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
YIKES! Today's the deadline to get the early registration rates for this year's SNUG meeting at the Red Lion Hotel in San Jose, California coming up in 5 1/2 weeks (Weds-Thurs-Fri on March 22-24)! (About 325 (+/-) users get together here to discuss Synopsys horror & success stories then. A lot of people also attend the Open Verilog Conference the following Mon-Tues-Weds March 27-29 and get weekend airfare rates.) SNUG rates are: On or Before Feb. 17 After Feb. 17th -------------------- --------------- SNUG registration: $100.00 $150.00 SNUG tutorials: $50.00 per $75.00 per To register for SNUG call (800) 344-SNUG, FAX (503) 690-6906 or e-mail "designinfo@synopsys.com" -- Personal checks, VISA, Mastercard and American Express all accepted (although I personally don't recommend sending credit card info via e-mail. Phone instead.) ---- ---- ---- ---- Synopsys is also doing something they're describing as "Mini-DACs" all over the country the next few weeks. This is a good opportunity to actually see some of their current & new tools at a site near you (and then you can tell me if you thought my review on DesignSource & HDL Advisor was on-the-money or not!) Bellow is the schedule for their Mini-DAC's in the American Southeast: Synopsys American Southeast "Mini-DAC" Schedule ----------------------------------------------- Austin, TX Feb 21: Austin Convention Center (3rd and Red River) Dallas, TX Feb 23: Harvey Bristol Suites (Coit and LBJ) RTP, NC Feb 28: Radisson Governors Inn (NC 54 and I-40) Atlanta, GA March 2: Atlanta Civic and Cultural Center (6400 Sugarloaf Pkwy) Orlando, FL March 7: Embassy Suites (8100 Lake Avenue) Ft. Lauderdale, FL March 9: Sheraton Executive Port Hotel (2440 West Cypress Creek) If you're interested in going to a Synopsys "Mini-DAC" near you, call your local Synopsys FAE real SOON (or you might miss it) and he'll get you in. - John Cooley the ESNUG guy P.S. For you Austin, Texas Synopsys Users: I'll actually be at your Mini-DAC to meet you and the Synopsys field staff there this upcoming Tuesday. I look forward to meeting *ya'll* face-to-face. See! -- I'm already getting my Southern twang ready! :^) ----------------------------------------------------------------------------- __)) "Glass ceilings? Name ANY goat farmer who's made it into management!" /_ oo (_ \ Holliston Poor Farm - John Cooley %// \" Holliston, MA 01746-6222 part time Sheep & Goat Farmer %%% $ jcooley@world.std.com full time contract ASIC & FPGA Designer Detailed SNUG info follows ------- Welcome to SNUG 1995 Synopsys invites you to the Fifth Annual Synopsys Users Group (SNUG=81) 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 5:00 - 7:00 fair 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 4:45 - 5:15 Keynote Speaker 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 Detailed Description: General Session Thursday, March 23 SNUG Keynote Speaker - John Gage of Sun John Gage works for Bill Joy, the Chief Technical Officer of Sun, and is responsible for Sun's reelationships with the world scientific and public policy communities, international scientific institutions and groups developing new forms of scientific research involving computing. He is on scientific and advisory panels of the United States National Science Foundation, the US Congress Office of Technology Assessment, the European Institute of Technology and the United States National Academy of Sciences. He has recently been appointed to the US National Research Council Mathematical Sciences Education Board. He is a member of ACM, IEEE, SIAM, AMS, AAAS, and SMPTE. He attended the Harvard Business School and the Harvard Graduate School of Public Policy. He did doctoral work in economics and mathematics at the University of Berkeley at the same time as Bill Joy. Gage subsequently left Berkeley with Joy start Sun in 1982. Gage is on the Board of Directors of Unicode, an industry consortium of IBM, Microsoft, Apple, Novell, Sun, GO Corporation, and others to provide multilingual capability in all world scripts for all documents and applications. 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. Preliminary topics include: Design Compiler Breakout Synthesis strategy examples will be presented in this session. Design Reuse Breakout This session provides real examples about how to effectively implement design reuse. Design Compiler Breakout Synthesis strategy examples will be presented in this session. Test Synthesis Breakout With the increasing complexity of current ASIC designs, you need to design-in testability from the beginning of your design cycle. This session will focus on design-for-test issues in high level design. Submicron Design Breakout The quality of a synthesized design is only as good as the timing accuracy of the models being used in the synthesis process. This session explores real world techniques for accurately modeling and building a submicron design. Simulation and Verification Breakout This session is intended to explore different verification techniques through different phases of the design cycle. 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. 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. Design Competition Sessions 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 Round Tables Meet members of Synopsys' services staff face to face. Confirmed topics include: 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. The Support Center Summit Meet with members of Support Center, and discuss ways for you to work together more effectively. 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. NOTE: The tutorial "Practical Design Reuse" has been cancelled. 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! Early Registration Drawing! Register early and you become eligible to win a Macintosh laptop! To qualify for the drawing and to receive the discount rate, your registration must be received by February 17, 1995. Conference packets will be mailed to all early registrants. Registrations received after February 17 will not be eligible for the drawing or discount rates and will have their conference packets held at the registration desk for pick-up. General session: Early registration - $100.00 Late registration - $150.00 Tutorials (per session): Early registration - $50.00 each 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. To Register via the WWW: * Open URL http://www.synopsys.com/events/snug/snug_reg_form.html * Fill in the form. * Click on the "Register" button. 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 SNUG attendees receive a special nightly rate of $105/single, $115/double, $125/triple, or $135/quad on rooms at the Red Lion Hotel. Call (408) 453-4000 to make yourself reservations; be sure and identify yourself as part of the Synopsys Users Group to receive this discounted rate. All reservations should be made as early as possible, but no later than March 3, 1995. Reservations received after that date will be on a space and rate availability only. 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. Design Compiler 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: $__________ All technical sessions and presentations are subject to change prior to conference dates. More Information: If you have questions or need more information on SNUG'95, please call 1-800-388-9125.Article: 726
The older "obsolete" Xilinx library had something called phfrcomp, which could be used to build a PLL. I've never tried it, so don't ask me for details. -tomArticle: 727
u8011620@cc.nctu.edu.tw wrote: > Hello, > I would like to know something diffrent among them? I was always cunfused > by them all. > In my previous impression, they are: > PAL: programmable AND, fixed OR > PLD: programmable AND, programmable OR > PLA: ???????????? AND, ???????????? OR > GAL=PLD ?? > Please correct the above, Thanks in advance! PLD = programmable logic device (any architecture) PAL = registered trademark of AMD (by purchase of Monolithic Memories) for their line of PLDs (Programmable Array Logic) GAL = registered trademark of Lattice semiconductor for their line of PLDs (Generic Array Logic) PLA = programmable logic array. I've usually seen this term used in ASIC design, and usually means programmable AND and programmable OR. -- ------- Bryan Butler butler@world.std.comArticle: 728
In article <3i284m$koa@debbie.cc.nctu.edu.tw> u8011620@cc.nctu.edu.tw () writes: >Hello, > >I would like to know something diffrent among them? I was always cunfused >by them all. > >In my previous impression, they are: > >PAL: programmable AND, fixed OR Correct! (PAL = Programmable Array Logic) >PLD: programmable AND, programmable OR PLD is a generic name for two-level (AND-OR) logic devices that *include* PALs, GALs, PLAs. (PLD = Programmable Logic Device) >PLA: ???????????? AND, ???????????? OR PLA: programmable AND, Programmable OR (your definition of PLD). (PLA = Programmable Logic Array) >GAL=PLD ?? GALs are PALs with programmable output functions (you need different PALs for registered or nonregistered outputs for example, but only one type of GAL). GALs where first introduced by Lattice (I think). Also generally GALs are electrically eraseable while PALs are not. (there are exceptions though). > >Please correct the above, Thanks in advance! > > Jason > Hope this helps, Mircea -- Mircea R. Stan | "Without immortality the whole world would UMass, ECE Dept. | be nonsense, all of creation an absurdity." Amherst, MA 01003 | Karl F. GaussArticle: 729
In Article <3i355p$a2m@quidnunc.qcktrn.com> biggs@qcktrn.com ( Tom Biggs ) writes: >The older "obsolete" Xilinx library had something called phfrcomp, which >could be used to build a PLL. I've never tried it, so don't ask me for details. > > > -tom > Absolutely you can. I did one in an Altera MAX 7000 a few years ago. There's no reason it couldn't be done in one of the FPGAs as well. The biggest limitation will be that your operating range and resolution may be limited by the FPGA's maximum clock rate in some applications. PHFRCOMP has been obsolete for a while, in fact I don't believe the latest Xilinx SW (V5.0) even supports it. I personally never used it, so I can't attest to how well it works. No matter though, you can just as easily build your own. As it is, many of the larger Xilinx macros can be done better with a little thought. Rolling your own out of the simpler macros and primitives also gives you a fall back position when the next rev calls your favorite macro obsolete. -Ray Andraka Chairman, the Andraka Consulting Group 401/884-7930 FAX 401/884-7950 email randraka@ids.net The Andraka Consulting Group is a digital hardware design firm specializing in obtaining the maximum performance from FPGAs. Services include complete design, development, simulation and integration of these devices and the surrounding circuits. We also evaluate, troubleshoot, and improve existing designs. Please call or write for a brochure.Article: 730
In article <D44sFL.G2t@oasis.icl.co.uk>, trev@ss11.wg.icl.co.uk writes... }cking@accutron.ie writes :- }> 1. The FPGA has 5000 internal gates. }> 2. It can run at 250Mhz internaly } ^^^^^^ }250MHz ! Are you sure about this, sounds like a toggle frequency to me, in which case }would I not expect your design to run at this speed. } }Please correct me if I am wrong and let me know who is making such a device. } }Cheers, }T.H. (trev@wg.icl.co.uk) It wouldn't surprise me at all if such a thing exists. Heck, DEC makes entire CPUs that run at 300MHz (the DECchip 21164). --- CarlArticle: 731
Here's a way to generate fractals with video hardware in real time. First, create a video feedback loop. Do this by aiming the video camera at the monitor. The monitor and camera settings are important, try the following general setup: monitor brightness all the way down, or at least very low (not completely black, obviously). Zoom the camera in so that the *image* of the monitor on the monitor is about the same size as the monitor (in other words, aspect ratio close to 1:1). Also, some rotation is desirable, else you just get a blob that is hard to stabilize. Hint: turn the monitor upside down, or put it on its side - this is much easier than trying to mount a camera upside down on a tripod! Oh yeah, tripods are good. At this point you should be able to create lots of swirling, spinning stuff. Try messing with the color, (B&W is usually easier to start with), phase, focus, zoom, iris, etc. If the camera and or monitor has automatic features, turn them all off, you get better control that way. The patterns that you see can exhibit spatiotemporal oscillation and chaos, and can be quite hypnotic. But there is usually a single fixed point on screen (that is either stable or unstable, depending on the zoom). To get more complicated patterns like fractals, you need to mix in another transform. The simplest way to do this is with a mirror. The setup should look like this: monitor | video camera | ___ +-->| >___|---+ | | ----------------- | | | mirror | | | +------------------------------------+ The mirror is positioned so that the camera sees *both* the monitor screen directly, and a reflected view of it. Now you have two transforms of the image to play with: the non-linear transform provided by the electronics of the camera+monitor, and the simple reflection of the mirror. With the setup above I have been able to create colorful, pulsating, fractal ferns, and some 'jellyfish' like shapes. Quite different from the usual video feedback fare, because of the break from circular symmetry. Almost like life-forms. Another way to do this (that I have not tried yet) is with a video mixer. Point two cameras at the screen, mix their output together and display it on the screen. With different rotation angles and zoom setting it should be possible to create lots of cool visual effects. Unfortunately, I do not have access to a video mixer, so I can't try this myself. I have tried to do it with the poor man's video mixer, a half-silvered mirror. I've been able to get some interesting patterns, but there are some serious constraints on this method. (Anybody know how I can get (or build) a cheap video mixer? Doesn't have to be fancy.) Video feedback is cool in general, and I encourage anybody with a camcorder to point it at the monitor screen! (Remember to rotate it, and zoom in to close to 1:1.) It is a great way to demonstrate nonlinear, self-organizing pattern formation. I have had people look at the screen and say "where does the pattern come from?" This is a great excuse to start telling them all about dissipative structures, reaction-diffusion systems, etc. etc.. Once I even got a pattern of spiral waves that looks a little like the BZ reaction. The WWW page with framegrabbed images is, alas, still "currently under construction" :-( Tom Holroyd Program in Complex Systems and the Brain Sciences The basis of Florida Atlantic University, Boca Raton, FL 33431 USA stability is tomh@bambi.ccs.fau.edu instability. The 9th Amendment of the U.S. Constitution: "The enumeration in the Constitution, of certain rights, shall not be construed to deny or disparage others retained by the people."Article: 732
Is there anybody who have written verilog models(behavioral or structural) for Xilinx LCAs or CLBs?Article: 733
carl@gergo.tamu.edu (Carl Perkins) writes :- In article <D44sFL.G2t@oasis.icl.co.uk>, trev@ss11.wg.icl.co.uk writes... >}cking@accutron.ie writes :- >}> 1. The FPGA has 5000 internal gates. >}> 2. It can run at 250Mhz internaly >} ^^^^^^ >}250MHz ! Are you sure about this, sounds like a toggle frequency to me, in which case >}would I not expect your design to run at this speed. >} >}Please correct me if I am wrong and let me know who is making such a device. >} >}Cheers, >}T.H. (trev@wg.icl.co.uk) > >It wouldn't surprise me at all if such a thing exists. Heck, DEC makes entire >CPUs that run at 300MHz (the DECchip 21164). > >--- Carl Not in a FPGA though (yet) . Cheers, T.H. (trev@wg.icl.co.uk)Article: 734
In article <3i0atg$t6l@jabba.ess.harris.com> dlanza@witch.ess.harris.com (David Lanza) writes: >In <3htibu$4vv@barnacle.iol.ie> coffey@iol.ie (Aedan Coffey) writes: > >>M. Movahedin (movahed@tumlis.lis.e-technik.tu-muenchen.de) wrote: >>: Hello, > >>: I have synthecized some combinational designs with 5 input and only 1 output >>: with FPGA Compiler from Synopsys, but they are made with more than one CLB in >>: Xilinx. Definitly, for a 5 to 1 combinational logic, only one CLB (XC4000) is >>: enough. >>: What is happening? >>i >If your using Xilinx PPR tool and the FPGA is no full, the router may partition >the design into multiple CLBs although it could fit into one. Check the >report file for occupied CLBs and packed CLBs. Occupied CLBs are those >containing some logic. Packed CLBs are the number of CLBs that the design >could be sqeezed into. The ppr statistics are badly broken, and I filed a bug report on that -- packed CLBs reported 1 packed CLB for 3 FG generators, for example, had 16 RAM16x1 in it and reported "0 packed CLBs" etc... Basically, under current circumstances you can forget all statistics in prr except the "Occupied CLBs" -- and those are not useful for any comparison, as ppr might just decide to partition CLBs if the chip is not full. I wish there were more serious statistics available, which really listed all the data... Packed CLBs is not of much use either, it's computed as follows: max (FFs, FGs) / 2 Which doesn't really take into account stuff like how to partition the design (there might be constraints preventing 2 FFs and 2 FGs in one CLB, other than routability, I think)... It also ignores H-generators, RAM cells, etc... Your best bet is to check the data from the design compiler and look what it has generated... Check out what goes in each cell and try to find what has happened. I've seen ppr optimize away 100's of gates, so the Synopsys FPGA compiler seems to be less than optimal, but all further tools do not generate reliable statistics... Hope this helps, m. -- Michael Gschwind, Institut f. Technische Informatik, TU Wien snail: Treitlstrasse 3-182-2 || A-1040 Wien || Austria email: mike@vlsivie.tuwien.ac.at PGP key available via www (or email) www : URL:http://www.vlsivie.tuwien.ac.at/mike/mike.html phone: +(43)(1)58801 8156 fax: +(43)(1)586 9697 Boycott Whaling!!! Boycott Norway!!! Boycott Norwegian Products!!!Article: 735
> Re: PLA? PAL? PLD? GAL? Calling Dave Greer or Stan Baker! If you're out there, could you post the letter Dave wrote to EE Times a few years back on naming PALs/GALs/CPLDs/FPGAs. A second suggestion is to try DataIO for a primer. I have a second edition copy of "Programmable Logic - A Basic Guide for the Designer" with a 1986 copyright on it. I don't know if they still print this. Finally, try books. "Digital Systems Design with Programmable Logic" by Martin Bolton is one. There are also plenty of others. You could also try full-line programmable vendors for a primer, but they probably emphasize their own products. --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer'sArticle: 736
The following is just one of many trivial questions posted here. In article <3i02ch$d85@msunews.cl.msu.edu> jimenez2@oscar.egr.msu.edu (Manuel Alejandro Jimenez-Cede) writes: >I've been looking for technical info about >fpgas (electrical characteristics, timing, pwr diss., packaging, etc.).Does anybody can suggest me where to find the right databooks? >Or at least a way to get info from manufacturers? > Calling all vendors! *Please* post contact information for these people! 800#s, literature request lines, application support numbers, web URLs, list server info, BBs, etc. Step 2 - Can someone collect these on a web site? --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer'sArticle: 737
cking@accutron.ie writes (heavily trimmed!): ... <lotsa snipping> >However, I was wondering if it is possable to use the FPGA to >generate a fractal image, in *real time*, to display via the >RAMDAC. I know that such a task is maths intensive, but: > 1. The FPGA has 5000 internal gates. > 2. It can run at 250Mhz internaly ^^^^^^ | Stop right there! A. In order to run at 250 MHz in CMOS, interconnect needs to be *very* short (no capacitance allowed!). In custom CMOS (e.g. DEC Alpha, Gennum GenLinx) this is possible with lots of effort. In FPGAs, this is very, very unrealistic (today). B. You may very well be able to handcraft small data paths to run at (maybe) up to 100 MHz. Maybe. But you won't be able to build big data paths at that cycle rate because of the power required to run at those rates. FPGA vendors guarantee their AC specs to either 70 or 85 deg junction temp. At 100 MHz, with available packaging, you will probably hit 85 deg junction at less than 1K gates, given typical operating environments, design practices, packaging, etc. CMOS slows down as it warms up. If you want 250 MHz, you're talking custom layout and/or exotic packaging and cooling. Almost all the FPGA packages currently offered are cavity-up configurations, so even heat sinks won't help you. If you have the time/money to buy dice and do custom packaging, then you'll want to do custom layout as well. This isn't trivial, even at 0.5u geometries. But it *is doable*, and it is being done. This just isn't a casual problem that can tolerate off-the-shelf design components. If you want to pursue custom CMOS, drop me a line. I know a couple of designers that can pull it off (250MHz) reliably. Bob Elkind, Tek TV Products (FPGA and CMOS/Bipolar ASIC designer) bobe@tv.tv.tek.comArticle: 738
In article <D46A3J.80H@world.std.com> butler@world.std.com (Bryan Butler) writes: >u8011620@cc.nctu.edu.tw wrote: >> Hello, > >> I would like to know something diffrent among them? I was always cunfused >> by them all. > >> In my previous impression, they are: > >> PAL: programmable AND, fixed OR >> PLD: programmable AND, programmable OR >> PLA: ???????????? AND, ???????????? OR >> GAL=PLD ?? > >> Please correct the above, Thanks in advance! > >PLD = programmable logic device (any architecture) >PAL = registered trademark of AMD (by purchase of Monolithic Memories) >for their line of PLDs (Programmable Array Logic) >GAL = registered trademark of Lattice semiconductor for their line >of PLDs (Generic Array Logic) >PLA = programmable logic array. I've usually seen this term used in >ASIC design, and usually means programmable AND and programmable OR. > That's what I've thought, but in his book FPGA Workout, 1994, David E. Van den Bout, Ph.D. writes (and I paraphrase): PLA: Developed first and consist of programmable AND and OR arrays. PAL: Developed next and consist of programmable AND and fixed OR arrays. PLD: A PAL with flip-flops added. CPLD: Complex PLD (several PLDs combined into a single IC). He does not acknowledge PAL as being a registered trademark of AMD. Also, he does not mention GALs.Article: 739
In article <3iaqd0$nuh@noao.edu>, Bill Hull <bhull@sunspot.noao.edu> wrote: >In article <D46A3J.80H@world.std.com> butler@world.std.com (Bryan Butler) writes: >>u8011620@cc.nctu.edu.tw wrote: >>> Hello, >> >>> I would like to know something diffrent among them? I was always cunfused >>> by them all. >> >>> In my previous impression, they are: >> >>> PAL: programmable AND, fixed OR >>> PLD: programmable AND, programmable OR >>> PLA: ???????????? AND, ???????????? OR >>> GAL=PLD ?? >> >>> Please correct the above, Thanks in advance! >> >>PLD = programmable logic device (any architecture) >>PAL = registered trademark of AMD (by purchase of Monolithic Memories) >>for their line of PLDs (Programmable Array Logic) >>GAL = registered trademark of Lattice semiconductor for their line >>of PLDs (Generic Array Logic) >>PLA = programmable logic array. I've usually seen this term used in >>ASIC design, and usually means programmable AND and programmable OR. >> > >That's what I've thought, but in his book FPGA Workout, 1994, David E. >Van den Bout, Ph.D. writes (and I paraphrase): > >PLA: Developed first and consist of programmable AND and OR arrays. >PAL: Developed next and consist of programmable AND and fixed OR arrays. >PLD: A PAL with flip-flops added. >CPLD: Complex PLD (several PLDs combined into a single IC). > >He does not acknowledge PAL as being a registered trademark of AMD. >Also, he does not mention GALs. > Well, sure enough PAL is a registered trademark of AMD as can be seen by the trademark they put by it on all their databooks. I have most often seen it used when refering to PLAs with fixed OR connections. Then the term PLD became more prevalent when PAL macrocells were pumped up with flip-flops and extra doo-dads for inverting SOP terms, handling the flip-flop control lines and clocking, and all types of feedback. As for GALs, I've never used them and I thought it was a marketing name for a particular brand of PLD (which looks to be the case). -- || Dave Van den Bout || || Xess Corporation ||Article: 740
XESS Corp. has just released the first chapter of "FPGA Workout II". This chapter covers the EPX780 FPGA architecture and timing. It's a hypertext document that will execute on a DOS machine with a VGA display. If interested, you can retrieve this file via anonymous FTP from ftp.vnet.net in directory pub/xess/hyperdoc. Get the files epxdata.exe and install.txt. -- || Dave Van den Bout || || Xess Corporation ||Article: 741
ATT and Xilinx recently announced (independently) that each company guarantees their AC specifications to 85 degrees C junction temp. Given that a typical 208-pin plastic quad flatpack has a thermal resistance of >30 degrees C per watt in still air, and that a typical ambient air temp for commercial grade applications is 55-70 degrees C, an FPGA design would be limited to 0.5 - 1 watt or less of power before the vendor's AC specs would have to be degraded. Is this interesting to anyone out there? Has anyone found their applications to be constrained to XX number of gates simply because they were running into power (heat) limits? Note that ATT and Xilinx databooks have historically been published with only "ambient" temperature limits at which their AC specs were guaranteed (70 degrees ambient in each of these cases), even though the fundamental limit was junction temperature. It was left to the designers to more or less guess how much heat they could push through a particular package before they reached some unspecified thermal bound at which the AC characteristics would put their design at risk. Both ATT and Xilinx are to be commended for (finally) stating their performance characteristics in unambiguous terms. Does anyone have equivalent commitments from other FPGA vendors? For what its worth, I believe that ATT has always guaranteed their devices to 85 deg junction (they just didn't put it in writing in the data sheet, for some inexplicable reason). I believe Xilinx had (until this announcement) guaranteed their devices to 70 degrees junction, and so the parts tested under the new spec will now be around 5% faster than they had been, a "free" performance enhancement/upgrade. Anyone from ATT, Xilinx, Altera, Motorola, Actel, etc. etc. care to comment? A few weeks ago I posted a questionnaire for FPGA designers concerned about thermal properties, etc. I got absolutely *zero* responses. Am I the only FPGA designer in the world running into power/thermal limits (as opposed to gate count or pin count)? If this is the case, then it will be very difficult for me to arouse interest from FPGA vendors in solving my petty little problems. If I'm *not* the only "high-power" designer out there in the not-so-cold-cold world, then I believe it is in our mutual best interest to share ideas, info about managing power and heat in these little critters while still getting some decent speed out of them, hopefully at a reasonable cost. When's the last time you asked your FPGA vendor for a power estimation tool? Bob Elkind, Tektronix TV Products bobe@tv.tv.tek.comArticle: 742
Has anyone used Cadence FPGA Designer? I read the EE Times article and have general information. User comments would be helpful. Comparisons with Synopsys and/or Exemplar would be interesting. --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer'sArticle: 743
>>PLD = programmable logic device (any architecture) > >That's what I've thought, but in his book FPGA Workout, 1994, David E. >Van den Bout, Ph.D. writes (and I paraphrase): > >PLD: A PAL with flip-flops added. It should be noted that the terminology/taxonomy is not completely agreed upon between everybody. Here at UofT (and elsewhere), we are defining the term as follows: PLD = programmable logic device -- includes CPLDs, FPGAs, but not ROM or RAM, can be one-time or reprogrammable. The general impression is that PLD shall be an umbrella term to categorize all types of user-programmable logic. CPLD, FPGA, PAL, PLA, GAL are all types of PLDs. Guy Lemieux University of TorontoArticle: 744
In <1995Feb21.210949.27248@jarvis.cs.toronto.edu> lemieux@eecg.toronto.edu (Guy Gerard Lemieux) writes: > >>>PLD = programmable logic device (any architecture) >> >>That's what I've thought, but in his book FPGA Workout, 1994, David E. >>Van den Bout, Ph.D. writes (and I paraphrase): >> >>PLD: A PAL with flip-flops added. > >It should be noted that the terminology/taxonomy is not completely >agreed upon between everybody. Here at UofT (and elsewhere), we >are defining the term as follows: > >PLD = programmable logic device -- includes CPLDs, FPGAs, but not > ROM or RAM, can be one-time or reprogrammable. > >The general impression is that PLD shall be an umbrella term to >categorize all types of user-programmable logic. CPLD, FPGA, PAL, >PLA, GAL are all types of PLDs. > >Guy Lemieux >University of Toronto > The 'commercial' programmable lgic industry also has problems with these definitions. The difference between FPGAs and PLDs/CPLDs is most often mis-represented. PLDs and CPLDs have FIXED routing resources and very (?) predictable timings accordingly. In an FPGA the delay is less predictable (whilst in most cases it can be accurately 'estimated' by static timing analysis tools) as it is a function of the route a net takes through the array. Post-Layout timing in an FPGA will be different than Pre-Layout and because of this, FPGAs have much more in common with ASICs than PLDs from an end user design perspective (both logic and metodology). To classify PLDs and FPGAs together, in anything other than the broadest sense of 'programable' logic, is demeaning to the added flexability that FPGAs offer the designer. Many PLD/CPLD vendors (mis-)market their products as FPGAs to appear to be in both markets (and to raise their ASP) - buyer beware!! Martin MasonArticle: 745
In article <3id386$hfb@elvis.vnet.net> devb@elvis.vnet.net (David Van den Bout) writes: >Well, sure enough PAL is a registered trademark of AMD as can be seen >by the trademark they put by it on all their databooks. I have most As someone else mentioned, MMI invented PALs and trademarked the name. This became AMD's property when they bought MMI. I was there at the time, in the sense of being an AMD employee. >often seen it used when refering to PLAs with fixed OR connections. >Then the term PLD became more prevalent when PAL macrocells were pumped >up with flip-flops and extra doo-dads for inverting SOP terms, handling >the flip-flop control lines and clocking, and all types of feedback. >As for GALs, I've never used them and I thought it was a marketing name >for a particular brand of PLD (which looks to be the case). GAL refers to Lattice's idea for Generic Array Logic. The original PALs were 16R8/16R6/16R4/16L8. Lattice came up with the idea for a single part, 16V8, which not only could replace any of these 4, but could do so with the original fuse map plus a few extra configuration bits. These bits were so easy to calculate that most programmers can do it for you. This allowed the use of GALs even with "legacy" JEDEC files where the source is unavailable. At the same time Lattice introduced another good idea, electrically erasable parts. Some people actually think that GAL means EE, but these are two orthogonal concepts. GAL is an architecture, EE is a device technology. You could have non-EE GALs (although I have never seen any) and EE PALs (AMD has EE 22V10s, as do many others). -- Question Authority, but never shoot back.Article: 746
****************************************************************************** * David L. Carroll Director of Communications * * International Telecommunications Center (ITC) "The only server on the * * Internet dedicated exclusively to telecommunications" * * PHONE:914-353-0311 FAX:914-353-0335 E-mail:dlc@telematrix.com * ****************************************************************************** February 23, 1995 Dear Sirs: This is to inform you that International Telecom Center (ITC), a storefront server devoted exclusively to telecommunications, datacommunications, networking and computer communications has set up a jobs and positions wanted bulletin board. For the next two weeks, starting February 23th, persons searching jobs (and companies seeking to fill positions) are invited to enter their listings on this board free of cost Access ITC at http://www.telematrix.com, then click on the Information Center icon Employment Agency and fill out the form as directed. ITC's server is currently looked at by between 30,000 to 40,000 telecom/datacom/computer professionals every month; more than half these visitors are from Fortune 500 companies. This is, in other words, an excellent opportunity to showcase your stuff for free of cost. But do it soon cause this opportunity ends soon. Good hunting, David Carroll Director of Communications, ITCArticle: 747
Can anyone please tell me if there is a newsgroup available which is dedicated to the topics of micro controllers in general and Motorola's MC683xx family in particular? Many thanks for your help. RolandArticle: 748
Phil Ngai (pngai@mv.us.adobe.com) wrote: > In article <3id386$hfb@elvis.vnet.net> devb@elvis.vnet.net (David Van den Bout) writes: > >Well, sure enough PAL is a registered trademark of AMD as can be seen > >by the trademark they put by it on all their databooks. I have most > As someone else mentioned, MMI invented PALs and trademarked the name. > This became AMD's property when they bought MMI. I was there at the > time, in the sense of being an AMD employee. > >often seen it used when refering to PLAs with fixed OR connections. > >Then the term PLD became more prevalent when PAL macrocells were pumped > >up with flip-flops and extra doo-dads for inverting SOP terms, handling > >the flip-flop control lines and clocking, and all types of feedback. > >As for GALs, I've never used them and I thought it was a marketing name > >for a particular brand of PLD (which looks to be the case). > GAL refers to Lattice's idea for Generic Array Logic. The original PALs > were 16R8/16R6/16R4/16L8. Lattice came up with the idea for a single > part, 16V8, which not only could replace any of these 4, but could do > so with the original fuse map plus a few extra configuration bits. > These bits were so easy to calculate that most programmers can do it > for you. This allowed the use of GALs even with "legacy" JEDEC files > where the source is unavailable. > At the same time Lattice introduced another good idea, electrically > erasable parts. Some people actually think that GAL means EE, but these > are two orthogonal concepts. GAL is an architecture, EE is a device > technology. You could have non-EE GALs (although I have never seen any) Some of the first 22V10s I used from Lattice and VTI were non-eraseable. > and EE PALs (AMD has EE 22V10s, as do many others). > -- > Question Authority, but never shoot back. -- ------- Bryan Butler butler@world.std.comArticle: 749
Tom Holroyd (tomh@bambi.ccs.fau.edu) wrote: : Here's a way to generate fractals with video hardware in real time. : First, create a video feedback loop. Do this by aiming the video camera at : the monitor. The monitor and camera settings are important, try the : following general setup: monitor brightness all the way down, or at least : very low (not completely black, obviously). Zoom the camera in so that the : *image* of the monitor on the monitor is about the same size as the monitor : (in other words, aspect ratio close to 1:1). Also, some rotation is : desirable, else you just get a blob that is hard to stabilize. Hint: : turn the monitor upside down, or put it on its side - this is much easier : than trying to mount a camera upside down on a tripod! Oh yeah, tripods are : good. This is also described at the beginning of Peitgen, Jurgens and Saupe's ``Chaos and Fractals: New Frontiers of Science'' (Springer-Verlag, 1992). : Another way to do this (that I have not tried yet) is with a video mixer. : Point two cameras at the screen, mix their output together and display : it on the screen. With different rotation angles and zoom setting it : should be possible to create lots of cool visual effects. With a single monitor, several cameras and a mixer you should be able to produce the attractor of an iterated function system of similtudes. With several monitors, several cameras and a mixer you get the attractor of a recurrent iterated function system of similtudes. -John Hart --- John C. Hart, Asst. Prof. School of EECS, Wash. St. Univ. Pullman, WA 99164-2752 (509)335-2343 fax:(509)335-3818
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