New Pacer--[CORRECTED] PLAINTIFFS’ CONSOLIDATED RESPONSIVE CLAIM CONSTRUCTION BRIEF

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION
ACER, INC., ACER AMERICA
CORPORATION and GATEWAY, INC.,
Plaintiffs,
v.
TECHNOLOGY PROPERTIES LIMITED,
PATRIOT SCIENTIFIC CORPORATION,
and ALLIACENSE LIMITED,
Defendants.
Case No. 3:08-cv-00877 JW
CORRECTED*
PLAINTIFFS’ CONSOLIDATED
RESPONSIVE CLAIM CONSTRUCTION
BRIEF
[RELATED CASES]
JURY TRIAL DEMANDED
Date: January 27, 2012
Time: 9:00 a.m.
Place: Courtroom 9, 9th Floor
Judge: Hon. James Ware
HTC CORPORATION, HTC AMERICA,
INC.,
Plaintiffs,
v.
TECHNOLOGY PROPERTIES LIMITED,
PATRIOT SCIENTIFIC CORPORATION,
and ALLIACENSE LIMITED,
Defendants.
BARCO N.V., a Belgian corporation,
Plaintiff,
v.
TECHNOLOGY PROPERTIES LTD.,
PATRIOT SCIENTIFIC CORP.,
ALLIACENSE LTD.,
Defendants.

TABLE OF CONTENTS

Page
Case Nos. 5:08-cv-00877, 5:08-cv-00882, 5:08-cv-05398 -i- [CORRECTED] PLAINTIFFS’ CONSOLIDATED
RESPONSIVE CLAIM CONSTRUCTION BRIEF
I. INTRODUCTION ................................................................................................................ 1
II. DISPUTED TERMS ............................................................................................................. 1
A. CPU Clock-Related Terms from the ’336, ’148, ’749 and ’890 Patents .................. 1
i. The “variable speed” clock of the patents-in-suit ......................................... 2
ii. Construction of “ring oscillator” (’336, ’148, ’749, ’890) ............................ 3
iii. “Non-controllable” and “variable based on the environment” is
consistent with TPL’s description of the ring oscillator during the
original prosecution ....................................................................................... 6
iv. Plaintiffs’ construction of the other clock-related terms should be
adopted .......................................................................................................... 9
v. Construction of “clocking said CPU” (’336 patent) .................................. 13
vi. Construction of “operates asynchronously to” (’336) ................................. 14
vii. Construction of “as a function of parameter variation” (’336/’148) ........... 16
B. Microprocessor Architecture Related Terms from the ’890 and ’749 Patents........ 17
i. Construction of “separate direct memory access central processing
unit” (’890) .................................................................................................. 17
ii. Construction of “push down stack” in “(first) push down stack
connected to said arithmetic logic unit” (’749) ........................................... 20
iii. Construction of “supply the multiple sequential instructions to said
central processing unit integrated circuit during a single memory
cycle” .......................................................................................................... 26
iv. Construction of “instruction register” ......................................................... 28
v. Construction of “multiple sequential instructions” ..................................... 30
III. CONCLUSION ................................................................................................................... 30
TABLE OF AUTHORITIES
Page(s)
Case Nos. 5:08-cv-00877, 5:08-cv-00882, 5:08-cv-05398 -ii- [CORRECTED] PLAINTIFFS’ CONSOLIDATED
RESPONSIVE CLAIM CONSTRUCTION BRIEF
CASES
Biovail Corp. Int’l v. Andrx Pharms., Inc.,
239 F.3d 1297 (Fed. Cir. 2001) ................................................................................................. 5
Cardiac Pacemakers, Inc. v. St. Jude Med., Inc.,
296 F.3d 1106 (Fed. Cir. 2002) ............................................................................................... 24
Chimie v. PPG Indus., Inc.,
402 F.3d 1371 (Fed. Cir. 2005) ................................................................................................. 4
Default Proof Credit Card Sys., Inc. v. Home Depot U.S.A., Inc.,
412 F.3d 1291 (Fed. Cir. 2005) ............................................................................................... 24
Desper Prods., Inc. v. QSound Labs, Inc.,
157 F.3d 1325 (Fed. Cir. 1998) ............................................................................................... 12
Dow Chem. Co. v. NOVA Chems. Corp. (Can.),
629 F. Supp. 2d 397 (D. Del. 2009) ........................................................................................ 15
Edwards Lifesciences LLC v. Cook Inc.,
582 F.3d 1322 (Fed. Cir. 2009) ......................................................................................... 13, 23
Elkay Mfg. Co. v. Ebco Mfg. Co.,
192 F.3d 973 (Fed. Cir. 1999) ................................................................................................... 4
Inpro II Licensing, S.A.R.L. v. T-Mobile USA, Inc.,
450 F.3d 1350 (Fed. Cir. 2006) ............................................................................................... 23
K-2 Corp. v. Salomon S.A.,
191 F.3d 1356 (Fed. Cir. 1999) ............................................................................................... 19
Microsoft Corp. v. Multi-Tech. Sys., Inc.,
357 F.3d 1340 (Fed. Cir. 2004) ................................................................................... 4, 5, 6, 23
Moleculon Research Corp. v. CBS, Inc.,
793 F.2d 1261 (Fed. Cir. 1986) ............................................................................................... 15
Nystrom v. Trex Co.,
424 F.3d 1136 (Fed. Cir. 2005) ............................................................................................... 10
Rheox, Inc. v. Entact, Inc.,
276 F.3d 1319 (Fed. Cir. 2002) ....................................................................................... 4, 5, 28
Salazar v. Procter & Gamble Co.,
414 F.3d 1342 (Fed. Cir. 2005) ........................................................................................... 5, 21
Page(s)
Case Nos. 5:08-cv-00877, 5:08-cv-00882, 5:08-cv-05398 -iii- [CORRECTED] PLAINTIFFS’ CONSOLIDATED
RESPONSIVE CLAIM CONSTRUCTION BRIEF
TIP Sys., LLC v. Phillips & Brooks/Gladwin, Inc.,
529 F.3d 1364 (Fed. Cir. 2008) ............................................................................................... 19
Trinity Indus. v. Road Sys.,
121 F. Supp. 2d 1028 (E.D. Tex. 2000) .................................................................................... 5
University of Pittsburgh v. Hedrick,
573 F.3d 1290 (Fed. Cir. 2009) ............................................................................................. 5, 6
STATUTES
35 U.S.C. § 112 ¶ 6 ....................................................................................................................... 24

[CORRECTED] PLAINTIFFS’ CONSOLIDATED RESPONSIVE CLAIM CONSTRUCTION BRIEF

TABLE OF ABBREVIATIONS
’148 patent or ’148 U.S. Patent No. 6,598,148, entitled “High Performance
Microprocessor Having Variable Speed System Clock,” issued
July 22, 2003
’336 patent or ’336 U.S. Patent No. 5,809,336, entitled “High Performance
Microprocessor Having Variable Speed System Clock,” issued
September 15, 1998
’749 patent or ’749 U.S. Patent No. 5,440,749, entitled “High Performance, Low Cost
Microprocessor Architecture,” issued August 8, 1995
’890 patent or ’890 U.S. Patent No. 5,530,890, entitled “High Performance, Low Cost
Microprocessor,” issued June 25, 1996
Plaintiffs Declaratory judgment plaintiffs Acer, Inc., Acer America
Corporation, Barco, N.V., Gateway, Inc., HTC Corporation and
HTC America, Inc.
Defendants or TPL Declaratory judgment defendants Technology Properties Limited,
Patriot Scientific Corporation and Alliacense Limited
HTC action HTC Corporation, HTC America, Inc. v. Technology Properties
Limited, Patriot Scientific Corporation, and Alliacense Limited,
Civil Case No. 5:08-cv-00882 JW
Acer action Acer, Inc., Acer America Corporation and Gateway, Inc. v.
Technology Properties Limited, Patriot Scientific Corporation,
and Alliacense Limited, Civil Case No. 5:08-cv-00877 JW
Opening Br. Defendants’ Opening Claim Construction Brief for the “Top Ten”
Terms, filed December 23, 2012 (HTC action Dkt. No. 339)
Chen Decl. Declaration of Kyle D. Chen in Support of Plaintiffs’
Consolidated Responsive Claim Construction Brief
Ward Order Memorandum Opinion and Order by Judge T. John Ward, filed
June 15, 2007 (Docket No. 259) in Technology Properties Ltd., et
al. v. Matsushita Electric Industrial Co., et al., Civil Action
No. 2:05-CV-494 (TJW), in the U.S. District Court for the Eastern
District of Texas, Marshall Division (“TPL v. Matsushita”).
Talbot U.S. Patent No. 4,689,581, entitled “Integrated Circuit Phase
Locked Loop Timing Apparatus,” issued August 25, 1987 to
Gerald R. Talbot
Edwards U.S. Patent No. 4,680,698, entitled “High Density ROM in
Separate Isolation Well on Single with Chip,” issued July 14, 1987
to Jonathan Edwards, et al.
Declaratory relief plaintiffs Acer, HTC and Barco entities as shown on the caption page
(collectively “Plaintiffs”) submit this joint brief in support of their claim construction positions.

I. INTRODUCTION

The four patents-in-suit (the ’336, ’148, ’749 and ’890 patents) share the same specification
and concern features of a commercially failed microprocessor called “Sh-Boom.” Plaintiffs’
proposed constructions are all based on the intrinsic record provided by the specification and
prosecution history. The patent owner’s own words, through the specification and file histories of
those patents, provide a clear picture of the true, narrow scope of the claims. When the patents
were challenged in reexamination, TPL was forced to characterize and amend their claims even
more narrowly to avoid prior art. TPL cannot now avoid its disclaimers, disavowals and
characterizations of the alleged invention by simply ignoring them or by trying to run away from
its own specification and file histories.
Rather than address the intrinsic record, TPL focuses on hearsay rhetoric regarding its Sh-
Boom microprocessor. However, even the article that TPL relies upon describes Sh-Boom as “a
bizarre processor” that was “never a commercial success.”1 Contrary to TPL’s rhetoric, the
intrinsic record shows that the patents-in-suit do not cover all microprocessors, but rather, only the
“bizarre” features of Sh-Boom that were not implemented by the Plaintiffs.

II. DISPUTED TERMS

A. CPU Clock-Related Terms from the ’336, ’148, ’749 and ’890 Patents
Five of the “top ten” disputed terms relate to mechanisms for timing or “clocking” a central
processing unit (“CPU”): (1) “ring oscillator,” (2) “providing an entire variable speed clock
disposed upon said integrated circuit substrate,” (3) “clocking said central processing unit,” (4)
“operates asynchronously to,” and (5) “as a function of parameter variation.” These closelyrelated
terms will be discussed together in this brief. Although these terms appear most
prevalently in the ’336 and ’148 patents, the term “ring oscillator” also appears in asserted claims
of the ’749 and ’890 patents. Because the clock-related terms are related and potentially
1 http://spectrum.ieee.org/semiconductors/processors/25-microchips-that-shook-the-world/5
dispositive of claims in all four patents-in-suit, this brief will address them first.
i. The “variable speed” clock of the patents-in-suit
The CPU in a commercial microprocessor consists of millions of transistors that work
together to interpret and execute instructions. To ensure that those millions of transistors work in
harmony instead of chaos, a CPU typically relies on a series of timing signals known as “clock
signals” to drive its operations. The clock signals, which are generated by a clocking device, are
akin to “heartbeats” that drive blood through a human body. The clock signals control (and in fact
equal) the speed at which the CPU operates.
To operate properly, a CPU’s transistors must have enough time between clock signals to
complete their operations before the next clock signal arrives. Accordingly, a CPU has a
maximum speed that depends on how fast its transistors can operate. To ensure proper operation,
the clocking device should never send clock signals “too fast” such that they exceed the CPU’s
maximum speed. See ’336, 16:67-17:2 (“CPU 70 will always execute at the maximum frequency
possible, but never too fast.”).2
Because the transistors in the CPU depend on electrical signals to operate, their maximum
speed for proper operation is constrained by how fast the electrical signals can transmit through
them, known as “transistor propagation delays.” According to the patents-in-suit, these delays
depend on varying environmental conditions such as temperature, voltage and manufacturing
process, which thus “determine” the CPU’s maximum speed. ’336, 16:47-50 and 59-60. For
example, if the temperature in the environment rises, the CPU’s maximum speed for proper
operation decreases. ’336, 16:59-67.
The patents-in-suit explain that, to avoid clocking the CPU at a rate faster than its maximum
speed, prior art systems constrain the clock speed to a fixed rate slow enough to “operate properly
in worse [sic] case conditions.” ’336, 16:48-53. The patents criticize this approach by claiming
that this constraint results in a CPU that operates at less than half of its theoretical maximum
2 All citations to “xx:yy-zz” refer to columns and lines in the referenced patent. As noted in the
text, the patents-in-suit share a common specification. For purposes of consistency, this brief will
cite to columns and lines in the ’336 patent (Chen Decl., Ex. 1) when discussing the five clockrelated
terms performance. ’336, 16:50-53.
The ’336 and ’148 patents are both entitled “High Performance Microprocessor Having
Variable Speed System Clock” and disclose a variable speed clock comprised of transistors on the
same integrated circuit as the CPU to provide higher performance when environmental conditions
permit. By placing a variable speed clock on the same integrated circuit as the CPU, according to
the patents-in-suit, the speed of the variable speed clock and the CPU’s maximum speed will “vary
together” in the same way according to changing environmental conditions. The result of this
allegedly improved approach is that “CPU 70 will always execute at the maximum frequency
possible, but never too fast.” ’336, 16:67-17:2.
The only variable speed clock disclosed in the patents-in-suit is a clock generating circuit
called a “ring oscillator” that is made of the same transistors on the same integrated circuit as those
in the CPU itself. ’336, 16:54-57. According to the patents-in-suit, because the ring oscillator and
the CPU are on the same integrated circuit, they are subject to the same environmental conditions
(temperature, voltage and process), resulting in the CPU “always” being clocked at its “maximum
frequency possible, but never too fast” under any environmental conditions. ’336, 16:54-17:10.
ii. Construction of “ring oscillator” (’336, ’148, ’749, ’890)
In the prior Texas action, Judge Ward construed “ring oscillator” as “an oscillator having a
multiple, odd number of inversions arranged in a loop.” Chen Decl., Ex. 2 at 11. The parties’
dispute turns primarily on whether the construction should incorporate statements made by TPL in
subsequent reexamination concerning the claimed “ring oscillator”:
Plaintiffs’ Construction TPL’s Construction
An oscillator having a multiple, odd number of inversions
arranged in a loop, wherein the oscillator is: (1) noncontrollable;
and (2) variable based on the temperature,
voltage, and process parameters in the environment.
An oscillator having a multiple,
odd number of inversions
arranged in a loop
Plaintiffs’ construction includes a “wherein” clause that incorporates explicit arguments and
disavowals that TPL made during reexamination after Judge Ward’s claim construction order and
after the dismissal of the Texas action. Specifically, in order to overcome a rejection of its claims
based on U.S. Patent No. 4,689,581 to Talbot (Chen Decl., Ex. 3), TPL argued that the voltage
controlled oscillator (“VCO”) of Talbot did not teach the “ring oscillator” of the patents-in-suit.
The examiner summarized TPL’s arguments, which were made in an in-person interview, as
follows:
Continuing, the patent owner further argued that the reference of Talbot does not
teach of [sic] a “ring oscillator.” The patent owner discussed features of a ring
oscillator, such as being non-controllable, and being variable based on the
environment. The patent owner argued that these features distinguish over what
Talbot teaches.
Interview Summary, 2/12/08, Control No. 90/008,227 (emphasis added) (Chen Decl., Ex. 4).
In light of TPL’s disavowing arguments made to the PTO after Judge Ward’s ruling, the
construction must be adapted to require that the claimed “ring oscillator” be (1) “noncontrollable,”
and (2) “variable based on the environment.”3 Federal Circuit law is clear that
“[a]rguments made during the prosecution of a patent application are given the same weight as
claim amendments.” Elkay Mfg. Co. v. Ebco Mfg. Co., 192 F.3d 973, 979 (Fed. Cir. 1999). It is
also black letter law that a court “cannot construe the claims to cover subject matter broader than
that which the patentee itself regarded as comprising its invention and represented to the PTO.”
Microsoft Corp. v. Multi-Tech. Sys., Inc., 357 F.3d 1340, 1349 (Fed. Cir. 2004). “The purpose of
consulting the prosecution history in construing a claim is to ‘exclude any interpretation that was
disclaimed during prosecution.’” Chimie v. PPG Indus., Inc., 402 F.3d 1371, 1384 (Fed. Cir.
2005) (citation omitted). “Accordingly, ‘where the patentee has unequivocally disavowed a
certain meaning to obtain his patent, the doctrine of prosecution disclaimer attaches and narrows
the ordinary meaning of the claim congruent with the scope of the surrender.’” Id. (citation
omitted); see also, e.g., Rheox, Inc. v. Entact, Inc., 276 F.3d 1319, 1325 (Fed. Cir. 2002) (“Explicit
arguments made during prosecution to overcome prior art can lead to narrow claim interpretations
because ‘the public has a right to rely on such definitive statements made during prosecution.’”)
3 Plaintiffs’ construction requires that the oscillator be “(1) non-controllable; and (2) variable
based on the temperature, voltage, and process parameters in the environment.” Part (2) of this
construction is based on TPL’s explanation of the term “environment” in its previous claim
construction briefing. See Doc. No. 221 in Acer action (02/11/2011 TPL Claim Construction
Brief), at 17:17-19 (“According to the ‘336 specification, ‘the ring oscillator frequency is
determined by the parameters of temperature, voltage and process.’ This is the only ‘environment’
that is disclosed in the specification.”) (citation omitted).
The examiner’s interview summary is a proper basis for finding a disavowal of claim scope.
It expressly reflects what TPL, the patent owner, argued. The Federal Circuit has repeatedly relied
upon patent owners’ arguments recorded in interview summaries to find that patent owners
disavowed claim scope to distinguish prior art. See, e.g., Rheox, Inc. v. Entact, Inc., 276 F.3d
1319, 1322 (Fed. Cir. 2002) (disavowal found based on patent owner’s arguments that the
examiner recorded in interview summary); see also Biovail Corp. Int’l v. Andrx Pharms., Inc., 239
F.3d 1297, 1302-04 (Fed. Cir. 2001) (same); Trinity Indus. v. Road Sys., 121 F. Supp. 2d 1028,
1044 (E.D. Tex. 2000) (“It is proper to consider the interview summary in claim construction as it
is part of the prosecution history.”) (citing Athletic Alternatives, Inc. v. Prince Mfg., Inc., 73 F.3d
1573, 1576 (Fed. Cir. 1996) (relying upon examiner’s interview summary of patent owner’s
statements in claim construction)).
The examiner had no motive to misstate TPL’s position, and TPL does not dispute the
accuracy of any aspect of the examiner’s summary of TPL’s argument. In its opening brief, TPL
cites its own self-serving amendment, written and filed after the examiner’s summary, but tellingly
that amendment did not dispute the examiner’s summary of TPL’s “ring oscillator” argument.
TPL’s speculation that the examiner did not rely upon TPL’s interview argument regarding
the claimed “ring oscillator” is unsupported and immaterial. The Federal Circuit has held “on
numerous occasions that patentee’s statements during prosecution, whether relied on by the
examiner or not, are relevant to claim interpretation.” Microsoft Corp., 357 F.3d at 1350.
TPL argues that Salazar v. Procter & Gamble Co., 414 F.3d 1342 (Fed. Cir. 2005), applies,
but it does not. Salazar held that “unilateral statements by an examiner” in a Notice of Allowance
did not give rise to a disavowal by the patent owner. The statements at issue here were not
“unilateral statements” by the examiner, but arguments made by TPL. The fact that the examiner
recorded TPL’s statements does not change the fact that it was TPL, not the examiner, who made
them.
TPL also misapplies University of Pittsburgh v. Hedrick, 573 F.3d 1290 (Fed. Cir. 2009),
which refused to give weight to a “terse” and ambiguous interview summary that was unclear
concerning which features of the claimed invention, if any, were being distinguished. Id. at 1297.
In the present case, however, TPL clearly argued that the claimed ring oscillator “distinguish[es]
over what Talbot teaches” because it has “features” such as “being non-controllable, and being
variable based on the environment.” Interview Summary, 2/12/08, Control No. 90/008,227 (Chen
Decl., Ex. 4). These disavowals clearly identify the claim language and the features on which it is
distinguished.
Finally, there is no merit to TPL’s suggestion that its disavowal is ineffective because it
occurred in the reexamination of the ’148 patent. The ’148 patent shares the same specification
and is directly related to the other three patents-in-suit, all of which claim a “ring oscillator.” The
Federal Circuit has made clear that “[a]ny statement of the patentee in the prosecution of a related
application as to the scope of the invention would be relevant to claim construction.” Microsoft
Corp., 357 F.3d at 1350. Accordingly, TPL’s arguments in the ’148 reexamination are relevant to
how common claim language should be interpreted in closely-related patents. TPL has not argued
that “ring oscillator” should be construed differently in the ’148 patent, nor would there be any
basis for TPL to do so. In light of TPL’s disavowing statements made to the PTO after Judge
Ward’s ruling, Plaintiffs’ proposal should be adopted.
iii. “Non-controllable” and “variable based on the environment” is
consistent with TPL’s description of the ring oscillator during the
original prosecution.
TPL’s reexamination disavowal as to the ring oscillator being “non-controllable” and
“variable based on the environment” was essentially a shorthand summary of the numerous
arguments the applicants made during the original prosecution of the ’336 patent to overcome
multiple prior art references. The original prosecution history underscores that the variable speed
clock is non-controllable because its frequency variation is based on environmental parameters.
TPL distorts the specification to argue the claimed ring oscillator is “controllable via these
[environmental] parameters” because “temperature, voltage and process are all controllable to one
degree or another.” Opening Br. at 18 (quoting ’336, 16:59-60). TPL is wrong. Nowhere does
the patent or prosecution history suggest using the environmental parameters to somehow control
the ring oscillator. Instead, as described by the patent and prosecution history, the claimed ring
oscillator naturally clocks the CPU at its maximum speed because they are comprised of the same
transistors on the same integrated circuit and respond to uncontrollable variations in temperature,
voltage and manufacturing process in the same way. See, e.g., ’336 PH 04/15/1996 Amend. at 8
(emphasis added) (Chen Decl., Ex. 5 at HTCMSJ000025) (“the microprocessor and clock will
naturally tend to vary commensurately in speed as a function of various parameters (e.g.,
temperature) affecting circuit performance”). No control of the ring oscillator is needed or
permitted. Indeed, any control of the ring oscillator would defeat the purpose of the alleged
invention by slowing the CPU from its maximum speed, as done in the prior art.
In the ’336 prosecution history, TPL repeatedly drew the distinction between (a) deliberate
“control” of the oscillator’s frequency through an input signal, crystal or other component of the
system and (b) the ability of the oscillator’s frequency to vary based on the “environmental
parameters” of temperature, voltage and process. For example, in response to rejections of claims
reciting a “variable speed clock,” a “ring oscillator variable speed system clock” and an
“oscillator,” TPL made the following argument:
A ring oscillator will oscillate at a frequency determined by its fabrication and design
and the operating environment. Thus in this example, the user designs the ring oscillator
(clock) to oscillate at a frequency appropriate for the driven device when both the oscillator
and the device are under specified fabrication and environmental parameters. Crucial to
the present invention is that since both the oscillator or variable speed clock and driven
device are on the same substrate, when the fabrication and environmental parameters
vary, the oscillation or clock frequency and the frequency capability of the driven
device will automatically vary together. This differs from all cited references in that
the oscillator or variable speed clock and the driven device are on the same substrate, and
that the oscillator or variable speed clock varies in frequency but does not require
manual or programmed inputs or external or extra components to do so.
’336 PH Amend. 07/07/1997 at 5 (Chen Decl., Ex. 5 at HTCMSJ000014) (emphasis added).
The patent owner continued to draw this “crucial” distinction between the prior art’s
concept of “control” (e.g., based on manual or programmed inputs or external components) and the
environmental factors discussed in the patent. For example, the patent owner contrasted the
“frequency controlled” clock in U.S. Patent No. 4,503,500 (“Magar”) with the claimed “variable
speed clock,” “ring oscillator variable speed system clock” and “oscillator” as follows:
[O]ne of ordinary skill in the art should readily recognize that the speed of the cpu and
the clock do not vary together due to manufacturing variation, operating voltage and
temperature of the IC in the Magar microprocessor . . . This is simply because the Magar
microprocessor clock is frequency controlled by a crystal which is also external to the
microprocessor. Crystals are by design fixed-frequency devices whose oscillation speed is
designed to be tightly controlled and to vary minimally due to variations in manufacturing,
operating voltage and temperature. The Magar microprocessor in no way contemplates
a variable speed clock as claimed.
See id. at 3-4 (Chen Decl., Ex. 5 at HTCMSJ000012-13) (italics in original; boldface and
underlining added). The patent owner further argued:
[C]rystals are by design fixed-frequency devices whose oscillation frequency is
designed to be tightly controlled and to vary minimally due to variations in
manufacturing, operating voltage and temperature. The oscillation frequency of a crystal
on the same substrate with the microprocessor would inherently not vary due to
variations in manufacturing, operating voltage and temperature in the same way as the
frequency capability of the microprocessor on the same underlying substrate, as claimed.
See id. at 4 (Chen Decl., Ex. 5 at HTCMSJ000013) (emphasis added).
In another example, the patent owner distinguished the “frequency control information” and
“clock control signals” in U.S. Patent No. 4,670,837 (“Sheets”) from the claimed variable speed
clocking mechanisms:
The present invention does not similarly rely upon provision of frequency control
information to an external clock, but instead contemplates providing a ring oscillator
clock and the microprocessor within the same integrated circuit. The placement of these
elements within the same integrated circuit obviates the need for provision of the type of
frequency control information described by Sheets, since the microprocessor and clock
will naturally tend to vary commensurately in speed as a function of various
parameters (e.g., temperature) affecting circuit performance. Sheets’ system for
providing clock control signals to an external clock is thus seen to be unrelated to the
integral microprocessor/clock system of the present invention.
’336 PH 04/15/1996 Amend. at 8 (Chen Decl., Ex. 5 at HTCMSJ000025) (emphasis added).
Specifically, the patent owner pointed out that the claimed oscillator will “naturally tend to
vary commensurately in speed as a function of various parameters (e.g., temperature) affecting
circuit performance.” Id. (emphasis added). Later, the patentee went even further to distinguish
Sheets’ clock “in the same integrated circuit” controlled by a “command input” as follows:
Even if the Examiner is correct that the variable clock in Sheets is in the same integrated
circuit as the microprocessor of system 100, that still does not give the claimed subject
matter. In Sheets, a command input is required to change the clock speed. In the present
invention, the clock speed varies correspondingly to variations in operating
parameters of the electronic devices of the microprocessor because both the variable
speed clock and the microprocessor are fabricated together in the same integrated
circuit. No command input is necessary to change the clock frequency.
’336 PH 01/03/1997 Amend. at 4 (Chen Decl., Ex. 5 at HTCMSJ000016) (emphasis added).
As the preceding discussion shows, the patent owners consistently characterized the claimed
“variable speed clock,” “ring oscillator variable system clock” and “oscillator” as environmentally
dependent, and expressly distinguished prior art clocks that were “controlled,” whether through
“clock control signals,” “frequency control information,” or “command inputs.” It should
therefore come as no surprise that, during reexamination, TPL again emphasized the “features of a
ring oscillator, such as being non-controllable, and being variable based on the environment” as
distinguishing the claims over the prior art. Interview Summary, 2/12/08, Control No. 90/008,227
(Chen Decl., Ex. 4).
iv. Plaintiffs’ construction of the other clock-related terms should be
adopted.
The term “providing an entire variable speed clock disposed upon said integrated circuit
substrate” should be construed together with the other two “ring oscillator” related terms: “an
entire ring oscillator variable speed system clock in said single integrated circuit” and “an entire
oscillator disposed upon said integrated circuit substrate.” Although the ’336 patent language uses
three different terms to claim the variable speed clock in the claims (i.e., “variable speed clock,”
“ring oscillator variable speed system clock” and “oscillator,”) each side has proposed parallel
constructions for each term with common limitations, as shown below:
Term Plaintiffs’ Construction TPL’s Construction
providing
an entire
variable
speed
clock
disposed
upon said
integrated
circuit
substrate
Providing a variable speed clock
that is located entirely on the same
semiconductor substrate as the CPU and
does not rely on a control signal or an
external crystal/clock generator to
generate a clock signal,
wherein the variable speed clock is: (1)
non-controllable; and (2) variable based
on the temperature, voltage, and process
parameters in the environment
Providing a variable speed system clock
that is located entirely on the same
semiconductor substrate as the CPU and
does not directly rely on a command
input control signal or an external
crystal/clock generator to generate a
clock signal
Term Plaintiffs’ Construction TPL’s Construction
an entire
ring
oscillator
variable
speed
system
clock in
said
single
integrated
circuit
A ring oscillator variable speed system
clock
that is located entirely on the same
semiconductor substrate as the CPU and
does not rely on a control signal or an
external crystal/clock generator to
generate a clock signal,
wherein the ring oscillator variable speed
system clock is: (1) non-controllable; and
(2) variable based on the temperature,
voltage, and process parameters in the
environment
A ring oscillator variable speed system
clock
that is located entirely on the same
semiconductor substrate as the CPU and
does not directly rely on a command
input control signal or an external
crystal/clock generator to generate a
clock signal
an entire
oscillator
disposed
upon said
integrated
circuit
substrate
An oscillator
that is located entirely on the same
semiconductor substrate as the CPU and
does not rely on a control signal or an
external crystal/clock generator to
generate a clock signal,
wherein the oscillator is: (1) noncontrollable;
and (2) variable based on
the temperature, voltage, and process
parameters in the environment
An oscillator
that is located entirely on the same
semiconductor substrate as the CPU and
does not directly rely on a command
input control signal or an external
crystal/clock generator to generate a
clock signal
The parties appear to agree that these three terms present common issues notwithstanding
differences in terminology. Both sides have treated the three terms in parallel fashion, reflecting
that they are supported by the same “ring oscillator” disclosure in the specification. See Nystrom
v. Trex Co., 424 F.3d 1136, 1143 (Fed. Cir. 2005) (“Different terms or phrases in separate claims
may be construed to cover the same subject matter where the written description and prosecution
history indicate that such a reading of the terms or phrases is proper.”).
The disputes regarding these terms fall into two categories. First, for the reasons explained
above, these terms should incorporate the requirement that the clock be “(1) non-controllable; and
(2) variable based on the temperature, voltage, and process parameters in the environment,” based
on the patent owner’s explicit arguments during prosecution and reexamination.
The remaining dispute turns on whether the claimed variable speed clock “does not rely on a
control signal” (Plaintiffs’ proposal) or whether the signal must be a specific “command input
control signal” that is “directly” relied upon, as TPL proposes. Plaintiffs’ construction
incorporates TPL’s arguments that the variable speed clock must be “non-controllable.” Logically,
a “non-controllable” clock cannot rely in any way – directly, indirectly, or otherwise – on any
“control signal,” whether it is based upon “clock control signals,” “frequency control information,”
or “command inputs,” which was disclaimed during the ’336 prosecution. Indeed, the
specification discloses no “control signal” for the claimed clocking mechanisms, and inclusion of
the word “directly” has no support in the intrinsic record. Plaintiffs’ proposed language, which
does not include “directly” or “command input,” should therefore be adopted.
TPL’s proposal also improperly attempts to recapture subject matter it surrendered when it
distinguished the Talbot reference. TPL now contends that a clocking circuit known as a “phase
locked loop” (“PLL”) infringes the “non-controllable” clocking mechanisms, despite the fact that
TPL previously argued that its claims do not cover such an arrangement in order to overcome the
Talbot reference. See generally HTC’s Motion for Summary Judgment of Non-Infringement (Doc.
No. 293 in HTC action). Talbot discloses a phase-locked loop (PLL), as confirmed by its title:
“Integrated Circuit Phase Locked Looped Timing Apparatus.” The PLL that TPL attempted to
distinguish is shown in Figure 1 of Talbot reproduced below (the PLL is numbered as 4):
A phase-locked loop provides a clock whose output frequency is controlled by locking the phase
of the output clock signal to the phase of the input clock signal provided by an external crystal
clock. For example, if the frequency of the crystal clock relied upon by a phase-locked loop is 10
MHz (10 million cycles per second), the phase-locked loop can multiply the crystal frequency by 2
or 3 to provide clock signal frequencies at 20 MHz or 30 MHz, respectively.
As noted above, TPL argued that the claimed ring oscillator was “non-controllable” and
“variable based on the environment,” and was distinguishable from Talbot. The particular
oscillator in Talbot was a “voltage-controlled oscillator,” shown as “VCO” in Figure 1 above
(item 12). See Amendment, 2/26/08 at 11 (Chen Decl., Ex. 6) and Interview Summary, 2/12/08,
Control No. 90/008,227 (Chen Decl., Ex. 4). The frequency of Talbot’s voltage-controlled
oscillator 12 is “controlled” by a “control signal” based upon the external clock signal (item 3).
See Talbot at 2:58-63, 3:7-16, 3:26-36 (Chen Decl., Ex. 3) (“[A] convertor and filter circuit 11 . . .
is arranged to convert the output pulses from the comparator 7 into a voltage signal for
controlling the frequency of oscillation of a voltage controlled oscillator circuit 12.”)
(emphasis added). Talbot’s voltage-controlled oscillator, therefore, relies on a control signal and
an external crystal/clock generator to generate its clock signal. See id.
TPL’s clear disclaimer of Talbot’s voltage-controlled oscillator confirms that the claimed
clocking mechanisms do not include a clock that relies on a control signal (voltage, current or
otherwise) or external crystal clock generator. In fact, absent its reliance on the control signal and
external clock, Talbot’s voltage-controlled oscillator 12 is structurally no different than an
“oscillator having a multiple, odd number of inversions arranged in a loop,” which is how TPL
proposes to construe the term “ring oscillator.” See Wolfe Decl. in support of Plaintiffs’ Sur-Reply
(Doc. No. 266 in Acer action).4
As noted, TPL now seeks to accuse the same type of voltage-controlled clocks it had to
disclaim during prosecution and reexamination. See Chen Ex. 7. It would be improper to permit
this. See Desper Prods., Inc. v. QSound Labs, Inc., 157 F.3d 1325, 1340 (Fed. Cir. 1998) (“Posthoc,
litigation-inspired argument cannot be used to reclaim subject matter that the public record in
the PTO clearly shows has been abandoned.”). Because the claimed clocking mechanisms are
non-controllable and cannot rely on any signal, directly or otherwise, the words “directly” and
4 Judge Fogel permitted the filing of the Wolfe Declaration during the prior briefing to rebut TPL’s
incorrect factual assertion that Talbot did not disclose an odd number of inversions in a loop.
“command input” should be removed from Judge Ward’s construction, and Plaintiffs’ proposals
should be adopted in their entirety.
v. Construction of “clocking said CPU” (’336 patent)
The disputed phrase, “clocking said CPU,” is in all asserted claims of the ’336 patent. The
dispute is whether the claimed variable speed clock will time the operation of the CPU at its
maximum frequency as disclosed by the specification:
Plaintiffs’ Construction [JCCS 20] TPL’s Construction
Timing the operation of the CPU such that it will always execute at
the maximum frequency possible, but never too fast
Timing the operation of
the CPU
Plaintiffs’ construction of “clocking said CPU” states that the CPU “will always execute at
the maximum frequency possible, but never too fast,” that is based directly on the clear statements
in the specification and prosecution history. As noted above, the specification criticizes prior art
approaches resulting in a CPU that operates at less than half of its theoretical maximum
performance. ’336, 16:50-53. The specification instead asserts that the alleged invention, “y
deriving system timing from the ring oscillator 430, CPU 70 will always execute at the
maximum frequency possible, but never too fast.” ’336, 16:59-17:2 (emphasis added).
TPL argues that Plaintiffs’ construction attempts to import limitations from the
specification. TPL is wrong. This patent is not entitled to claims broader than the sole
embodiment in the specification. When the embodiment “is described in the specification as the
invention itself, the claims are not necessarily entitled to a scope broader than that embodiment.”
Edwards Lifesciences LLC v. Cook Inc., 582 F.3d 1322, 1330 (Fed. Cir. 2009) (citation omitted).
Moreover, “[w]here the specification makes clear that the invention does not include a particular
feature, that feature is deemed to be outside the reach of the claims of the patent, even though the
language of the claims, read without reference to the specification, might be considered broad
enough to encompass the feature in question.” Id. at 1329 (citation omitted). And finally, when
the specification “describes a feature of the invention ... and criticizes other products ... that lack
that same feature, this operates as a clear disavowal of these other products....” Id. at 1333.
All of these principles apply here because the specification emphatically declares that the
CPU of the alleged invention “always” executes at the maximum frequency and criticizes products
that lack that feature. The patent owner also relied on this feature to distinguish the Sheets
reference during prosecution of the ’336 patent, arguing that “CPU 70 executes at the fastest speed
possible using the adaptive ring counter clock 430.” Amendment, 4/15/96 at 8-9 (Chen Decl.,
Ex. 8). The term “clocking said CPU” should therefore be construed to require “timing the
operation of the CPU such that it will always execute at its maximum frequency possible, but
never too fast.”
vi. Construction of “operates asynchronously to” (’336)
The phrase “operates asynchronously to” appears at the end of claims 11, 13 and 16 and is
part of the longer phrase: “wherein said central processing unit operates asynchronously to said
input/output interface.” The dispute is whether operating “asynchronously” excludes synchronous
operation using independent clocks:
Plaintiffs’ Construction [JCCS 29] TPL’s Construction
operates without a timing relationship to/with timed by independent clock signals
As discussed above, the patent discloses a variable speed ring oscillator that clocks the CPU
at its maximum frequency possible while varying its frequency based on the environmental
conditions. However, for the CPU to communicate with outside components, “[t]he external world
must be synchronized to the microprocessor 50 for operations such as video display updating and
disc drive reading and writing.” ’336, 17:22-25. To synchronize the microprocessor with the
external world, a second, fixed speed clock for timing the I/O interface is provided. “This
synchronization is performed by the I/O interface 432, speed of which is controlled by a
conventional crystal clock 434.” ’336, 17:25-27. The specification explains that this “dual clock
scheme” has the additional advantage of not dragging down the CPU’s speed with the typically
slower I/O interface. ’336, 17:12-21.
To allow the CPU to always execute at the maximum frequency possible and not be dragged
down by the speed of the I/O interface, the CPU must operate “asynchronously,” i.e., without a
timing relationship with, the I/O interface. Indeed, it is logically impossible for the CPU’s
environmentally dependent “variable speed clock” to have any timing relationship with the I/O
interface’s fixed frequency clock.
TPL’s proposed construction, “timed by independent clock signals,” is contrary to the plain
meaning of “asynchronous” because “independent” clock signals can nevertheless have a timing
relationship with one other – in other words, be “synchronized.” Anything that is synchronized, by
definition, is not “asynchronous.” A simple example of “independent” yet synchronized clocks
comes from old war movies in which soldiers synchronize their “independent” wrist watches.
During reexamination, TPL actually dedicated an entire section entitled “Synchronism Does Not
Preclude Independence” to distinguish the Kato prior art by arguing that “independent” clocks may
nonetheless be synchronous. See Amendment, 9/8/08, pp. 21-22 of 28 (Chen Decl., Ex. 9). TPL’s
argument that two “independent” clocks can nonetheless operate “synchronously” fatally
undermines its current litigation position on the meaning of “asynchronously.” TPL’s proposed
construction, as admitted by TPL, improperly includes both asynchronous and synchronous
operations, contrary to the plain claim language.
TPL’s definition is derived entirely from an excerpt of an extrinsic reference, Computation
Structures, that TPL submitted to the PTO in the reexamination. See Opening Br. at 12. TPL’s
reliance on this textbook is problematic. Because the excerpt was submitted to the PTO during
this litigation, perhaps in anticipation of claim construction, it should be given little weight. See
Moleculon Research Corp. v. CBS, Inc., 793 F.2d 1261, 1270 (Fed. Cir. 1986) (observing that
documents submitted to PTO during litigation “might very well contain merely self-serving
statements which likely would be accorded no more weight than testimony of an interested witness
or argument of counsel.”). TPL’s reliance on its own submission also improperly attempts to use
the prosecution history to broaden the scope of its claims. See, e.g., Dow Chem. Co. v. NOVA
Chems. Corp. (Can.), 629 F. Supp. 2d 397, 415 (D. Del. 2009) (“[Plaintiff] does not cite any
authority, and the Court is not aware of any, suggesting that the prosecution history can be used to
broaden the scope of claims beyond that which is supported by the specification.”).
A more relevant portion of that textbook, which TPL failed to submit to the PTO or this
Court, shows that by “independent clocks,” the textbook actually describes separate clocks with no
timing relationship. In a section entitled “Multiple-Clock Systems,” the book describes a situation
involving “multiple asynchronous clocks, each clock a free-running oscillator generating [clock
signals] independently of the others.” Chen Decl., Ex. 10 at 175 (emphasis added). The book goes
on to explain: “This relationship is common among large, independently designed subsystems; as
an extreme example, the interconnection of two separate computers (each of which may run
synchronously with its single clock) constitutes a system with at least two unsynchronized clocks.”
Id. Two separate computers, which might have been powered on at different times and may be
separated by great distances, present a clear example of two things that operate without a timing
relationship with each other, or in other words, asynchronously. This passage clarifies that when
Computation Structures uses the term “independent” in the context of asynchronous operations, it
is referring to the lack of a timing relationship.
vii. Construction of “as a function of parameter variation” (’336/’148)
The ’336 and ’148 patents require that the CPU’s maximum speed for proper operations and
the “oscillator” vary in the same way “as a function of parameter variation” in fabrication or
operational parameters. The two sides’ competing proposals are below:
Plaintiffs’ Construction TPL’s Construction
in a determined functional relationship with parameter variation based on parameter variation
The specification explains that the temperature, voltage and process parameters in the
environment “determine” the CPU’s and the oscillator’s frequencies in a “functional relationship:”
The ring oscillator[’s] frequency is determined by the parameters of temperature,
voltage, and process. At room temperature, the frequency will be in the
neighborhood of 100 MHZ. At 70 degrees Centigrade, the speed will be 50 MHZ.
The ring oscillator 430 is useful as a system clock, . . . because its performance tracks
the parameters which similarly affect all other transistors on the same silicon die.
’336, 16:59-67 (emphasis added). By disclosing that the ring oscillator’s frequency is “determined
by” the environmental parameters and claiming that the CPU’s processing frequency is a
“function” of the parameters’ variation, the claims require that the frequency of the CPU and the
on-chip oscillator have a specific and unique value for any given combination of temperature,
voltage and process. Put another way, for a given combination of temperature, voltage and process
parameters, the CPU’s and the on-chip oscillator’s frequencies should be reproducible. The
numerical example provided by the specification in fact suggests such a determined functional
relationship. ’336, 16:60-63 (“At room temperature, the frequency will be in the neighborhood of
100 MHZ. At 70 degrees Centigrade, the speed will be 50 MHZ.”). Plaintiffs’ proposed
construction captures this requirement of a “determined” value. Plaintiffs’ construction is also
consistent with and interpretive of the example in the specification discussed above.
TPL’s proposed construction is too vague and claims, as environmental parameters vary,
non-reproducible, even random (i.e., undetermined) CPU and oscillator frequencies for a given
combination of temperature, voltage and process. Thus, TPL’s proposal should be rejected.
B. Microprocessor Architecture Related Terms from the ’890 and ’749 Patents
The ’890 and ’749 patents, both entitled “High Performance, Low Cost Microprocessor
Architecture,” disclose different aspects of a specialized microprocessor system. The following
five related terms from the ’890 and ’749 patents will be discussed together: “separate direct
memory access central processing unit,” “(first) push down stack connected to said arithmetic logic
unit,” “supplying the multiple sequential instructions to said central processing unit integrated
circuit during a single memory cycle,” “instruction register” and “multiple sequential instructions.”
i. Construction of “separate direct memory access central processing unit” ’890)
The ’890 patent purports to describe aspects of the specialized microprocessor architecture
intended to allow faster access to certain memory locations. Claim 11, the only independent claim
of the ’890 patent following the reexamination,5 recites “[a] microprocessor, which comprises a
main central processing unit and a separate direct memory access central processing unit in a
single integrated circuit….” ’890, Reexam. Cert., Claim 11 (Chen Decl., Ex. 11) (emphasis added).
The term “direct memory access” or “DMA” is a well-known technology for improving the
performance of computer systems. DMA allows certain subsystems or components within a
computer (such as a disk drive or other device) to transfer data to memory without the main CPU
having to perform the actual data transfer, allowing the CPU to perform other tasks. The ’890
patent acknowledges that conventional “DMA controllers can provide routine handling of DMA
5 Claim 1 was canceled in the reexamination and new claim 11 was added.
requests and responses, but some processing by the main central processing unit (CPU) of the
microprocessor is required.” ’890, 1:55-58. The ’890 patent purports to address this problem by
claiming a “separate direct memory access central processing unit” (“separate DMA CPU”), for
which the parties have proposed the following constructions:
Plaintiffs’ Construction TPL’s Construction
a separate central processing unit that fetches and
executes instructions for performing direct memory
access without using the main central processing unit
electrical circuit for reading and
writing to memory that is separate
from a main CPU
Plaintiffs’ proposed construction is the only one that comports with the specification and
claim language of the ’890 patent. The claim language itself recites “a separate direct memory
access central processing unit,” which is “separate” in the sense that it is physically and
functionally distinct from the main CPU. As explained in the specification: “The DMA CPU 72
controls itself and has the ability to fetch and execute instructions. It operates as a co-processor to
the main CPU 70 (FIG. 2) for time specific processing.” ’890, 8:22-24. The specification criticizes
“conventional microprocessors” that use “DMA controllers” because “some processing by the main
central processing unit (CPU) of the microprocessor is required.” ’890, 1:52-58. The specification
identifies as an object of the invention a processor “in which DMA does not require use of the main
CPU during DMA requests and responses and which provides very rapid DMA response with
predictable response times.” ’890, 2:2-5. The specification confirms, therefore, that a separate
DMA CPU is a separate CPU that fetches and executes instructions for performing DMA without
using the main CPU, as Plaintiffs have proposed.
TPL’s proposal should be rejected because it ignores the “CPU” in the claim term “DMA
CPU.” TPL relies on the disclosure of a “DMA controller” embodiment that, as a matter of plain
claim language, is unclaimed. The DMA CPU, unlike a conventional DMA controller, has the
ability to fetch and execute instructions. TPL concedes in its opening brief that “DMA controllers”
are different from the claimed “DMA CPU” because: “This ‘more traditional DMA controller’ is
one that functions more as a traditional state machine, without the ability to fetch its own
instructions that characterizes a CPU.” Opening Br. at 9:24-26 (emphasis added). But the
ability to fetch instructions – a feature that even TPL concedes “characterizes a CPU” – is
conspicuously missing from its construction of DMA CPU. TPL’s construction attempts to rewrite
the claim to remove “CPU” from the claim term “DMA CPU.” This would be improper. See K-2
Corp. v. Salomon S.A., 191 F.3d 1356, 1364 (Fed. Cir. 1999) (“Courts do not rewrite claims;
instead, we give effect to the terms chosen by the patentee”).
TPL attempts to equate the DMA CPU 314 of Figure 9 with a DMA controller, but TPL is
wrong. Figure 9 shows “a layout diagram of a second embodiment of a microprocessor” that has a
“DMA CPU 314.” ’890, 4:61-63 and Fig. 9. A separate passage appearing eight columns later in
the specification describes a different and unclaimed embodiment in which “the DMA processor 72
of the microprocessor 50 has been replaced with a more traditional DMA controller 314.” ’890,
12:62-13:4. That passage makes no reference to Figure 9 or the DMA CPU described earlier in the
specification, and in fact, actually supports Plaintiffs’ position. By disclosing an alternative system
in which a DMA CPU has been “replaced with a more traditional DMA controller 314” (’890,
12:62-13:4 (emphasis added)), the specification actually confirms that a DMA CPU is different
from a DMA controller.
TPL’s assertion that Plaintiffs’ construction would exclude a preferred embodiment is
similarly without merit. The Federal Circuit has repeatedly recognized that a specification can
disclose subject matter not covered by the claims. See TIP Sys., LLC v. Phillips &
Brooks/Gladwin, Inc., 529 F.3d 1364, 1373 (Fed. Cir. 2008) (“Our precedent is replete with
examples of subject matter that is included in the specification, but is not claimed.”). “Therefore,
the mere fact that there is an alternative embodiment disclosed in the [patent-in-suit] that is not
encompassed by [a proposed] claim construction does not outweigh the language of the claim,
especially when [that] construction is supported by the intrinsic evidence.” Id. Because the
specification describes the DMA CPU as an improvement and replacement over the conventional
DMA controller, it makes sense that the claims exclude the DMA controller. Because TPL’s
construction improperly seeks to lay claim over the DMA controller that the specification
distinguishes from the claimed DMA CPU, it should be rejected.
Finally, TPL asserts that Acer’s claim construction expert, Dr. Wolfe, acknowledged that the
main CPU can initiate memory transfers. Opening Br. at 10:3-8.6 This argument misses the point
completely: the issue is what a DMA CPU can do without the main CPU, not what the main CPU
can do (with or without a DMA CPU). Although a main CPU can initiate a memory transfer
request, the specification makes clear that the DMA memory transfer is actually performed by the
DMA CPU – not the main CPU. ’890, 2:2-5. Plaintiffs’ proposed language, “performing direct
memory access without using the main [CPU],” is therefore accurate and should be adopted.
ii. Construction of “push down stack” in “(first) push down stack connected
to said arithmetic logic unit” (’749)
The ’749 patent claims a specialized microprocessor architecture with a “first push down
stack.” Chen Decl., Ex. 13 (’749 at claims 1 and 9). The operation of a push down stack is often
explained by analogy to a spring-loaded stack of plates at a cafeteria in which the most recently
stored plate is pushed onto the top of the plate stack. The top item in the stack, the one that was
most recently added, is also the first to be removed. A push down stack, therefore, operates in a
“last-in-first-out” manner. When a new item is placed on the top of the stack, it “pushes” the other
items down by one storage space, causing the other items to move towards the bottom of the stack
by one space. This everyday analogy is consistent with how the term “push down stack” is used in
the ’749 patent, and is captured by Plaintiff’s proposed construction:
Plaintiffs’ Construction TPL’s Construction
data storage elements organized from top to bottom to provide last-in
first-out access to stored items, wherein any previously stored items
propagate towards the bottom by one data storage element when a
new item is stored in the top data storage element
data storage elements organized to provide
last-in first-out access to items
The term “push down stack” is a component of the larger phrase, “first push down stack
connected to said arithmetic logic unit,” which is among the top ten terms and addressed separately
below.7 Plaintiffs’ construction of “push down stack” should be adopted because it is consistent with the
intrinsic record and, unlike TPL’s construction, explains for the jury what “last-in-first-out” means.
6 TPL mischaracterizes the testimony of Dr. Wolfe on this point. When asked if the main CPU can
perform any DMA-related operations, he testified: “Any? I don’t think so.” Wolfe Depo. at
167:11-13 (Chen Decl., Ex. 12). Dr. Wolfe did testify that the main CPU can request a single
element of data from memory, but made clear that such a request is not a “DMA-related” function.
Id. at 167:19-168:10. 7 Because “push down stack” is a component of “push down stack connected to said arithmetic logic unit,” the parties are in agreement that these terms count as a single term.
Figure 2
During prosecution of the ’749 patent, the examiner described a push down stack as follows:
“Note that a stack is such that inputted items propagate from one end of the stack to another via the
stages in the stack.” ’749, Office Action 12/31/92 at 3 (Chen Decl., Ex. 14). “Statements about a
claim term made by an examiner during prosecution of an application may be evidence of how one
of skill in the art understood the term at the time the application was filed.” Salazar, 414 F.3d at
1347. Plaintiffs’ proposed construction accordingly incorporates the examiner’s concept that the
inputted items “propagate” from the top of the stack towards the bottom, which accurately
describes how a “push down stack” in the ’749 patent operates.
Construction of “(first) push down stack connected to said arithmetic logic unit”
The ’749 patent recites a “first push down stack connected to said arithmetic logic unit.”
The parties’ proposed constructions are set forth below:
Plaintiffs’ Construction TPL’s Construction
(first) push down stack comprising a top item
register and a next item register, both directly
coupled to the ALU such that source and
destination addresses are not used
data storage elements organized to provide lastin
first-out access to items connected to convey
signals to a digital circuit that performs both
arithmetic and logical operations
Plaintiffs have separately addressed the meaning of “push down stack” in the preceding
section of this brief, and the parties have stipulated that an “arithmetic logic unit” (“ALU”) is a
digital circuit that performs both arithmetic and logical operations. The remaining disputes with
respect to this term relate to the structure of the first push down stack and the manner in which it is
connected to the ALU. The specification and file history make clear that the “first push down
stack” includes a top item and next item register directly coupled to the ALU such that source and
destination addresses are not used.
The “first push down stack” is depicted in
Figure 2 of the ’749 patent (of which the relevant
portion is reproduced at right), which shows how the
first push down stack (74) is structurally connected
to the ALU. TPL acknowledges in its opening brief
that “Figure 2 discloses a push down stack (74) connected to separate top and next item registers
(76 and 78).” Opening Br. at 21:13-14; id. at 21:2-3 (“Figure 2 illustrates dedicated registers that
provide inputs to the ALU.”). The specification explains: “The microprocessor 50 architecture has
the ALU 80 (FIG. 2) directly coupled to the top two stack locations 76 and 78.” ’749, 19:6-8
(emphasis added). This direct coupling is not merely a design choice of the disclosed embodiment,
but an essential aspect of the claimed invention.
The ’749 patent explains that prior art microprocessors rely on instructions that have to
specify (a) the logical or arithmetic operation to be performed by the ALU and (b) the locations
(i.e., addresses) of the two “sources” of the data to be used and one “destination” where the result
of the operation will be held. See ’749, 25:68-26:3. To take a simplified example, suppose an
instruction specifies a computation in which a first number (X) is added to a second number (Y) to
yield a third number (Z) (i.e., X+Y=Z). Such an instruction might require: (a) 8 bits to specify the
“add” arithmetic operation to be performed, (b) 8 bits to specify the address of the first number (X),
(c) 8 bits to specify the address of the second number (Y) and (d) 8 bits to specify the address
where the computed value (Z) will be stored. ’749, 25:68-26:3 (“Many 32-bit architectures use 8-
bits to specify the operation to perform but use an additional 24-bits to specify two sources and a
destination [because each requires 8-bits for addressing].”).
The need to specify the source and destination addresses (b, c, and d above) is eliminated by
the fact that the ALU is “directly coupled” to the top and next item registers (76) and (78). In
particular, the top item and next item registers (76) and (78) hold the two sources for the operation.
After the arithmetic or logic operation is completed, the top item register (76) serves as the
destination holding the result of the operation. ’749, 15:30-32 (“A math or logic operation always
uses the top two stack items as source and the top of stack as destination.”). Because the top item
and next item registers are “directly coupled” to the ALU, the ALU can exchange data with them
without the need for explicit addresses. ’749, 7:19-22 (“The push down stack allows the use of
implied addresses, rather than the prior art technique of explicit addresses for two sources and a
destination.”). Using the push down stack of the ’749 patent, therefore, saves 24 bits.
The advantages of using an 8-bit instruction instead of a 32-bit instruction by eliminating the
24-bits used to specify the two sources and one destination were repeatedly emphasized throughout
the specification:
“For math and logic operations, the microprocessor 50 exploits the inherent advantage of
a stack by designating the source operand(s) as the top stack item and the next stack item.
The math or logic operation is performed, the operands are popped from the stack, and the result is
pushed back on the stack. The result is a very efficient utilization of instruction bits as well as
registers.” ’749, 26:4-11 (emphasis added).
“Most microprocessors use on-chip registers for temporary storage of variables . . . A few
microprocessors use an on-chip push down stack for temporary storage. A stack has the advantage
of faster operation compared to on-chip registers by avoiding the necessity to select source
and destination registers.” ’749, 15:24-30 (emphasis added).
“The availability of 8-bit instructions also allows another architectural innovation, the
fetching of four instructions in a single 32-bit memory cycle.” ’749, 26:16-18.
By touting the use of implicit addressing and criticizing the prior art’s use of explicit
addressing, the patent owner told the public that not using explicit addressing for the top and next
item locations of the first push down stack was essential to the invention. See Edwards
Lifesciences LLC, 582 F.3d at 1334 (when the patent owner “describes a feature of the
invention . . . and criticizes other products . . . that lack that same feature,” a clear disavowal of
those other products results); see also Microsoft Corp., 357 F.3d at 1347 (construing “connected
to” as requiring direct connection based on description of invention in specification); Inpro II
Licensing, S.A.R.L. v. T-Mobile USA, Inc., 450 F.3d 1350, 1354-56 (Fed. Cir. 2006) (construing
claim to require a “direct connection” between components based on statements in specification
touting performance advantages of such a direct connection).
This point was reiterated during prosecution where the patent owner told the PTO that the
connection between the ALU and the top and next item locations of the first push down stack are
“in addition to the conventional construction of the first push down stack . . . .” ’749 File History,
7/6/93 Amendment at 9 (Chen Decl., Ex. 15). The importance of the recited “additional”
connections was clear: “The [first push down] stack 74 in fact allows arithmetic operations to be
carried out on operands supplied from it to the ALU and receives ALU results as a result of the
recited connections.” Id. (emphasis added).
TPL’s assertion that Plaintiffs’ proposed construction improperly imports limitations from
the specification also ignores the use of narrow “means-plus-function” language in the claim itself
defining the “first push down stack.” Claim 1 recites, in relevant part:
first push down stack connected to said arithmetic logic unit,
said first push down stack including
means for storing a top item connected to a first input of said arithmetic logic
unit to provide the top item to the first input and
means for storing a next item connected to a second input of said arithmetic logic
unit to provide the next item to the second input, . . . ,
said arithmetic logic unit having an output connected to said means for storing a
top item; […]
As shown above, the claim language expressly defines the “first push down stack” as including the
“means for storing a top item” and the “means for storing a next item,” and specifies the precise
connections between them and the ALU. All parties agree that these top item and next item
elements are written in means-plus-function format under 35 U.S.C. § 112 ¶ 6. Joint Claim
Construction Statement, Ex. B at 4 (Doc. No. 305-2 in Acer action). Federal Circuit law is clear
that a court must look to the specification to identify the corresponding structure for a means-plusfunction
element. See Cardiac Pacemakers, Inc. v. St. Jude Med., Inc., 296 F.3d 1106, 1113 (Fed.
Cir. 2002). Critically, “[a] structure disclosed in the specification qualifies as ‘corresponding’
structure only if the specification or prosecution history clearly links or associates that structure to
the function recited in the claim.” Default Proof Credit Card Sys., Inc. v. Home Depot U.S.A., Inc.,
412 F.3d 1291, 1298 (Fed. Cir. 2005). The top item and next item registers that are directly
coupled to the ALU in Figure 2 precisely match the function recited in the claim language. The use
of means-plus-function claim language defining the “first push down stack” and its connection to
the ALU reinforces the critical importance of the specification in construing this term.
TPL ignores the disclosures in the specification and the claim language discussed above.
TPL instead points to unrelated details in the specification that have nothing to do with the first
push down stack or how it is connected to the ALU. TPL first contends that Figure 13 shows an
alternative embodiment that does not disclose the dedicated registers of Figure 2. But TPL ignores
that Figure 13 shows exactly the same arrangement
as Figure 2, with precisely the same “direct
coupling” between the top and next item registers
and the ALU, as shown in the figures.
TPL’s assertion that the “STACK
POINTER” in Figure 13 (of which the relevant
portion is reproduced at right) is connected to the
first push down stack is similarly baseless. See
Opening Br. at 21:4-5. The stack pointer shown
in Figure 13 is not used to communicate to the
ALU at all. It is instead pointing to the bottom of
the first push down stack. See ’749, Fig. 13.
Finally, TPL’s argument based on the stack
architecture in Figure 21 is similarly inapposite.
Nothing in Figure 21 shows an ALU, let alone
any connection between any push-down stack
with top and next item registers and the ALU.
Instead, the stack pointer is used only to manage inter-stack operations of the “triple cache stack
architecture” illustrated in Figure 21. ’336, 18:23-27.
The primary flaw in TPL’s arguments regarding Figure 13 and 21 is that it ignores the
specific term at issue here, the “first push down stack connected to said arithmetic logic unit.”
The portions of the figures cited by TPL relate to other stacks in the specification that are not the
first push down stack. One such example is the “second push down stack” that is separately recited
in claim 10. But the claim language itself confirms that the “first push down stack” is the one
depicted in Figures 2 and 13 as item 74, because it is the only push down stack in the specification
that is “connected to said arithmetic logic unit” and has a “top item” register and a “next item”
register connected to inputs of the ALU, as expressly recited in the claim language. TPL’s attempt
to point to details of other stacks that are not the “first push down stack” is unavailing.
iii. Construction of “supply the multiple sequential instructions to said
central processing unit integrated circuit during a single memory cycle”
The microprocessor described in the ’749 patent operates by fetching “instructions” (which
specify CPU operations) from memory into an instruction register, which supplies them to the
CPU for execution. The ’749 patent explains, however, that “[t]he slowest procedure the
microprocessor 50 performs is to access memory. Memory is accessed when data is read or
written. Memory is also read when instructions are fetched.” ’749, 22:14-17. “The bottleneck in
most computer systems is the memory bus. The bus is used to fetch instructions and fetch and
store data.” ’749, 5:54-56. The ’749 patent purports to address this issue by fetching multiple
instructions from memory and then supplying them to the CPU during “a single memory cycle.”
According to the ’749 patent, because the CPU can execute instructions much faster than
they can be fetched from memory, multiple instructions can be executed during a single memory
cycle. The alleged invention allows fetching and execution of instructions to be overlapped,
resulting in performance improvements. See ’749, 22:17-40. The specification repeatedly touts
the advantages of this feature, including:
“The microprocessor 50 fetches 4 instructions per memory cycle . . . System speed is
therefore 4 times the memory bus bandwidth. This ability enables the microprocessor to break the
Von Neumann bottleneck of the speed of getting the next instruction.” ’749, 7:12-18.
“The bottleneck in most computer systems is the memory bus. The bus is used to fetch
instructions and fetch and store data. The ability to fetch four instructions in a single memory bus
cycle significantly increases the bus availability to handle data.” ’749, 5:54-58.
“The microprocessor 50 fetches up to four instructions in a single memory cycle and can
perform much useful work before requiring another memory access.” ’749, 18:10-12.
Claim 1 captures this requirement by reciting the following limitation (boldface type
showing the disputed term): “said means for fetching instructions being configured and connected
to fetch multiple sequential instructions from said memory in parallel and supply the multiple
sequential instructions to said central processing unit during a single memory cycle.” ’749,
claim 1. The parties have proposed the following constructions:
Plaintiffs’ Construction TPL’s Construction
provide the multiple sequential instructions in parallel (as
opposed to one-by-one) to said central processing unit
integrated circuit during a single memory cycle without using
a prefetch buffer or a one-instruction-wide instruction
buffer, that supplies one instruction at a time
provide the multiple sequential
instructions in parallel to said
central processing unit
integrated circuit during a single
memory cycle
The additional language in Plaintiffs’ proposed construction comes directly from TPL’s
express disclaimer made during reexamination of the ’749 patent. In particular, in attempting to
distinguish U.S. Patent No. 4,680,698 to Edwards (“Edwards”), TPL argued:
Edwards describes the way the Transputer decodes and executes instructions. As
described in Edwards, see, e.g., Fig. 8, below, instructions are supplied to a oneinstruction-
wide instruction buffer, one at a time, and are there decoded. Fetching
multiple instructions into a prefetch buffer and then supplying them one at a time is
not sufficient to meet the claim limitation – the supplying of ‘multiple sequential
instructions to a CPU during a single memory cycle.’
1/19/10 Amendment at 26 of 58 (Chen Decl., Ex. 16) (emphasis added).8 TPL further made a
similar disavowal in attempting to distinguish an article entitled The Motorola MC68020 by Doug
MacGregor et al. (“MacGregor”):
However, [MacGregor] does not disclose fetching “multiple sequential instructions from
said memory in parallel and supply the multiple sequential instructions to said central
processing unit integrated circuit during a single memory cycle”. MacGregor might imply
that it fetches two instructions from memory at a time, but the instructions are supplied to
the CPU one at a time. Such non-parallel supplying of instructions to the CPU is not
supplying them to the CPU during a single memory cycle as required by the claim.
Id. at 45 of 58 (emphasis added). Then, following an interview with the Examiner, TPL filed a
written response summarizing the substance of the interview and further disclaimed systems that
supply instructions to the CPU one at a time:
Next the MacGregor reference was discussed [during the interview]. Mr. Henneman [TPL’s
counsel] explained that although two instructions might be fetched at the same time,
only one instruction is supplied to the CPU at a time. The second instruction is stored in
a temporary register. Because MacGregor only discloses providing instructions to the CPU
one-at-a-time, Examiner Pokrzwya indicated that he would reconsider this rejection.
11/29/2010 Amendment at 19-20 of 35 (Chen Decl., Ex. 18) (emphasis added). Emphasizing how
8 Citing no evidentiary support, TPL asserts that “the instructions [in Edwards] were not supplied
to the instruction register until a second memory cycle.” Opening Br. at 14. Nothing in Edwards
supports TPL’s assertion. See Chen Decl., Ex. 17.
important this feature is, TPL made this point yet again:
As discussed in the interview and elaborated on above with respect to the May/Edwards
rejections, the “during a single memory cycle” limitation is not satisfied by supplying
only one instruction to a CPU at a time. Rather, the “multiple sequential instructions”
must be supplied “during a single memory cycle.”
Id. at 31 of 35 (Chen Decl., Ex. 18) (emphasis added).
As such, this term must be construed consistently with the multiple clear and unmistakable
disavowals and disclaimers that TPL made to the PTO. See Rheox, 276 F.3d at 1325. This phrase
should be construed as providing the multiple sequential instructions “in parallel (as opposed to
one-by-one) to said central processing unit integrated circuit during a single memory cycle without
using a prefetch buffer or a one-instruction-wide buffer, that supplies one instruction at a time.”
Doc. No. 243 in HTC action, at 25:11-14.
iv. Construction of “instruction register”
Computer instructions generally include two components, known as “opcodes” and
“operands.” An “opcode” is generally used to specify a specific logical or arithmetic operation to
perform, while “operands” specify the data that will be subject to the operation. In a theoretical
instruction in which two numbers are added together, i.e., A+B, the opcode is “+” and the two
operands identify A and B. This theoretical instruction could then be provided to an “instruction
register,” which supplies the instruction to circuits that interpret and execute the instruction (in this
case, by adding the two numbers). The dispute here turns on how the operands in the “instruction
register” are arranged. The parties’ proposed constructions are set forth below:
Plaintiffs’ Construction TPL’s Construction
register that receives and holds one or more instructions for
supplying to circuits that interpret the instructions, in which
any operands that are present must be right-justified in
the register
register that receives and holds one
or more instructions for supplying to
circuits that interpret the instruction
Plaintiffs’ proposed construction of “instruction register” should be adopted because it alone
comports with the undisputed intrinsic evidence. The ’749 patent describes a specialized
instruction register that, according to the specification, provides significant advantages over prior
art systems. The specification explains that, unlike prior art microprocessors, the processor in the
’749 patent can handle operands of variable sizes using the same opcode. ’749, 18:35-37. The
specification describes this accomplishment as “magic,” and explains that: “This magic is possible
because operands must be right justified in the instruction register. This means that the least
significant bit of the operand is always located in the least significant [i.e., right-most] bit of the
instruction register.” ’749, 18:43-47 (emphasis added). The specification makes clear, therefore,
that right justified operands in the instruction register are not an optional design choice of one
embodiment, but a required feature – something that “must” be present in order to accomplish the
“magic” of the alleged invention.
TPL further emphasized this “magic” during the original prosecution of the ’749 patent in
an attempt to distinguish U.S. Patent No. 5,127,091 to Boufarah. In a summary of an in-person
interview with the examiner on October 25, 1994, the examiner noted with respect to claim 1:
“operand width is variable and right adjusted.” 10/25/1994 Interview Summary at 1 (Chen
Decl., Ex. 19) (emphasis added).) The interview summary, which was never disputed, is
consistent with the specification’s description of the alleged invention.
Judge Ward and the Federal Circuit have also addressed this issue in connection with the
prior Texas litigation that involved U.S. Patent No. 5,784,584 (“’584 patent”), a division of the
’749 patent. The key issue involving the ’584 patent was whether any “operands” in the
“instruction register” must be “right justified.” The claim language of the ’584 patent did not
expressly recite that instruction operands had to be right justified, so TPL argued -- as it does
here – that “right justified operands are a feature of the preferred embodiment.” Ward Order at 22
(Chen Decl., Ex. 2). Judge Ward rejected TPL’s argument and noted that “[t]he specification and
prosecution history refer to the fact that operands in the instruction register must be right justified.”
Id. at 23.9 Because it was clear that the accused processors in that case did not have right justified
9 Judge Ward also construed the term “operand” as “an input to a single operation specified by
an instruction that is encoded as part of the instruction where the size of the input can vary.” Id.
at 24. Judge Ward also noted that TPL “appear[ed] to agree” that the size of the operand in the
specification was variable. Id. TPL’s previous claim construction briefing in this case, however,
argued that the specification discloses “fixed length” operands that need not be right justified, but
TPL appears to have abandoned that position – and for good reason. This issue was specifically
litigated in the Federal Circuit appeal that TPL lost. See ARM Appeal Brief at 23-24 (Chen
Decl., Ex. 20) (“The Specification Confirms The Right Justified Operands Are the Only
operands, TPL stipulated to a judgment of non-infringement and appealed to the Federal Circuit.
See TPL Appeal Brief at 2 (Chen Decl., Ex. 21). The Federal Circuit rejected TPL’s arguments
and summarily affirmed Judge Ward’s decision. See Fed. Cir. Ruling (Chen Decl., Ex. 22).
Following the Federal Circuit’s ruling, TPL granted a covenant-not-to-sue to the Plaintiffs herein
and the ’584 patent was dismissed from this litigation. The issue of whether the operands in the
instruction register must be right justified has been correctly settled, and those rulings should not
be disturbed.
v. Construction of “multiple sequential instructions”
As explained above, the specification and file history of the ’749 patent, as well as rulings
from Judge Ward and the Federal Circuit, confirm that the instruction operands must be right
justified in the instruction register. This same requirement should also apply to the term “multiple
sequential instructions” from the ’749 patent:
Plaintiffs’ Construction TPL’s Construction
Two or more instructions in sequence, in which any operands that
are present must be right-justified in the instruction register
Two or more instructions
in a program sequence
Judge Ward’s prior construction of the ’584 patent construed a closely-related phrase,
“instruction groups,” as “sets of from 1 to a maximum number of sequential instructions, each set
being provided to the instruction register as a unit and having a boundary, and in which any
operand that is present must be right justified.” Ward Order at 22-23 (Chen Decl., Ex. 2)
(emphasis added). An “instruction group” is synonymous with “multiple sequential instructions,”
as recognized by Judge Ward’s construction. For all of the reasons explained above, the
requirement that operands be right justified in the instruction register should be incorporated into
the construction of this term.

III. CONCLUSION

For the foregoing reasons, Plaintiffs’ constructions should be adopted in their entirety.
Embodiment Described.”) (boldface and underlining in original). TPL cannot cite a single
instance of any operand in the specification that is not right justified.
Dated: January 9, 2012 K&L GATES LLP
By: /s/ Timothy Walker
Timothy P. Walker, Esq.
Howard Chen, Esq.
Harold H. Davis, Jr., Esq.
Jas Dhillon, Esq.
Jeffrey M. Ratinoff
K&L Gates LLP
Four Embarcadero Center, Suite 1200
San Francisco, CA 94111
Attorneys for Acer, Inc., Acer America Corp.
and Gateway, Inc.
Dated: January 9, 2012 COOLEY LLP
By: /s/ Kyle D. Chen
Kyle D. Chen, Esq.
Heidi L. Keefe, Esq.
Mark R. Weinstein, Esq.
Cooley LLP
3000 El Camino Real
Five Palo Alto Square, 4th Floor
Palo Alto, California 94306
Attorneys for HTC Corporation and HTC
America, Inc.
Dated: January 9, 2012 BAKER & MCKENZIE
By: /s/ Edward Runyan
Edward Runyan, Esq.
Baker & McKenzie
130 East Randolph Drive
Chicago, IL 60601
Attorneys for Barco, N.V.

ATTESTATION PER GENERAL ORDER 45

I, Kyle D. Chen, am the ECF User whose ID and password are being used to file Plaintiffs’
Consolidated Responsive Claim Construction Brief. In compliance with General Order 45, X.B., I
hereby attest that the counsel listed above have concurred with this filing.
Dated: January 9, 2012 By: /s/ Kyle D. Chen
Kyle D. Chen