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
I should be more specific. I am looking at using DSP chips for a graphics application, because of their high arithmetic power, and I'd like to know if I can implement fast polynomial, and other, calculations in FPGA with better price/performance. In article <3jgvu4$bee@vanbc.wimsey.com>, Rob Semenoff <rob@vanbc.wimsey.com> wrote: > >Hi everybody, > >I want to implement a special purpose DSP algorithm in FPGA, >hoping that it will be faster,cheaper than the >alternatives which are: > > A general purpose DSP microprocessor > An ALU chip using an external state machine > -rob --Article: 826
Hi, I am looking for someone with whom I could share a room in Seattle on May 1,2 and 3. I will be presenting a paper in ISCAS-95. Thanks, Aurobindo DasguptaArticle: 827
I am interested in learning more about implementations of bit-serial multipliers in lookup-table based FPGAs. Input operands for my design are eight bits long. A pointer to any relevant literature with implementation examples would be appreciated. Thanks Russ Tessier MIT Lab for Computer ScienceArticle: 828
Hi, Are there any tools available to translate BLIF FORMMAT to Xilinx XNF format, and vice versa? Any comment and hints? Thanks for your help! Chih-chang LinArticle: 829
Yep, I heard $995 as $99.50. Oops. As an aside, the lower-priced version is the same CD and books that sell for $5K or so, just without the 4K/5K support since it's missing the dongle. I guess the 5.1 version will run 2K/3K and EPLD parts without the infamous dongle (if you're looking for a second copy, this is a hint). Not that I'm suggesting you void your user agreement on this seriously overpriced package. And yeah, we fell for it and bought BOTH systems, foolishly thinking there was more of a difference than merely a missing dongle. Silly us.Article: 830
I may be in the wrong forum for this, but I've seen FFT and the like mentioned from time to time; so I thought I'd give it a try. My application involves generating complex sound waves by adding and mulitplying together sines waves of many frequencies and relative phases. I thought I would generate the waves using Forth (for speed) and feed the amplitude stream to a DAC. But, having not tested the idea yet, I fear that bandwidth might still be a problem event on a 90 Mhz Pentium, since I have to do alot of math in real time to generate 44,100 16-bit amplitude values per second. So, can anyone suggest a less borderline way of doing this, say, with a DSP card? I know that DSPs can be implemented with FPGAs, but I'm not certain this is necessary. I guess the best way is to get the DSP to do all the math (that solves the bandwith problem). Can anyone recommend a DSP card and development software, or point me to a forum? Thanks. LeeArticle: 831
In article <lfadden.69.0015CDF9@harris.com>, lfadden@harris.com (Lee Fadden) writes: |> I may be in the wrong forum for this, but I've seen FFT and the like |> mentioned from time to time; so I thought I'd give it a try. |> |> My application involves generating complex sound waves by adding and |> mulitplying together sines waves of many frequencies and relative phases. |> I thought I would generate the waves using Forth (for speed) and feed the |> amplitude stream to a DAC. But, having not tested the idea yet, I fear that |> bandwidth might still be a problem event on a 90 Mhz Pentium, since I have |> to do alot of math in real time to generate 44,100 16-bit amplitude values |> per second. |> |> So, can anyone suggest a less borderline way of doing this, say, |> with a DSP card? I know that DSPs can be implemented with FPGAs, but I'm |> not certain this is necessary. I guess the best way is to get the DSP to |> do all the math (that solves the bandwith problem). Can anyone recommend |> a DSP card and development software, or point me to a forum? Thanks. |> |> Lee |> Check out comp.dsp for an answer to this question. You could do this with an FPGA but since you are interested in audio rate processing a DSP may be better, depending on the number of multiplications you expect to be doing per output sample. - Russell Petersen petersr@fpga.ee.byu.eduArticle: 832
In article <3jl66b$5pe@GRAPEVINE.LCS.MIT.EDU>, tessier@ROEBLING.LCS.MIT.EDU (Russ Tessier) writes: |> |> |> I am interested in learning more about implementations of bit-serial |> multipliers in lookup-table based FPGAs. Input operands for my design |> are eight bits long. |> |> A pointer to any relevant literature with implementation examples would |> be appreciated. |> |> |> Thanks |> |> Russ Tessier |> MIT Lab for Computer Science |> |> I have been working on results of FPGA multiplication for part of my thesis. My experience so far with bit-serial mults. is that they are simple to build if both operands are unsigned but the implementation can get a bit ugly with FPGAs if you want two's complement multiplication, depending on how you want to feed in your operands. With unsigned bit-serial multipliers I have achieved 80 Mhz clock rates. Lyon's paper on bit-serial multiplication is pretty much a standard reference for bit-serial mult. but the paper does have an error. Look for "Two's Complement Pipeline Multipliers" by R.F. Lyon in IEEE transactions on Communications, April 1976 pgs. 418-424. ___________________________ ----- Russell Petersen || BYU || Brigham Young University || || Reconfigurable Hardware Lab ----- petersr@fpga.ee.byu.edu --------- voice: (801) 378-7206 | - - | ___________________________ ---------Article: 833
Hi, If anyone would like to receive a PostScript copy of the following papers, please e.mail me. 1. ISCAS'94 - Virtual Hardware and the Limits of Computational Speed-up 2. ICCD'94 - Area & Time limitations of FPGA-based Virtual Hardware (both by Albaharna, Cheung, & Clarke) -------------------------------------------------------------------------- -- Osama T. Albaharna <a.osama@ic.ac.uk> -- -- -- -- Information Engineering Section -- -- Department of Electrical and Electronic Engineering -- -- Imperial College, Exhibition Road, SW7-2BT, London, U.K. -- -- -- -- Tel: +44-171-589-5111(ext 56210) Fax: +44-171-581-4419 -- --------------------------------------------------------------------------Article: 834
linchih@guitar.ece.ucsb.edu (Chih-chang Lin) wrote: > > I am looking for the information (call for paper, > registration)for "FPGA Custom Computing Machine workshop". > See info in http://www.super.org:8000/FPGA/caf.html -- Osama Albaharna <a.osama@ic.ac.uk>Article: 835
tessier@ROEBLING.LCS.MIT.EDU (Russ Tessier) wrote: > I am interested in learning more about implementations of bit-serial > multipliers in lookup-table based FPGAs. Input operands for my design > are eight bits long. > > A pointer to any relevant literature with implementation examples would > be appreciated. Hi Russ, See "Implementation and Performance Evaluation of Cellular Array Multipliers using FPGAs" by Arindam Saha & Rangasayee Krishnamurthy (Missisippi State University) in Proc. of the 26th Southeastern Symposium on System Theory, 20-22 March, 1994. ----------------------------------- Osama Albaharna <a.osama@ic.ac.uk> Information Engineering Section Imperial College, U.K. -----------------------------------Article: 836
Oops! I oversimplified my brief description of the bit serial multiplier in my earlier post. The bit cell is a full adder with registers on the sum and carry outputs and with the carry output wrapped back around to the input as I stated. One of the other inputs to the cell comes from the sum output of the previous cell, and the other is fed with the AND of the serial input and the corresponding bit in the parallel input. For a fixed parallel input, the ANDs get eliminated and the serial input gets fed only to cells corresponding to ones in the parallel input. The zero bits can be reduced to a single register The individual 'carry-save' adders can be extremely fast. The performance of the bit serial multiplier is limited by the distribution of the serial input across the cells in most of the FPGA architectures. A sixteen bit (on the parallel input) multiplier in the Atmel/NSC parts can do better than a 30 MHz bit rate, even with the local/global bus connections. My apologies if I caused any confusion in the earlier post. -Ray Andraka Chairman, the Andraka Consulting Group 401/884-7930 FAX 401/884-7950 email randraka@ids.netArticle: 837
There have been a few requests on the work of Razdan and Smith. You can access their papers thru Prof. Smith's homepage. http://das-www.harvard.edu/users/faculty/Michael_Smith/Michael_Smith.html well, if that is too long, follow the link from http://das-www.harvard.edu/ and find out new and exciting research at Harvard :-) -- - Kee kee@das.harvard.eduArticle: 838
Hi everybody, Does anybody around here use the XACT program from xilinx ? .... I need some help. I need to know that if we design the circuit that fit on xc4010. But we want to separate it into several chips such as 2 of xc4005. How can we do ? Can we use PPR program to do that for us ? Or we must cut the circuit down into 2 circuits ourself ? Any help would be appreciate. Regard, Wichai TangArticle: 839
In <3joqgq$9f@paperboy.ids.net> randraka@ids.net (Ray Andraka)writes: stuff deleted..... >The individual 'carry-save' adders can be extremely fast. The performance of >the bit serial multiplier is limited by the distribution of the serial input >across the cells in most of the FPGA architectures. A sixteen bit (on the >parallel input) multiplier in the Atmel/NSC parts can do better than a 30 MHz >bit rate, even with the local/global bus connections. The critical path for this type of macro is usually in routing to the arithmetic logic and this can be 'reduced' by registering (pipelining) the incoming/outgoing data to push data rates extreamly high (>70Mhz in some cases). Atmel is currently building a bit serial multiplier macro generator for its SRAM reconfigurable FPGA family, to compliment its growing set of arithmetic parameterizable macros such as : pipelined parallel multipliers; several different flavors of accumulator and adder; FIR filters; etc., more are being added - let us know what you would like to see. Questions, comments or literature requests should be sent to martin@atmel.com. Martin Mason FPGA Apps Engineer Atmel Corp. (San Jose) martin@atmel.com or fpga@atmel.comArticle: 840
In article <3j034s$rn6@newshound.uidaho.edu>, lharold@mrc.uidaho.edu (Len Harold) writes: >David Hough (dave@sectel.com) wrote: >: Most designers I know are willing to work pretty hard to avoid analog things >: like VCOs. >Most definitely. Looks like you are unfortunate to know few designers competent in analog design. ;-) In The Real World synchronization by a PLL & VCO sometimes is unavoidable: In a system collecting samples, the phase jitter of +- one tick introduced by digital PLL-less synchronization is the last thing you want. In a digital system receiving asynchronous inputs, there is always the problem with setup and hold times getting violated. The designer must address the problem of metastable states and be up to the job of verifying the error rate to an acceptable figure. (It can't be zero.) The digital part of a PLL can be as simple as a single XOR gate. I am presently designing a system architecture where the ~50MHz clock for entire digital system including the microprocessor is phase locked to a 136 kbit/s telemetry channel. Metastable states are avoided and equidistant sampling is assured. The PLL will most likely be executed as part of an FPGA. ========================================================================= Goran Olsson Alfven Laboratory, Plasma Physics, Space Group, Electronic Engineering, KTH (Royal Institute of Technology), Stockholm, Sweden F1 - The Electric Field Experiment on the FREJA satellite Two years in orbit 6 Oct 1994 olsson@plasma.kth.se http://www.plasma.kth.se/alfven-lab.html =========================================================================Article: 841
AMD: I have the data books of the Mach devices of AMD, and in it the Mach215 is expressively designated for asynchronous operation. Each output macrocell can be clocked by its own product term. The polarity of the clock is programmable for each macrocell. Individual asynchronous reset and preset and three-state product terms for each macrocell. Furthermore all the Mach3 and Mach4 devices are able to do asynchronous operation. They have everything the Mach215 has except the programmable clock polarity, and only one of asynchrounous reset or preset is (alternatively) available at a macrocell. And they are able to build RS- and JK-flipflops, not only D-, T-flipflops and Latch. So it gets down to this: if individual clocks and either a reset or preset are good enough for you, you have a wide choice, if you need both asynchronous reset and preset, only the Mach215 applies. XILINX: Their EPLD line (XC7xxx) is not designed for asynchronous operation. They have no individual clock. But they have one individual product term for asynch. reset and preset, so it is remotely possible to use it asynchronously. The XC3000 devices have individual clocks and individual reset (one for both flipflops in a CLB/macrocell), but no preset. The XC4000 family is somewhat comparable with the Mach3 and Mach4 family in that it has either an asynchronous reset or preset, but not both in a flipflop. Hope this helps, Holger. -- Holger Hellmuth at Uni Karlsruhe <hellmuth@ira.uka.de>Article: 842
In article <lfadden.69.0015CDF9@harris.com> lfadden@harris.com (Lee Fadden) writes: >My application involves generating complex sound waves by adding and >mulitplying together sines waves of many frequencies and relative phases. >I thought I would generate the waves using Forth (for speed) and feed the >amplitude stream to a DAC. But, having not tested the idea yet, I fear that >bandwidth might still be a problem event on a 90 Mhz Pentium, since I have >to do alot of math in real time to generate 44,100 16-bit amplitude values >per second. > >So, can anyone suggest a less borderline way of doing this, say, >with a DSP card? I know that DSPs can be implemented with FPGAs, but I'm >not certain this is necessary. I guess the best way is to get the DSP to >do all the math (that solves the bandwith problem). Can anyone recommend >a DSP card and development software, or point me to a forum? Thanks. > >Lee > This may be very out of line depending on your particular application, but have you considered doing the same thing but using Walsh functions as your basis instead? They're very easy to generate using FPGAs (you just need lots of XORs) and to sum, and there are relationships with sine/cosine functions ('cal' and 'sal') you can use for your phase shift... I once built a little system with old TTL parts for this Graham -- -------------------------------------------------------------------- Graham Seaman, School of Computer Science, University of Westminster, 115 New Cavendish St. London W1M 8JS email: seamang@wmin.ac.uk www: http://www.wmin.ac.uk/~seamangArticle: 843
I presume vendors are increasing FPGA density by going for bigger dies and smaller feature size. Are any trying to get more logic into a package for less money by interconnecting dies in a package in some programmable, or customer-specified way ? -robArticle: 844
Head's Up! SNUG's happening in 12 days! OVI's happening in 17 days! NOW's the time to register for these conferences, reserve those hotel rooms and book those flights to San Jose, California before they sell out! (I've included the SNUG '95 info sheet in this message; you can get the OVI info from comp.lang.verilog on the Internet.) - John Cooley Part Time EDA Consumer Advocate Full Time ASIC, FPGA & EDA Design Consultant P.S. For you SNUG '95 attendees: if you need any materials to do a quick Verilog/VHDL -> Synopsys design (like coding style templates, etc.) make sure you bring it with you to SNUG '95. I'm judging the Design Contest where you'll be given 1 1/2 hours to create a quickie design from a spec and synthesize it to the fastest or smallest gate design you can get. This will be a good test of not only synthesis prowess but how you handle HDL coding styles. P.P.S Oh, yea, the winner or the winning team gets $1000.00! =========================================================================== Trapped trying to figure out a Synopsys bug? Want to hear how 3196 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." ----------- Welcome to SNUG 1995 Synopsys invites you to the Fifth Annual Synopsys Users Group (SNUG) Conference at the Red Lion Hotel in San Jose, California, on March 22-24, 1995. What is SNUG? SNUG is an open forum that provides Synopsys' users an opportunity to exchange ideas, discuss problems and explore solutions. It also gives users a chance to meet with Synopsys executives, developers and application engineers. The program is highly technical and developed by users with an emphasis on customer issues. The SNUG Technical Program Committee is responsible for the content of all breakout sessions. If you would like to participate in this process, contact Jeff Flieder at the address shown below. Why is it important? SNUG provides a unique opportunity to meet other Synopsys users. The program is designed to address a variety of design issues. User breakout sessions provide unique solutions to various design problems. ASIC vendor sessions provide specific techniques for building successful chips. EDA tool vendor presentations explore other EDA products which leverage Synopsys High-Level Design tools. What's new this year? SNUG 1995 is the site of the First Annual Design Competition. This competition gives you an opportunity to match your design skills against other Synopsys users. Winners will share their design tricks with you. You will also get a sneak preview of upcoming Synopsys Solutions, and a chance to examine some third party products as well. Feel free to pass this program on to any of your colleagues who are familiar with Synopsys. This year's program promises to be extremely exciting. Don't miss it! Jeff Flieder SNUG Technical Program Chair (719)528-7718, fax (719)528-7635 fmicos!splinter!flieder@uunet.uu.net Ken Nelsen Synopsys Applications (415)694-1771, fax (415)694-1864 knelsen@synopsys.com Schedule Wednesday, March 22 Tutorial registration 12:00 - 2:00 Tutorial sessions 2:00 - 5:00 Welcoming cocktail party/ Synopsys new product fair 5:00 - 7:00 Thursday, March 23 General session registration/breakfast 7:30 - 9:00 Welcome/agenda 9:00 - 9:15 Synopsys Direction (Aart de Geuss, President) 9:15 - 9:45 Breakout sessions 10:00 - 11:30 Topics include: User sessions, Design competition Lunch 11:45 - 1:00 Breakout sessions 1:15 - 2:45 Topics include: User sessions, Design competition Breakout sessions 3:00 - 4:30 Topics include: User sessions, Design competition Services Round Tables 4:40 - 5:15 Topics include: Surfing the Internet, Talk Back to Training Support Center Summit, Conversations with Consulting Keynote Speaker (John Gage) 5:30 - 6:15 Wrap-up 6:15 - 6:30 Cocktail party/EDA tool vendor fair 6:30 - 8:30 Friday, March 24 Tutorial registration / breakfast 7:00 - 8:00 Tutorial sessions 8:00 - 11:30 General Session Thursday, March 23 User Breakout Sessions These sessions are always the hit of the conference. Hear Synopsys users' experiences on specific topics. Each user breakout session will consist of two presentations, thirty minutes each, with another thirty minutes for questions and answers. Topics include: Design Compiler Breakout Moderater: John Cooley Jerry McGoveran, Synthesis of a 200K Gate DSP ASIC Steve Golson, My Favorite dc_shell Tricks Design Reuse Breakout Moderator: Kurt Baty Kurt Baty, DesignWare for Fun and Profit Jeff Freeman, Design Reuse Design Compiler Breakout Randy Bolling, Register Balancing : When and Where? Ron Crevier, Rapid Prototyping Using Behavioral Compiler Test Synthesis Breakout Moderator: John Vogel Rob Maffit, Test Compiler Methodology Submicron Design Breakout Moderator: Volker Kiefer Stefan Rusu, Advanced Timing and Wire Models in Synopsys Simulation and Verification Breakout Moderator: Erik Magdanz Brian Balthazor, General Instrument Case Study Erik Magdanz, Logic Modeling Overview Chris Beggy, Synopsys in a Mixed Tool-Suite Design Methodology Semiconductor Vendor Sessions Back by popular demand! Feedback from last year's sessions was very positive. This year we will expand to two sessions. This gives you an opportunity to hear about the latest technologies available from the semiconductor vendors and their recommended customer design flow using Synopsys tools. Participants include: Altera, Actel, IBM, LSI Logic, VLSI and Xilinx. EDA Tool Vendor Sessions New this year! These sessions have a format similar to the semiconductor sessions. They are an opportunity to see other EDA vendors show how they can help you leverage your high level design tools from Synopsys. Participa= nts include: Compass Design Automation, Summit Design, Inc., Intergraph, Mentor Graphics, Chronologic Simulation, Veda Desgin Automation Ltd., Viewlogic, High Level Design Systems, ArcSys, Inc., Speed Electronics, Vista Technologies and Knowledge Based Silicon. Design Competition Sessions Coordinator: John Cooley New this year! These sessions provide you with an opportunity to work with other high level designers and compete to see who can build the best design. Due to hardware limitations, the competition participation will be limited. Teams will be formed on a first come, first serve basis. A $1000 prize will be awarded to the winning team. Sign up for the competition when you register. Participants will be notified of final team composition and details as soon as they become available. Services Roundtables Surfing the Internet Find out what tools are available to help you explore the Internet, and get a glimpse into the future of automated customer services. A Coversation with Consulting Talk to our consulting team about how they can help you complete projects on time and within budget. The Support Center Summit Meet with members of Support Center, and discuss ways for you to work together more effectively. Talk Back to Training We are always working to improve our training workshops. Here's your chance to talk to our courseware developers and hear what they have planned for the future. Half-Day Tutorials Wednesday, March 22 New for SNUG '95: Speed Optimization Symposium This session is intended for experienced synthesis users who are interested in a comprehensive methodology for speed based optimization using Design Compiler. A "Baseline" methodology for speed optimization will be presented and then a panel of various experts from Synopsys will discuss their favorite tricks for optimizing high performance designs. This session is repeated on Friday. New for SNUG '95: Behavioral Synthesis Using VHDL This session is intended for any synthesis user interested in behavioral synthesis. Hardware designers are frustrated with lengthy IC design cycles for increasingly complex, data-intensive, algorithmic applications. This tutorial will show you how to enter your design specifications at the behavioral level and use Behavioral Compiler to explore implementation alternatives and determine the optimal architecture. Content is the same as the tutorial on Friday, but examples are presented in VHDL. VHDL Synthesis Techniques and Recommendations This tutorial will present a wide range of VHDL synthesis coding styles and issues. Caveats based on real-world VHDL synthesis models will be explored in full detail. Topics will be presented in the following areas: VHDL synthesis fundamentals, importance of VHDL coding styles, relying on hardware design experiences, disparities between efficient simulation models and optimum synthesized hardware, potential simulation and synthesis mismatches, and VHDL coding style differences between targeted CMOS and FPGA technologies. Pins-Out Verification This tutorial was first offered at EuroSNUG in August. "Pins-out" verification offers a solution to the problem of ensuring ASICs are right the first time. As the complexity of ASICs increases, validation of functionality and the trading off of design parameters becomes more difficult. First time right design depends on correct interactions between the ASIC and surrounding subsystems. Thus, it is important to simulate the IC in its system context; in other words, pins-out. This tutorial introduces the pins-out methodology and shows its benefits. Friday, March 24 New for SNUG '95: Speed Optimization Symposium This session is intended for experienced synthesis users who are interested in a comprehensive methodology for speed-based optimization using Design Compiler. A "Baseline" methodology for speed optimization will be presented and then a panel of various experts from Synopsys will discuss their favorite tricks for optimizing high performance designs. This session is a repeat of the Wednesday session. New for SNUG '95: Behavioral Synthesis Using Verilog This session is intended for any synthesis user interested in behavioral synthesis. Hardware designers are frustrated with lengthy IC design cycles for increasingly complex, data-intensive algorithmic applications. This tutorial will show you how to enter your design specifications at the behavioral level and use Behavioral Compiler to explore implementation alternatives and determine the optimal architecture. Content is the same as the tutorial on Wednesday, but examples are presented in Verilog. New for SNUG '95: DSP Design Techniques using COSSAP This session reviews DSP basics and applications. It provides a good overview of DSP issues and particularly focuses on the COSSAP design methodology to solve these issues. This session is intended for the designer who is new to using EDA tools for solving DSP applications. It is not intended for the designer who is proficient in the use of COSSAP tools. Advance Registration Please complete a separate registration form for each attendee. Specify breakout sessions for Thursday, and your choice(s) of tutorials. General Session Registration Registration includes all materials for the general and breakout sessions. Due to limited space, breakout session sign-up will be done on a first-come-first-serve basis. Both cocktail parties , breakfasts on Thursday and Friday, and the Thursday lunch are included. Tutorial Session Registration Registration for each tutorial includes the session and all related course material. Tutorials are available only to general session registrants. Space is limited - register early! General session: Late registration - $150.00 Tutorials (per session): Late registration - $75.00 each Full payment in U.S. dollars must accompany registration. Company and personal check, VISA, MasterCard, and American Express cards are accepted. All checks must be made payable to: Synopsys Users Group. For Credit Card Registration: * call 1-800-388-9125 * e-mail the information to designinfo@synopsys.com * fax your completed form to SNUG at 503-690-6906 To Register By Mail: Detach the form and mail it with your check to: Synopsys P.O. Box 310 Beaverton, OR 97075-9962 Cancellations/Refunds A letter of cancellation must be received 7 days prior to the beginning of the conference to qualify for a refund. Travel/Housing Hotel Due to the heavy response, the Red Lion Hotel is full. You can try these alternative locations: San Jose Biltmore 408-988-8411, Santa Clara Marriot 408-988-1500 Air Travel Synopsys has negotiated an American Airlines discount for SNUG attendees. You'll receive 5% off the lowest applicable fare or 10% off if the ticket is purchased 7 days in advance. A Saturday night stay-over will increase your savings. When you or your travel agent makes reservations be sure and give the American Airlines agent the code STAR NBR 02 35 TA to receive the discount. Car Rental Hertz has been appointed the official car rental company for SNUG'95. Special discount rates, with unlimited mileage, start from $35.99 per day to a weekly rate of $131.99 for a sub-compact to a $43.99 per day to a weekly rate of $181.99 for a full-size, 4 door sedan. These rates are guaranteed from March 15th through March 31st. When making reservations use the SNUG ID number, CV#14168, and Hertz will automatically compare the SNUG guaranteed rate to other Hertz published rates to give you the best comparable rate available. For reservations call Hertz at (800) 654-2240. SNUG'95 Registration Form Name:________________________________________________________ Title:_________________________________________________________ Company:_____________________________________________________ Mailing Address:______________________________________________ _____________________________________________________________ Phone:_____________________FAX:______________________________ e-mail Address:_______________________________________________ Credit Card Name:______________Credit Card Number:_____________ Exp. Date:__________Name Printed on Card:_______________________ Signature:____________________________________________________ ____ I prefer vegetarian meals General Registration - $100 Breakout Session Choices (Select one per session plus alternate): 1st Session: first choice #A___, alternate #A____ A1. Design Compiler A2. Design Reuse A3. Design Competition 2nd Session: first choice #B___, alternate #B____ B1. Synthesis B2. Test Synthesis B3. Semiconductor Vendor Sessions B4. EDA Tool Vendor Sessions B5. Design Competition 3rd Session: first choice #C___, alternate #C____ C1. Sub-micron Design C2. Simulation and Verification C3. Semiconductor Vendor Sessions C4. EDA Tool Vendor Sessions C5. Design Competition Wednesday VHDL Based Tutorial Registration - $50 Tutorial Session Choice (Select one): #W____ W1. Speed Optimization Symposium W2. Behavioral Synthesis Using VHDL W3. VHDL Synthesis Techniques and Recommendations W4. Pins-Out Verification Friday Verilog Based Tutorial Registration - $50 Tutorial Session Choice (Select one): #F____ F1. Speed Optimization Symposium F2. Behavioral Synthesis Using Verilog F3. DSP Design Techniques Using COSSAP Registration Total: $__________ All technical sessions and presentations are subject to change prior to conference dates. More Information: If you have questions or need more information on SNUG'95, please call 1-800-388-9125.Article: 845
Check out Altera's 50,000 gate Multi-Chip Module.Article: 846
In article <3jqfis$a4o@vanbc.wimsey.com>, Rob Semenoff <rob@vanbc.wimsey.com> wrote: >Are any trying to get more logic into a package for >less money by interconnecting dies in a package >in some programmable, or customer-specified way ? Altera has a device called the 8050M. it is a multichip module containing 4 FLEX 81188 FPGAs (1188 flip flops, 1008 4-input LUTs) and an Aptix FPIC (FP interconnect chip). i think they claim 50,000 gate capacity. it costs $5k US. i think other vendors are pursuing similar directions. the goal is to get ultra-high-capacity devices that are easy to partition/program. guyArticle: 847
In article <3jqfis$a4o@vanbc.wimsey.com>, rob@vanbc.wimsey.com (Rob Semenoff) says: > >I presume vendors are increasing FPGA density by going >for bigger dies and smaller feature size. > >Are any trying to get more logic into a package for >less money by interconnecting dies in a package >in some programmable, or customer-specified way ? > >-rob I understand that Altera has such kind of Multi-chip module (MCM) for its FLEX8000 series FPGA. But it doesn't seem cheap. In fact, such MCMs are very expensive as fabrication process is not as easy as we think. Futhermore, the logic utilization of MCM is not proportion to the number of chips integrated. Actually the more chips you put in the MCM, the lower the utilization. Since there will not be enough routing resources among the chips, the MCM becomes unroutable as it grows too big. Therefore, MCM is not really practical for high gate count FPGAs. However, if you want to get more logic into a package, it is not necessary to go for MCM. Reduce process technology to 0.5um or less, definitely, the number of logic gates on the same die will be more. Not only you can get more logic, the performance (speed) will also be increased. Also, add more metal layers for routing resources to improve the routing so as the logic utilization. This is very important for FPGA greater than 10K gates, otherwise, the FPGA will become not routable. Simon.Article: 848
lemieux@eecg.toronto.edu (Guy Gerard Lemieux) writes: >Altera has a device called the 8050M. it is a multichip module >containing 4 FLEX 81188 FPGAs (1188 flip flops, 1008 4-input LUTs) >and an Aptix FPIC (FP interconnect chip). i think they claim >50,000 gate capacity. it costs $5k US. Is this really a multichip module? Last time I glanced at the brochure, it looked more like a small motherboard with four normally-packaged '1188's and the Aptix chip soldered on, with some pins on the bottom, probably with all the I/O pad and pin-to-pin delays that implies (especially when passing through the Aptix). I was under the impression than an MCM was built in such a way that interconnect speeds between chips were nearly as fast as intra-chip connects (take the Intel P6 for example). Also, I was at an Altera workshop a few months ago and the spin that they put on this device was that it was mainly intended for fast gate-array prototyping, not production use. -Paul -- Paul Secinaro (Paul.Secinaro@UNH.edu OR pss1@christa.unh.edu) Synthetic Vision and Pattern Analysis Laboratory UNH Dept. of Electrical and Computer EngineeringArticle: 849
I am using VIEWLogic's PROSeries as a front end tool for my FPGA designs. A question came up with regards to command files for silulation with PROsim. I know that command files can also call other command files, but is there a possibility to transfer arguments from one command file to the other? I would like to write 'macros' which simulate micro processor write and read cycles and then call these with the address and data values as aruments from within a 'main' command file. Any help and hints are greatly appreciated. Thanks. Roland
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