jss@sjuvax.UUCP (Jonathan Shapiro) (11/30/84)
[Aren't you hungry...?] A lot of talk has been exchanged here about microcode bugs in the movsu class of instructions. Recently I have heard that things are worse than that and that a few groups of instructions are known to be buggy. Is this still the case? Would anybody be kind enough to mail me a summary of which classes of opcode appear to have problems? Thanks in advance Jon Shapiro P.S. I have heard the same about problems with the MMU. Any comments on that one?
srm@nsc.UUCP (Richard Mateosian) (12/02/84)
In article <693@sjuvax.UUCP> jss@sjuvax.UUCP (Jonathan Shapiro) writes: > > A lot of talk has been exchanged here about microcode bugs in the movsu >class of instructions. Recently I have heard that things are worse than >that and that a few groups of instructions are known to be buggy. > >P.S. I have heard the same about problems with the MMU. Any comments on >that one? If anyone can help me track down the source of this kind of rumor, I'd appreciate it. It goes beyond the limits of normal competitive give and take. Jonathan, the bug lists supplied with chips are accurate to the best of our ability to make them so. We don't hide or hold back anything. And while they've been serious in the past, the bugs in the current chips are few and easily avoided. -- Richard Mateosian {cbosgd,decwrl,fortune,hplabs,ihnp4,seismo}!nsc!srm nsc!srm@decwrl.ARPA
davet@oakhill.UUCP (Dave Trissel) (12/04/84)
In article <1938@nsc.UUCP> srm@nsc.UUCP (Richard Mateosian) writes: >In article <693@sjuvax.UUCP> jss@sjuvax.UUCP (Jonathan Shapiro) writes: >> >> A lot of talk has been exchanged here about microcode bugs in the movsu >>.... I have heard the same about problems with the MMU. Any comments on >>that one? > >If anyone can help me track down the source of this kind of rumor, I'd >appreciate it. It goes beyond the limits of normal competitive give and >take. ............ > >-- >Richard Mateosian >{cbosgd,decwrl,fortune,hplabs,ihnp4,seismo}!nsc!srm nsc!srm@decwrl.ARPA Dear Net.viewers, It may interest you all to know about the pot calling the kettle. It seems lately that our customers are telling us that someone at National is spreading the rumor that the MC68020 is actually an NMOS, and *NOT* a CHMOS machine. (We won't go into a discussion on how much power 200,000 NMOS transistors would take versus the .5 watts consumption of the MC68020.) Also, the same felon has had the nerve to personally post this weak excuse of a rumor to me. I'll give ya one guess who it is, and the first one doesn't count! (We'll just chalk Jonathan's comments up to "normal competitive give and take" eh, Richard?) Motorola Semiconductor Dave Trissel Austin, Texas 32-bit Applications Engineer
jss@sjuvax.UUCP (Jonathan Shapiro) (12/04/84)
[Aren't you hungry...?] This is in part a response to Mr. Mateosan's comments, and in part a justification of the bug list I requested. For a number of months I have been trying to get National to cough up working sample chips. I gather from the conversations on the net that I am in the same boat as a lot of people in netland. Funny, but I was led to believe that samples were ready to ship.... They aren't here yet. I firmly believe that if a product doesn't live up to its claims, it is the obligation of the manufacturer to inform the purchaser of this *AND* to see to it that parts which were provided which are bad are replaced when the bugs are ironed out. This is true regardless of the size of the customer. National has not done this. Now someone might well argue that replacing the amount of bad silicon National has produced in these chips would be prohibitively expensive, and that is probably true. Reprinting the manuals is not so expensive. I find it particularly vexing when a representative of National Semiconductor has the gall to take me to task for covering my small net worth by trying to find out things his company should have told me in the first place. Until I posted the request, I was not even aware that a buglist existed. On the other hand, National also seems to me to be the best crowd around to be in bed with, and Mr. Mateosan and National are of course separate and distinct. His opinions may not represent those of his employer. Perhaps if you have a minute, Mr. Mateosan, you would be kind enough to send me a copy of this buglist, now that the public has informed me of my ignorance. I would like to see if National's buglist is keeping up with the bugs people are known to have found. While you are at it, what about those samples.....? Jonathan S. Shapiro Haverford College Haverford Pa. 19041 (215) 649-3929 ..!sjuvax!jss
srm@nsc.UUCP (Richard Mateosian) (12/06/84)
In article <255@oakhill.UUCP> davet@oakhill.UUCP (Dave Trissel) writes: > It may interest you all to know about the pot calling the kettle. It seems >lately that our customers are telling us that someone at National is >spreading the rumor that the MC68020 is actually an NMOS, and *NOT* a CHMOS >machine. > >Also, the same felon has had the nerve to personally post this weak excuse >of a rumor to me. Well, I haven't been telling any of Motorola's customers this story, but I did send Dave some mail asking if it were true. The source of the *rumor* that reached me was a Motorola public seminar on the 68020 given here in Sunnyvale on Oct 16th 1984. Someone from MOTOROLA -- obviously misinformed -- said that the 68020 was in HMOS, not HCMOS. When questioned closely on this point, the same MOTOROLA person -- obviously halucinating -- told a long involved story about a failure of the HCMOS project in Feb 84 and a parallel backup effort in HMOS that produced the current parts. Let me repeat that I haven't told any of Motorola's customers this story. Furthermore, every Motorola person I've discussed it with has denied it emphatically and convincingly. In fact, until Dave's recent posting, I'd forgotten about it and decided it wasn't true. -- Richard Mateosian {cbosgd,decwrl,fortune,hplabs,ihnp4,seismo}!nsc!srm nsc!srm@decwrl.ARPA
jss@sjuvax.UUCP (Jonathan Shapiro) (12/10/84)
[Aren't you hungry...?] Now, now, gentlemen, before we get real flames going, I spoke to Richard Mateosian about his comments, and let me back up what he said. His interpretation of my complaint was that my information had come from competitors and some dissatisfied customers who have been particularly nasty about false rumors. If competitors had fed me bad stories, I would agree with him. The fact is that they didn't. Dissatisfied customers, on the other hand, speak for themselves and are right to do so. My information had come from the net and from several almost but not quite satisfied National customers. Why there are unsatisfied National customers is subject to comment, but I will refrain. On the other hand, it seems to me that Motorola and National wasting my cpu cycles and phone bills pointing fingers at each other and saying "You are telling lies about me" is as bad as the misinformation that has come out in the first place. That stuff should have been left in kindergarden. If I had seriously considered using a 68000 series chip I would have made sure I got the information from Motorola, not from National. Other folks out there are at least as smart as I am and have the benefit of greater experience. Unfortunately, Motorola won't do much of anything by way of samples or availability commitments for anyone smaller than IBM. I haven't seen any 68020 samples... It seems to me that National is for the moment the way to go. Its a shame, really, because they are both potentially elegant chips, and if Motorola actually had silicon for the 68020 I would do a design with it if only because the 68000 did so well. As to availability and samples, it would be worth their while. You never know who is going to design the next Machine. Jon Shapiro