however. > Also, if it >is a microchannel card, what are the chances of it being high performance? >Hopefully, it is not a PS/2 card.... It's a special card for the 6000; the card is a MicroChannel "bus master" and should thus be a good performer. Opinions are my own. Sam Drake / IBM Almaden Research Center Internet: drake@ibm.com BITNET: DRAKE at ALMADEN Usenet: ...!uunet!ibmarc!drake Phone: (408) 927-1861
hascall@cs.iastate.edu (John Hascall) (03/06/90)
In article <50437@lll-winken.LLNL.GOV> brooks@maddog.llnl.gov (Eugene Brooks) writes: }Well boys? The top IBM 6000 series Killer Micro turns in Killer *MICRO* ??? It's as big as a 'fridge!!! } NO ONE WILL SURVIVE THE ATTACK OF THE KILLER MICROS! Especially if it falls over on you! ;-) John