decoration decoration
Stories

GROKLAW
When you want to know more...
decoration
For layout only
Home
Archives
Site Map
Search
About Groklaw
Awards
Legal Research
Timelines
ApplevSamsung
ApplevSamsung p.2
ArchiveExplorer
Autozone
Bilski
Cases
Cast: Lawyers
Comes v. MS
Contracts/Documents
Courts
DRM
Gordon v MS
GPL
Grokdoc
HTML How To
IPI v RH
IV v. Google
Legal Docs
Lodsys
MS Litigations
MSvB&N
News Picks
Novell v. MS
Novell-MS Deal
ODF/OOXML
OOXML Appeals
OraclevGoogle
Patents
ProjectMonterey
Psystar
Quote Database
Red Hat v SCO
Salus Book
SCEA v Hotz
SCO Appeals
SCO Bankruptcy
SCO Financials
SCO Overview
SCO v IBM
SCO v Novell
SCO:Soup2Nuts
SCOsource
Sean Daly
Software Patents
Switch to Linux
Transcripts
Unix Books

Gear

Groklaw Gear

Click here to send an email to the editor of this weblog.


You won't find me on Facebook


Donate

Donate Paypal


No Legal Advice

The information on Groklaw is not intended to constitute legal advice. While Mark is a lawyer and he has asked other lawyers and law students to contribute articles, all of these articles are offered to help educate, not to provide specific legal advice. They are not your lawyers.

Here's Groklaw's comments policy.


What's New

STORIES
No new stories

COMMENTS last 48 hrs
No new comments


Sponsors

Hosting:
hosted by ibiblio

On servers donated to ibiblio by AMD.

Webmaster
IBM's Unsealed Sur-Reply Memo in Opposition to SCO's 3rd Amended Complaint
Saturday, May 07 2005 @ 11:58 PM EDT

Here's another of IBM's newly released unsealed documents, IBM's Unsealed Sur-Reply Memorandum in Further Opposition to SCO's Motion for Leave to File a Third Amended Complaint [Docket 407] [PDF], and it gives us a peek, even before the transcript of the hearing is released, what the main IBM arguments likely were that day.

There are so many cases submitted by IBM to support its arguments listed at the beginning, IBM doesn't even start to argue until page 7. As you may remember, IBM had to file a motion to be able to do a sur-reply, and got it, and then had to seek leave to file an overlength sur-reply [PDF], and got it [PDF]. The reason they asked for a chance to do a sur-reply is because SCO brought up new issues in its Reply that IBM had had no opportunity to rebut. That's like hitting below the belt, so the court granted IBM its chance to speak.

Normally it goes like this: someone submits a motion, the other side files a memorandum in opposition, and then the submitter gets to reply. Then the judge rules on the motion. Here, SCO filed a motion for leave to file a 3rd amended complaint (its memorandum in support of the motion was filed under seal and is still sealed, number 323 on the list), IBM opposed the motion (sealed, number 337 on the docket), and SCO then filed its reply (sealed still, number 369 on the docket). Because new issues were raised by SCO in the reply, IBM got another chance to speak, and here it answers the new issues raised. So, that is the context, and now let's break it down into the essential arguments.

Their preliminary statement gives us the overview of their position:

SCO seeks improperly to extend and expand this case just as it should be coming to a close. SCO asks the Court to add a complicated, new claim that cannot (by contract) be brought in this Court and that SCO could seek to file in another forum without causing much of the prejudice threated by the present motion. SCO's motion to amend should be denied for two primary reasons.

First, SCO's motion should be denied because SCO cannot establish good cause under Rule 16 of the Federal Rules of Civil Procedure for its failure to make this motion before the deadline for amending the pleadings. SCO's reply papers dismiss the documentary evidence showing that it knew of the proposed claim years ago on the grounds that it is unclear or ambiguous, and on the grounds that SCO knew nothing of the evidence in any event. As we show below, SCO's papers distort the record. There is no reasonable doubt that SCO knew of the proposed claim years ago. Moreover, even if the documentary evidence were ambiguous and SCO did not actually know of the proposed claim, it clearly should have known of the claim. At a minimum, SCO was on inquiry notice, based upon the public record and the documents in SCO's own files, of a claim it failed to investigate. Having failed to investigate the proposed claim. SCO cannot establish good cause to bring it at this late date in this litigation. . . .

Second, SCO's motion should be denied under Rule 15 of the Federal Rules of Civil Procedure because (1) it is unduly delayed; (2) adding the proposed claim would be unduly prejudicial to IBM; (3) denying the motion would result in no prejudice to SCO as it could seek to bring the claim in another forum; and (4) the proposed claim is futile insofar as it is barred by contractual venue and limitations provisions. SCO's reply papers raise new and contradictory arguments that, in the end, fail to support its motion for leave to amend....

While the proposed claim is demonstrably without merit, we do not dwell on the merits in deciding this motion. The primary issue here is whether SCO should be allowed to amend its complaint to add a claim that will complicate and delay an already complicated case, where that claim could be -- indeed, by contract, must be -- asserted in a New York forum and there would be no prejudice to SCO in requiring it to assert the claim there.

You can read Rule 16 of the FRCP here and Rule 15 is here. IBM is referring particularly to 16(b)'s pertinent language, "A schedule shall not be modified except upon a showing of good cause and by leave of the district judge or, when authorized by local rule, by a magistrate judge." All the Federal Rules of Civil Procedure are here, if you want to bookmark them. You will probably want to have the Project Monterey contract handy too, to follow along with the arguments. Now let's decipher what all IBM is saying. Their arguments boil down to this:

  • It's too late to bring it up now. SCO knew or should have known about the facts underlying their proposed copyright infringement claim many years ago, at least two years before it brought this action. A copyright claim "accrues when the alleged copyright holder 'knows or reasonably should have known of the infringement.'"

  • It's also too late with respect to the scheduling for this case, because the deadline to amend the pleadings has come and gone, and SCO has failed to show good cause under Rule 16 of the Federal Rules of Civil Procedure to amend. SCO is saying that all the evidence showing that they coulda, shoulda have had knowledge about SVR4 for Power was confusing or ambiguous. They only just learned about this copyright issue, they claim.

    Nonsense, says IBM. This is a software company, with sophisticated knowledge, the "alleged successor" of a partner in Project Monterey and "alleged owner of UNIX". Their predecessor in interest helped develop AIX 5L. IBM's submitted evidence showed clearly that it was known back then that AIX 5L operated on Power and contained SVR4 code. Most of the evidence IBM submitted, by the way, came from materials SCO itself turned over in discovery, they say, so how can they plead ambiguity?

    SCO has apparently submitted a statement by employee, Jay Petersen, saying he didn't know about it, but IBM counters that others at Santa Cruz surely did. IBM submitted evidence that Santa Cruz employee John Boland back in 2000 distributed internally an IBM press release mentioning AIX 5L for Power and adding that it included key "System V [SVR4]" technologies. Petersen, they point out, can't speak for the company as a whole, anyhow, only for himself. He claims his first awareness of AIX 5L on Power including SVR4 code was in reading IBM's opposition papers. In a footnote, IBM lists an IDC report as well, that also mentioned the same information, dated August 2000. How could SCO not know?

    I find the Peterson argument particularly difficult to swallow, simply because we know SCO reads Groklaw, and Groklaw wrote an article on AIX 5L on Power and about it including SVR4 back in July of 2003, when Groklaw was only two months old, and this information was repeated in August of 2004, with additional evidence Groklaw members collected. I did the July 2003 article when I found Santa Cruz had put up a web page in praise of AIX 5L, copyrighted 2000, that said this:

    AIX technology will help contribute to the success of the future systems running AIX 5L on Power and IA-64. . . .

    Among the UNIX System 5 technologies to be incorporated in this release is the SVR4 printing subsystem providing industry-standard print administration and drivers.

    Mr. Petersen may claim he doesn't personally read Groklaw and only reads the funny papers, not the trade press, so his first inkling was when he read IBM's opposition papers in this litigation, but we know SCO's CEO reads Groklaw, because of his public remarks about it and because the company acknowledged to the Washington Post's Jonathan Krim in June of 2004 that their lawyers used Groklaw as a resource. Frankly, I think Groklaw gave SCO notice back then, in 2003, certainly in August of 2004, after they said publicly they use Groklaw as a resource, even if they missed the earlier Groklaw article and all the Santa Cruz evidence IBM lists. So how do they explain missing the deadline to amend pleadings? (By the way, Krim has written a valuable article on patent "reform" you don't want to miss.)

  • SCO, today's SCO, has argued that they can't be charged with the knowledge of Santa Cruz, oldSCO. Wrong, as a matter of law, IBM argues, and cites three cases, all of which say the successor in interest is charged with the knowledge and actions of a predecessor in interest.

  • Even if they could bring it up now, they can't bring it up here in this litigation, because the contract says all disputes are to be handled in New York, under NY law.

  • SCO, even if it could win the ambiguity argument, had enough evidence from the public record and from their own files that it triggered a duty to investigate back then. Copyright holders do have a duty to investigate possible infringement when evidence of an alleged infringement comes to their knowledge. Failure to do so results in the copyright holder being "charged with knowledge of facts constituting infringement which could have been discovered in the exercise of reasonable diligence." It is a simple matter of fairness. Those of you who have been here with Groklaw a while will remember that in our first article about Judge Kimball years ago, we noted that he ruled against a copyright holder, represented by Brent Hatch, for waiting too long to assert a copyright. Having failed to exercise reasonable diligence, IBM says, SCO can't now excuse its delay in bringing the claim.

  • Allowing SCO to bring the claim in Utah is prejudicial to IBM, which negotiated terms and rights under the venue and limitations provisions of the Joint Development Agreement (JDA). SCO didn't even try to argue that its rights would be prejudiced if not allowed to bring the claim in Utah; IBM is arguing that its rights *would* be prejudiced by allowing them to. Where is there any harm to SCO if they are required to bring a new and separate action in New York, as stipulated in the contract?

  • SCO has tried to argue that IBM's Ninth Counterclaim encompasses their proposed copyright claim, but IBM has already answered that and has stipulated to the limited scope of its claim. Its counterclaim seeks only a declaration that IBM's continued distribution of AIX after SCO's "termination" does not infringe. It wasn't seeking any ruling about AIX on Power, because at the time the counterclaim was entered, nobody had said a word about that subject.

  • Allowing SCO to amend its complaint now will involve more time and more discovery, something SCO admits. The law, says IBM, is clear that a need for additional discovery can be sufficiently prejudicial to justify denial of a motion to amend. (You can understand why, if you think about what it could be like with a plaintiff like SCO, who could otherwise just keep coming up with new and additional claims 'til the end of time. Remember the 9-year-old Habie case?)

  • SCO has previously argued that IBM waived its right to enforce the venue clause by asserting its Ninth Counterclaim, and now instead says that IBM waived its right by failing to raise "the 'defense of improper venue' in its responsive pleading. Courts normally don't consider arguments that are raised in a reply brief for the first time or that reflect a change in position from that given in the opening brief, IBM notes, providing some cases to prove that point, and here SCO has done both. Anyway, IBM neatly slices through that argument by simply pointing out that they haven't filed their responsive pleading yet, because SCO has yet to file a 3rd Amended Complaint. It's still working on getting the right to file it. So IBM can hardly be argued to have waived anything in its responsive pleading. (This is all about the fact that if you are charged in a civil case with a claim, you have to raise all your defenses the first time you answer, or you waive them. SCO cited that rule, but it either forgot that IBM hasn't yet had a chance to file a response, or hoped IBM and the judge wouldn't notice. Nope. IBM noticed.)

    In any case, IBM did raise the defense of improper venue in its answer to SCO's original complaint, its amended answer to the complaint, its answer to SCO's amended complaint, and its answer to SCO's 2nd Amended Complaint.

  • The clause in question in the JDA governs "[a]ny legal or other action related to a breach of this Agreement." SCO and IBM differ as to whether the language encompasses a claim for copyright infringement. IBM argues it's broad enough to cover (see especially their footnote 12 on page 16), and SCO says it only covers breach of contract. That requires removing language from the clause, IBM says, or rewriting it. SCO, in its opening brief, framed their claim as a breach of contract-related claim, resulting in a copyright infringement, did it not? SCO said that "IBM converted SCO's copyrighted code for IBM's own use, in violation of the specific restrictions of the parties' Joint Development Agreement" and that IBM "ignored the [JDA's] restrictions on its use of SCO's SVR4 code" and that, IBM says, is breach of contract language, so that too brings it within the JDA's venue clause, even if that clause only applied to breach of contract. Again IBM cites multiple cases, including several where copyright claims were held to come within such forum selection clauses, one in particular where the ruling was that "analysis of the copyright infringement claims will inevitably require reference to rights and duties defined in the Agreement", as they will here ( Direct Mail Prod. Servs. Ltd. v. MBNA Corp).

  • Asserting a counterclaim doesn't preclude IBM from asserting a forum selection clause, but even if it did, it doesn't here, where the JDA specifically states that there can be no waiver of any provision of the agreement "unless it is set forth in a writing which refers to the provisions so affected and is signed by an authorized representative of each Party," and entering a counterclaim doesn't meet the requirements in that clause. Again, IBM presents multiple cases to the judge, including one that found that "the fact that [the defendant] filed a counterclaim and participated in discovery does not render its objections to venue abandoned," which is right on point. (Rogen v. Memry Corp., 886 F.SUpp. 393, 396 (S.D.N.Y 1995).)

    Several times in this document, IBM mentions that SCO has offered only one case on a certain point, including this one. If you read between the lines, I think you can see what happens when you cap legal fees. Legal research is one of the most expensive aspects of litigation, and when there is no money, it's the first thing that gets cut back.

    Here, the case SCO relied on, IBM says, talked about waiver only in dicta. That is a big NO. You don't cite dicta in a case. Dicta is, roughly explained, where the judge chats about stuff he isn't ruling on directly. It's not supposed to be used as precedent, as SCO apparently has here in its only case. See what I mean? See also IBM's footnote 20, where they massacre another SCO case.

  • As a "fallback position", SCO has argued that "the principle of continuing infringement" should govern their copyright claim, even though they admit that the doctrine has never been recognized in the Tenth Circuit. The JDA clearly sets a 2-year limitation period to bring a claim and doesn't recognize a "continuing infringement" exception.

On all those bases, IBM asks that the court deny SCO's motion for leave to file a 3rd Amended Complaint.

So what is the takeaway?

To me, it's the irony that SCO, who has piously claimed to be the innocent victim of a breach of contract, now tries like mad to escape the same contract's wording about venue and time-to-sue limitations, when it is in *its* own interest to do so.


  


IBM's Unsealed Sur-Reply Memo in Opposition to SCO's 3rd Amended Complaint | 189 comments | Create New Account
Comments belong to whoever posts them. Please notify us of inappropriate comments.
Corrections go here Please.
Authored by: Hiro Protagonist on Sunday, May 08 2005 @ 12:58 AM EDT
Corrections go here Please.

---
I Grok... Therefore... I am.

[ Reply to This | # ]

Off topic here please
Authored by: fudisbad on Sunday, May 08 2005 @ 12:59 AM EDT
For current events, legal filings and Caldera® collapses.

Please make links clickable.
Example: <a href="http://example.com">Click here</a>

---
See my bio for copyright details re: this post.
Darl McBride, show your evidence!

[ Reply to This | # ]

One point that needs to be clear
Authored by: Anonymous on Sunday, May 08 2005 @ 01:14 AM EDT
I think just in case any idiots are reading this (there are always a few), the
points raises about venue and timing (2 year limit), are NOT an admission that
SCO's claim has any basis or merit.

Indeed IBM points out that IBM's believes that SCO's underlying claim (whether
there is a copyright infringement in AIX5L by containing SVR4 code) does NOT
have any basis or merit, and is contradicted by the plain text of the JDA.

(while it's also true that the timing issue is an *additional* reason why SCO's
claim is without basis, that is not the central point here).

Rather, IBM is pointing out that the venue and timing issues, are additional
reasons not to allow SCO to raise the claim here in *this* case in Utah.

Furthermore, IBM invites SCO in this document (and several other documents
before and after), to raise the claim, if SCO wants to, in the correct venue,
i.e. New York

Quatermass
IANAL IMHO etc

[ Reply to This | # ]

Shall we revisit the laws of Agency again?
Authored by: Anonymous on Sunday, May 08 2005 @ 02:18 AM EDT
The Laws of Agency and Contract law are taught in Business Law 101. Consumers
have protections from Agents that are out of control. Companies can be held to
the agreements made by agents that are Actual, Apparent, or Ostensible. A deal
is a deal. And if by acting or failing to act, one acts in certain ways, then
you have the makings of a deal.

And the word Estoppel comes to mind as well.

SCO is asking for a JURY to decide these matters? Fair is Fair and a jury sees
that better than anyone. SCO has gotta be scratching their heads and looking at
the cap on legal fees and looking at starting over again in NY this time?
Ouch.

Anyway, this is just the type of waters where the IP law and Consumer law and
contract law collide. Which one takes the higher ground on shore above another?
This should be interesting? Any guesses?

[ Reply to This | # ]

Let me get this straight
Authored by: Anonymous on Sunday, May 08 2005 @ 03:34 AM EDT
SCO are arguing this:

Step 1: SCO say that sky is green.
Step 2: IBM say that the sky is blue.
Step 3: SCO say that horses eat coal, and that IBM can't refute that, because
they failed to do so in step 2.

Do they really, truly think that the judge and IBM will let that sort of thing
slide? I can almost smell the desperation from here.

[ Reply to This | # ]

Possible SCO Strategy?
Authored by: sproggit on Sunday, May 08 2005 @ 04:10 AM EDT
This latest from SCO is starting to seem somewhat transparent to me...

In the last few weeks we've seen them throw every possible obstacle, up to and including the kitchen sink, in the path of IBM's legal Team. They are, we could argue, bringing "nuscience motions" into this trial in an attempt to make it as unpleasant and expensive as possible for IBM. Witness the ludicrous discovery requests.

So what's with the need to include this latest amended complaint as a reply?

Simple - it's cost reduction on Darl's part. If we go back to the terms of the "cap of legal costs" that SCO reached in agreement with their own legal people, then I believe I recall correctly that SCO are covered for this litigation and appeal. They are not covered for any new claims or new cases, such as a new complaint filed in New York.

See, that pesky earlier contract they signed with IBM has wording in it which would force SCO to open up a new front in the IBM war - an attack they cannot afford to fund.

So this attempt at sneaking a new strike against IBM in under the RADAR is merely an attempt by SCO to twist the terms of their legal terms with BSF.

Now this is going to be interesting. BSF - especially Boies - are nobody's fools. If we get to a point where this legal Firm are convinced that SCOs willingess to "game the system" knows no bounds - right up to the point where the company starts to play fast-and-loose with their own counsel, then right around that point things might get interesting.

I've suspected for some time now that when Boies signed up for this case he did so on the mistaken belief that he had a chance to play "David" against IBMs "Goliath" and win. I suspect that ego had a lot to do with the decision to take the case. Where does that put BSF now? I'd argue it's got them in "loss of face" territory. They are wedged in a trap of SCOs own making, out-foxed by their own cunning.

They will continue to argue, litigate, squirm, to look for any avenue of escape, for any crumb of "success" that they can extract from this nightmare. IBM aren't going gracefully. IBM want total victory, because they have total confidence in their innocence in law. In other words, IBM won't give BSF an escape route, which is why that law firm are stuck with their client...

Interesting

[ Reply to This | # ]

Successor in Interest
Authored by: tangomike on Sunday, May 08 2005 @ 09:48 AM EDT
It's nice to see TSCOG get caught on this one. Groklaw commentary has repeatedly
noted that TSCOG's claims about succession are unsubstantiated. Here IBM uses
the claims against them. Another lose-lose for TSCOG.

---
Nothing screams 'poor workmanship' like wrinkles in the
duct tape.

[ Reply to This | # ]

Interesting TIdbit.
Authored by: rsteinmetz70112 on Sunday, May 08 2005 @ 11:14 AM EDT
One of the things about the JDA what has intrigued me since I first read it is
that the JDA refers to Project Supplements. They were to describe the
contributions of the parties but none of them have surfaced so far. They have
always seemed to me to be the key to what the parties thought they were getting
from each other.

IBM's sur-reply, for the first time, refers to Project Supplement. It is an
apparently unsigned Supplement C (Page 5 Footnote 5) which, I gather, SCOG
admits did apparently unequivocally grant IBM a SYS V r4 license.

I wonder what's in Supplements A and B and if there is a supplement D or later?
In Supplement C SCOG seems to have been willing to give the license, I wonder if
IBM actually needed to accept it? I wonder if the failure to sign Supplement C
is a serious defect? I wonder if it matters that the parties continued to act as
if it was valid or whether it was actually necessary to include the
contributions in a supplement for the license to be valid?

We may never know, because the Third Amended Complaint may not be allowed. If it
is not allowed SCOG may not have the money to bring the claim in NY, since as I
understand it a new suit would not be covered under the fee agreement. On The
Other Hand perhaps BSF committed a major error by not including this claim in
the original (or prior amended) complaint and SCOG may be able to make them
include this under the fee agreement.

"The world wonders ..." October 25, 1944

---
Rsteinmetz

"I could be wrong now, but I don't think so."
Randy Newman - The Title Theme from Monk

[ Reply to This | # ]

Extraordinary Writ Availability?
Authored by: Adam B on Sunday, May 08 2005 @ 11:38 AM EDT

This seems like a situation tailor-made for the use of an extraordinary writ. If I were IBM, I would be applying for a writ of prohibition against SCO raising its AIX of Power claims in this case. It's both very important to this case, very important to other cases, and it relates to a jurisdictional matter, which are frequently taken up on mandamus or prohibition.

But, IBM is not doing this. And IBMs lawyers are much smarter than me. So can someone explain to me exactly what way they are smarter than me? Why is it that we don't see the writ application?

[ Reply to This | # ]

Does SCO have standing to sue?
Authored by: RFD on Sunday, May 08 2005 @ 03:05 PM EDT
I notice that IBM has not conceded that "The SCO Group, Inc." is the
successor-in-interest to "The Santa Cruz Operation, Inc." IBM's
sur-reply refers to Santa Cruz as "SCO's alleged
predecessor-in-interest."

IBM is still not giving an inch, and may challenge SCO's standing to sue.

---
Eschew obfuscation.

[ Reply to This | # ]

moore's law, computer widths, and i86
Authored by: jeleinweber on Sunday, May 08 2005 @ 06:14 PM EDT
While it's true that microprocessors went from 4 to 8 to 16
to 32 and now 64 bits, as measured by the size of the principle registers, there
has been a lot more than that going on.

First, note that the big campaign for 32-bit processors started in the 1960's
with the IBM 360 line of mainframes. It is by no means a new phenomenon,
although on microprocessors it had to wait for transistor counts per chip to get
past 200,000 or so before it became practical. Historically mainframes and
minicomputers used lots of chips in parallel to get wider results, which was one
reason why they were so expensive.

Next, note that graphic chips are already much wider internally than general
purpose CPU chips. The graphics folks jumped to 128 or 256 bits internally a
few years ago already. That's because computations involving 24 or 32 bit color
times 3 channels plus a z-buffer are so much easier, when you can manipulate an
entire pixel as a single unit.

Even in the Intel CPU-compatible world, Transmeta's chips use 128 or 256 bit
instruction issue widths, which is wider than the principle registers.

Third, note that width isn't a single number in modern computer systems. Since
all modern processors have on-chip cache memory, one important width is the
cache line size for the L2 or L3 cache, which-ever one is doing transfers with
RAM memory. Cache line size is usually in the range of 4-32 bytes, which in the
presence of load pipelining, is usually much wider than the width of the memory
data bus. The memory address bus is often narrower than the data bus, as
physical memory sizes are usually smaller than virtual, with notable exceptions
such as DEC PDP 11's or Intel 8086's. The L2 to L1 cache bus is on-chip, and
not constrained by pin counts, so that may well be much wider than the physical
memory data bus, particularly since overall CPU performance is quite sensitive
to inter-cache latency. Any special purpose functional units on-chip may have
larger internal sizes, particularly floating point units, which spend a lot of
time working on quantities which are 2x the main register size, or more. A
processor with 64 bit integer registers might be using 140 bits or more at
places inside its floating point units.

In the midst of all these different widths, the typical size quoted for a
microprocessor chip is the width of the general purpose registers. CPU
designers pick that based on tradeoffs between addressing needs, floating point
arithmetic, and the transister budget.

Note that the original IBM PC used intel 8088 chips, which had an 8 bit memory
bus width (cheap), a 16 bit register size (typical), and 20 bit addressing
(new). At the time the IBM PC shipped, most computers had 64 Kibibytes or less
of memory, but that was starting to pinch. The fact that the IBM PC scaled past
that while maintaining binary compatibility was a key element in its success.
Today, lots of server computers have more than 4 Gibibytes of RAM, which makes
32-bit registers pinch. Fiddling segmentation registers or virtual memory
tables to get to objects bigger than your address offsets is a ruddy nuisance.

Scientific processing is mostly happy with 64-bit floating point
representations, and as with graphics, it's very convenient to manipulate those
as single entities. So high end scientific computers from the likes of CDC went
to 60 or 64 bit word sizes already in the 1970's. There isn't much pressure to
go past that.

Pin counts and on-chip data buses are still expensive, even as transistors get
cheaper, so now that Moore's law allows single chip designs of 64-bit CPU's, we
have probably hit the point of diminishing returns on increases in general
purpose register width. How to usefully spend transistors to buy increased
performance in the face of more or less static clock speeds and rapidly
increasing heat dissipation is the main challenge for current CPU designers.
The upshot is that it's possible that the reign of the 64-bit computers may last
longer than the 4 decade reign of their 32-bit forebears,

What made the IBM PC line so popular was a combination of:
relatively inexpensive implementation (due in part to the use of a chip with and
8 bit bus), breaking the 16-bit memory addressing barrier, and 80 character wide
monochrome graphics. The use of the magic 3 letters on the front was ironic, as
none of the reasons which made them traditionally appealing in the mainframe
market had any actual application to the PC's.

I'm not aware of any widely used CISC chips designed since the Intel 486. Even
the original Pentium chips from Intel were RISC designs under the hood, with an
abomination of an x86 instruction set decoder tacked on the front. Subsequent
designs from Intel, AMD, Cyrus, Via, and Transmeta have continued that design
theme. In general, microprocessors have been re-capitulating the history of
mainframe and scientific computers, adding features such as pipelining, virtual
memory, speculative execution, multiple functional units, and so on. Those were
well known by the mid-1970's.

The market for embedded computers is many times the PC market, and that is
almost entirely dominated by non-x86 chips. It's much easier to do a system on
a chip for a cell phone with decent battery life if you don't drag around all
the PC baggage with you. Which is why even Intel sells ARM design CPU's, too.

---
-- Jim Leinweber (Madison, WI)

[ Reply to This | # ]

PJ
Authored by: Anonymous on Monday, May 09 2005 @ 09:29 AM EDT

I commend you on your attitude, but if you just raise the flag, there are quite
a few of us who are just watching from the side lines who will lend a hand in
any way if you just ask. Keep up the good work for the good cause.

[ Reply to This | # ]

Groklaw © Copyright 2003-2013 Pamela Jones.
All trademarks and copyrights on this page are owned by their respective owners.
Comments are owned by the individual posters.

PJ's articles are licensed under a Creative Commons License. ( Details )