Site Home Archive Home FAQ Home How to search the Archive How to Navigate the Archive
Compare FPGA features and resources
Threads starting:
Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Peter wrote: > > >With switched current circuits you can do it on a pure digital process. > >With switched capacitor circuits you can do it on a digital process with > >some analog features. [Good resistors and capacitors.] > > Plus a comparator. > > And with "good resistors" and a comparator you could do an ADC as > well, using just a successive approximation register as the only > logic. A switched cap ADC is just that, instead of resistors it uses > caps. > > I think someone is pulling our leg here, unless this "model" is a > freebie, or unless it contains the pretty complex algorithm used to > self-calibrate the switched-cap ADCs. I've been making SPICE models of ADC's that way for years - and teaching others how to do it in my modelling training courses. The question of calibration never arises, since the capacitors can have any tolerance you care to specify. Nope, he's telling you the truth. Want to join my class...? -- Best regards, Mark ------------------------------------------------------------------------------ Mark Sitkowski 笑比哭好! Australian Computing and Communications Institute 723 Swanston Street Carlton Victoria 3053 ------------------------------------------------------------------------------ Phone: (613-9) 282-2530 E-mail: marks@acci.com.au Fax: (613-9) 282-2534 WWW: http://www.acci.com.au/People/marks.html Home phone: (613-9) 729-0731 给我打电话! Home fax: (613-9) 720-1487 ------------------------------------------------------------------------------Article: 3651
Felix K.C. CHEN <flxchen@diig.dlink.com.tw> wrote: >** internal error: vhdlsim: >Please Report (No attributes on signal /T_BENCH/TO_SW/ESWCLK) >FAULT CONTEXT > program: 'vhdlsim' > release: '3.3b' > Architecture: 'sparc' > phase: 'Run-time' > last UI Command: run 200 > simulation time = 0ps >FAULT ID: >'2304524 374576 454228 323452 324296 79012 3049100 3046720 300676 8300' > >Have any of you encountered similar troubles before? I need help. Felix, you really haven't given enough information here to enable anyone but the Synopsys Hotline to help you. That is, no one but them can use just FAULT ID's to determine what a bug is. Good luck! - John Cooley Part Time EDA Consumer Advocate Full Time ASIC, FPGA & EDA Design Consultant =========================================================================== Trapped trying to figure out a Synopsys bug? Want to hear how 4488 other users dealt with it ? Then join the E-Mail Synopsys Users Group (ESNUG)! !!! "It's not a BUG, jcooley@world.std.com /o o\ / it's a FEATURE!" (508) 429-4357 ( > ) \ - / - John Cooley, EDA & ASIC Design Consultant in Synopsys, _] [_ Verilog, VHDL and numerous Design Methodologies. Holliston Poor Farm, P.O. Box 6222, Holliston, MA 01746-6222 Legal Disclaimer: "As always, anything said here is only opinion."Article: 3652
Subject: Ref:n.5972 - ASIC/VHDL, Graphics Chips Development, to 35k, CAMBRIDGE This young, dynamic company is at the cutting edge of the high profile sector of specialist semiconductor development, primarily in graphics applications. They seek a number of high calibre ASIC Designers to progress the next generation of projects. With a good degree in a relevant subject, you will have extensive experience developing devices of 30/40K gates and upwards and have solid VHDL skills. All the roles will involve high level design and testing and will be extremely challenging. These are excellent opportunities to develop your career in this exciting development environment at the cutting edge of this high profile sector. -- For further information on ECM visit www.ecmsel.co.uk Please contact us by Email (Cvs in plain ASCII text - not coded!) topjob@ecmsel.co.uk ------ Alternatively Snail, Fax or Phone: ECM Selection Ltd, The Maltings, Burwell, Cambridge, CB5 0HB. Tel: 01638 742244 Fax: 01638 743066Article: 3653
Subject: Ref:n1275 - Advanced Network Products,Top UK Co., to c55k, SOUTHERN ENGLAND ADVANCED NETWORK PRODUCTS Negotiable to c55,000 PLEASE QUOTE REF: 1275. With one of the UK's most exciting, high-growth companies engaged in the research and product development of 'next generation' computer communication & network technologies. These new appointments, reporting at Board level, are created as part of a planned development programme. SOFTWARE DEVELOPMENT MANAGER Responsible for all aspects of Software Product Development, from concept to release, for both stand alone and PC based products. You should have a sound background gained in lifestyle Software Development - ideally for Network Management, Switching or related products, coupled with excellent Project Management skills and the ability to manage and motivate a team of c12 software specialists. HARDWARE DESIGN MANAGER To take responsibility for Product Hardware Design & Innovation. Ideally you should have a strong background in the design of digital products for medium/high volume manufacture, coupled with excellent Team Leadership & Design Management skills. Experience with ASIC/FPGA based Network, Bridges, Switches, Graphics plug-ins or related products would be particularly valuable. SOFTWARE & HARDWARE ENGINEERS 20k -45k Additionally we seek six top-calibre Software & Hardware Specialists - from fresh Graduate to Technical Specialist - to work at the forefront of high-speed networking technology. Ideally you should have an excellent academic track record, enjoying the challenge of working at the forefront of technology and have experience of either low-level Embedded Asynchronous Device Driver, or similar software, or digital FPGA-based hardware design. A knowledge of Ethernet, Token Ring or ATM Networking; IP routing; PBX Core Switching; DPNSS; ITU Qxxx or Hxxx Standards; QSIG or related technologies would be a significant plus. The importance placed upon Product Innovation as the key to continued market leadership is indicated by the Upper Echelon Salaries, Comprehensive Benefits Package, excellent Facilities and Career Development Opportunities which are offered. -- For further information on ECM visit www.ecmsel.co.uk Please contact us by Email (Cvs in plain ASCII text - not coded!) topjob@ecmsel.co.uk ------ Alternatively Snail, Fax or Phone: ECM Selection Ltd, The Maltings, Burwell, Cambridge, CB5 0HB. Tel: 01638 742244 Fax: 01638 743066Article: 3654
testArticle: 3655
Hi, I am embarking on an XC4013E design and have a couple of questions... 1. I am bringing a bidirectional 32-bit bus into the device which essentially looks like a microprocessor bus. It is used to load and read back registers, read back status, etc. There are two ways to design this. One is to create a single 32-bit bus using the internal tri-states, and everything gets dumped onto that bus. The second is two create one 32-bit bus coming in from the pins (writing to internal device registers) and lots of little buses which get muxed together for the readback path. The question is, which of these is better? It seems to me that the tri-state uses less routing resources since it is only one bus, but that the placement (and hence routing) is more critical due to the restricted placement on the TBUFs. The muxed approach requires more routing resources and CLBs. Any opinions from people who have had to make this decision? 2. The second question is regarding the preferred method for using the floor planner. Is it better to floor plan before PPR or after? I realize that the answer to this is design dependent, so here are some details about the design: It has some arithmetic elements (adders, subtracters) and many large counters. It contains some logic which must run at a fairly reasonable speed (33 MHz). It is fairly datapath intensive. Thanks for your responses! -- ############################################################################ # # # Dave Matthews djm@papillonres.com # # # # Papillon Research Corp. Providing Full Product Development # # (formerly Chrysalis Research Corp.) including: # # 52 Domino Dr., Concord, MA 01742 * System Architecture # # Phone - (508)371-9115 * ASIC and FPGA Design # # Fax - (508)371-9175 * VHDL/Verilog Modeling & Synthesis # # * EMI - EMC Consulting Services # # # # (Chrysalis Research Corp. is not affiliated with Chrysalis Symbolic # # Design, Inc.) # ############################################################################Article: 3656
---------------------------- Achim Gratz writes: >What do I have to tell our distributor here in Europe to get a kit? >If they don't have it, what would be the non-800 number to call in >the U.S.? ---------------------------- Just ask your distributor for the DSP Design Kit from Altera. Otherwise, there are two options: Contact our main European Sales office in the UK: 44 1628488811 or, contact us here in the US: (408) 894-7000, ask for the Literature Department. Regards, JackArticle: 3657
Hi, Has anyone received their order of Xilinx XC4000E FPGA's lately? We have ordered 5 XC4025E's, and it's been about 2 month since we placed the order, and they keep on telling us the shipment is being delayed. -Joey ChenArticle: 3658
Hi, Anyone of you out there had experience using both Xilinx and Altera FPGA's? Since our Xilinx FPGA (XC4025E) order has been delayed yet again, I would like to know how the Altera Flex 10K series compares with the Xilinx XC4000E in terms of capacity. The Xilinx XC4025E claims capacity of 25,000 "typical" gates, and the Altera Flex 10K-100 claims 100,000 "typical" gates. Are they comparing the same thing here? Thanks. -Joey ChenArticle: 3659
> Sure, but for those of us in the US, is there a printed or web > reference you could point me to? I am an EE, doing digital chips, but > am very interested in broadening my horizons. > > TIA > Garfield Hi Garfield, I'm afraid analog modelling is a bit of a black art and, since it's perceived as 'difficult', few people acquire the skills to do it (in fact, there's only about 50 of us in the world get involved in it in a major way). If you're an EE, you're 70% of the way there, but the reading material is a bit thin. There's only one book on macro-modelling ever been published (its title is something original, like 'Macro-Modelling') by Prentice-Hall, but I wouldn't use their ADC model. It's a software implementation of the hardware. If the method you're using permits it, try this: First make a DAC, using weighted current sources, switched (multiplied) by a set of digital inputs ('1' and '0', not '5' and '0'!). Run this at a clock rate of about 3GHz, and put the analog output into a comparator, whose other input is the analog input. Differentiate (or whatever...) the comparator output to give you a 'CONVERSION COMPLETE' strobe, and drive a latch with this. If you have an edge triggered latch, you don't need to differentiate. The data inputs to the latch are the DAC outputs, and the latch output is the ADC output. The clock rate is equal to the inverse of the conversion time, and the monotonicity is set by the accuracy with which you specify the weighting of the current sources. The conversion glitches can be introduced by bending the response times of the various latch elements. Hope this helps. -- Best regards, Mark ------------------------------------------------------------------------------ Mark Sitkowski 笑比哭好! Australian Computing and Communications Institute 723 Swanston Street Carlton Victoria 3053 ------------------------------------------------------------------------------ Phone: (613-9) 282-2530 E-mail: marks@acci.com.au Fax: (613-9) 282-2534 WWW: http://www.acci.com.au/People/marks.html Home phone: (613-9) 729-0731 给我打电话! Home fax: (613-9) 720-1487 ------------------------------------------------------------------------------Article: 3660
In article <31DA0270.41C67EA6@eede.ericsson.se> Andreas Hofmann <eedanho@eede.ericsson.se> writes: > There seems to be some problems with the implementation of their user > interface. I couldn't see the problem you described, but I have a nice > oddity, too. If I try to get the online help from the menu HELP the > message appears: [help won't come up] I get the same message occasionally. IMHO the new user interface sucks. I think it's actually from another company so we can't really blame Altera too much. I'm sure it makes their job easier, and it makes it easier for them to keep the PC version in line with the UNIX version (which at 5.1 was missing a whole lot of stuff that was in the PC version). If the people that write the "Windows-Under-X" code would get their act together it would be OK. As it is you have to run another program to reserve enough colors for the app (i guess they never heard of private colormaps) and of course it doesnt understand my virtual windowmanager (FVWM), so it's always stuck to the glass, which is pretty annoying. And when iconified it doesnt follow the rules either. However, I love MaxPlus2 6.x. rob not speaking for SunArticle: 3661
> If the people that write the "Windows-Under-X" code would get their > act together it would be OK. As it is you have to run another program > to reserve enough colors for the app (i guess they never heard of > private colormaps) and of course it doesnt understand my virtual > windowmanager (FVWM), so it's always stuck to the glass, which is > pretty annoying. And when iconified it doesnt follow the rules either. > > However, I love MaxPlus2 6.x. You're speaking from the bottom of one's heard! Even if I've accepted lot's of this oddities now, I expect them to improve their user interface soon. It would make the live of a designer much less boring if you don't have to fight against a strange behaving application on your desktop (speaking from the X point of view). Let's see! This morning I've found the Rev. 6.2 on my desk. I'll be back with comments when I've installed it. Otherwise let's wait until Rev. 7.x? -- _/ _/ _/ EED/E/X/A/ Andreas Hofmann Phone :+49-5121-707-378 _/ _/ _/ Ericsson Eurolab Deutschland FAX :+49-5121-707-333 _/ _/ _/ Daimlerring 9 Memo :EED.EEDANHO _/ _/ _/ 31135 Hildesheim - Germany mailto:eedanho@eede.ericsson.seArticle: 3662
Hi there! I've got a problem generating a Xilinx netlist (.xnf) with physical pin assignments from my Veribest design capture (.sbk) schematic using Veribest synthesis capture. What I want: Is a .xnf-file with the LOC property on the EXT signals, e.g. EXT,IN,I,,LOC=P46 What I've tried: In the top level design, I have assigned a hierarchical pin the "hier pin name" property, e.g. "IN". Connected to each pin is a wire which has the "Net name" property set to exactly the same name i.e. "IN". In addition to this the wire has the "LOC" property set to a pin number e.g. "P46". Next I synthesize the design using the Veribest synthesis capture and then I try to export the project to Xilinx format using the 3164APC84-3 device. What I get: Is a .xnf-file where everything seems right except that the LOC parameter is missing on the EXT signals e.g. EXT,IN,I What now? I've read all the manuals I've got, I've talked to all my friends, I've called the Veribest hot-line, I try very hard not to go berserk. Please, if you have any suggestions, mail me. /PerArticle: 3663
>I've been making SPICE models of ADC's that way for years - and teaching >others how to do it in my modelling training courses. The question of >calibration never arises, since the capacitors can have any tolerance >you care to specify. Nope, he's telling you the truth. Mark, I would certainly like to know how to make a 16-bit ADC using non-precise caps, with no calibration required. I am probably too far away geographically to join your class :) Peter.Article: 3664
1997 ACM/SIGDA Fifth International Symposium on Field-Programmable Gate Arrays Sponsored by ACM SIGDA, with support from Altera, Xilinx, and Actel Monterey Beach Hotel, Monterey, California February 9-11, 1997 (Web page: http://www.ece.nwu.edu/~hauck/fpga97) The annual ACM/SIGDA International Symposium on Field-Programmable Gate Arrays is the premier conference for presentation of advances in all areas related to the FPGA technology. The topics of interest of this symposium include, but are not limited to: o Advances in FPGA architectures, including design of programmable logic blocks, programmable interconnects, programmable I/Os, and development of new FPGAs and field-configurable memories. o New CAD algorithms and tools for FPGAs, including new algorithms for sequential and combinational logic optimization, technology mapping, partitioning, placement, routing, and development of new FPGA synthesis or layout systems. o Novel applications of FPGAs, including rapid prototyping, logic emulation, reconfigurable custom computing, and dynamically reconfigurable applications. o Advances in field-programmable technology, including new process and fabrication technologies, and field-programmable analog arrays. Authors should submit 20 copies of their original work by September 27, 1996. Each submission should include an 100-250 words abstract, and is limited in length to 12 pages (including figures and tables, minimum point size 10). Notification of acceptance will be sent by November 18, 1996. A proceedings of accepted paper will be published by ACM. Authors must assign copyright of their accepted papers to ACM as a condition of publication. Final versions of accepted papers will be limited to seven pages, and must be submitted by December 6, 1996. All submissions should include the e.mail addresses of the authors, as all correspondence with authors will be done via e.mail. Submissions should be sent to: Prof. Jason Cong FPGA'97 Program Chair UCLA Computer Science Department 4711 Boelter Hall Los Angeles, CA 90095 Phone: (310) 206-2775, Fax: (310) 825-2273, E.mail: fpga97@cs.ucla.edu Organizing Committee: General Chair: Carl Ebeling, University of Washington Program Chair: Jason Cong, UCLA Publicity Chair: Scott Hauck, Northwestern University Finance Chair: Jonathan Rose, University of Toronto Local Chair: Pak Chan, UC Santa Cruz Program Committee: Michael Butts Quickturn Pak Chan UCSC Jason Cong UCLA Carl Ebeling U. Washington Masahiro Fujita Fujitsu Labs Scott Hauck Northwestern Univ. Dwight Hill Synopsys Brad Hutchings BYU Sinan Kaptanoglu Actel David Lewis U. Toronto Jonathan Rose U. Toronto Richard Rudell Synopsys Rob Rutenbar CMU Gabriele Saucier Imag Martine Schlag UCSC Tim Southgate Altera Steve Trimberger Xilinx Martin Wong UT Austin Nam-Sung Woo Lucent Technologies +-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+ | Scott A. Hauck, Assistant Professor | | Dept. of ECE Voice: (847) 467-1849 | | Northwestern University FAX: (847) 467-4144 | | 2145 Sheridan Road Email: hauck@ece.nwu.edu | | Evanston, IL 60208 WWW: http://www.ece.nwu.edu/~hauck | +-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+Article: 3665
I would be interested to hear from anyone who has interfaced a Xilinx 4000 FPGA with a static RAM chip. My problem lies with creating bidirectional tri-state buffers in the FPGA to access the RAM data bus. I am creating designs using behavioural VHDL and the Synopsys Design Analyzer to create a Xilinx netlist. Many thanks for any advice ! Regards, Steven Acock Department of Electronic Engineering, University of Kent, Canterbury, Kent, United Kingdom sjba@ukc.ac.ukArticle: 3666
If you are using a mac, a good freeware logic emulator (circuit design software is called "DigSim"). The program has the ability to use circuits which you have already designed as sub-circuits for your new designs. It comes complete with all the usual gates (AND,NAND,OR,XOR) plus it even has flip flops and adders/subtracters. You can find the program in the following ftp sites: ftp.umich.edu ftp.info.apple.comArticle: 3667
In article <13472@eagle.ukc.ac.uk>, sjba@ukc.ac.uk (S.J.B.Acock) writes: |> |> I would be interested to hear from anyone who has interfaced a Xilinx 4000 |> FPGA with a static RAM chip. My problem lies with creating bidirectional |> tri-state buffers in the FPGA to access the RAM data bus. I am creating |> designs using behavioural VHDL and the Synopsys Design Analyzer to create |> a Xilinx netlist. I've done that, in theory, it's simple. Declare the data-bus as inout: ... data : inout STD_LOGIC_VECTOR(7 downto 0); Then you can get the data from the RAM via 'data'. To put internal values on the data-bus, use something like data<= internal_data when output_enable='0' else "ZZZZZZZZ"; This infers 8 OBUFTs with lowactive enable (that's a primitive). That's how it should work... But there seems to be bug in Synopsys (3.3b), that frequently causes crashes at 'insert_pads' at inout-pins. That doesn't happen, if you instantiate the IBUF/OBUFT/OFD etc. manually and exclude them from the port_is_pad-attribute. -- Bye Georg Acher +--------------------------------------------------------------+ | Georg Acher, acher@informatik.tu-muenchen.de | | "Oh no, not again !" The bowl of petunias | +--------------------------------------------------------------+Article: 3668
A number of users have asked me to seek out the Avant! court filing (instead of just the Avant! press release) so they could judge for themselves what was going on. Here it is in its gory detail! - John Cooley Avant!'s Court Filing (part 1 of 3) -------------------------- INTRODUCTION Cadence asks this Court to enjoin Avant!'s sale of highly sophisticated, break-through technology that is used by every major computer chip manufacturer in the world and which is clearly superior to Cadence's products. The motion should be denied because no Avant! product infringes any Cadence copyright or contains any Cadence trade secret. Cadence's copyright claims are based on allegations that Avant! has copied certain low level code from old database, graphic utilities and related software. Cadence's experts can identify only a handful of the alleged similarities, and they admit they have not made the findings necessary to support a claim of copyright infringement. This low level code, moreover, is covered by a 1994 settlement agreement that released all claims against Avant!. Nevertheless, to avoid any challenge to the code containing these similarities, Avant! has removed it from its products and replaced it with code developed in a clean room. Aware that the similarities were found in low level code covered by the release, Cadence has tried to make it appear to the Court, customers and investors that Avant!'s products contain place and route code allegedly stolen by former Cadence employee Mitsuru Igusa. This is absolutely untrue. No Avant! product uses code that is linked in any way to Mitsuru Igusa. Cadence's reliance on cloak-and- dagger melodrama cannot change this essential fact. Unable to support its copyright case, Cadence is relegated to charging that Avant!'s place and route software is "derived" from Cadence trade secrets. This, too, is untrue. Cadence's analysis and conclusions are contradicted by nationally recognized experts. In scattergun fashion, Cadence claims as trade secrets various algorithms, techniques and methods that are the subject of hundreds of published papers and books -- a vast literature that Cadence's experts never examined. Cadence's experts instead base their opinions on demonstrably false assumptions about the structure and content of Avant!'s code, claiming similarities that simply do not exist. Cadence, moreover, ignores completely the overwhelming evidence of Avant!'s independent development of its place and route software, a detailed record of hard work and creativity. Incredibly, Cadence's experts reached their conclusions without even reviewing the overwhelming evidence of independent development by Avant! engineers. Finally, Cadence impermissibly delayed in bringing this lawsuit. Such a delay vitiates Cadence's claim of irreparable harm and sharply tips the balance of hardships in Avant!'s favor. Cadence argues that this case is like no other in Silicon Valley history. Just the opposite is true; this case is like many others where a large company grows complacent and finds its technology surpassed by a new, smaller company staffed in part by former employees. Unable to win in the marketplace, the larger company resorts to litigation and smear tactics. When the Court cuts through Cadence's misdirection and innuendo, it will conclude that the preliminary injunction motion must be denied. STATEMENT OF FACTS I. AVANT!'S ENGINEERS AND PRODUCTS Avant!'s two main product lines are place and route software and verification software. Place and route software is crucial to manufacturers of integrated circuits ("ICs" or "chips"). This software enables designers to place and connect (or "route") millions of microscopic circuit elements on a chip. Verification programs test whether the design works as planned. With the advent of submicron and deep submicron technology, IC manufacturers are designing increasingly smaller, faster and more complex chips. Traditional physical design software, based on technology developed 7 to 12 years ago, cannot meet these new challenges. Manufacturers are demanding more sophisticated tools to survive in extremely competitive markets. Avant!'s products were designed to meet these demands. Cadence's depiction of Avant! as a group of conspirators selling stolen technology could not be further from the truth. Avant!, which has 215 employees, dedicates a large portion of its annual budget to R&D and employs some of the world's most talented place and route specialists. Almost three-quarters of the 48 engineers and scientists working on its place and route tools have Ph.D.s. In all, Avant!'s research teams include 37 Ph.D.s, and Avant!'s engineers and scientists have published more than 200 technical papers. II. THE UNDISPUTED SUPERIORITY OF AVANT!'S PRODUCTS. Cadence is Avant!'s main competitor in place and route and verification software. Cadence is older and much larger, with 1995 annual revenue of over $540 million, compared to Avant!'s revenue of $38 million. Avant!'s place and route products are known as ArcGate, ArcCell and ArcCell-XO. ArcCell enables the design of deep submicron cell-based ICs. ArcCell-XO is an upgrade, extending ArcCell to the most complex, high performance IC designs. Cadence's place and route products include Cell Ensemble, Cell3 Ensemble and Gate Ensemble. ArcCell and ArcCell-XO compete directly with Cadence's Cell Ensemble and Cell3 Ensemble, while ArcGate competes directly with Cadence's Gate Ensemble. Avant!'s place and route software frequently competes head-to- head with Cadence's products (principally Cell3 Ensemble). Typically, a customer finds that the Cadence product cannot produce a layout from a certain design or that the resulting layout will make the chip's size (known as the "die size") too large. The customer then asks Avant! to run its software with the same design to see if it will produce an efficient layout and smaller die size. Avant! has identified 86 such head-to-head competitions in the past few years. Avant!'s software produced a smaller die size in 74 of the 83 instances where the customer could obtain results from the Cadence product. On average, the Avant! tool produced die about 13% smaller than the Cadence package. These results are significant: a 10% reduction in die size can produce a 20% or greater reduction in per-die production costs. The outcomes of these competitions have translated into sales. Industry analyst Wessels, Arnold & Henderson reports that Avant!'s share of the 1995 deep submicron market exceeded Cadence's share by 23.3% to 18%. Wessels confirms that the deep submicron market is the future for IC physical design products, projecting 36% annual growth rates. Wessels predicts that Avant!'s share of this market in the year 2000 will exceed Cadence's share by 33% to 12%. ARGUMENT III. CADENCE'S CLAIMS CAN BE DIVIDED INTO CLAIMS REGARDING CODE CREATED BEFORE AND AFTER THE RELEASE. The Avant! products that Cadence asks this Court to enjoin are massive, extraordinarily complex computer programs. ArcCell consists of 1,454,338 lines of code organized into 2,260 files. (C. H. Chang Decl. & 14.) ArcCell-XO comprises 1,241,357 lines organized into 2,048 files. (Id. & 13.) Cadence's motion is devoid of any effort to specify what claims apply to which parts of this immense body of code. On February 21, 1996, Magistrate Judge Trumbull ordered Cadence to identify the Avant! code that represents trade secret misappropriation or copyright infringement. (Snyder Decl. & 4; Exh. C.) Cadence ultimately produced a list of 117 files. Attached hereto as Exhibit A is a chart listing the 117 Avant! files with the corresponding Cadence files. For the Court's convenience, the chart is color coded. Files created before the May 6, 1994 Confidential Settlement Agreement ("Release") are designated in blue. This code ("Pre-Release Code") is used in ArcCell and ArcCell-XO. Files created after the Release date are designated in green. These files ("Post-Release Code") are used only in ArcCell-XO. Files that are not used in any Avant! product have no color designation. The chart also designates which Avant! expert will testify on each file. The chart serves as an outline of Avant!'s defenses to Cadence's claims. With regard to Pre-Release Code, the Release bars all claims, including claims for equitable relief, based on Avant!'s continued use of any Cadence code or trade secrets found in those files. All Pre-Release Code containing questionable similarities, moreover, has been removed from Avant!'s products and the functions replaced with code written in a clean room. With regard to Post-Release Code, Cadence's claims relate to trade secret misappropriation. As detailed below, the alleged trade secrets are either in the public domain or they are not found in Avant!'s products. Moreover, overwhelming evidence shows that all of these Avant! files were independently developed. IV. CADENCE IS NOT ENTITLED TO A PRELIMINARY INJUNCTION BASED ON CODE CREATED BEFORE THE RELEASE. A. Cadence Released All Claims To Enjoin Avant!'s Continued Use Of Cadence Source Code And Trade Secrets Allegedly Acquired Before The Release. 1. The 1994 Release Of Injunction Claims. In early 1994, Cadence threatened to sue Avant! and its current president, Gerry Hsu, for trade secret theft after Hsu announced his intention to leave Cadence to join Avant!. After weeks of intensive negotiations, the parties entered into a Confidential Settlement Agreement ("Release") on June 6, 1994. (Brill Decl. && 2, 21.) By the terms of the Release, the parties agreed to forever release and discharge each other from any claim which they may have against each other at the time of the execution of this Agreement, known or unknown, including but not limited to any claims arising out of, or in connection with, or relating directly or indirectly to . . . any anticompetitive activity or unfair competition or trade secret misappropriation by Cadence, Hsu or ArcSys with respect to Cadence, Hsu or ArcSys . . . ." (Release at 8, & H(1) (emphasis added).) As the circumstances surrounding the settlement demonstrate, the Release discharged all of Cadence's claims to enjoin Avant!'s continued use in its products of any Cadence source code and trade secrets acquired before the Release. In the negotiations leading up to the Release, Cadence threatened to enjoin the sale of Avant! products that Cadence believed contained stolen intellectual property. (Brill Dep. 137:2- 21, 138:6-24, 150:24-151:17; Brill Decl. && 7-15; Kagle Decl. & 5.) Cadence threatened to sue if an acceptable settlement could not be reached and provided Avant! the complaint it intended to file. (Brill Decl. && 7-9, 16; Kagle Decl. & 5; Healy Dep. 52:10-54:21, 69:6-15; Sherwood Dep. 52:3-17, 53:11-24.) Cadence's complaint alleged that Avant! was formed by Cadence employees who "had access to and had assisted in developing Cadence Software Products." (Complaint & 7.) These individuals allegedly used Cadence's trade secrets -- broadly defined to include "Cadence software systems and all related confidential materials" -- to "develop ArcCell and other products. . . ." (Id. && 29-32; Brill Dep. 116:10-117:1; Salmon Dep. 15:12-16:17, 19:16-21:2, 24:15-26:13.) The complaint sought an injunction to prevent Avant! from "using, reproducing or disseminating" Cadence intellectual property in these products. (Id. Prayer for Relief & 1, & 34; Brill Decl. && 14-15.) In a clear reference to the incorporation of misappropriated source code into Avant! products, Cadence requested a constructive trust requiring Avant! to hold in trust "all Cadence's trade secrets and confidential information and all resulting products and technologies containing same in whole or in part . . . ." (Complaint & 55) (emphasis added). Faced with threats to enjoin the sale of its products, Avant! entered into a settlement. (Brill Decl. && 16-21; Kagle Decl. && 6- 7.) In exchange, Cadence forever released its threatened claims to enjoin the continued sale of Avant!'s products containing or using Cadence software or other confidential materials allegedly acquired before the date of the Release. 2. The Law Governing The Release. Ignoring the fact that it had specifically asserted and then released its claims to enjoin continued use of its software and confidential information, Cadence now insists that the Release only protected Avant!'s use of such material prior to the date of the Release. (Br. 31-32.) Under this view, Cadence was free to sue Avant! to enjoin the sale of its products the day after the Release was executed. The law does not permit such an interpretation. Construing the Release to extinguish claims based on continuing use is the only reasonable interpretation. See Kennewick Irrigation Dist. v. U.S., 880 F.2d 1018, 1032 (9th Cir. 1989). At a minimum, Avant! bargained for an agreement to settle the claims framed by Cadence's 1994 complaint. (Brill Dep. 73:15-76:2, 114:2-8, 118:3- 119:8, 123:19-124:14, 137:2-21, 138:6-24, 143:14-144:7; Kagle Dep. 68:4-69:10; Healy Dep. 69:6-15; Brill Decl. & 19.) The parties entered the agreement expressly to "avoid future disputes and disagreements" and accordingly discharged all "known and unknown" claims. (Release at 8, & H(1).) Thus, the Release was intended to permit Avant! to market its products free from injunction threats based on any pre-Release acquisition of confidential material. (Brill Dep. 137:2-21, 138:6-24; Kagle Decl. & 6; Brill Decl. && 26- 27; Salmon Dep. 75:3-17; Coxe Dep. 76:2-15.) Interpreting the agreement to allow Cadence to reassert its injunction claims -- either the day after the Release was signed or today -- would eliminate the repose for which Avant! had bargained. See Henning v. Kitchen Art Foods, Inc., 127 F. Supp 699, 702 (S.D. Ill. 1954) (holding that release precluded claim based on post-release use of the alleged trade secret). California law provides that a trade secret misappropriation claim arises only once, at the time of the initial misappropriation; successive uses of the information -- in this case the Pre-Release Code -- are part of that single claim. See Cal. Civ. Code $ 3426.6; Whittaker Corp. v. Execuair Corp., 736 F.2d 1341, 1345 (9th Cir. 1984) (quoting Monolith Portland Midwest Co. v. Kaiser Aluminum & Chem. Corp., 407 F.2d 288, 293 (9th Cir. 1969)) (holding that trade secret misappropriation "is not a continuing tort and that `[t]he cause of action arises but once'"); Ashton-Tate Corp. v. Ross, 916 F.2d 516, 524 (9th Cir. 1990); Intermedics, Inc. v. Ventritex, Inc. ("Intermedics II"), 822 F. Supp. 634, 657 (N.D. Cal. 1993). Courts reason that successive uses constitute a single claim because the tort of misappropriation occurs only with the initial breach of confidence. See Monolith, 407 F.2d at 293 (holding that "[t]he fabric of the relationship once rent is not torn anew with each added use or disclosure, although the damage suffered may thereby be aggravated") (emphasis added). Once a trade secret owner releases a party from liability, any claim of continuing misappropriation is extinguished. The cases Cadence cites are irrelevant to both the plain meaning and interpretation of the Release under California law. In Remington Rand Corp. v. Amsterdam-Rotterdam Bank, N.V., 68 F.3d 1478, 1483-85 (2d Cir. 1995), the court determined the meaning of the disputed release pursuant to New York law. New York courts endorse a "continuing tort" theory of trade secret misappropriation whereby each use of a trade secret creates a separate cause of action -- just the opposite of California law. See Lemelson v. Carolina Enterprises, Inc., 541 F. Supp. 645, 659 (S.D.N.Y. 1982). The result mandated by California law is also supported by the absence of any reservation of rights in the Release to enjoin future use of the allegedly acquired trade secrets. Despite claiming at the time that Avant!'s trade secret use would continue unless enjoined, Cadence did not reserve the right to prosecute such future claims based on Avant!'s continuing use of Cadence software and confidential information. See NTN Communications, Inc. v. Interactive Network, Inc., 1993 WL 266663, at *4 (N.D. Cal. July 7, 1993) (discussing comparable release in which patent holder declared "except that IGP reserves its prospective claims of infringement"), aff'd, 41 F.3d 1520 (Fed. Cir. 1994), cert. denied, 115 S.Ct. 2000 (1995) ; Realex Chem. Corp. v. S.C. Johnson & Son, Inc., 849 F.2d 299, 302 (8th Cir. 1988) (holding that party could have easily reserved copyright claim asserted); Monon Corp. v. Wabash Nat'l Corp., 780 F. Supp. 577, 583 (N.D. Ind. 1991) (ruling that party could have easily reserved patent claim asserted). Until now, Cadence's conduct has been completely consistent with the interpretation that Avant! was released from all claims based on continued use of code and other confidential information acquired before the Release. Despite the allegations in Cadence's 1994 complaint, no one from Cadence asked Avant! to pull its products off the market after the Release was signed. Nor did Cadence ever request Avant! to rewrite its source code to remove any Cadence confidential materials. (Sherwood Dep. 173:13-175:1; Healy Dep. 61:22-62:9; Brill Decl. & 26.) The reason Cadence took no such actions is clear: the Release discharged Cadence's claims against Avant!'s continued use in its products of the information allegedly acquired before the Release date.Article: 3669
A number of users have asked me to seek out the Avant! court filing (instead of just the Avant! press release) so they could judge for themselves what was going on. Here it is in its gory detail! - John Cooley Avant!'s Court Filing (part 2 of 3) -------------------------- Cadence's attempt to revive its released claims by alleging copyright infringement is unavailing. The Release bars all claims -- however styled -- to enjoin continued use of code and other confidential information acquired before the Release date. Courts consistently resist efforts by plaintiffs to circumvent releases by reformulation of claims. In Press Mach. Corp. v. Smith R.P.M. Corp., 727 F.2d 781 (8th Cir. 1984), the court found that an agreement releasing a trade secret claim also barred patent claims based on subsequent use of the trade secrets, even though the release did not mention patent claims. The court reasoned that permitting the patent claims would deny the defendant "the right to use the trade secrets and confidential information for which it had paid [and] would strip the agreement of much of its meaning." Id. at 785; see also Realex, 849 F.2d at 302 (interpreting settlement agreement to bar future copyright claims); Gallowhur Chem. v. Schwerdle, 37 N.J. Super. 385, 117 A.2d 416 (1955). In a decision barring a continuing use claim, Magistrate Judge Brazil emphasized the importance of repose as a matter of public policy and the danger of a plaintiff manipulating the definition of its trade secrets in an attempt to assert fresh claims. Intermedics II, 822 F. Supp. at 654-57. The court explained: [The] facts of cutting edge technological life, to say nothing of the limitations of language generally, have the effect of giving the person asserting trade secret claims . . . very wide definitional latitude [and] considerable power to manipulate the articulation of claims for tactical purposes . . . [S]uch an approach would make it extremely difficult for persons who might be targets of trade secret litigation ever to achieve . . . repose . . . [Repose is] characterized as a kind of confidence about legal circumstances that permits and encourages people to move forward energetically and boldly with new economic development, free from fear that the fruits of their labor will be stripped from them as a result of trade secret litigation. The law's ability to provide people with that kind of confidence would be badly compromised if parties could escape [preclusion] by artful manipulation of the definitions of related trade secrets. Id. at 656. Although the court was referring to the repose secured by the statute of limitations, the same policy applies here. Avant! and Cadence executed the Release specifically to permit Avant! to move forward free of claims based on Avant!'s use in its business of Cadence trade secrets, which Cadence defined to include software and related confidential information. Cadence cannot be allowed to escape the natural, foreseeable consequence of the Release through an "artful manipulation" of its claims or by an untenable distinction between Avant!'s use of the same material in products before and after the Release date. B. Cadence's Experts Have Not Made The Findings Necessary To Support Claims Of Copyright Infringement Against Pre- Release Code. Cadence experts Randall Davis and Margaret Johnson were assigned to review the files that make up the Pre-Release Code. They did not analyze these files for evidence of trade secret misappropriation. (Kelly Decl. && 26-27.) Instead, they limited their review to identifying "similarities" or "correspondences" between the Avant! and Cadence code. (Davis Dep. 40:1-41:24; 139:2-14; Johnson Dep. 126:24-127:19.) To prevail in this motion on copyright grounds, Cadence must show a high likelihood that it will prove infringement. The methodology and findings of Cadence's experts are flawed and cannot satisfy the standards imposed by law. As Cadence acknowledges, to establish infringement it must show that Avant! had access to its programs and that Avant!'s code is substantially similar to Cadence's copyrighted programs. (Br. 27:14- 16.) Cadence experts, however, did not perform the required analysis. When asked: "Were you asked [by Cadence's lawyers] to do an analysis of substantial similarity as to any Avant! and Cadence Code?", Professor Davis replied: "I don't recall that question being put to me." (Davis Dep. 168:8-12.) Dr. Johnson admitted that she did not analyze the significance of any similarities: "I was merely asked to provide files where correspondences existed. That's what I did." (Johnson Dep. 127:2-6.) Other than an 85 line "[ ]" file in Post-Release code that she incorrectly believed was copied (see footnote 11, infra), Dr. Johnson had no opinion about whether Avant! had copied any Cadence code. (Johnson Dep. 133:12-15.) The actual work performed by these experts starkly contrasts with the sweeping conclusions Cadence attributes to them. Cadence identified some 89 files in the database, graphics editor, human interface and related areas as allegedly containing copied code. (Exhibit B hereto lists these Pre-Release files and indicates which files were examined by Cadence's experts.) Professor Davis and Dr. Johnson think they looked at about 54 of these files. (Kelly Decl. & 14.) It is impossible to reconstruct their analysis; they did not even record what files were examined much less the results of their examination. (Kelly Decl. & 15.) Ultimately, they could only discuss similarities in 10 files. (Kelly Decl. & 14.) Neither expert examined public domain or third party sources to determine if they accounted for the alleged similarities. (Kelly Decl. && 19-24.) Although he maintained that he eliminated the possibility that some similarities were the result of common programming practices or ideas, Professor Davis admitted that he did not understand even the rudimentary functionality of the code. (Davis Dep. 143:16-24.) Dr. Johnson admitted that she did not consider the impact of common programming constraints. (Johnson Dep. 26:21-25.) When provided with the source code from several Avant! files referred to in his declaration as containing "code copied from Cadence," Professor Davis could not identify evidence of copying. (Kelly Decl. && 16-17.) In short, Cadence has not provided this Court with the expert findings necessary to support an injunction on the ground that Avant!'s Pre-Release Code infringes Cadence code. C. Any Code Containing Questionable Similarities Has Been Removed And Replaced With Code Written In A Clean Room. Not only are Cadence's claims barred by the Release and unsupported by expert analysis, they are moot. Any code on Cadence's target list containing questionable similarities has been eliminated from Avant!'s products. After Cadence produced its list of Avant! files, Avant!'s experts began a meticulous, line-by-line comparison of the Avant! code and the corresponding Cadence code. Avant!'s expert, Professor John Kelly, was conservative in his approach. Professor Kelly did not try to link similarities to third party or public domain sources. If there was any doubt about code containing a questionable similarity, it was designated to be deleted or replaced with new code. (Kelly Decl. && 28-32.) New code was written under exacting conditions in a clean room by engineers who had no access to Cadence or Avant! source code. (Case Decl. && 11-16; Tsou Decl. && 14-16; Liu Decl. && 12-14; Yang Decl. && 6-8; Liao Decl. && 6-8; Lo Decl. && 82-85.) All information provided to the clean room engineers was rigorously screened and recorded by an independent engineer who acted as the "filter." (Case Decl. && 12-19.) The clean room engineers wrote the new code based entirely on their own knowledge and experience, certain designated public domain materials provided by the filter and specifications giving "only a basic English description of the higher level ideas that needed to be embodied in the new code." (Case Decl. && 13-16.) Professor Kelly has reviewed and approved the clean room protocol and the procedures by which the new code was integrated into Avant!'s existing programs. (Kelly Decl. & 45-46.) His declaration includes an exhibit detailing his analysis and designating which files have been deleted or replaced and which files were not changed because there was no evidence of questionable similarities. (For the Court's convenience, Exhibit C, attached hereto, is a chart summarizing these findings and describing the status of all 117 files on Cadence's list.) Because Avant! has removed any questionable code, Cadence is not entitled to a preliminary injunction unless the resulting code is substantially similar to the Cadence code. Computer Assoc. Int'l Inc. v. Altai, Inc., 775 F. Supp. 544, 561-62 (E.D.N.Y. 1991), aff'd in pertinent part, 982 F.2d 693 (2d Cir. 1992). Professor Kelly has carefully reviewed the Avant! code that is the result of the deletion of old code and the integration of clean room code. Professor Kelly's opinion is that the resulting code is not substantially similar to the corresponding Cadence code. (Kelly Decl. & 47.) Thus, the clean room is yet another reason no injunction should issue based on Cadence's claims of similarities between its code and Avant!'s code. Alexander & Alexander Consulting Group, Inc. v. W.F. Corroon Corp., 1995 WL 46373, at *7 (N.D.Cal. Jan. 23, 1995); Computer Assoc., 775 F. Supp. at 561-62. V. CADENCE IS NOT ENTITLED TO A PRELIMINARY INJUNCTION BASED ON THE POST-RELEASE CODE. Based on the alleged presence of Cadence trade secrets in the Post-Release Code, Cadence seeks to enjoin the sale of Avant!'s two premier place and route products. To obtain an injunction, Cadence must show that Avant!'s current products use Cadence trade secrets. California Civil Code $ 3426.2. As Cadence's experts discovered, ArcCell does not use any of the Post-Release Code. (C. H. Chang Decl. && 16-20; Smith Decl. & 13, Exh. 12 (Navas Dep. 205:4-18, Exh. 19).) ArcCell-XO, as demonstrated below, contains no Cadence trade secrets. The Post-Release Code (designated in green on the chart in Exhibit A) covers three parts of Avant!'s ArcCell-XO place and route tool: global routing, detail routing and the Design Size Optimization program. (C. H. Chang Decl. && 16-18.) None of these Post-Release Files contains a Cadence trade secret. A trade secret, by definition, must not be "generally known to the public or to other persons who can obtain economic value from its disclosure or use." California Civil Code $ 3426.1. In contrast, "[i]nformation that is public knowledge or that is generally known in an industry cannot be a trade secret." Ruckelshaus v. Monsanto Co., 467 U.S. 986, 104 S.Ct. 2862, 2872 (1984). Every "trade secret" claimed by Cadence is either not used in the Post-Release Files or is described in the public literature, and frequently both. The presence of Cadence's so-called "trade secrets" is so pervasive in public literature that Avant!'s experts have cited, merely as examples, hundreds of books and articles discussing in detail the allegedly misappropriated techniques. Many of the features that Cadence claims as proprietary, moreover, were used by Avant! engineers while they were students. Many of the techniques also appear in Avant!'s ArcGate product that Cadence has not challenged. The routing code in ArcGate was written by engineers who never worked at Cadence and who never had access to Cadence code. (S. C. Chang Decl. && 1, 7, 10.) These engineers employed techniques they learned in school, at previous employment, or developed at Avant!. The following discussion summarizes the independent sources for the claimed trade secrets and some of the major differences between the Avant! and Cadence programs. For the Court's convenience the declarations of Professors Sarrafzadeh and Robins, Dr. Paul Lo, and Dr. Ping-San Tzeng address the issues in the same order presented below. A. Avant!'s Global Router Was Independently Developed And Contains No Cadence Trade Secrets. The global router in ArcCell-XO, known by its directory name [A], is the product of intense effort by an experienced and talented place and route engineer, Dr. Sheng Chung ("Paul") Lo. It is not the result of software theft or trade secret misappropriation. (Lo Decl. & 23.) Dr. Lo contemporaneously documented his development of [A] in his lab notebook, weekly reports, thousands of pages of test results, archived copies of unused versions, and the line by line computer record of changes made to [A] after it had been checked in to Avant!'s software version control system. This material confirms Dr. Lo's testimony regarding his independent development of [A]. (Lo Decl. && 14-47, Exhs. A-C.) Incredibly, Cadence withheld this information from Steven Teig, the only Cadence expert to review this source code. (Teig Dep. 352:14-353:10.) Mr. Teig's failure to review the immense evidence of independent development is fatal to his conclusions. Avoiding the evidence confirming the code's independent creation, Cadence purports to identify certain "trade secrets" used by [A]. (Br. 24:7-18.) Yet Cadence and its experts do not present a clear or consistent statement of the trade secrets at issue. To assist the Court's analysis of these issues and to insure that each of Cadence's baseless allegations receives a response, Avant! prepared the following comprehensive list of issues raised by Cadence: a Using the [#1] algorithm a Using the [#1] algorithm in a [#2] a The " [#3] algorithm" a Using [#4] global router a [#5] a [#6] in global routing a An array of [#7] a [#8] a A certain [#9] function a Nomenclature a Net ordering a [#10] a Removing violations Each of these issues is separately addressed below. In every instance, Cadence's experts are either mistaken about [A] or are describing a well-known technique. Using the [#1] Algorithm. Even Cadence's Professor Sechen admits that both the [#1] algorithm and its use in global routing are in the public domain. (Sechen Decl. & 37.) Avant!'s experts agree. (Sarrafzadeh Decl. & 49; Robins Decl. && 21-34.) Using the [#1] Algorithm in a [#2] Maze. Cadence claims that the use of [#2] global routing is a trade secret, even while admitting that it is "common" in the related context of detailed routing. (Sechen Decl. & 36.) The logical use of [#2] routing in multi-layer designs, for both global and detailed routing, has been described in the literature and has been used for several years by Avant! in its ArcGate place and route tool (detailed routing). (Sarrafzadeh Decl. && 53-54; Robins Decl. && 26-29; S. C. Chang Decl. & 17.) Six years ago, Cadence touted its own Cell3 Ensemble global router as including [#2] global routing. (Robins Decl. & 28; Smith Decl. & 14, Exh. 13 (Nequist Dep. 78:25- 79:14, 90:19-91:25, Exh. 3).) If it were ever a trade secret, it certainly is not one now. The [#3] Algorithm". Cadence claims great advantages to having [#3] aspects of its use of the [#1] algorithm. (Teig Decl. & 35; Sechen Decl. & 39.) This technique has been published. (Robins Decl. & 30.) Although Cadence claims that Avant! uses this technique, Cadence's experts could not find it in the code. (Teig Dep. 340:5- 341:23, 360:7-24; Sechen Dep. 188:17-20.) The reason is simple: ArcCell-XO's global router does not use the alleged [#3] . (Lo Decl. & 51; Sarrafzadeh Decl. && 56-58.) The absence of [#3] actually reflects a difference in the two global routers. Using an [#4] Global Router. Cadence also apparently claims that the combination of global routing approaches is unique to Cadence and constitutes a trade secret. (Teig Decl. & 25; Sechen Decl. & 45.) Dr. Lo decided on this combination of publicly documented approaches only after conducting numerous experiments and implementing what he believed would create the best program. (Lo Decl. && 56-57.) Cadence's experts, moreover, are apparently unaware that every aspect of the approach they describe has been published, separately and in combination. (Sarrafzadeh Decl. && 59-60; Robins Decl. && 31-34.) [#5] . Cadence's Professor Sechen claims that storing wiring information in the form of a [#5] is unusual and assumes that, like Cadence, [A] uses the same approach. (Sechen Decl. & 42; Sechen Dep. 193:13-194:7; Robins Decl. & 35.) Sechen is wrong: [A] stores wiring information in the form of a [ ], the admittedly more common approach. (Lo Decl. & 58; Sarrafzadeh Decl. && 61-63.) This differs entirely from the form used by Cadence. (Teig Dep. 322:4-7; Vuong Dep. 47:6-12.) According to Cadence's corporate designee on the subject and Avant!'s experts, this difference affects every aspect of the companies' global routers. (Vuong Dep. 86:10-14; Sarrafzadeh Decl. && 61-63; Lo Decl. && 28-29.) [#6] in Global Routing. Although Professor Sechen claims that "[t]he notion of using the so-called `[#6] ' in global routing [is] not found in the published literature," (Sechen Decl. & 49), this standard approach has been published in several books and papers. (Sarrafzadeh Decl. & 65; Robins Decl. & 36.) Dr. Lo, who wrote [A], and Dr. Tzeng, who wrote the ArcCell-XO detailed router, both learned this technique before working at Cadence. (Lo Decl. & 59; Tzeng Decl. & 44.) Avant! has also used this technique in its ArcGate product since 1992. (Sarrafzadeh Decl. & 64; S. C. Chang Decl. & 20.) Dr. Tzeng even published a paper using the idea in 1988. (Tzeng Decl. & 44, Exh. B.) Using an Array of [#7] . Cadence claims that this basic method of storing data is a "proprietary technique." (Teig Decl. & 39.) This method, however, has been repeatedly published in the place and route field, as Cadence's Steven Teig ultimately admitted. (Sarrafzadeh Decl. && 68-70; Robins Decl. & 38; Teig Dep. 349:1-4.) Dr. Lo learned this technique as an undergraduate (Lo Decl. & 61), and Avant! first used it in its ArcGate product long before the misappropriation alleged by Cadence. (Lo Decl. & 61; S. C. Chang Decl. & 23.) Cadence also avoids disclosing that Cadence and Avant! implement the basic idea of a [#7] differently. Cadence uses an array of [ ]. (Teig Dep. 350:2-25.) Avant! uses a different data structure, an array of [ ], as Cadence's Steven Teig eventually realized. (Teig Dep. 350:20-351:2; Lo Decl. & 62.) Again, Cadence's claim of similarity actually reveals a difference. [#8]. Cadence falsely claims that using "[#8]" "was invented by and is proprietary to Cadence." (Teig Decl. & 43; Sechen Decl. & 41.) The author of [A], Dr. Lo, learned this technique as an undergraduate. (Lo Decl. & 65.) Dr. Tzeng, who wrote the ArcCell-XO detail router, also learned this technique as a computer science student and incorporated it in precisely this context in the routing program written for his Ph.D. dissertation. (Tzeng Decl. & 71.) This technique is so basic and obvious that it was "invented" by one-third of the students on a recent exam. (Robins Decl. && 39-41; Sarrafzadeh Decl. && 72-73.) A Certain [#9] Function. Cadence claims to use a "unique" method for estimating the distance to multiple targets in its application of the [#1] algorithm. (Teig Decl. & 46; Sechen Decl. & 46.) This technique has been published. (Sarrafzadeh Decl. && 74-76; Robins Decl. && 42-44.) Dr. Lo adopted it based on his non-Cadence experience and his recognition that a more accurate, but more complicated, approach would merely slow down the program. (Lo Decl. && 66-68; Sechen Dep. 179:24-180:3.) Confirming Dr. Lo's assessment, Avant!'s ArcGate tool has used this technique for years. (S. C. Chang Decl. & 25.) Nomenclature. The terms identified by Cadence as evidence of misappropriation are common in the literature and are used by Avant! in files unchallenged by Cadence. (Robins Decl. && 46-50; Sarrafzadeh Decl. & 77; S. C. Chang Decl. && 26-27.) Even Cadence's experts admitted that the terms "[ ]," "[ ]," "[ ]," "[ ]", "[ ]" and "[ ]" are common in the literature. (Teig Dep. 334:22-335; Sechen Dep. 206:19-207:6.) The only terms which Cadence's experts consider to be an indicia of copying, "[ ]" and "[ ]," are not found in [A] or [B].(Sarrafzadeh Decl. & 77; Robins Decl. && 51-52; Lo Decl. & 71.) Net Ordering. Cadence's Professor Sechen and Avant!'s experts agree that ordering the nets for initial routing by some estimate of the [ ] is a standard technique. (Sechen Dep. 202:6-16; Robins Decl. && 54-55.) Cadence claims only that the formula it uses to estimate the [ ] is a trade secret. (Teig Decl. && 51- 52; Sechen Decl. & 48.) Cadence is apparently unaware that its "secret" formula was published in Mathematics for Elementary School Teachers, and several papers on the subject of maze routing. (Robins Decl. && 57-58; Sarrafzadeh Decl. & 80.) Cadence is also incorrect that Avant! uses this formula. In fact, Avant! uses a completely different formula. (Sarrafzadeh Decl. && 80-82; Lo Decl. & 73.) Once again, examination of the so-called "similarity" reveals another difference. [#10] Routing. [#10] routing has been described in the literature for more than 20 years, including by Cadence. (Sarrafzadeh Decl. & 83; Robins Decl. && 59-63; Smith Decl. & 14, Exh. 13 (Nequist Dep. 78:25-79:14, Exh. 3 at 68).) Professor Sechen admits that this technique is quite common in detailed routers; indeed, the detailed router for ArcCell-XO allows [#10] routing, and ArcGate's detailed router has used this technique for several years. (Sechen Dep. 189:22-190:7; Tzeng Decl. & 72; S. C. Chang Decl. & 28.) Cadence apparently claims that this technique is a trade secret only in global routing. (Teig Decl. & 55; Sechen Dep. 190:8-19.) Yet, the ArcCell-XO global router [A] does not allow [#10] routing. (Sarrafzadeh Decl. & 85; Lo Decl. & 75.) Again, the Cadence and Avant! products do not share this alleged similarity. Removing Violations. According to Cadence, improving the results of global routing by using [ ] to remove violations is a trade secret. (Sechen Decl. & 44.) This is surprising, because the extensive literature on [ ] includes a book by one of Cadence's experts, Prof. Sechen, entitled "VLSI Placement and Global Routing Using Simulated Annealing." (Robins Decl. && 65-67; Sarrafzadeh Decl. & 86.) In any event, Avant! does not use [ ] in global routing. (Lo Decl. & 80; Sarrafzadeh Decl. & 90.) Of the dozen Cadence purported "trade secrets" allegedly found in [A], every one has been published. At least half are not even found in [A]. Cadence's claims of similarity in these areas actually reveal fundamental differences in the structure and approach of the two programs. (Sarrafzadeh Decl. && 91-96.) Cadence has demonstrated no likelihood of proving that the ArcCell-XO global router contains any Cadence trade secrets. B. Avant!'s Detailed Router Was Independently Developed And Contains No Cadence Trade Secrets. The ArcCell-XO detailed router, known by its directory name [B], is not the result of software or trade secret theft. This code is the product of nearly a year of dedicated work by an engineer with a decade of experience in routing technologies, Dr. Ping-San Tzeng. Before receiving his Ph.D. in electrical engineering, Dr. Tzeng developed two place and route tools as part of his academic work.The code in [B] includes many of the techniques he developed while he was a student. Like his colleague Dr. Lo, Dr. Tzeng recorded his development of [B] in an original abstract description of the future router, a proposed schedule, weekly reports, archived copies of unused versions, and the line by line computer record of changes made to [B] after it had been checked in to Avant!'s version control system. This material confirms Dr. Tzeng's independent development of [B]. (Tzeng Decl. && 17-40; Exhs. C-E.) Perhaps because of this compelling evidence, Cadence did not deign its claims regarding [B] to be worthy of more than one sentence in its brief. (Br. 24:19-20.) Instead, Avant! and the Court are left to glean from the declarations of Steve Teig and Carl Sechen what supposed "trade secrets" are allegedly found in [B]. The following is a complete list of issues relating to [B]: a [#6/ #11] a [#12] a [#13] a [#14] a [#15] a [#16] a [#17] a [#8] a [#10] As in the case of the ArcCell-XO global router, each of these alleged similarities rests on either an incorrect description of Avant!'s [B] or on the common use of public domain programming techniques, many of which Dr. Tzeng had used in his own programs while still a student. [#6/#11] . Cadence claims that using a collection of [#6] to define the area (called a [#11] ) in which detailed routing will be performed is a trade secret. (Teig Decl. & 59-60; Sechen Decl. & 49.) Not only has this technique been published (Robins Decl. && 70-71), Avant!'s ArcGate has used it since January 1993. (S. C. Chang Decl. && 31-33; Sarrafzadeh & 101.) Even earlier, Dr. Tzeng used this technique as a student and published a paper using the term "[#11] ." (Tzeng Decl. & 45, Exh. B.) [#12] . Cadence claims as trade secrets the process of searching for violations and defining a new region of [#6] in which to correct them, as well as the term "[#12] " to describe this process. (Teig Decl. & 62; Sechen Decl. & 49.) Apparently unknown to Cadence's experts, a 1990 article describing Cell3 Ensemble referred to this technique by the name "[#12] ." (Smith Decl. & 14, Exh. 13 (Nequist Dep. 78:25-79:14, Exh. 3 at 68); Robins Decl. && 73-75.) Prior to 1990, Dr. Tzeng used this technique as a student, and Avant!'s ArcGate has used both this technique and the term "[#12] " since 1993, well before the alleged misappropriation. (Tzeng Decl. && 51-54; S. C. Chang Decl. & 34; Sarrafzadeh Decl. && 102-103.) [#13] . Although Cadence claims that [#13] was a Cadence trade secret (Teig Decl. & 63; Sechen & 43), an original author of Cadence's Gate Ensemble global and detailed routers admits that the idea was originally taken from the literature. (Nequist Dep. 153:18-154:23.) In 1990, Cadence described this feature in the public literature. (Smith Decl. & 14, Exh. 13 (Nequist Dep. 78:25-79:14, Exh. 3 at 68).) Several other papers address this subject, including a 1988 paper by Dr. Tzeng. (Sarrafzadeh Decl. & 104; Robins Decl. & 77; Tzeng Decl. & 55, Exh. A at 32.) This concept is used by other routers, including ArcGate's detail router and programs written and studied by Dr. Tzeng in graduate school. (Tzeng Decl. & 55; S. C. Chang Decl. & 35.) [#14] . According to Cadence, allowing [#11] to [#14] is a proprietary technique. (Teig Decl. & 66; Sechen & 49.) This trivial detail has appeared in the literature. (Robins Decl. & 79.) The idea has also been used for several years in the ArcGate detail router. (Sarrafzadeh Decl. && 106-107; S. C. Chang Decl. && 36-40.) [#15] . Cadence's experts claim that the use of [ ] in modern IC designs makes Avant!'s decision to use a [ ] unusual. (Teig Decl. && 67-69; Sechen Decl. & 50.) Most routers, however, use a [ ] approach, including ArcGate. (Sarrafzadeh Decl. & 109; Robins Decl. & 80; S. C. Chang Decl. && 41-44.) Dr. Tzeng considered [ ] in his Ph.D. dissertation and concluded that [ ] were superior. (Tzeng Decl. & 60, Exh. B at 55-57.) All of the routers he has developed in his career have been [ ] . (Tzeng Decl. & 60.) Dr. Tzeng, however, has created new techniques in [B] for handling [ ] , techniques apparently unknown to Cadence. (Tzeng Decl. & 62.) [#16] . Mr. Teig claims that Cadence and Avant! share a common bug as a result of using the [#1] algorithm and attempting to handle [#16] . (Teig Decl. && 70-72.) Mr. Teig admits that this "bug" exists only if a router uses [#1]. (Teig Dep. 408:9-22.) Significantly, [B] does not use [#1]. (Tzeng Decl. & 69; Sarrafzadeh Decl. & 113.) Moreover, Dr. Tzeng invented a new approach for handling [#16] for [B]. (Tzeng Decl. & 68.) Cadence has identified a bug in its own program that ArcCell-XO does not have. [#17] . Cadence's Mr. Teig also claims that Cadence and Avant! share a common bug related to the use of [#17] to avoid violating certain design rules. (Teig Decl. & 74-75.) Avoiding such violations is a common problem in place and route generally and has been discussed repeatedly in the literature. (Sarrafzadeh Decl. && 117-118; Robins Decl. & 85.) Once again, however, Dr. Tzeng adopted a novel approach that avoids the most significant aspect of this problem. (Tzeng Decl. & 70.) The so-called "bug" is only in Cadence's code. [#8] and [#10] .These issues as they relate to both global and detailed routing were discussed in the previous section on global routing. Again, these are common programming techniques widely used in the field. (Sarrafzadeh Decl. && 119-120; Robins Decl. && 86-87.) As in the case of the global router, every one of the claimed "trade secrets" allegedly found in [B] has been published. In asserting certain similarities, Cadence has repeatedly mischaracterized [B]. Cadence's claims, when corrected, reveal differences in the programs. Cadence has also ignored that [B] differs fundamentally from the Cadence detailed routers. (Sarrafzadeh Decl. && 121-129.) Cadence has demonstrated no likelihood of proving that the ArcCell-XO detailed router contains any Cadence trade secrets.Article: 3670
A number of users have asked me to seek out the Avant! court filing (instead of just the Avant! press release) so they could judge for themselves what was going on. Here it is in its gory detail! - John Cooley Avant!'s Court Filing (part 3 of 3) -------------------------- C. Avant!'s Design Size Optimization Program Was Independently Developed And Contains No Cadence Trade Secrets. The Design Size Optimization ("DSO") programs in ArcCell-XO and Cadence's VSize automate the process of establishing and adjusting die size. VSize was written by Eric Cheng. DSO (which includes the directory [ ]) was written by Eric Cheng and Dr. Venky Ramachandran. (Ramachandran Decl. & 6.) Before writing any source code, both Mr. Cheng and Dr. Ramachandran wrote proposals that were presented to the Avant! technical staff. (Ramachandran Decl. && 11-14, Exhs. B-D.) After incorporating the comments of the Avant! technical staff, Mr. Cheng and Dr. Ramachandran began writing code. (Ramachandran Decl. & 18.) Over the next months they communicated regularly to improve the quality and features of DSO. (Ramachandran Decl. && 19-25.) As finally released to customers, DSO represents the collaboration over several months by its authors Mr. Cheng and Dr. Ramachandran. Cadence's trade secret claims with respect to DSO rest on the analysis of Dr. Johnson, who concludes that Avant! misappropriated "unique, proprietary trade secret algorithms and methods." (Johnson Decl. & 2.) Professor Sechen offers some views on the alleged novelty of certain aspects of DSO, but he conceded that he relied on Dr. Johnson's examination of the code. (Sechen Dep. 130:9-131:8.) This reliance was entirely misplaced. Dr. Johnson admitted that she is not an expert in computer aided design, that she had never read a paper in the field, and that she had no familiarity at all with DSO's functionality, the data it used, or even the most basic concepts in place and route technology. (Johnson Dep. 11:19-12:9, 20:22-21:4, 22:14-23:2, 58:6-60:3, 59:1-4; 80:4-8; 119:2-8.) Not surprisingly, Cadence's allegations of trade secret misappropriation with respect to DSO are entirely wrong. All of Cadence's claims and Avant!'s expert rebuttals are summarized below. Implementing a [#18A] By [#18B] . Dr. Johnson calls this a "very unusual approach in the field in general, and finding it in both systems is even more unusual." (Johnson Decl. & 15.) When asked, however, Dr. Johnson could not explain the concept of a [#18A] and admitted that she had no knowledge of what approaches were usual or unusual in the field. (Johnson Dep. 20:19-23:2.) As Professors Sarrafzadeh and Robins explain, the concept of an "inherently [#18B] " is nonsensical: all routers use a [ ] during the routing process. (Robins Decl. && 90-94; Sarrafzadeh Decl. && 133-136.) The common function of VSize and DSO is to automate the otherwise manual process of adjusting the size of the die in a fixed-die place and route tool. (Robins Decl. && 97-98; Sechen Dep. 158:11-159:10.) Cadence was not even the first to introduce this approach commercially. (Robins Decl. & 99.) This is not a novel approach for place and route tools and certainly does not constitute a trade secret. (Robins Decl. & 99.) Although the authors and the general purpose overlap, VSize and DSO accomplish their tasks in different ways. In the earliest stages of development, fundamental aspects of DSO were specified that would improve its utility to users and allow it to be used with more types of designs. (Robins Decl. && 138-141; Ramachandran Decl. && 26-30.) VSize shares none of these characteristics. (Robins Decl. && 138- 141.) Use of [#19] . In her declaration, Dr. Johnson asserts that the use of [#19] in the programs is "unique." Under questioning, however, she admitted that she had no personal knowledge to support her conclusion. (Johnson Dep. 27:1-9.) In fact, [#19] are common in computer science and particularly in place and route tools. (Sechen Dep. 163:6-165:2; Robins Decl. && 101-107; Sarrafzadeh Decl. & 137.) In both VSize and DSO, they are used to [ ] for the purpose of estimating the proper chip size. Cadence's Professor Sechen and Avant!'s experts agree that this is a "standard use of [#19] ." (Sechen Dep. 166:25-167:2; Robins Decl. & 107.) Years before Cadence even began the VSize project, source code written at Avant! by Dr. Yang Cai, who never worked at Cadence, used [#19] in this very context. (Cai Decl. && 7-8.) [#19] Data Structures. Dr. Johnson concludes that the similarity in the terms used in [#19] data structures in VSize and [ ] is "beyond coincidence." (Johnson Decl. & 28.) Dr. Johnson admitted, however, that she looked only for similarity, not its significance, and that she has no familiarity with the field and cannot say whether these terms reflect common usage. (Johnson Dep. 58:20-25, 59:18-23.) Indeed, the terms that she identified, such as "[ ]," "[ ]," and "[ ]," are all common in this field. (Robins Decl. & 110.) Rather than "evidence of derivation," these similarities result from the same author writing code for a similar purpose at a different time without access to the original code. (Robins Decl. && 109-111.) Dr. Johnson admitted that this would explain all of the similarity she found. (Johnson Dep. 65:8-15, 66:3-14, 61:9-14, 81:11-20.) Cadence withheld this information from Professor Sechen, who never even considered the possibility. (Sechen Dep. 146:11-147:25; 168:9-169:5.) [#19] Construction. Again ignoring that the same author wrote both programs, Dr. Johnson questions the [ ] way that VSize and DSO build [#19]. (Johnson Decl. && 33-34.) Admitting that she did not even try to search the relevant literature, Dr. Johnson was unable to say whether or not this is the standard approach. (Johnson Dep. 76:24-78:5.) In fact, this is a standard way of building [ ], and a similarity that one would expect to find in code independently written by the same person at different times, as Dr. Johnson ultimately admitted. (Robins Decl. && 113-114; Johnson Dep. 81:11-20.) The [#20] . At her deposition, Dr. Johnson was asked to describe the [#20] . Her reply was: "No, I can't." She was asked whether she knew if any other commercial products used [#20]. Her answer was "No." When she was asked whether she made "any effort to find out," she again answered "No." (Johnson Dep. 10:16-11:24.) Nevertheless, in her declaration she confidently asserts that "the Cadence VSize and Avant! [ ] systems are unique in their use of the [#20] ." (Johnson Decl. & 36.) Professor Sechen and Avant!'s experts agree that [#20] has been widely discussed in the literature. (Sechen Dep. 167:17-21; Robins Decl. & 116; Sarrafzadeh Decl. & 138.) Cadence itself concluded that [#20] was in the public domain after Dr. Johnson inquired, but withheld this information from its own expert. (Rogoyski Dep. 113:1- 11, 117:21-119:21.) Admitting that [#20] has been generally discussed, Professor Sechen questions its use in the context of [ ] in an integrated circuit design. (Sechen Decl. & 32.) As Professor Robins explains in detail, and Professor Sarrafzadeh confirms, [#20] has been frequently used in this context. (Robins Decl. && 125-137; Sarrafzadeh Decl. & 138.) Many of these implementations use [#19] , as do both VSize and DSO. (Robins Decl. && 127-132.) The foregoing demonstrates that many of the supposed "trade secrets" common to VSize and DSO have been published. The remaining similarities merely result from the same engineer approaching a similar programming task at a different time. None of these is evidence of misappropriation. Cadence has demonstrated no likelihood that it will succeed in proving that the DSO program in ArcCell-XO contains any Cadence trade secrets. VI. CADENCE'S INNUENDO REGARDING IGUSA DOES NOT SUPPORT ITS CLAIM FOR INJUNCTIVE RELIEF. A. The Code Related To Igusa Does Not Support Injunctive Relief. In discovery, Avant! produced a diskette with a limited amount of code related to Igusa. (Snyder Decl. & 2.) These files, as identified on Cadence's list of 117, begin with the letters "[ ]" or "[ ]" and have no color designation on the chart in Exhibit A. This code in no way supports Cadence's preliminary injunction motion. First, neither this code nor any other code related to Igusa is, or ever has been, used in any Avant! product. (C. H. Chang Decl. && 19-20; Tsou Decl. & 18.) Cadence's experts independently reached this conclusion, although Cadence did not disclose this fact to the Court. (Smith Decl. & 13, Exh. 12 (Navas Dep. 199:12-23, 204:9-20, Exh. 16 ([ ]/ files), Exh. 18 ([ ]/ file)).) Second, as Cadence's expert has now conceded, Cadence's assertion that the [ ] code on the diskette contains Cadence code is simply wrong, as is Cadence's assertion that Igusa's trash contained "a printout of source code taken from Cadence's QplaceT product." (Br. 11:11-12, 25:18-19.) (Sechen Decl. & 69; Robins Decl. && 151-152.) Third, the code demonstrates that Igusa was doing legitimate work unrelated to the code he allegedly took from Cadence. (Robins Decl. && 151-152.) When questioned, Professor Sechen admitted that every similarity he found would be explained by Igusa being the author of both sets of files. (Sechen Dep. 226:8-13, 230:22-231:1, 234:15-23.) B. Cadence Is Using Igusa To Divert Attention From The Fact That Avant!'s Products Contain No Infringing Code Or Misappropriated Trade Secrets. To divert attention from the absence of evidence showing infringement or trade secret misappropriation by any Avant! product, Cadence attempts to construct an elaborate conspiracy among Avant! employees and Mitsuru Igusa. Nothing Cadence alleges about Igusa can justify an injunction of Avant! products that do not infringe or misappropriate Cadence intellectual property. More than that, Cadence's conspiracy theory is groundless. One of the few things on which the parties in this case agree is that Mitsuru Igusa is a brilliant engineer. Cadence CEO Joseph Costello describes Igusa as the best place and route engineer at Cadence. (Costello Dep. 67:6-18.) Costello "made several extraordinary offers" to try to convince Igusa to stay at Cadence, including the opportunity for Igusa to define his own compensation package. (Costello Decl. & 5; Snyder Decl. & 3, Exh. B.) Allegations of stolen code aside, any place and route company would jump at the chance to employ Igusa purely for his skills. Igusa also is a friend of a number of Avant! employees. (Hsu Decl. & 10.) There is nothing surprising or improper about the fact that Avant! hoped that Igusa would come work at Avant! in 1995. (Hsu Decl. & 10.) Nor was there anything wrong with Avant! employees having contact with Igusa. (Snyder Decl. & 3.) After the search of Igusa's home, members of Avant!'s Board of Directors concluded that it would not look right for Avant! to hire Igusa before the allegations against him were resolved. The directors also believed (correctly, as this suit demonstrates) that Cadence would use any association between Igusa and Avant! as an excuse to sue Avant!. (Kagle Decl. & 8.) One of the issues discussed by Avant!'s board members in connection with hiring Igusa was the fact that Igusa faced financial pressures. The directors concluded that, while Avant! should not hire Igusa, Igusa's friends at Avant! could assist him personally with their own funds if they wished. Also present was Avant!'s corporate attorney from Brobeck, Phleger & Harrison. (Kagle Decl. & 9; Hsu Decl. & 10; Spurlock Decl. & 3.) Avant! employees Hsu, Cho, Wuu, Liao, and Huang discussed the possibility of investing in a venture with Igusa that would not involve Avant!. Such an investment would both assist Igusa and carry through on earlier discussions among these Avant! employees about investing in a start-up company. Avant!'s initial public offering in June 1995 generated hundreds of thousands of dollars of income for Cho, Wuu, and Liao, who each agreed with Hsu to contribute $50,000 as seed money for a start-up company. In the meantime, they also agreed that some of this personal money could be used to assist Igusa and his family. Shiao-Li ("Leigh") Huang agreed to collect and maintain the funds until something more formal was established. (Hsu Decl. & 11; Cho Decl. && 10-11.) On June 23, 1995, Leigh Huang spoke with a corporate attorney about funding a start-up company involving Mitsuru Igusa, with $50,000 initial investments from Hsu and Avant!'s founders. (Brill Decl. & 32.) The Avant! employees did not set up a formal company, but there is nothing illicit about that. To describe 1995 as frenetic for Avant!'s principals would be a vast understatement. The company went public, went through a merger, and dramatically increased its sales and customer base. Then the District Attorney's search, this lawsuit, and Cadence's smear campaign caused Avant!'s stock to lose half a billion dollars in value in a matter of days and compelled Avant!'s customers to reconsider their relationships. Avant!'s principals have been working night and day, first to build and then to save their company. Formalizing the structure of a possible start-up company was hardly a priority. Nor did these four friends and close business associates feel any urgency to document their agreement. (Hsu Decl. & 11; Cho Decl. & 11.) All of Cadence's "evidence" of a grand Igusa conspiracy is completely consistent with the explanation that the Avant! employees were doing no more than attempting to follow the board's decision that Avant! should not be associated with Igusa. This simple explanation sharply contrasts with Cadence's convoluted conspiracy theory. VII. CADENCE CANNOT MEET THE REQUIREMENTS FOR PRELIMINARY INJUNCTIVE RELIEF. A. Cadence Cannot Meet Its Burden Of Establishing Its Likelihood Of Success On The Merits. A plaintiff's likelihood of success on the merits is the most important element of the Ninth Circuit's test for preliminary injunction. RasterOps v. Radius, Inc., 861 F. Supp. 1479, 1482 (N.D. Cal. 1994). As demonstrated above, Cadence will not succeed on the merits of its claims. Moreover, any doubt regarding Cadence's likelihood of prevailing must result in the denial of its motion: "Probability of success" implies that the plaintiff must have a very clear and strong case. Some courts have stated this in terms by the maxim that in considering preliminary injunction relief "to doubt is to deny." That is, if there is doubt as to the probability of plaintiff's ultimate success on the merits, the preliminary injunction must be denied. Direx Israel, Ltd. v. Breakthrough Medical Corp., 952 F.2d 802, 813 (4th Cir. 1991) (citation omitted). B. Cadence Cannot Establish Irreparable Harm. To obtain a preliminary injunction, regardless of the Ninth Circuit formulation applied, Cadence "must demonstrate that there exists a significant threat of irreparable injury." Oakland Tribune, Inc. v. Chronicle Pub. Co., 762 F.2d 1374, 1376 (9th Cir. 1985). Cadence cannot demonstrate irreparable harm. First, the Pre-Release Code in dispute has been removed from Avant!'s products. Second, Cadence's long delay in bringing this action belies its claim of irreparable harm relating to any post-Release misappropriation. Finally, any harm resulting from such misappropriation is readily quantifiable and can be remedied through money damages. 1. Cadence Cannot Demonstrate Irreparable Harm From Any Pre-Release Misappropriation Because Avant! Has Removed All Such Code Through A Clean Room Process. Avant! has replaced the Pre-Release Code through a clean room process. As a result, there can be no irreparable injury and an injunction is not appropriate. Alexander & Alexander Consulting Group, Inc. v. W.F. Corroon Corp., 1995 WL 46373, at *7 (N.D. Cal. Jan. 23, 1995); accord, Kepner-Tregoe, Inc. v. Leadership Software, Inc., 12 F.3d 527, 538 (5th Cir.), cert. denied, 115 S.Ct. 82 (1994); Speech Technology Assocs. v. Adaptive Communication Systems, Inc., 1994 U.S. Dist. LEXIS 11660, at *29 (N.D. Cal. Aug. 16, 1994) (quoting Cal. Civ. Code $ 3426.2); Computer Assoc., 775 F. Supp. at 561-62; aff'd in pertinent part, 982 F.2d 693, 701 (2d Cir. 1992). 2. Cadence's Assertion Of Irreparable Harm Is Contradicted By Its Own Delay In Bringing Suit. A "[p]laintiff's long delay before seeking a preliminary injunction implies a lack of urgency and irreparable harm," and thereby negates the whole basis for injunctive relief -- the need for speedy action to protect the plaintiff's rights. Oakland Tribune, 762 F.2d at 1377; accord Ampex Corp. v. Abekas Video Sys., Inc., 1990 U.S. Dist. LEXIS 11309, at *14-*15, 15 U.S.P.Q.2d 1219 (BNA) (N.D. Cal. Feb. 7, 1990). Cadence knew that Eric Cheng moved from Cadence to Avant! in February 1995 to work on a competing product. (Given Dep. 112:14-117:1.) Cadence also knew in February 1995 that Cheng had made a copy of some of the source code at issue in this case. (Woo Decl. && 2, 4-6, 8- 12.) Cadence concedes that it learned no new facts about Cheng's alleged misappropriation prior to filing this lawsuit. (Given Dep. 120:21-121:11.) Despite having in February 1995 the same evidence it now cites to establish misappropriation of its trade secrets, Cadence took no action. Courts routinely reject injunctive relief when faced with similar, and often shorter, delays. Citibank, N.A. v. Citytrust, 756 F.2d 273, 276 (2d Cir. 1985) (reversing injunction in trademark infringement action where plaintiff delayed ten weeks after it learned directly of defendant's plan to begin using allegedly infringing mark and more than nine months after it indirectly learned of such plans); Cuddle Wit, Inc. v. Chan, 1989 U.S. Dist. LEXIS 14412, at *2, 14 U.S.P.Q.2d (BNA) 1572, Copyright L. Rep. (CCH) & 26,507 (S.D.N.Y. Dec. 1, 1989) (seven-month delay in instituting copyright infringement action warranted denial of preliminary injunction); Business Trends Analysts v. Freedonia Group, Inc., 650 F. Supp. 1452, 1459 (S.D.N.Y. 1987) (plaintiff's delay of at least six months before commencing action vitiated presumption of irreparable harm). 3. The Harms Cadence Asserts Are Not Irreparable. Cadence concedes that its direct injuries resulting from lost sales of other products and services are quantifiable. (Teece Decl. & 10- 11.) It argues, however, that those lost revenues will produce more subtle impacts that are "less tangible" and "more difficult to monetize." (Id. & 11). Not only are the additional harms claimed by Cadence readily quantifiable, (Silverman Decl. && 20-34), they cannot constitute irreparable injury as a matter of law. Cadence's argument that its lost sales inhibit its ability to reposition itself as a service-oriented company is speculative and does not constitute irreparable harm. (Br. 34; Teece Decl. & 9(o)). Goldie's Bookstore v. Superior Court of State of Cal., 739 F.2d 466, 472 (9th Cir. 1984) (loss of "untold" customers too speculative); Bell Atl. Business Sys., Inc. v. Storage Technology Corp., 1994 U.S. Dist. LEXIS 4471, at *7-*9 (N.D. Cal. Mar. 31, 1994) (finding assertions that the moving party will have greater difficulty obtaining contracts in the future to be speculative); Brooktree Corp. v. Advanced Micro Devices, Inc., 705 F. Supp. 491, 493-96 (S.D. Cal. 1988) (refusing to grant injunction based on "lost ability to expand or go public due to lost profits"). Cadence's own witness admits that it is too early to say whether there will be any effect on that aspect of its business. (Douglas Dep. 104:9-105:4.) Cadence's assertion that loss of reputation due to lost sales is grounds for an injunction is erroneous. (Teece Decl. & 9(h)). See Los Angeles Memorial Coliseum Comm'n v. National Football League, 634 F.2d 1197, 1202 (9th Cir. 1980); accord Virginia Carolina Tools, Inc. v. International Tool Supply, Inc., 984 F.2d 113, 120 (4th Cir. 1992), cert. denied, 508 U.S. 960 (1993). Nor, as Cadence's expert maintains (Teece Decl. & 9(g) and (l)), is diminished research and development due to lost profits an irreparable injury. Eli Lilly and Co. v. American Cyanamid Co., 82 F.3d 1568, 1578 (Fed. Cir. 1996). Characterizing such collateral impacts as irreparable harm would improperly "convert the 'extraordinary' relief of a preliminary injunction into a standard remedy." Id. C. The Balance Of Hardships Overwhelmingly Favors Avant!. A critical consideration in granting an injunction is the relative hardship to the parties. International Molders' and Allied Workers' Local Union No. 164 v. Nelson, 799 F.2d 547, 551 n.2 (9th Cir. 1986). "Traditional standards for granting a preliminary injunction impose a duty on the court to balance the interests of all parties and weigh the damage to each." Los Angeles Memorial Coliseum Comm'n, 634 F.2d at 1203. In 1995, Cadence had revenues of over $540 million. (Silverman Decl. & 10.) Any harm to Cadence prior to a permanent injunction hearing can be compensated by monetary damages. (Silverman Decl. && 18-34; Huyett Decl. & 2.) On the other hand, a preliminary injunction would seriously harm Avant!, whose place and route products account for more than 60% of its income. (Huyett Decl. & 3.) Thus, the balance of hardships tips sharply in Avant!'s favor. See Apple Barrel Productions, Inc. v. Beard, 730 F.2d 384, 390 (5th Cir. 1984); Farmers Indep. Tel. Co. v. Thorman, 648 F. Supp. 457, 459 (W.D. Wis. 1986). Injunctive relief is particularly inappropriate in this case because of the harm to Avant! caused by Cadence's delay. Cadence filed its complaint as Avant! was about to release its next generation product, ArcCell-XO, on which Cheng had been working for most of 1995. (Jeong-Tyng Li Decl. & 22.) Avant! invested tremendous time and resources in the development of ArcCell-XO while Cadence waited to sue, including eight man years of work from its most talented engineers. (Hsu Decl. & 13.) Under similar circumstances, where the plaintiff had delayed until after defendant had printed 12,000 copies of a book, the court rejected injunctive relief stating "plaintiff must bear the consequences of its [strategic] delay." New Era Publications Int'l, ApS v. Henry Holt & Co., 684 F. Supp. 808, 811 (S.D.N.Y. 1988), aff'd, 873 F.2d 576 (2d Cir. 1989). This Court also has rejected preliminary injunctive relief where the defendant has invested substantial resources and developed a significant client base during the plaintiff's delay. Integral Systems, Inc. v. PeopleSoft, Inc., 1991 U.S. Dist. LEXIS 20878, at *53 (N.D. Cal. July 19, 1991). D. An Injunction Would Severely Harm Avant!'s Customers And The General Public. This Court also must consider the harmful effect that an injunction would have on third persons and on the general public. Caribbean Marine Services Co. v. Baldrige, 844 F.2d 668, 674 (9th Cir. 1988). Enjoining Avant!'s ArcCell-XO -- indisputably the most advanced product in the field -- would seriously harm Avant! customers, which include every major chip maker in the world, and the public. (Eastman Decl. && 4, 8, 11, 15 & 19.) Avant!'s place and route products are essential to chip makers' producing better, more powerful ICs, faster and more efficiently. (Id.) Place and route tools have fallen behind the technology curve, creating a bottleneck for many IC designers. (Silverman Decl. & 16.) The injunction Cadence seeks would seriously exacerbate this bottleneck. An injunction also would extend Cadence's monopoly to upward of 80% of the place and route market and leave chip makers dependent on Cadence's outdated technology. (Douglas Decl. & 19.) Given that applications employing ICs have expanded dramatically from their traditional uses in computers to telecommunications, multimedia, automotive, industrial automation, and consumer electronics industries, the public would suffer greatly if Avant!'s products are enjoined. (Silverman Decl. & 16; Eastman Decl. & 6.) Courts are not hesitant to deny injunctive relief under such circumstances. See Abend v. MCA, Inc., 863 F.2d 1465, 1479 (9th Cir. 1988), cert. granted, 493 U.S. 807 (1989), aff'd 495 U.S. 207 (1990); Intel Corp. v. Advance Micro Devices, Inc., 1994 WL 621665, at *2 (N.D. Cal. Oct. 31, 1990). As Judge Lynch observed in denying a preliminary injunction, intellectual property law "must strike a balance between the broad monopoly power to exclude . . . and the invariable need for competition in our free market economic system." Ampex Corp. v. Abekas Video Systems, Inc., 1990 U.S. Dist. LEXIS 11309, at *16-*17, 15 U.S.P.Q.2d 1219 (BNA) (N.D. Cal. Feb. 7, 1990). Here, an injunction's harm to chip makers and the public can be avoided entirely given the adequacy of money damages. CONCLUSION For the foregoing reasons, Avant! respectfully requests that the Court deny Cadence's Motion for Preliminary Injunction.Article: 3671
Joey Y. Chen wrote: > > Hi, > > Anyone of you out there had experience using both Xilinx and Altera > FPGA's? Since our Xilinx FPGA (XC4025E) order has been delayed yet again, > I would like to know how the Altera Flex 10K series compares with the Xilinx > XC4000E in terms of capacity. > The Xilinx XC4025E claims capacity of 25,000 "typical" gates, and > the Altera Flex 10K-100 claims 100,000 "typical" gates. Are they comparing > the same thing here? > Thanks. > > -Joey Chen Basically, No. We are currently doing a comparison of the XILINX 4025e vs. ALTERA 10K50. Altera computes the gate count as a function of the logic cells and the memory cells that are available in the 10K parts. Altera publishes an app-note that describes how they compute gate count. XILINX uses the standard gate count they have had for all of the XC4000 parts. - ScottArticle: 3672
David J. Matthews wrote: > > Hi, > > I am embarking on an XC4013E design and have a couple of questions... > > 1. I am bringing a bidirectional 32-bit bus into the device which > essentially looks like a microprocessor bus. It is used to load > and read back registers, read back status, etc.<snip> As you point out, there are tradeoffs. My larger 4000 designs have all had a single wide read path(32 bit results) and a narrow write path(config & control), so I used a unidirectional internal tristate bus for the wide read data path, and muxes on the lower bits for register readback. It is helpful not to need bus alignment for non-critical control registers. I sometimes use a RAM block as a "register shadow RAM" to eliminate the need for a readback data path from non-volatile control registers. > > 2. The second question is regarding the preferred method for using > the floor planner. Is it better to floor plan before PPR or after? > My experience with the XACT floorplanner is nonexistent, other than running it once to look at the results of an unconstrained PPR run (not pretty), but I would say floorplanning on a big, high-performance, high-utilization design should come very early, in fact during the design feasibility stage. Efficiently designing, placing, packing and aligning blocks on the CLB grid is not an easy problem. I prefer to do it myself with handmade RPMs and RLOC_ORIGINs at the schematic level, while I am designing the blocks, with occasional PPR runs to make sure things are still copacetic. Not exactly a pushbutton design flow, but it gets the job done, with fewer layers of design translation going on to confuse things, and fewer pages in the "Known Issues" release document to search when things go wrong. (grin) Best wishes - TomArticle: 3673
Hello. I'm not quite suttisfied from the israeli EastRonics Altera vendor services and availiability, and since they are the only one in israel, i'm searching for a second source of altera's chips e.g AT&T for Xilinx. So please, if you know about it anything please remail me to: mailto:gabby@isv.dec.com -- [#]---[#]---[#]---[#]-<*>-<*>-<*>-<*>-<*>-[#]---[#]---[#]---[#] `\ /' 0 \ 0 / Gabby Shpirer /[ ]\ [ ] VLSI DEC site at Jerusalem Israel / [ ] \ [ ] email:gabby@isv.dec.com --- --- _/ \_ _/ \_ Opinions expressed are mine. [#]---[#]---[#]---[#]-<*>-<*>-<*>-<*>-<*>-[#]---[#]---[#]---[#]Article: 3674
Is There anybody out there who knows anything new about the Xilinx XC6200 partially reconfigurable fpga ? The data available from Xilinx has not changed for months and it looks like if they are not pushing the device ahead anymore. I hope I'm wrong because this FPGA is a must in the field of reconfigurable computing. It will be really Dommage ! Thanks for any info on that . Cheers, Christian. ((((( ( ) ( 0 0 ) ( ! ) | \_/ | |------------------------------ooO---------Ooo-------------------------------| | De/From Christian Gamrat (MIND-1024) | | Commisariat a l'Energie Atomique / Atomic Energy Commission | | Groupe Architectures Paralleles / Parallel Computing Group | | DTA/LETI/DEIN/SLA/GAP | O| CE-SACLAY 91191 Gif-sur-Yvette Cedex - FRANCE |O | e-mail : cgamrat@gaap.saclay.cea.fr | | phone : +33 1 69 08 71 94 | | fax : +33 1 69 08 83 95 | |----------------------------------------------------------------------------|
Site Home Archive Home FAQ Home How to search the Archive How to Navigate the Archive
Compare FPGA features and resources
Threads starting:
Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z