dfh@ecs.UUCP (07/20/83)
Thanks to all who responded to me DEC Professional marketing strategy query. Now I know I'm not the only one with serious questions about what DEC is trying to do with the Pro series. Here are the responses I received. As a final note, DEC was *supposed* to deliver our Pro 350 this afternoon. Need I say more? David Hinnant NCECS ===================== >From unc!mcnc!unc-c!rrb Fri Jul 8 18:52:04 1983 remote from tucc Although we have only recently ordered our first DEC 350, we have been looking at them pretty seriously for some time, and talking with DEC marketing about them. On that basis plus previous contacts with DEC marketing people, I will offer the following observations: DEC is trying to sell the 325/350 as a distributed workstation, and really does expect people to buy them to run software developed on the VAX. The selling point of the 325/350 is that it is basically a PDP-11/23 in a single user box. Users (or OEM's) will want to develop software for this powerful (true 16 bit) machine, and the VAX is the sytem on which to do it (if you can put up with VMS, since they don't have a version of the Programmer's Tool Kit for Unix(tm)). DEC is working very hard to control both the hardware and software for the 325/350 and is strongly resisting efforts by third parties to develop and market sofware OR hardware for the machines. They are not encouraging third parties at all. The reason is that they don't want to have to support all of the hackers who are the traditional buyers/users of 11's, Unix, micros, etc. -- they're trying to move into the Fortune 500, business marketplace where all the money is! I find it particularly interesting to note that IBM, traditionally strong in the Fortune 500 large machine MVS environment, has been successful beyond their wildest expectations (or plans for production and support capabilities) with a machine which is clearly intended to be extendable and usable by hackers! The IBM PC is neither revolutionary nor inexpensive, but it sure has expanded the market for micros. My overall view is that the DEC 325/350 are great machines, but that DEC marketing doesn't really know what it is or how to market it. I would agree that there's a strong possibility that DEC will completely blow the opportunity it has. ======================= >From mcnc!burl!spanky!hocda!ihnp4!uw-beaver!microsof!decvax!genrad!grkermit!marks Mon Jul 11 13:35:29 1983 remote from tucc I can see two reasonable explanations of the DEC "Marketing" strategy: 1) DEC decided it could behave like IBM in the late 60's, come in late with the !!!NAME!!! and take over without regard to technical merits. If one believes this, one can then accept many limits on the product to ensure it won't compete with existing products. DEC's emphasis on "infinite PDP-11 software" implies it believed this to some extent. 2) The group doing the Professional fouled up. Note that they did a lot of things right (various hardware engineers I know swoon at the thought of the PRO's keyboard, packaging, connecting cables, etc.) However, they DEC marketing types were trying for a new market, one that wouldn't accept (they thought) the RSX-11 interface. To keep control of software (and make it easier to fit) DEC assumed you would use a big RSX-11M system to develop software for the PROs. If this means that the "cottage software" houses couldn't buy one and develop software for it, so what? This was going to be a PROFESSIONAL product. In practice, P/OS came in very slow since PDP-11's don't really have the address space for graphical monitors and it had to use slow floppies and the 150mSec access time Winchesters some huckster foisted on them. Everyone at their OEMs (like us) asked "How do you escape to RSX?" and lost interest when told we couldn't. At this point, the PRO has got a bad reputation. DEC can still rescue it by beefing it up. Notably, they have an 11/70 chip that will probably show up running UNIX at some point. Soon enough? Well, maybe. Personally, I have no intention of buying any machine with 16 bit registers. ======================= >From mcnc!burl!spanky!hocda!ihnp4!uw-beaver!microsof!fluke!kurt Sat Jul 16 16:52:45 1983 remote from tucc The operating system P/OS is really RSX/11 with a menu driven command interpreter. You can get at RSX if you REALLY want to. If you are clever enough, you might even get standard RSX-11 programs to run on the 300. You ought to know by now that DEC always soaks the customer for much more than the general market rate on everything from hardware to software to service contracts. They can do it because, bad as it is, IBM was even worse on mainframes. That may soon change as micros take over. ===================== >From duke!decvax!genrad!mit-eddie!nessus Tue Jul 12 06:01:38 1983 remote from tucc %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % In short, let me pose the question that many people are probably % % asking themselves: Why should I buy a DEC Professional (if I can get % % one!), when I can buy an IBM-PC for half the price, and have access to % % twice as much software? % % ... % % A question for people who actually *have* a Professional: Do you like % % it? Do you think there is adequate software from DEC? Are you using % % the so-called Tool-Kit? Is there a free-standing C compiler yet? Should % % this article go in net.flame? % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% We have several PRO350s. We did not buy them, though. They were given to us by DEC. I'd rather have a PRO350 than an IBM-PC, if for no other reason than an LSI-11/23 is a much nicer/faster processor than an 8088. I don't really think, though, that a PRO350 is worth the price, and would rather have a nice 68000 system. A really bad feature of the PRO350 is its floppy disk drives. They are slow, noisy, do not cram a whole lot of data on a disk, and worst of all, they only work with specially formatted disks that you must buy from DEC (out outrageously inflated prices). The software that DEC gave us (P/OS) is pretty sorry, but we don't worry about that all that much because we are running Venix (a Unix for PDP-11s supported by Venturcom). Venix is of course much much nicer than P/OS. It has decent editors (e.g. The Final Word), troff, nroff, a C compiler, etc. ===================== >From tucc!jerryr Sat Jul 9 00:58:59 1983 Dave, After having read your note about the 350, I'd thought you'd be interested to know that the 350 communications package does not work at 7-bit, even parity, and 1 stop bit. I am using a loaned demo model now but at 8-no parity. 7-bit odd is not working either. I would assume that the hardware is defective except that the DECmate II we had for demo had the same problems. Not only do the two packages not work with parity enabled but the cursor control keys do not work as a VT100's (in fact not at all). We have a Dataswitch which essentially autobauds and autoparitys and the request prompt does not display properly with parity enabled. The Dataswitch also does not recognize anything typed in even though it is a valid request. Since 8-bit no parity works fine, I'm becoming more convinced of a design flaw. (Note: the Rainbow 100 does not do this, but that VT102 emulation is in ROM, essentially taken directly from the VT102 since it is a Z80 based terminal.) I haven't tested the Rainbow software communications package, but can next week. If and when your 350 arrives please let me know if this bug appears. Thanks in advance.
BILLW@sri-kl@sri-unix.UUCP (07/28/83)
Somebody in a recent message said something like "Id rather have
a DEC 350 than an IBM PC because the 11/23 is nicer/faster than
an 8088". I don't beleive this to be a true statement. True
the 11/23 has a much more elegant architechture, BUT:
a) I overheard a conversation at the airport in St Louis
(just before DECUS) to the effect that the 11/23 is
actually about the same speed as a 4 MHz Z80 processor
[with the advantages of greater memory addressing capability,
of course, plus the availability of a floating point chip]
b) I have a detailed spec sheet of the DEC T11 chip (thats
the 11/23 on a single chip - supposed to be about the same
speed as a /23) that includes instructin timing information:
T11 (7.5 Mhz)
ADD R0,R1 1.6 uS
ADD R0,(R1) 2.8 uS
ADD #n,R1 2.4 uS
8088 (5Mhz)
ADD BX,CX 0.6 uS
ADD (SI),BX 5.8 uS
ADD BX,n 1.6 uS
Notes: the 8088 suffers from its 8 bit bus. Memory operands that are
only 8 bits speeds things up, while it doesnt effect the T11
times. Also the 8088 being weird, times can depend on exactly
which registers you use and the like.
8088 times do not include instruction fetch times. T11 times
do. 8088 has 4 byte instruction prefetch, 11s dont.
8088s have no memory protection or mapping.
11s restrict single process addressing to 64K.
All things considered, I suspect that the 8088 and the 11/23 are
actually about the same speed. If you could really run all the
existing 11 software on the 350, it would be a clear winner. You
can't though, and an ugly architecture with available software beats
an elegant architecture with little software.
BillWmel@houxm.UUCP (07/30/83)
I agree with BILLW in his assessment of the PC350 vs 8088. Not only is the 8088 likely to be faster, but there is a lot more useable software available for the 8088. That is why I got a Rainbow 100 rather than a PC 350. But the real surprise is that the Z-80 in the Rainbow is faster yet for all the things one does with a PC. Perhaps bigger isn't better. Does anyone have timing figures on the PC 350 vs the Rainbow for real- world PC applications? or a source of any PC-like software for the 350 ? or the Rainbow ? I sure wish I had bought a Kaypro or a Marrow. Mel Haas , houxm!mel
ron@brl-bmd@sri-unix.UUCP (08/03/83)
From: Ron Natalie <ron@brl-bmd> The T11 is not the same at all as the 11/23. Do not use such in your comparisons. The T11 is a single chip old style LSI 11. -Ron