`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`MICRON TECHNOLOGY, INC.,
`Petitioner,
`
`v.
`
`VERVAIN, LLC,
`Patent Owner.
`
`
`
`IPR2021-01549
`U.S. Patent No. 9,997,240
`
`
`
`DECLARATION OF SUNIL P. KHATRI
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Vervaiin Ex. 2001, p. 1
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`TABLE OF CONTENTS
`
`INTRODUCTION ...............................................................................................................1
`
`BACKGROUND AND QUALIFICATIONS .....................................................................1
`
`SCOPE OF ASSIGNMENT AND MATERIALS CONSIDERED ..................................10
`
`PERSON OF ORDINARY SKILL IN THE ART .............................................................12
`
`RELEVANT
`THE
`BACKGROUND OF
`GENERAL
`TECHNOLOGY ................................................................................................................13
`
`A.
`
`B.
`
`C.
`
`D.
`
`SLC and MLC Flash ..............................................................................................16
`
`Address Table ........................................................................................................17
`
`Data Integrity Tests ................................................................................................18
`
`Hot and Cold Data .................................................................................................18
`
`THE PATENTS-IN-SUIT .................................................................................................18
`
`TECHNICAL OPINIONS .................................................................................................20
`
`CONCLUSION ..................................................................................................................26
`
`i
`
`Vervaiin Ex. 2001, p. 2
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`DECLARATION OF SUNIL P. KHATRI, PH. D
`
`I, Sunil P. Khatri, do hereby declare as follows:
`
`INTRODUCTION
`
`I have been retained on behalf of Vervain, LLC (“Vervain”), and its
`
`counsel, McKool Smith, P.C., as an expert in this proceeding. I am personally
`
`knowledgeable about the matters stated herein and am competent to make this
`
`declaration.
`
`I understand that Vervain will submit this Declaration in connection
`
`with their Patent Owner’s Preliminary Response in Micron’s petition for inter partes
`
`review, No. IPR2021-01549.
`
`I receive compensation at an hourly rate of $700 per hour for my time
`
`working on this matter, plus expenses. I have no financial interest in Vervain or in
`
`the patents involved in this litigation, and my compensation is not dependent on the
`
`outcome of this litigation. The conclusions I present are due to my own judgment.
`
`BACKGROUND AND QUALIFICATIONS
`
`I have over thirty-five years of experience with electronics, electrical
`
`engineering, and computer engineering. A copy of my latest curriculum vitae (CV)
`
`is attached hereto as Appendix A and provides further details regarding my
`
`background and qualifications. During my career, I have acquired extensive
`
`knowledge and experience with VLSI circuits, computer architecture, testing,
`
`1
`
`Vervaiin Ex. 2001, p. 3
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`computer-aided design (CAD) algorithms and algorithm acceleration, logic
`
`synthesis, semiconductor memory, redundancy, synchronous and asynchronous
`
`circuits, and related software and hardware topics. Most relevant to the challenged
`
`patents, my technical expertise includes extensive work with semiconductor memory
`
`devices such as DRAM, SRAM and flash. My work with semiconductor memory
`
`devices has included work on 3D integration and novel ring-based memory
`
`architectures, power and speed tradeoffs using selective body bias, architectures and
`
`circuit approaches for processing-in-memory, radiation hardening analysis for
`
`memories, the use of flash transistors for designing logic circuits (such as ternary
`
`Content-addressable Memories (CAMs), Field Programmable Gate Arrays
`
`(FPGAs), and traditional binary-valued as well as ternary-valued digital logic), and
`
`clocking and source-synchronous design. I recently was awarded a research grant by
`
`the Air Force Research Laboratory (AFRL) in Rome, NY, to conduct research in
`
`secure digital circuits using flash-based digital design approaches. Additionally, I
`
`will be submitting a book chapter on the use of flash transistors in novel Very Large
`
`Scale Integrated (VLSI) design applications. My MS thesis involved designing a
`
`memory interface for a multi-threaded Reduced Instruction Set Computing (RISC)
`
`microprocessor.
`
`The following describes some of my relevant experience. I earned my
`
`Bachelor of Science in Electrical Engineering in 1987 from the Indian Institute of
`
`2
`
`Vervaiin Ex. 2001, p. 4
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`Technology, Kanpur, India. After graduating with my B.S. degree, I was a candidate
`
`for a Master of Science degree in Electrical and Computer Engineering at the
`
`University of Texas from 1987–89. At the University of Texas, I held the
`
`Microelectronics and Computer Development (MCD) Fellowship from 1987–89. I
`
`also conducted my M.S. research and wrote my thesis on the design of the METRIC
`
`memory interface and memory system. METRIC was one of the first super-scalar
`
`processors that was developed in the world. I earned an M.S. degree in 1989 from
`
`the University of Texas, Austin.
`
`
`
`After leaving the University of Texas, I worked at Motorola Inc. from
`
`1989–93 as a design engineer for the MC88110 reduced instruction set computing
`
`(RISC) microprocessor team. My duties included the design of digital and analog
`
`circuitry, test logic and circuits, JTAG boundary scan design, input/output driver
`
`design, and clock phase-locked loop (PLL) logic. During my time at Motorola, I was
`
`independently responsible for the design of the factory test controller of the
`
`MC88110 microprocessor. I performed all attendant tasks in a “vertical” VLSI
`
`design methodology, which included high-level modeling, circuit and layout design
`
`and verification, as well as global and detailed routing. I also helped in the design of
`
`the Translation Lookaside Buffer (TLB) unit, which included a static random-access
`
`memory (SRAM) block.
`
`
`
`3
`
`
`
`Vervaiin Ex. 2001, p. 5
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`In 1999, I earned a Doctor of Philosophy degree in Electrical
`
`Engineering and Computer Sciences from the University of California, Berkeley.
`
`While at Berkeley, I held the California Microelectronics (MICRO) Fellowship in
`
`1993.
`
`I joined the faculty at the University of Colorado, Boulder, in 2000 as
`
`an Assistant Professor of Electrical and Computer Engineering. At the University of
`
`Colorado my research focused on VLSI logic design automation, VLSI layout design
`
`automation, and VLSI design methodologies to address Deep Submicron (DSM)
`
`issues such as crosstalk and power.
`
`I joined the faculty at Texas A&M University in 2004 as an Assistant
`
`Professor in Electrical and Computer Engineering. In 2010 I was promoted to
`
`Associate Professor in Electrical and Computer Engineering. In 2015, I was
`
`promoted to full Professor in Electrical and Computer Engineering. My research
`
`focuses on three primary areas: the first is computer systems, including computer
`
`architecture from the circuits up, and algorithm acceleration using GPUs, FPGAs
`
`and custom ICs. The second is logic and its applications, while the third area consists
`
`of interdisciplinary extensions of the first two. Some specific recent research topics
`
`include circuit design using floating gate devices, wireless power delivery, battery-
`
`less electronic systems, machine learning architectures, secure computing
`
`approached from both a hardware and software perspective, and the mathematics of
`
`4
`
`Vervaiin Ex. 2001, p. 6
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`scalable cryptocurrency. One of my new focus areas is intelligent and secure
`
`computing viewed from the hardware (circuit) as well as the software levels. I am
`
`conducting research on tamper-proof memory techniques, as well as multi-row read
`
`architectures for SRAM, DRAM and flash memory arrays.
`
` At Texas A&M I teach classes that cover memories extensively,
`
`featuring thorough discussion of sense amplifiers, row and column decoders, and
`
`different types of memory circuits. For example, in Electrical and Computer
`
`Engineering (ECEN) 752 “Advances in VLSI Circuit Design,” a graduate level
`
`course, I cover all aspects of VLSI design, including memory design. In ECEN
`
`449/749 “Microprocessor System Design,” and in ECEN 752, I cover memories,
`
`including flash memories, the design of flash memory cells, the organization of
`
`memories into multiple banks, and the division of data across multiple banks of
`
`memory, as well as other design techniques that can be used to optimize and manage
`
`memories. This course is attended by both undergraduate (ECEN 449) and graduate
`
`(ECEN 749) students. In ECEN 454 “Digital Circuit Design,” a senior
`
`undergraduate course, I cover circuit design techniques for memory in detail. The
`
`Ph.D. thesis of one of my recent doctoral students dealt with the use of flash
`
`transistors to design logic circuits. The research of a recent M.S. student entailed a
`
`new ring-based source synchronous architecture for 3D DRAM technologies, which
`
`has been published at a conference and in a journal and is being submitted for
`
`5
`
`Vervaiin Ex. 2001, p. 7
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`dissemination as a research monograph. In the past, I have conducted research into
`
`new topologies for efficient memory redundancy as part of a course project for my
`
`graduate course.
`
` Since 2000, I have earned 24 research contracts from funders including
`
`Intel, the National Science Foundation, the National Security Agency, Altera
`
`Corporation,
`
`the National Center
`
`for Atmospheric Research, National
`
`Semiconductor Corporation, and several private sources. The total amount for these
`
`research grants is $17.53 million, of which my portion is $2.85 million.
`
`
`
`I have a total of over 268 peer-reviewed publications. Among these
`
`papers, five received a best paper award, while six others received best paper
`
`nominations (including one journal best paper nomination). An additional three
`
`journal papers and two conference paper are currently undergoing peer review. I
`
`have co-authored nine research monographs, one edited research monograph, and
`
`three book chapters. Additionally, I have six awarded U.S. Patents (one of which
`
`was filed during my tenure at Texas A&M), two filed provisional U.S. Patents, and
`
`another U.S. Patent which is currently under review and was also submitted during
`
`my tenure at Texas A&M. I have co-authored one invited journal paper and 13
`
`invited conference or workshop papers (including one from Design Automation
`
`Conference (DAC) and one from Allerton). Moreover, I was invited to serve as a
`
`panelist at a conference seven times and have presented two conference tutorials. I
`
`
`
`6
`
`
`
`Vervaiin Ex. 2001, p. 8
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`received the “Outstanding Professor Award” in the ECE Department at Texas A&M
`
`University in 2007 and also in 2020. My H-index is 33 (per Google Scholar).
`
` Since 2003, I have published numerous research monographs, journal
`
`papers, and conference papers on flash transistors and memory systems, as detailed
`
`in my attached CV. A few papers on relevant subject areas authored or co-authored
`
`by me include:
`
`•
`
`Fast, Ring-based Design of 3D Stacked DRAM, IEEE Transactions
`
`on Very Large Scale Integrated Circuits (TVLSI), IEEE Transactions
`
`on Very Large Scale Integration (VLSI) Systems. Vol 27 number 8,
`
`•
`
`•
`
`•
`
`•
`
`•
`
`Aug 2019. pp 1731-1741.;
`
`Fast, Ring-Based Design of 3D Stacked DRAM, IEEE International
`
`Conference on Computer Design 2017: pp 665-672;
`
`Selective Forward Body Bias for High Speed and Low Power SRAMs,
`
`Journal of Low Power Electronics, Vol. 5, No. 2, Aug. 2009, pp. 185-
`
`95;
`
`Low Power and High Performance SRAM Design using Bank-based
`
`Selective Forward Body Bias, IEEE/ACM Great Lakes Symposium on
`
`VLSI, May 10-12, 2009, Boston, MA, pp. 441-44;
`
`Modeling Dynamic Stability of SRAMs in the Presence of Single Event
`
`Upsets (SEUs), IEEE International Symposium on Circuits and
`
`Systems, May 18-21, 2008, Seattle, WA, pp. 1788-91;
`
`“Design of a Flash-based Circuit for Multi-valued Logic”, Proceedings
`
`of the Great Lakes Symposium on VLSI (GLSVLSI) 2017, pp 41-46,
`
`May 10-12, 2017. Banff, Canada.
`
`7
`
`Vervaiin Ex. 2001, p. 9
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`•
`
`"SAT-Based Optimization
`
`for Flash-Based Digital Designs",
`
`IEEE/ACM Design Automation Conferenec (DAC), Jun 18-22 2017,
`
`Austin, TX.
`
`•
`
`"A Flash-based Digital Circuit Design Flow", IEEE/ACM International
`
`Conference on Computer-Aided Design (ICCAD) 2016, Austin, TX,
`
`Nov 2016.
`
`•
`
`"Implementing low power digital circuits using flash devices", 2016
`
`IEEE 34th International Conference on Computer Design (ICCD), pp
`
`109-116, Oct 3-5, 2016, Phoenix, AZ.
`
`•
`
`•
`
`•
`
`"Exploring Flash Devices to Implement Digital Circuits", IEEE/ACM
`
`Design Automation Conference (DAC), June 2016, Austin, TX.
`
`FTCAM: An Area-efficient Flash-based Ternary CAM Design, IEEE
`
`Transactions on Computers, Vol. 65, No. 8, Aug. 2016, pp. 2652-58;
`
`An Area-efficient Ternary CAM Design Using Floating Gate
`
`Transistors, IEEE International Conference on Computer Design, Oct.
`
`19-22, 2014, Seoul, S. Kor., pp. 55-60; and
`
`•
`
`A Fast Ternary CAM Design for IP Networking Applications,
`
`International Conference on Computer Communications and Networks,
`
`October 22, 2003, Dallas, TX, pp. 434-39 (awarded best paper).
`
`
`In addition to my work on the papers listed above, I have also served as an
`
`editor for IEEE Transactions on Computers, ACM Transactions on Design
`
`Automation of Electronic Systems, and MDPI Journal of Electronics.
`
`
`
`I have served as EDA Track Co-Chair for ICECS 2014, Panel Chair for
`
`TexasWISE 2014, Track Co-Chair (VLSI Systems, Applications and Computer
`
`
`
`8
`
`
`
`Vervaiin Ex. 2001, p. 10
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`Aided Design track) for ICECS 2013, Poster Session Chair for TexasWISE 2013,
`
`Advisory Committee for HotPI 2013, Panel Session Chair for SLiP 2013, Track
`
`Chair (Logic track) for ICCAD 2009-10, 2015-17, Track Chair (logic track) for DAC
`
`2016-17, General Chair for IWLS 2009, Technical Program Chair for IWLS 2008,
`
`Track Co-Chair, Computer Aided Network DEsign (CANDE) Track, for ISCAS
`
`2008-10, Track Co-chair for the DSP track for ISCAS 2022, Track Co-Chair, Test
`
`and Methodologies Track, for ICCD 2007, Panel Chair for ITSW 2009, Publicity
`
`Co-Chair for GLS-VLSI 2009, and as a member of the TPC for several conferences.
`
`
`
`I am generally familiar with the analysis of patents. I am a named
`
`inventor on the following U.S. Patents:
`
`•
`
`•
`
`•
`
`•
`
`•
`
`Data Processing System Having Serial Self Address Decoding and
`
`Method of Operation, United States Patent No. 5,347,523, issued
`
`September 13, 1994;
`
`Circuit Identifier for Use with Focused Ion Beam Equipment, United
`
`States Patent No. 5,408,131, issued April 18, 1995 (“the ’131 patent”);
`
`Driver Circuit with Self-Adjusting Impedance Matching, United States
`
`Patent No. 5,448,182, issued September 5, 1995 (“the ’182 patent”);
`
`Circuit Identifier for Use with Focused Ion Beam Equipment, United
`
`States Patent No. 6,156,579, issued December 5, 2000 (“the ’579
`
`patent);
`
`Datapath Design Methodology and Routing Apparatus, United States
`
`Patent No. 6,598,215, issued July 22, 2003;
`
`9
`
`Vervaiin Ex. 2001, p. 11
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`•
`
`Low Power Reconfigurable Circuits with Delay Compensation, United
`
`States Patent No. 7,880,505, issued February 1, 2011.
`
`
`
` The ’131 and ’579 patents are directed to an identification means for
`
`redundant circuits that distinguishes said circuits by respective function. This allows
`
`for identification by focused ion beam equipment, which can then repair, replace, or
`
`supplement circuits as necessary. These patents disclose a scheme for replacing
`
`defective cells or circuits within a larger circuit.
`
` The ’182 patent relates to a driver circuit capable of switching from one
`
`driving impedance to a second in response to the output signal of a first driver portion
`
`reaching a predetermined voltage. It discloses a driver circuit capable of adjusting
`
`circuit configurations depending upon the current output state.
`
` SCOPE OF ASSIGNMENT AND MATERIALS CONSIDERED
`
`
`
`I have been retained by Vervain (“Patent Owner”) to provide an
`
`explanation to the Board regarding the inventions described in the Patents-In-Suit. I
`
`understand Petitioner Micron Technology, Inc. (“Micron” or “Petitioner”), has filed
`
`petitions for inter partes review, Nos. IPR2021-01547, -01548, -01549, and -01550,
`
`against each of the patents-in-suit. I have been retained to provide my opinions
`
`regarding various technical issues relating to the validity of the patents-in-suit over
`
`the prior art references identified by Micron’s petitions.
`
`
`
`10
`
`
`
`Vervaiin Ex. 2001, p. 12
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`In preparing this Declaration, I am relying on my own knowledge and
`
`expertise as well as the following documents:
`
`• The Patents-In-Suit:
`
`U.S. Patent No. 8,891,298 to G.R. Mohan Rao (“298
`patent”)
`
`U.S. Patent No. 9,196,385 to G.R. Mohan Rao (“385
`patent”)
`
`U.S. Patent No. 9,997,240 to G.R. Mohan Rao (“240
`patent”)
`
`U.S. Patent No. 10,950,300 to G.R. Mohan Rao (“300
`patent”);
`
`• Prosecution histories for the Patents-In-Suit, which I understand
`
`constitute the exchange of correspondence between the Patent Office
`
`and the applicant;
`
`• Micron’s petitions for inter partes review, Nos. IPR2021-01547, -
`
`01548, -01549, and -01550, as well as the declarations of Dr. David Liu
`
`submitted in support thereof;
`
`• The alleged prior art references cited in Micron’s petitions, including
`
`U.S. Patent Application Publication Nos. 2011/0099460 (“Dusija”),
`
`2008/0140918 (“Sutardja”), and 2010/0017650 (“Chin”); and
`
`• The other references cited within this Declaration.
`
`11
`
`Vervaiin Ex. 2001, p. 13
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`PERSON OF ORDINARY SKILL IN THE ART
`
` When interpreting a patent, I understand that it is important to view the
`
`disclosure and claims of that patent from the level of ordinary skill in the relevant
`
`art at the time of the invention. My opinion of the level of ordinary skill in the art
`
`with regard to the Patents-in-Suit is based on my personal experience working and
`
`teaching in the fields of electrical engineering and computer science, including work
`
`with DRAM technologies, my knowledge of the background and education of
`
`colleagues and others working in that general field as of and for several years prior
`
`to the 1999 to 2001 time frame, my study of the Patents-in-Suit, and its file history,
`
`and my knowledge of:
`
`• The level of education and experience of persons actively working in the
`
`field at the time the subject matter at issue was developed;
`
`• The types of problems encountered in the art at the time the subject matter
`
`was developed;
`
`• The prior art patents and publications;
`
`• The activities of others working in that same technical field;
`
`• Prior art solutions to the problems addressed by the relevant art; and
`
`• The sophistication of the technology at issue in this case.
`
`
`
`In determining the level of ordinary skill in the art, I also considered
`
`the following factors: (1) the sophistication of the relevant technology; (2) the
`
`12
`
`Vervaiin Ex. 2001, p. 14
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`rapidity with which innovations are made in that field; and (3) the educational level
`
`of active workers in that field. It is my further understanding that these factors are
`
`not exhaustive and are merely a useful guide to determining the level of ordinary
`
`skill in the art.
`
` Taking the above factors into account, in my opinion a POSITA in the
`
`technology field of the Patents-in-Suit would be a person with at least a Bachelor of
`
`Science degree in electrical engineering, computer engineering, or a closely related
`
`field, with at least 3-5 years of experience in the design of non-volatile memory
`
`devices. An individual with an advanced degree in a relevant field would require
`
`less experience in the design of non-volatile memory devices.
`
`
`
`I understand that Micron’s expert, Dr. David Liu, agrees with this
`
`definition. Ex. 1009, ¶ 7. This accords with my experience. Many of the individuals
`
`hired by semiconductor manufacturing companies at the time of the invention did
`
`not have graduate degrees, and thus the level of ordinary skill in the art should
`
`specifically include such individuals.
`
` GENERAL BACKGROUND OF THE RELEVANT TECHNOLOGY
`
` Volatile memory, such as static random access memory (“SRAM”) and
`
`dynamic random access memory (“DRAM”), lose memory when power is turned
`
`off. However, more persistent memory is needed for many applications, such as
`
`photos in a digital camera, bootable code or settings in circuits, or a wide range of
`
`
`
`13
`
`
`
`Vervaiin Ex. 2001, p. 15
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`data storage needs. Non-volatile memories (e.g., thumb drives, hard drives, and
`
`compact discs) can store information after the system is powered off. Traditionally,
`
`media such as hard disks, floppy disks, compact discs, or magnetic tapes was used
`
`for persistent storage in computing systems, but these media are large, bulky, and
`
`slow, and use a large amount of power. Flash memory is a specific type of non-
`
`volatile memory, where data is stored in “blocks” of “pages.” Flash memory chips
`
`have come to be used for persistent data storage in a wide range of applications.
`
`Data stored in flash memory persists across power on/off cycles and has a small size,
`
`high performance, and low power consumption.
`
` Flash memory uses a special type of transistor that has both a “floating
`
`gate” and a “control gate.” In a floating gate transistor, charge is stored on an
`
`isolated conductor, called the floating gate. This charge has no path to dissipate.
`
`The charge on the floating gate controls the current flowing between the source and
`
`drain. This current allows the user to determine the value stored in the cell.
`
` To erase the cell, a reverse voltage is applied between the drain and the
`
`control gate. Charge is then dissipated through tunneling.
`
`14
`
`Vervaiin Ex. 2001, p. 16
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
` Floating gate transistors individually are limited in endurance.
`
`Depending on the type of transistor and how it is configured, an individual transistor
`
`may not work reliably after it has been programmed and erased too many times. For
`
`example, endurance for some flash transistors is on the order of 10,000 to 100,000
`
`program-erase cycles.
`
` Because of the limited endurance of floating gate transistors, and
`
`because some locations in a memory may be very frequently rewritten, it is not
`
`always feasible to use the same transistor each time for each memory location.
`
`Rather, flash memory typically includes a controller that uses a flash translation
`
`layer to map logical addresses presented to the host to physical addresses used to
`
`address the physical flash memory. This flash translation layer allows for wear to
`
`be leveled across all the transistors on a device, and for bad blocks to be managed
`
`and avoided.
`
`15
`
`Vervaiin Ex. 2001, p. 17
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`A.
`
`SLC and MLC Flash
`
`Early flash memory stored only a “0” or a “1” in each transistor. This
`
`is today known as “SLC,” or “single level cell,” flash. Later on, to increase the
`
`density of storage in flash memory, a technique known as “MLC”, or “multiple level
`
`cell,” was introduced, where multiple threshold voltage levels in each transistor
`
`would represent multiple bits of data. For example, the following diagram shows
`
`how in SLC flash, large voltage ranges are assigned to the “1” and “0” bits
`
`respectively, whereas in MLC flash, smaller voltage ranges are assigned to multiple-
`
`bit values such as “11,” “10,” “01,” and “00.”
`
`Chen et al., Ultra MLC Technology Introduction, Advantech Technical White Paper
`
`(Oct. 5, 2012), 3.
`
` The primary difference between SLC and MLC is what data each
`
`voltage represents. With SLC flash, the transistor stores only a 1 or 0, so a wide
`
`range of voltages can be allotted to a single bit. This allows for faster and more
`
`reliable memory access. On the other hand, MLC flash must be slowly and carefully
`
`programmed using a narrow, precise range of voltages, with each voltage range
`
`16
`
`Vervaiin Ex. 2001, p. 18
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`representing a specific sequence of bits (see the figure above, which shows four
`
`sequences of bits—11, 10, 01, and 00—corresponding to different ranges of
`
`threshold voltages).
`
` Flash memories sometimes use dual-mode flash, where the cells can be
`
`configured as SLC or MLC. When the cell is in SLC-mode, it does not use multiple
`
`levels per cell, but instead uses a single dividing line between ‘1’ and ‘0’.
`
`B.
`
`Address Table
`
`To provide wear leveling, garbage collection, and bad block
`
`management, a translation layer is used to map logical addresses to actual physical
`
`locations. As part of this translation layer, tables are widely used in order to map
`
`sectors and pages from logical to physical. These tables map logical blocks to
`
`physical blocks. Using a “block” or similar granularity is important, since flash
`
`memory is arranged so that when erasing and rewriting data, all the memory in a
`
`block is erased together. Therefore, Dr. Rao explained that “[t]he address ranges
`
`within the translation table will assume some minimum quantum, such as, for
`
`example, one block, although a smaller size, such as one page could be used, if the
`
`NAND flash has the capability of erasing the smaller size quantum.” 240 patent,
`
`5:46-50. Dr. Rao further explained that memory is written and mapped on the
`
`granularity of a “quantum,” such as a block or page. 240 Patent, Figs. 3A-B.
`
`17
`
`Vervaiin Ex. 2001, p. 19
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`C.
`
`Data Integrity Tests
`
`When data is stored in MLC memory, it is more prone to errors, and
`
`some data is more prone to errors than other data. Errors can occur when writing or
`
`reading the data. Errors can also be caused by the data stored in neighboring cells.
`
`A data integrity test is a test that checks the integrity of the data (i.e., whether errors
`
`have occurred). This test can be run immediately after data is written, or at a later
`
`time. If the test reveals a problem such as corrupt data, the data can be remapped to
`
`higher-performance SLC or less-used MLC, and the address table is modified
`
`accordingly. 240 Patent, 4:24-30.
`
`D.
`
`Hot and Cold Data
`
`One of the key features of Dr. Rao’s invention is how it distinguishes
`
`between “hot” blocks (which receive more frequent writes), and “cold” blocks
`
`(which receive less frequent writes). 240 Patent, 6:46-52. Because SLC flash has
`
`the endurance to handle frequent writes, “hot” blocks can be allocated to SLC flash
`
`to increase the lifetime of the system. “Cold” blocks, on the other hand, can be
`
`allocated to MLC flash to take advantage of its higher density storage.
`
`THE PATENTS-IN-SUIT
`
` The Patents-in-Suit relate to systems for storing data using SLC and
`
`MLC flash memory.
`
`18
`
`Vervaiin Ex. 2001, p. 20
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
` U.S. Patent No. 8,891,298 (the “’298 patent”), entitled “Lifetime Mixed
`
`Level Non-Volatile Memory System,” issued on November 18, 2014. I understand
`
`the Asserted Claims from the 298 patent include claims 1, 3-5, and 11.
`
` U.S. Patent No. 9,196,385 (the “’385 patent”), entitled “Lifetime Mixed
`
`Level Non-Volatile Memory System,” issued on November 24, 2015. I understand
`
`the Asserted Claims from the 385 patent include claims 1, 3-5, and 11-13.
`
` U.S. Patent No. 9,997,240 (the “’240 patent”), entitled “Lifetime Mixed
`
`Level Non-Volatile Memory System,” issued on June 12, 2018. I understand the
`
`Asserted Claims from the 240 patent include claims 1-2 and 6-7.
`
` U.S. Patent No. 10,950,300 (the “’300 patent”), entitled “Lifetime
`
`Mixed Level Non-Volatile Memory System,” issued on March 16, 2021. I
`
`understand the Asserted Claims from the 300 patent include claims 1-12.
`
`G.R. Mohan Rao is the sole named inventor of the four Patents-in-Suit.
`
`
`
`In the Asserted Claims, data is stored in non-volatile memory using
`
`single level cell (SLC) memory that stores 1 bit per cell, and multiple level cell
`
`(MLC) memory that stores more than 1 bit per cell. There are pros and cons to SLC
`
`and MLC flash. SLC is faster and less prone to errors, but requires more space and
`
`power to store a given amount of data. The opposite is true of MLC. MLC flash is
`
`slower and more prone to errors, but stores data more densely with less power
`
`consumption.
`
`19
`
`Vervaiin Ex. 2001, p. 21
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
` The Asserted Claims are directed to specific techniques for efficiently
`
`using SLC and MLC flash to improve the overall performance of the memory. For
`
`example, if certain data is more prone to errors, or is used more frequently, then it is
`
`transferred to higher-performance SLC or less-used MLC. By doing so, the number
`
`of errors is reduced, and the overall endurance of the memory is increased.
`
`TECHNICAL OPINIONS
`
` When data is stored in MLC memory, it is more prone to errors, and
`
`some data is more prone to errors than other data. One reason for this is that the
`
`threshold voltage intervals for MLC memory are smaller than the threshold voltage
`
`intervals for SLC memory, and thus, more errors can occur when writing or reading
`
`the MLC data. Errors can also be caused by the data stored in neighboring cells. A
`
`data integrity test is a test that checks the integrity of the data (i.e., whether errors
`
`have occurred). This test can be run immediately after data is written, or at a later
`
`time.
`
` Flash memory systems typically have two forms of addressing: a
`
`“logical address,” which is an address from the perspective of an executing
`
`application program and a “physical address,” indicating the specific location in a
`
`flash memory chip where the data is stored.
`
` During operation of the flash memory, logical addresses are frequently
`
`remapped to new physical locations. Over time, a particular logical address may be
`
`20
`
`Vervaiin Ex. 2001, p. 22
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`
`
`mapped or associated with many different physical locations (blocks). And multiple
`
`logical addresses may point to the same block over time, so there is not a one-to-one
`
`correspondence between the logical addresses and the blocks over time.
`
` A system that determines how frequently data is written to each of the
`
`logical addresses will not necessarily determine which of the physical blocks are
`
`accessed most frequently.
`
` A logical address may be remapped to a new physical address due to,
`
`for example, wear leveling, garbage collection, or bad block management. When
`
`this occurs, the frequency of writes to the logical address may be different from the
`
`frequency of writes to the physical block that it is currently mapped to. If, for
`
`example, “LogicalAddressA” is remapped from Block1 to Block2, there may be 100
`
`writes to LogicalAddressA, but only 70 writes to Block1 and 30 writes to Block2.
`
` Additionally, over time, multiple logical addresses may point to the
`
`same block. When this occurs, the frequency of writes to the logical addresses may
`
`be different from the frequency of writes to the blocks. If, for example,
`
`“LogicalAddressA” is mapped to Block1; and “LogicalAddressB” is first mapped to
`
`Block2, and then “LogicalAddressC” is subsequently mapped to Block2, the
`
`frequency of the writes to the Logical Addresses may provide a distorted view of the
`
`frequency of the writes to the blocks. For example, Block2 may have 150 writes and
`
`Block1 may have 100 writes, while LogicalAddressA may have 100 writes, and each
`
`
`
`21
`
`
`
`Vervaiin Ex. 2001, p. 23
`Micron v. Vervaiin
`IPR2021-01549
`
`
`
`of LogicalAddressB and LogicalAddressC have 75 writes. In this example,
`
`LogicalAddressA may have the most writes, but Block2 will have more writes than
`
`Block1.
`
`
`
`In Sutardja