gnu@hoptoad.uucp (John Gilmore) (10/02/88)
(I would've crossposted to comp.parallel, but it's moderated, and crossposts
to moderated groups do not work. Their tough luck, I guess.)
I was in the Sunnyvale Patent Library this week and looked up recent
patents issued in general purpose computers (class 364, subclass 200).
Unfortunately the library stopped sorting their patents by topic,
starting Jan 1988, so I have no data from after then.
Transputer 4,704,678 3 Nov 87 Inmos
"Function Set for a Microcomputer", Michael D. May
Motorola DSP? 4,709,324 24 Nov 87 Motorola
"Data Processor Control Unit Having an Interrupt Service using
Instruction Prefetch Redirection", Kevin L. Kloker
This describes how to get low-latency interrupts in a pipelined
machine, without losing cycles, by "instruction jamming" a couple
of instructions into the prefetch stream when an interrupt occurs,
without flushing the pipeline or changing the program counter.
This makes things like CPU-mediated DMA work great, since an
interrupt can take e.g. 1 cycle total.
Connection Machine 4,709,327 24 Nov 87 (individuals)
"Parallel Processor / Memory Circuit", W. Daniel Hillis et al.
This looks like the meat of the connection machine design.
It probably has lots of stuff that up til the patent was considered
proprietary. Have at it!
MC68010 or 20 4,710,866 1 Dec 87 Motorola
"Method and Apparatus for Validating Prefetched Instruction",
John Zolnowski, Les Crudele, Michael E. Spak
This is how they cope with page faults on prefetched instructions.
The 680xx avoids a bus error on prefetch unless you actually use
the instruction. They also seem to avoid starting to execute an
instruction until the whole thing has been prefetched.
Spectrum? 4,713,755 15 Dec 87 HP
"Cache Memory Consistency Control with Explicit Software Instructions",
W. S. Worley, Jr, Wm. Bryg, Allen Baum
XT/370 4,714,993 22 Dec 87 IBM
"Apparatus and Method for Effecting Dynamic Address Translation
in a Microprocessor Implemented Data Processing System", David L.
Livingston, Daniel J. Sucher, Brue M. Walk.
This one describes the DAT implementation on the IBM XT/370,
a "370 on an IBM PC card". The XT/370 is implemented with a
remicrocoded 68000, remicrocoded float chip, and a regular 68000
to handle odd instructions and talking to the IBM PC. This
patent includes a COMPLETE SOURCE CODE listing for the prom
code the support 68000 runs in assisting the emulation.
--
John Gilmore {sun,pacbell,uunet,pyramid,amdahl}!hoptoad!gnu gnu@toad.comcasey@admin.cognet.ucla.edu (Casey Leedom) (10/03/88)
In article <5511@hoptoad.uucp> John Gilmore (gnu@hoptoad.uucp) writes
about some of the interesting U.S. patents that were granted in 1987.
Some of the patents were very fundamental in nature, addressing points
such as ``how to reduce interrupt latency to 1 cycle'' (Motorola, patent
4,709,324, 24 Nov 87).
Not being a lawyer, I don't understand what the exact limits of patents
are, but I would guess that if one were to want to incorporate some of
these ideas into a new processor design, you'd have to negotiate a
contract with the patent holder and then pay royalties. On the surface I
suppose this sounds great, and justly paying others for their creative
work, but it strikes me that we're at a point now where there are a hell
of a lot of good ideas out there (most of which are patented by IBM or
soon will be) and one would be forced to negotiate separate contracts for
each one that you chose to use in your processor (for example).
Moreover, if a patent holder decided not to allow others to use their
technology, that would give them a strangle hold on a possibly
fundamental idea forcing all others to use a substandard approach.
(Again, not being a lawyer, I don't know whether the anti-trust laws
would apply in this case).
For example (and I should point out that I don't know about the factual
nature of this, but it still makes a good example), I once heard that DEC
patented the idea of using a general register for the CPU's Program
Counter. This allowed them to design a very clean orthogonal instruction
set without special purpose instructions to access PC relative
quantities. I heard further that they refused to license that
``technology'' out to any third parties and as a result all other CPU's
are forced to have the PC set aside as a special register, not accessible
in the same way as their general registers, etc. As I said, I'm not sure
of the factual nature of this story, so I apologise in advance for any
mis-information.
But the story demonstrates the point I'm trying to make, which is in
fact simply a question:
Do patents stifle progress?
This is not at all a clear question for me. I won't jump on the Richard
Stallman bandwagon, but on the other hand, I also won't jump on the (pick
your favorite patent pusher) Bandwagon. My feelings are very ambivalent
about this. On the one hand I do believe that people should be paid for
their creativity, but on the other, I don't want to see progress stifled
by a greedy individual.
Note that since corporations are considered to be people, the above is
intended to cover them. And that brings up yet one more card in the
question game: many patents are granted to corporations for the work of
individuals employeed by the corporations. Does the fact that someone is
employeed by a corporation justify the corporation claiming all creative
work of the individual? Certainly I'm less worried about my feeling that
individuals should be paid for their creative work if in fact the
individual at hand is merely a corporation that has completely usurped a
person's creative work ...
In any case, this is hardly a question for comp.arch, but since it was
inspired by a comp.arch article and since I can't figure out what the
legal questions group is, I'm posting it here ... Feel free to flame me,
but I do feel it's an important topic and am personally concerned with
the computer technology aspects of the question.
Caseyjohnl@ima.ima.isc.com (John R. Levine) (10/03/88)
In article <16406@shemp.CS.UCLA.EDU> casey@cs.ucla.edu (Casey Leedom) writes: > Do patents stifle progress? Yes and no. The theory behind patents is that you have to disclose your invention, but then you get exclusive rights to it for 17 years. After that, anybody can use it. This gets inventions disclosed rather than being held forever as trade secrets, so further work can be built upon them. Patent protection is much stronger than copyright or trade secret, because nobody can use your patent without your permission even if he reinvents your thing completely independently. (Comments in the trade press suggest that a lot of people in the computer industry who should know better seem to think that independent reinvention is a defense, which should give rise to some interesting lawsuits.) Admittedly, the patent office and the related courts work at a pace better suited to the technical climate of the 1780s than to the 1980s, so that in many cases by the time a patent is granted it seems obsolete. On the other hand, lots of patents work the way they're supposed to -- the entire computer industry in the 1950s was built on (and paid license fees under) the technology in the Eniac patents. -- John R. Levine, IECC, PO Box 349, Cambridge MA 02238-0349, +1 617 492 3869 { bbn | think | decvax | harvard | yale }!ima!johnl, Levine@YALE.something Rome fell, Babylon fell, Scarsdale will have its turn. -G. B. Shaw
darin@nova.laic.uucp (Darin Johnson) (10/04/88)
In article <16406@shemp.CS.UCLA.EDU> casey@cs.ucla.edu (Casey Leedom) writes: > But the story demonstrates the point I'm trying to make, which is in >fact simply a question: > > Do patents stifle progress? Originally, patents were designed to promote progress, rather than stifle it. Without patents, you would risk a lot to publish any invention, since anyone could legally copy your idea. In fact, if you start selling your invention, people can copy you to their hearts content. In order to promote exchange of ideas, patents were setup to allow inventions to be published, but with a certain time period during which only the inventor (and licensed parties) could market the inventions. The time period involved was long enough for most patent holders to develop their inventions and get a foot into door of the market before the rest of the world is allowed to copy their ideas (but not so long that the patent holder gets a virtual monopoly). However, in the computer industry, the length of patents (I don't remember what it is, 7 or 15 years perhaps) is so long, that the technology is pretty much out of date by the time the patent expires. This seems to run counter to the original purpose of patents (if my description above is accurate). Comments? Darin Johnson (...pyramid.arpa!leadsv!laic!darin) (...ucbvax!sun!sunncal!leadsv!laic!darin) "All aboard the DOOMED express!"
root@sbcs.sunysb.edu (root) (10/04/88)
In article <352@laic.UUCP>, darin@nova.laic.uucp (Darin Johnson) writes: > In article <16406@shemp.CS.UCLA.EDU> casey@cs.ucla.edu (Casey Leedom) writes: > > But the story demonstrates the point I'm trying to make, which is in > >fact simply a question: > > > > Do patents stifle progress? Yes. I look at it from this perspective: patents create an unfair monopoly (on sometimes trivial ideas, eg the Sun MMU Ras/Cas timing). When patents are awarded for such concepts they serve only to stifle competition. I suppose that if the patent laws enforced some concept of fair use such that people were guaranteed to be able to license a particular patent at a government established rate (based perhaps on projected volume of sales), they might not prove so objectionable. > Comments? I would modify computer patents laws to much shorter terms, eg 5 years, and enforce fair use. If a concept is still in use after 5 years then it probably has become a fundamental technique in the field and should not continue to line the inventors pocketbook. After all I would surely hate to be paying the estate/descendents of Fourier, Steinmetz, etc everytime I directly applied one of their methods/techniques, etc :-) > > Darin Johnson (...pyramid.arpa!leadsv!laic!darin) > (...ucbvax!sun!sunncal!leadsv!laic!darin) > "All aboard the DOOMED express!" Rick Spanbauer SUNY/Stony Brook
henry@utzoo.uucp (Henry Spencer) (10/07/88)
In article <1663@sbcs.sunysb.edu> root@sbcs.sunysb.edu (root) writes: > ... patents create an unfair > monopoly (on sometimes trivial ideas, eg the Sun MMU Ras/Cas > timing). When patents are awarded for such concepts they serve > only to stifle competition... The patent laws already forbid patents on obvious ideas; it is not at all clear that the Ras/Cas patent would hold up to a serious challenge. One should beware, however, of ideas that are "obvious" only after somebody else has pointed them out -- that's not quite the same thing. > I suppose that if the patent laws > enforced some concept of fair use such that people were guaranteed > to be able to license a particular patent at a government established > rate... While I tend to agree that mandatory licensing would be a good idea, I think it's a mistake to create a government bureaucracy to set the rates. In an ideal world that might work well; in this one it won't. There's got to be a better way. > I would modify computer patents laws to much shorter terms, eg 5 > years... That sounds sensible, provided it is applied specifically to fast-moving fields. In other areas -- drugs, for example -- it can take longer than that just to get government approval to sell the stuff. -- The meek can have the Earth; | Henry Spencer at U of Toronto Zoology the rest of us have other plans.|uunet!attcan!utzoo!henry henry@zoo.toronto.edu
baum@Apple.COM (Allen J. Baum) (10/08/88)
[] >In article <>henry@utzoo.uucp (Henry Spencer) writes: >In article <1663@sbcs.sunysb.edu> root@sbcs.sunysb.edu (root) writes: >> I suppose that if the patent laws >> enforced some concept of fair use such that people were guaranteed >> to be able to license a particular patent at a government established >> rate... > >While I tend to agree that mandatory licensing would be a good idea, I >think it's a mistake to create a government bureaucracy to set the rates. >In an ideal world that might work well; in this one it won't. There's >got to be a better way. There are several countries that have this kind of policy:e.g. Korea & Brazil. If you want to patent something there, you are required to license it at some low price that the govt. sets. Needless to say, there are a lot of companies that won't file patents in those countries because it is tantamount to giving them away. Now you know why Brazil produces some many illegal clones. -- {decwrl,hplabs}!nsc!baum@apple.com (408)973-3385
ghn@munsell.UUCP (Gary Newman) (10/08/88)
In reply to the comments about computer patents being out of date by the time they expire, note that the original paging patent dates to 1965. This expired in 1983, yet paging is the mainstay of memory management today.
mslater@cup.portal.com (10/09/88)
Can someone post the patent number of this RAS/CAS Sun patent being discussed? Michael Slater, Microprocessor Report mslater@cup.portal.com
dtynan@sultra.UUCP (Der Tynan) (10/11/88)
With all this talk of a Sun RAS/CAS pseudo-patent, my curiosity is piqued.
Would someone care to give a description of the patent? I mean, what *is*
it that's patented. I think it's fair to say that anyone interested in
this already understands the basic RAS/CAS operation. All we need is a
brief summary. Thanks...
- Der
--
Reply: dtynan@sultra.UUCP (Der Tynan @ Tynan Computers)
{mips,pyramid}!sultra!dtynan
Cast a cold eye on life, on death. Horseman, pass by... [WBY]henry@utzoo.uucp (Henry Spencer) (10/13/88)
In article <2567@sultra.UUCP> dtynan@sultra.UUCP (Der Tynan) writes: >With all this talk of a Sun RAS/CAS pseudo-patent, my curiosity is piqued. >Would someone care to give a description of the patent? ... The basic notion is, since the RAMs want the address in two parts, one at a time, if you choose things like page size carefully, you can hand the RAMs the untranslated low-order bits first, and run the high-order bits through the MMU while the RAMs are contemplating the low-order bits. The result is that the MMU imposes no time penalty, because you have to wait until the RAMs are ready for the second half of the address anyway. Of course, with modern RAMs this does mean that your MMU has to be pretty damned fast. It also, unfortunately, means that your pages have to be pretty large, since half the chip-level address has to go through without translation. (For example, with 1Mb RAMs you need 10 untranslated bits, and if you address to the byte and have 32-bit memory, two more bits go for byte number, so you need 12 bits of address within a page and your minimum page size is 4KB. This gets worse if you want to plan for 4Mb RAMs or for a faster-but-compatible machine with 64-bit-wide memory.) If you've ever wondered why the pages on the Suns are so big, now you know... -- The meek can have the Earth; | Henry Spencer at U of Toronto Zoology the rest of us have other plans.|uunet!attcan!utzoo!henry henry@zoo.toronto.edu
kuan@cpsvax.cps.msu.edu (Yih-pyng Kuan) (10/29/88)
Can anybody tell me the E-mail address of Dr. Dobry, T.P. in Berkeley ? Thanks in advance. kuan@cpsvax.msu.edu