chaim@nsc.nsc.com (Chaim Bendelac) (05/29/90)
This message was prompted by the following recent threads: 32016 etc, Micro- processors and market forces, Bubble Memory (was Re: 32016 etc.). I quote (strongly condensed): mash@mips.COM (John Mashey) quotes Dave Mason: > > mason@tmsoft.uucp (Dave Mason) in his response to Henry Spencer: > > henry@utzoo.uucp (Henry Spencer): > H "It wasn't for want of trying; a lot of people lost their shirts. > H The successful ones can be counted on your thumbs." > > D "That's how you do it Henry, you have extra hands :-) I count at least > D Encore, Sequent, Whitechapel,... in the Unix market. 32000s are also > D very popular in Laserprinters". And says: > J "I'm interested in the second part: could somebody give some examples > J of laser printers that have 32000's in them? I don't remember seeing > J the 32000 mentioned" In spite of its elegant architecture, the Series 32000 family was mostly unsuccessful in its attempt to carve out part of the "general purpose" market. It is no fun to be third. However, the Series 32000 is not dead. Far from it. When we realized that we had a very slim chance to succeed in the limited workstation market we looked around, and repositioned ourselves into what we call the "desktop imaging" market. The classic example of this market is the laser-printer; other examples include various (integrated) office-peripherals, such as servers, fax-machines, scanners, voice-mail equipment, and combinations of the above. I am not a marketing person, but even I can see that the potential in this market is better. And when we started, not as crowded too :-). The "desktop imaging" market is characterized by high-volume, low-cost (that's the bad part), high-performance computing requirements. To hit the right balance, the original Series 32000 processors were redesigned, resulting in the Series 32000/EP family of Embedded Processors. The first member, the NS32CG16, was introduced in 1988. On top of the 32000 architecture (binary compatible), the CG16 adds some 18 imaging instructions (variations of bitblt and other bit operations). It is probably no secret that our largest customer for the 32CG16 is Canon, whose Mark III/LBP-8 and LBP-4 laser printers use it (I am told that these are now the most popular laser printers in Europe). Other announced customers include Facit, Ollivetti, Everex (which sells printers under various additional labels), and Mannesmann-Tally. We have many more design-wins, many which I cannot announce. (This market is undoubtly much less glamorous than the WS market). To further answer "alvitar@xavax.com (Phillip Harbison): > I'm sure National would like to THINK it is popular in things like > laser printers, but I trust John's analysis more than NSC > marketing hype. Laser printer manufacturers certainly can't be > too pleased with the 32CG16' performance. I'd imagine Postscript > would be slower than usual. :-)" The 32CG16 is not an R3000. But it fits the market, in terms of cost/ performance. In fact, at 15 MHz running PostScript, it is equivalent to the 8PPM Apple LaserWriter NTX, which sports a 16.6 Mhz 68020. At a fraction of the cost, I may add. All major PS vendors/clones have announced products around the 32CG16. The second member of the Series 32000/EP, the NS32GX32, was introduced in 1989. Since it is derived from the NS32532, the 32GX32 is in a higher performance bracket. It is about 6-8 times the Apple NTX in performance. Alps-America has introduced a 20 PPM 32GX32-based PostScript printer; there are several more GX32 design wins, with at least one color PostScript printer that I am aware of. Now after two years, with an established laser printer market, we are about to expand our focus to the entire office peripheral market. To that end we proudly introduced last week three more members in the Series 32000/EP family, the 32CG160, the 32FX16, and the 32GX320. The 32CG160 is a more integrated version of the 32CG16 (DMA, counters, ext bitblt, interrupt control, etc). It runs at 25 Mhz. (And yes, we have learned our lessons: samples are available NOW, full spec. We have PS running on it, dev boards and tools released). The CG160 simply provides a higher performance/cost ratio, and is in that sense an evolutionary product over the CG16. The 32FX16 is our new flagship: our first imaging signal processor. On top of the CG16's core, we have implemented a DSP block (incl on-chip memory for DSP coefficients). This unique combination (of a general 32-bit processor with DSP capability) makes the FX16 (and its new big brother, the GX320), the only embedded processors to have enough computational power to implement high speed fax transmission in SOFTWARE alone (all in C!). We demonstrated this capability at our product launch. As you may know, currently existing fax products usually have at least two processors: one dedicated fax-chip (Rockwell, Yamaha, etc), and another one to do control, compression/decompression, error correction, grey-scales and other stuff. The FX16 can do all of the above, and PostScript capability thrown in (think about PostScript/plain-paper fax/printer products). The FX16 and the GX320 are revolutionary products. We have several designs in progress, some with *very* big companies. The FX16 is available in samples (has been since January), runs at a full 25 MHz. The 32GX320 is our new top-of-the-line. It is based on the 32GX32 core and has a re-designed hardware multiplier plus 4 new (complex mul and complex mul-accumulate)in addition to other system integration functions (DMA, counter/timers, and interrupt control). It is targeted to fax/printer server types of products. It is offered at 20, 25 and 30 Mhz. Samples available. I am hesitant to make this posting any longer than it already is, by adding more technical details. If there is interest, I will be happy to oblige :-). To summarize: The Series 32000 no longer competes with the likes of sparc, mips and the i860. As such, we have disappeared from the day-to-day spotlights. But our technology has matured, we have refocused and we are doing well. --------------------------------------------------------------------------- Footnote: On a regular basis, whenever a message in this group mentions National's Series 32000, it is followed by a predictable stream of (sometimes uninformed, often irrelevant) bashing/flaming. A recent thread that started with a question about the 32000's limited success in the workstation/Unix market soon turned into reminiscences of De Gaulle's actions in the fifties, Hitachi-scandals, sewage treatment, the number of thumbs on Henry's hands, a story about a bozo NS salesman and Intel's bubble memories. I will not counter the various claims that were made here on "companies that went out of business" because of the 32000 or of National. Frankly, whatever I say, some will continue to stick to their beliefs. Several readers have pointed out corrections regarding Sequent, Siemens, Encore, TI, WhiteChapel, etc. Please folks, try to stick to facts, not rumours! --- chaim@nsc.nsc.com (Chaim Bendelac; National Semiconductor)
mash@mips.COM (John Mashey) (05/29/90)
In article <14013@nsc.nsc.com> chaim@nsc.nsc.com (Chaim Bendelac) writes: >mash@mips.COM (John Mashey) quotes Dave Mason: >And says: >> J "I'm interested in the second part: could somebody give some examples >> J of laser printers that have 32000's in them? I don't remember seeing >> J the 32000 mentioned" <lots of good information; thanx to Chaim> >To further answer "alvitar@xavax.com (Phillip Harbison): >> I'm sure National would like to THINK it is popular in things like >> laser printers, but I trust John's analysis more than NSC >> marketing hype..... Just to clarify, recall that my assertion was that I didn't see the 32000 much in the lists I had, but there were reasons why it could well be popular and still not show up at that particular show. I.e., my assertion was that I hadn't seen data that supported the 32K, not that I had data that showed it didn't exist.... In any case, this actually appears to be an interesting strategy for NSC, and I've heard well of their recent presentation on the new products. -- -john mashey DISCLAIMER: <generic disclaimer, I speak for me only, etc> UUCP: mash@mips.com OR {ames,decwrl,prls,pyramid}!mips!mash DDD: 408-524-7015, 524-8253 or (main number) 408-720-1700 USPS: MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086
alvitar@xavax.com (Phillip Harbison) (06/02/90)
In article <14013@nsc.nsc.com> chaim@nsc.nsc.com (Chaim Bendelac) writes: > Footnote: > On a regular basis, whenever a message in this group mentions National's > Series 32000, it is followed by a predictable stream of (sometimes > uninformed, often irrelevant) bashing/flaming. A recent thread ... > soon turned into ... a story about a bozo NS salesman and Intel's > bubble memories. ... > Please folks, try to stick to facts, not rumours! I would suggest that Mr. Bendelac check with people via mail before accusing them of dealing in rumor. In the case of the three stories I have related, I have dealt in fact, not rumor. These are my sources, given for your examination. [1] The story of the bozo salesman is firsthand. I was the engineer present in the meeting when the salesman said NSC was dropping the RGP and suggested we use the 32CG16 instead. [2] The person who got burned by National on bubble memories worked for Itec (a Huntsville manufacturer of telecom equipment) at the time. We worked together at a company he started in Huntsville. [3] The company that was burned on the 32016's long bug list and poor performance was Intergraph, a Huntsville manufacturer of worksta- tions. I know several people who work there, and I was told that at one point, the 32016 required about 5 PALs to make it behave. I'm sure that Intergraph would rather not admit that the Interpro 32 (based on the 32xxx) was a failure, since no company likes to admit failure. However, the fact remains that production was hin- dered by the bug list, customers were not satisfied with performance, and the Interpro 32C is based on the Clipper rather than the 32xxx. I now kindly suggest that you retract your accusation that I've dealt in rumors. In my humble opinion, discrediting someone's character is just as serious as discrediting a companies performance. -- Live: Phil Harbison, Xavax, P.O. Box 7413, Huntsville, AL 35807 Uucp: alvitar@xavax.com Bell: 205-539-1672, 205-880-8951