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
Michael J. Wirthlin (wirthlim@fpga.ee.byu.edu) wrote: : This newsgroup has already : discussed the reasons why FPGA/PLD vendors do not make the : back-end programming information available. I missed this discussion. Is there an archive somewhere, or could somebody summarize it via email? (wware@world.std.com) Thanks. : Perhaps some of the hardware vendors will be willing to : sell the back end tools at a significantly reduced price (i.e. sell the : bitstream generation program at a nominal cost) in order to maintain : control of the proprietary information, yet allow hobbiests to use the : tools. If they did so (or published enough info to write freeware back-end tools), wouldn't they gain a lot in hardware sales? Would this be offset by their loss in software sales? Would it hurt them by facilitating their competitors' efforts to reverse-engineer their parts? -- ------------------------------------------------------------- Will Ware <wware@world.std.com> web - http://world.std.com/~wware/ PGP fingerprint 45A8 722C D149 10CC F0CF 48FB 93BF 7289Article: 1451
I just got a statement from xilinx that at end of this year they will sell the new series 6200, which is a followup of the algotronix chips, some of you might remember. a sram based fpga with fine granularity, not like the 3000 or 4000 series. with this family they will provide the programming information, this means how to generate the bitstream, because the chips are also partial reconfigurable while running. maybe it would be a good idea, to check this and if it holds true, the use this chips on the low cost board. maybe we could get the alliance vhdl run with this chips, from vhdl down to the bitstream. no more money for inefficent tools !! -------------------------------------------------------- Andreas Kugel Chair of Computer Science V Phone:(49)621-292-5755 University of Mannheim Fax:(49)621-292-5756 A5 D-68131 Mannheim Germany e-mail:kugel@mp-sun1.informatik.uni-mannheim.de --------------------------------------------------------Article: 1452
Thanks for you input. What do you mean by "redundant gates"? Could you elaborate? David J. StarrArticle: 1453
M A N A G E W A R E Create NetWare Loadable Modules and NetWare utilities with an easy, Visual Basic-like language, then compile to generate stand-alone, royalty-free NLM and EXE files that run on clients (DOS EXE) and servers (NLM). call Robin Cyrus @ 602 970 1025 HiTecSoft at your service! -- M A N A G E W A R E THE BEST NLM makerArticle: 1454
In article f4b@senator-bedfellow.MIT.EDU, jsmith@red-branch.MIT.EDU () writes: >How many people out there would be willing to put some time and effort into >a project to develop good solid tools, perhaps something akin the the FSF ( >gnu stuff). I personally would LOVE to see a suite of good solid front end >tools that interfaced to various backends depending on what FPGA/CPLD/etc >that one choose to use. > >I realise getting the programming information for the backends might take >some doing, however.. the front end we can get started on now. I suggest someone summarize this thread, repost it as Low Cost FPGA/CPLD Tools and crosspost on comp.lang.verilog. There are several CAE vendors who have free and/or cheap versions of verilog. Both CAE vendors and FPGA/CPLD vendors presumably would be interested in decent low cost tools. Starting a shareware toolkit, library, utilities etc. somewhere is interesting. Perhaps the kind folks who support the archive could help? Seems like a great university project. >I am more than willing to put some effort into programming and organizing >such a project, if I can get a few other people (at least) willing to spend >some time at it. Sorry, I have good tools in my job and a life outside my job. In order to pull working people into this, you might try coordinating with user groups for CAE tools and FPGA/CPLDs. This might help in general. --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer'sArticle: 1455
In article 2Gu@world.std.com, wware@world.std.com (Will Ware) writes: >Is there an archive somewhere http://www.super.org:8000/FPGA/caf.html COMP.ARCH.FPGA Archive --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer'sArticle: 1456
Intergraph seeks EDA sales person for the Research Triangle Park, NC. Must have 3-5 years experience in EDA, solid communications and presentation skills a must. Phone or fax resume to 603-664-5523 or respond via email to dcui@ingr.com Dan CuiArticle: 1457
Hi. I sympathize with the low cost tool idea. I write FPGA place and route software for QuickLogic for a living. It is possible for one person to write good quality tools including placement, routing, and generating programming information. However, I believe you will find resistance from many of the vendors (not QuickLogic). Altera, Actel, and Xilinx jelously guard details required to implement a netlist in their parts. In truth, there are few vendors other than Altera who really want interoperable tools. Xilinx bought NeoCAD partly to eliminate generic tools. Also, almost half of the work (but only 15% of the code) in FPGA automatic tools are in handling wacky special cases specific to an architecture, so significant work must be done to support each new one. I would be willing to contribute to front-end tool developement. A simple schematic capture, and a simple synthesis language and simulator shouldn't be too hard with a good group of programmers. As for backend tools, what if someone offered a service over the internet to place and route designs for you using vendor supplied tools at a minimal charge? -- Bill -- Bill Cox cox@qlogic.comArticle: 1458
Hi. I'm looking for the Orcad library that comes with AMD's Palasm toolchain. I have an orcad schematic that has been built with this library and i can't display it because i don't have this library.Only the wires show up.Not the symbols since the library is missing. regards Vincent -------------------------------------------------------------- Vincent Himpe ///// Internet : O *) vincent.himpe@ping.be / vi_himpe@mietec.be \__/ Fido : 2:291/1912.8 http://www.ping.be/~ping0751 --------------------------------------------------------------Article: 1459
I'm willing to design a small ASIC (5k) using a PC package. I'm considering either Viewlogic or Integraph's Veribest for schematic & simulation and perhaps some synthesis jobs. I know quite well the Viewlogic tools. Veribest look good, but... Have anyone used Veribest for a REAL ASIC design? What are the pitfalls and advantages? What about sign-off certification? I appreciate any feedback. M. ZalcbergArticle: 1460
I really hope I am not going to regret this........ This message is only posted to comp.arch.fpga. When (if) the issues are resolved, I will post announcements into other appropriate groups. What follows is a request for suggestions for the creation of a cheap (but not free) system for non-corporate users to get access to place and route tools for FPGAs. I have been thinking about providing such a service for about a year and a half, but have finally realized that if I don't tell anyone that I am thinking about this, I will be stuck with just my own ideas on the subject, and clearly, after a year and a half I have not managed to resolve all the issues that I need to resolve to offer such a service. Philosophy: The current situation for hobbyists and students who don't have access to the current state of the art technology is tragic. When I can go to FRY's (an electronics store in the Silicon valley) and buy a floppy disc controller board including cables and brackets for less that I need to pay to buy the floppy disk controller chip on the board, I have no trouble understanding why the home hardware hacker of computer and digital design is a dying breed. Go look in alt.comp.hardware.homebuilt to see what I mean. As an ex-lecturer of comp.sci and elec.eng I know that the students that do 'stuff on the side' as opposed to only course work tend to do far better in the practical work and actually have an understanding of what they are learning, and are much better at applying it. They also tend to be the only ones who can handle the cross discipline work required in integrating hardware with software. Todays state of the art digital technology that is an ideal platform for digital systems design for proffesionals, hobbyists, and students, are the FPGAs. I do not believe that they are too expensive to use, especially the reprogrammable RAM base products from Xilinx. You can order them in one-off quantities from companies like Digi-Key and JDR, for as little as $16.74 (current Digikey catalog, page 83, XC3020-70PC68C). There are two barriers for the above mentioned target audience that I want to serve: The cost of the place and route tools, and the learning curve to use them. The service I would like to offer: I would like to offer a service to the students and hobbyists who want to use these products, but can't afford the place and route tools. As a hobbyist myself, I paid for a set of these tools some time ago and have used them for many projects. I am now a consultant, and still use these tools, and am about to double the number of machines I will be running with these place and route tools. Since they are not running 24 hours a day (sometimes) there are unused place-and- route hours available. Here's what I have in mind: 1) People email me either XNF files or SCH/SYM files 2) I run the appropriate tools to convert and compile the design to a bitstream (.BIT/.RBT) 3) I email the results and reports back to you. 4) You pay me. What help do I need: I need opinions and ideas on all of the following: 1) How do I set this up so that it requires almost zero effort for me to run. I.E. if I have to baby sit each run, then I don't have the time for this. 2) How do the users of the service create XNF files. I don't know of any public domain schematic capture system that runs on PCs that includes Xilinx compatible libraries and netlist generators. 3) How do the users of the service debug their designs when they don't work. 4) How much should I charge for the service $5 ?? $50 ?? or what...??? Should it be per run? Should it be per CPU hour, should it be by device type .... How do I collect. 5) How do I respond to people who complain about the above question ?? 6) When this all falls flat on its face (no interest or no easy way for people to get XNF to me or this all degenerates into a flame war about proprietory software, and how the vendors should just give it away, and couldn't a few of us net critters just get together like the GNU folks and write a better place and route software anyway, and .....) how do I save face. 7) When this becomes wildly successful, and it is too much for me to look after, how do I hand over the service to someone else, or shut it down. 8) How do I set myself up as a clearing house for the PD software that will be needed to generate the XNF. 9) How do I find people who want to help in creating and running this service. 10) Are there any legal things I should be worrying about. 11) Should I ever consider bulk buying of these chips and selling them to users of the service. 12) If things are successful, should there also be an archive of design modules: PCI, UART, VME, PC-AT, SCSI and other things that people want to share. 13) What would be needed to make this TOTALLY automatic, so that I didn't need to do anything to run someone's design. 14) I DO NOT WANT TO DEBUG PEOPLES DESIGNS FOR THEM. After making this clear to all users of the service, how should I deal with people who flame me publically for not helping them with their designs, and charging them for place and route runs that didn't work. 15) Which products should I support: XC3000, XC4000, XC5200, XC6200, XC7000, XC8000, MAX5000, MAX7000, MAX9000, FLEX8000, AT6000, MACH100/200/300/400, QLxxXyy, ACT10xx, ACT12xx, ACT14xx, 22V10 16) Please enter your own question here that you feel others need to think about and answer, and which I should have thought about too. 17) Am I nuts? Why can't someone else compete with me at a cheaper price: Sounds like a great idea to me. All I want is for hobbyists and students to have access to the current state of the art just as I did when I was a lad (oh so many years ago). Once (when...if) I have the service set up, I can't see why I wouldn't want others to also offer the same or better service. I expect that the only proprietory/licensed software that will be involved will be the place and route software. The rest must all be PD, so that the cost of entry into all this for someone is having access to a reasonable PC, buying a chip, downloading the appropriate PD software, and paying me for the 'run'. Others who have also purchased licenses for the place and route tools could offer similar services. Why don't the vendors themselves offer such a service: I'll probably find out the hard way. Why the service is not free: Because I had to buy the computers, the software, the net access, and it's going to require my time. Who is this Philip Freidin anyway: I am an independent consultant, with a little business called Fliptronics. I do not currently work for any FPGA company, although I have in the past. I make my living doing digital systems design, and Windows programming. I am quite knowledgeable about these products. I am still a hw/sw hobbyist. Thankyou for your interest in reading all of this Awaiting your input. praying to avoid a flame-fest Philip Freidin fliptron@netcom.comArticle: 1461
Hi ! Everybody we are now using the Synopsys FPGA compiler and Xilinx XACT 5.1, there is a probelm that we counter. We use the VHDL to describe our design and we use a bidirection port(InOut). The problem is how to insert a bidirection I/O with Synopsys FPGA Compiler. We will appreciate your replay in the tin or mail jean@pds.cis.nctu.edu.tw -- Feng-Chen Chang - (035) 712712-56668 - jean@pds.cis.nctu.edu.tw National Chiao Tung University - Computer and Infomatiom Science. Dept. Hsinchu, Taiwan 30050, Republic of China Parallel and Distribute LaboratoryArticle: 1462
Dave, I am using the XSI interface (Xilinx Synopsys Interface), and for bidirectional ports, I am instantiating the IOBUF (bidirectional buffer) component. Definition: component IOBUF port ( O :out std_logic; I :in std_logic; T :in std_logic; IO :inout std_logic -- signal type INOUT ! ); end component; Instantiation: (If I apply it to your illustration) I1:IOBUF port map (sig_in, sig_out, out_en, data); > > > sig_out |-------| > --------------->| obuft |-----> > |-------| | > out_en | | |-------| > -------->| |<---->| Bipad | "data" > | |-------| > sig_in |-------| | > <---------------| ibuf |<---- > |-------| > Yuce BeserArticle: 1463
"Save Your Computer" Surge sentry computer grade with fax/modem. four outlet surge/noise cube with two RJ-11 jacks for fax/modem line and power protection. 3 stage/line circuitry. 1 picosecond clamp response time. Emi/rfi noise rejection to 35db. UL 1449 rating at 330v. 347 joules. Max surge 11,505 amps. Lifetime warranty and power protection. Putty. 2"h x 4"w x 6.5"d. MFG# ds41psg1. Shipping weight 1.721 lbs. Price: $69.95, shipping and handling included. --------------------------------------------------------- "Repair your Computer" Computer tool kit. By Curtis. Can be used to repair maintain and upgrade virtually any microcomputer system. All tools are demagnetized. Lifetime warranty. Zipper vinyl case. Black 6"h x 9"w x 1.37"d. MFG# tk3ctf 20 piece set (l lb). Price $29.95. Shipping and handling included. MFG# tk4ctf 55 piece set (3 lbs). Price $52.99. Shipping and handling included. ----------------------------------------------------- "Super discount" Voice activated telephone recorder. By TT Systems. One button record, pause, fast forward/rewind, volume control, tape counter, cue and review, LED record light, built-in condenser mic for non-phone recording. AC or battery powered. Uses standard cassettes. Black. 1.5"h x 7"w x 4"d. MFG# vtr500tts. Shipping weight 2.28 lbs. Special price $199.95. Shipping and handling included. Save $100 dollars! Send your check or money order to: Britestar Enterprises 3701 Inglewood Ave Suite #304 Redondo Beach, CA 90278 Please list MFG# with your order. Please allow 7 - 10 days for delivery. ------------------------------------------------------- Money Manager Software "MoneyManager," from Times Publishing does virtually everything Quicken does, and even surpasses it in some respects due to a more powerful language it uses. (IBM compatible only). "MoneyManager" for Windows is the software that allows a person to take control of their personal finances by organizing, tracking and reporting income, expenses, assets, liabilities and investments on a daily basis on his PC. "MoneyManager" prints checks, pays bills and reconciles even the most complex bank statement in seconds. "MoneyManager" tracks your checking, savings, credit card and money management. You can record and categorize your expenses, compare them to budgets and know exactly where your money goes. Keep your investment of stocks, bonds and mutual funds portfolio up-to-date with "MoneyManager." Calculate and analyze your gains, losses and returns on investments and tax consequences in an instant. Thinking of buying a home? "MoneyManager" will create a home affordability analysis for you. Lease-or-buy decisions were never easier. Want to pay off that mortgage faster? It will produce amortization schedules and show you the whopping savings in interest you can realize by making small prepayments of principal. How about income statement, balance sheet, cash flow forecasts, and net worth projections? All of these are available at your fingertips with "MoneyManager." Sale price: $24.99 - Shipping and handling included. 7 - 10 days for delivery. Comes in 3 1/2 or 5 1/4 diskettes, (choose one). Send your check or money order to: Britestar Enterprises 3701 Inglewood Ave Suite #304 Redondo Beach, CA 90278 -------------------------------------------------------- Geographics GEO4000 Moving Message Color Sign, from "Geographics" Multi-color: 16 vibrant colors/hues. Displays maximum 27 characters, minimum 14 characters. 3 fonts, 4 text styles, 10 preprogrammed graphics, 46 European characters. "Magic" interface. Network up to 254 signs. 90-day memory backup. Display area: 3"H x 37.6"W. Indoor use only. Includes wireless remote, mounting brackets and adapter. 5.5"H x 39.6"W x 1.4"D. Shipping weight 11 lbs. MFG# 08386EZL. Sale price $370.50, shipping and handling included. --------------------------------------------------- "Fresh Air" "Bionaire" 2 speed personal air purifier/ionizer. Cleans air in a room 10' x 10' x 8' in 12 minutes. 2-speed, low noise, turbo system. 4 stage filtration system. Effective on tobacco smoke, pollen, dust, bacteria. Ion on/off switch. Flame retarded, injection-molded plastic casing. One year limited warranty. Black. 12.5" x 7.5" x 8"d. MFG# f7obio. Price $94.95, shipping and handling included. -------------------------------------------------------- 18 Karat Gold "Cross" 18 Karat Gold Fountain Pen. 18 Karat Gold Fountain Pen available in four 18 Karat nib widths. Piston converter included. Gift boxed. Comes with Lifetime Mechanical Guarantee. Shipping weight 1 lb. MFG# 8036bcro Broad nib MFG# 8036mcro Med. nib MFG# 8036fcro Fine nib MFG# 8036xcro X-fine nib Price $990.64, Shipping and handling included. Save $279.36. Send your check or money order to : Britestar Enterprises 3701 Inglewood Ave Suite #304 Redondo Beach, CA 90278 Please list MFG# with your order. Please allow 7 - 10 days for delivery. --------------------------------------------------------- Snail Mail Order Form Send your check or money order to: Britestar Enterprises 3701 Inglewood Ave Suite #304 Redondo Beach, CA 90278 Name: ________________________________________________________ Address: ______________________________________________________ City/State/Zip:_______________________________________________ Phone: _______________________________________________________ I've enclosed my check or money order for: "MoneyManager" software for Windows. Diskette Size: 3 1/2 ____ 5 1/4 ____ Surge sentry computer grade with fax/modem MFG# ds41psg1 Price 69.95, shipping and handling included. [ ] Computer took kit. By Curtis. MFG# tk3ctf 20 piece set (l lb) Price $29.95, shipping and handling included. [ ] MFG# tk4ctff 55 piece set (3 lbs). Price $199.95, shipping and handling included. [ ] Voice activated telephone recorder. By TT Systems. MFG# vtr500tts shipping weight 2.28 lbs. Special price $199.95. Shipping and handling included. [ ] GEO 4000 Moving Message Color Sign, from :Geographics" MFG# 08386EZL Sale price $370.50, shipping and handling included [ ] "Bionaire" 2 speed personal air purifier/ionizer MFG# f7obio. Price $94.95, shipping and handling included. [ ] "Cross" 18 Karat Gold Fountain Pen MFG# 8036bcro Broad nib [ ] MFG# 8036mcro Med. nib [ ] MFG# 8036fcro Fine nib [ ] MFG# 8036xcro X-fine nib [ ] Price $990.64. Shipping and handling included. "Never leave home to shop" A 544 page catalog, with over 12,356 name brand items, filled with office/electronic products. Names like Sony, Sharp, Magnavox, Panasonic, JVC, Packard Bell, Brother, Cobra, Toshiba, Pioneer, Casio, and over 216 others. Price $8.00, shipping and handling included. [ ] ................................ . BriteStar, Inc. . 3701 Inglewood Ave . Suite 304 . Redeondo Beach, CA . 90278 ..................................Article: 1464
A little apology. When questions about PREP benchmark winners went up on this newsgroup some days ago I responded with the URL for the PREP homepage. And we got lots of action at prep.org. But not everyone got through to the data. Unfortunately, at the same time the PREP page was being moved to a new site and not all of the pages were available. By Tuesday morning they should all be working fine. I'm sorry for inconveniencing anyone searching for the PREP certified programmable logic benchmark data. By tomorrow you will be able to access the data at "http://www.prep.org". If you have any more troubles getting rhrough or if you have any further comments about the page, please let me know. Stan Baker President, PREP Corp. sbaker@prep.orgArticle: 1465
Our low cost ISA board is comming along. Thanks to everyone who E-mailed me their comments:) So far the design looks like: ISA plug&paly (Fujistu part windows 95 ready) Xilinx 5210 160pqfp EEPROM 32Kx8 Sram Socket Sockets for old Drams Programmable Oscillator (opt) Analog prototype area near backplate Large Digital protoptye area All Xilinx pins broken out To include: Schematic Capture Logic simulation and timing Xilinx place and route tools Target Price $995 Steve Casselman Virtual ComputerArticle: 1466
Great idea. I hope the compilcations don't spoil it. Lack of tools is a terrible problem, one that I'm convinced is blocking what would be a real return of independent amateur hardware design. Would Woz have built the Apple if he'd had to buy $$ tools just to use the chips at all? Unfortunately FPGA vendors do have defensable reasons for keeping their codes proprietary, besides optimization of short-term shareholder value from sales of tools. Many commercial FPGA users would object to the ability to reverse-engineer hardware designs that published configuration codes could enable, for one example. Remote execution of FPGA tools on a pay-per-use server could solve this deadlock. A paper at this year's DAC may be worth a look in this context. A system called "Henry" is proposed, for doing EDA across the Web, with remote execution of tools. Design files in the form of "active documents" traverse the Web between user and tool server, and the tool runs at the remote Web site, paid for per-use. The tool server could be at the tool vendor and/or some third party. There are many more details, issues, and such raised by the paper. *Please* let's not start debating all that here. Just go read the paper. Maybe it's even readable on DAC's Web site, I know a few papers were (http://www.dac.com). I'd check but have no browser handy here. I think what Philip proposes will work fine with ordinary email or ftp or telnet, and should start that way for now. Keep it simple. But I think it's interesting that the Web is prompting some to propose just such a model for all EDA tool usage. Maybe the FPGA tool server will be an early example of Web-based design. The paper is: Mario J. Silva, Randy H. Katz, "The Case for Design Using the World Wide Web", Proc. 32nd ACM Design Automation Conference, June 1995, p. 579-585. (The authors are at UC-Berkeley.) --Mike -- Mike Butts, Portland, Oregon mbutts@netcom.comArticle: 1467
In article <3slvoo$i9l@debbie.cc.nctu.edu.tw>, jean@cis.nctu.edu.tw (Feng-Chen Chang) wrote: > Hi ! Everybody we are now using > the Synopsys FPGA compiler and > Xilinx XACT 5.1, there is a > probelm that we counter. > We use the VHDL to describe > our design and we use a bidirection > port(InOut). > The problem is how to insert a > bidirection I/O with Synopsys FPGA > Compiler. > Did you remember to place the InOut pad into 'Z' when OE = 0? That will do it every time. I have no trouble declaring it as port INOUT with Synopsys. Jason Hou, Qualcomm Inc., yjhou@qualcomm.comArticle: 1468
fliptron@netcom.com (Philip Freidin) writes: [Long article, heavily snipped to save bandwidth] >What follows is a request for suggestions for the creation of >a cheap (but not free) system for non-corporate users to get >access to place and route tools for FPGAs. >Philosophy: > The current situation for hobbyists and students who >don't have access to the current state of the art technology >is tragic. >There are two barriers for the above mentioned target audience >that I want to serve: The cost of the place and route tools, >and the learning curve to use them. > Here's what I have in mind: > 1) People email me either XNF files or SCH/SYM files > 2) I run the appropriate tools to convert and compile > the design to a bitstream (.BIT/.RBT) > 3) I email the results and reports back to you. > 4) You pay me. >What help do I need: > I need opinions and ideas on all of the following: > 1) How do I set this up so that it requires almost > zero effort for me to run. I.E. if I have to > baby sit each run, then I don't have the time > for this. With expert users supplying the XNFs, probably little problem. Could be bad with newbies (no fault of theirs), whose runs may do unexpected things. We all took some time to learn how to drive these things properly. > 2) How do the users of the service create XNF files. > I don't know of any public domain schematic > capture system that runs on PCs that includes > Xilinx compatible libraries and netlist generators. One answer would be to build such a library, that does work on a PD schematic capture. Not a trivial task, to be sure. > 3) How do the users of the service debug their designs > when they don't work. This (IMHO) is the big one. I spend plenty of time running ViewSim on my design, before it ever gets near the hardware. You (as a service bureau) cannot do this: it needs the circuit designer hands-on. No, I don't know of a PD version of ViewSim. > 4) How much should I charge for the service $5 ?? $50 ?? > or what...??? Should it be per run? Should it be > per CPU hour, should it be by device type .... > How do I collect. I would suggest per CPU hour - it's all the same to you what devoce library they are reading from (IMHO). The payments issue could open that famous worm-bucket called "payments via Internet". > 5) How do I respond to people who complain about the > above question ?? If they don't like it/you, they don't have to use same. > 6) When this all falls flat on its face (no interest or > no easy way for people to get XNF to me or this all > degenerates into a flame war about proprietory > software, and how the vendors should just give it > away, and couldn't a few of us net critters just get > together like the GNU folks and write a better place > and route software anyway, and .....) how do I save > face. Hmm, is that really an issue? You have done your best to help people (thereby doubtless creating much good karma :-), if they refuse to use your service, not your problem. I am here assuming that you will get some _positive enjoyment_ out of setting this thing up - if you are doing it as a money-making venture (nothing wrong in that, just different priorities) then "face" is not involved, just dollars. > 7) When this becomes wildly successful, and it is too > much for me to look after, how do I hand over the > service to someone else, or shut it down. You might publish it as shareware - so you still get "royalties", and the workload is spread. > 8) How do I set myself up as a clearing house for the > PD software that will be needed to generate the XNF. You'll need to input plenty if time, I suspect. As to the mechanics, the GNU project people could probably give you some tips. > 9) How do I find people who want to help in creating and > running this service. Ditto. (Your post here is a good start). > 10) Are there any legal things I should be worrying about. I don't know: just read your software license carefully. > 11) Should I ever consider bulk buying of these chips and > selling them to users of the service. Do you really want to compete with the Fry's of this world? I don't see how you would make anything out of this side of it. > 12) If things are successful, should there also be an > archive of design modules: PCI, UART, VME, PC-AT, SCSI > and other things that people want to share. Hopefully, this will evolve naturally. (An excellent idea). > 13) What would be needed to make this TOTALLY automatic, > so that I didn't need to do anything to run someone's > design. You would need to limit your service (in the Xilinx paradigm) to just running XMAKE and XDELAY in a hands-off mode. And make it clear that is all you do. > 14) I DO NOT WANT TO DEBUG PEOPLES DESIGNS FOR THEM. Assuredly. And this is the nub of it: I pity anyone trying to debug a design "blind" in the hardware. For the scheme to work, we do need a PD logic simulator. > 15) Which products should I support: XC3000, XC4000, XC5200, > XC6200, XC7000, XC8000, MAX5000, MAX7000, MAX9000, > FLEX8000, AT6000, MACH100/200/300/400, QLxxXyy, ACT10xx, > ACT12xx, ACT14xx, 22V10 I use XC3100 and XC4000, but that's just my one vote. > 16) Please enter your own question here that you feel > others need to think about and answer, and which I > should have thought about too. > 17) Am I nuts? Not at all: I have often wished a similar thing existed somewhere. >Philip Freidin fliptron@netcom.com -- David R. Brooks <daveb@perth.DIALix.oz.au> Tel/fax. +61 9 434 4280 "Government is not reason. It is not eloquence. It is a force. Like fire, a dangerous servant and a fearful master." - G. WashingtonArticle: 1469
Mike Butts (mbutts@netcom.com) wrote: : A paper at this year's DAC may be worth a look in : this context. A system called "Henry" is proposed, : for doing EDA across the Web, with remote execution : of tools. Design files in the form of "active : documents" traverse the Web between user and tool : server, and the tool runs at the remote Web : site, paid for per-use. The tool server could be : at the tool vendor and/or some third party. The question of how to charge fairly for this service over the net is a perplexing one, but this looks like a great place to start. Here's a related thought: Hand out passwords that act as debit cards. It would work like this. I want to use your service, so before I do, I send you a check and in response you create an account for me, and send me the username and password. I use the name and password to telnet to your tool server, and my account has been assigned some number of CPU hours. When my design has been satisfactorily placed and routed, I can redeem the remaining hours on my account, sell or donate my account to somebody else, or leave the account open if I'm planning more designs in the future. People on limited budgets might pool their resources and share an account, provided they can find some suitable arrangement among themselves to insure an equitable division of CPU hours. If I sell, donate, or share my account, it doesn't add any bookkeeping for the service provider. -- ------------------------------------------------------------- Will Ware <wware@world.std.com> web <http://world.std.com/~wware/> PGP fingerprint 45A8 722C D149 10CC F0CF 48FB 93BF 7289Article: 1470
Here are some more thoughts on my subject (following up my own article :-) 1) My thinking is that I want this to be a service that a hobbyist can use without feeling it is too expensive. My intent is NOT to make money on this, Not to pay for my computer and software, and not to pay for my time. My hope is that it would cost a hobbyist $5 to get a 3020 routed. 2) given the problems that others have highlighted for me, and given that the target audience is probably not highly familiar with FPGAs, I dont think I want to support large-complex designs. How do the following constraints strike you (as examples) A) only 3020, 3030, 3042, 4002/3/4/5/6 B) no more than 70% full C) Only 1 clock net, all FFs must use this net, via a global buf D) No async reset/presets E) No locked pinouts F) Clock rates no faster than (pick a number: 1MHz, 5MHz, 10MHz..) 3) It seems that the PD tools MUST include simulation, and that it should be a pre-requisite before a design is submitted. I certainly thoroughly simulate all my designs before I get near the place-n-route tools. Thanks for your interest. Philip Freidin fliptron@netcom.comArticle: 1471
In article <fliptronDAtLr9.BHw@netcom.com>, fliptron@netcom.com (Philip Freidin) writes: ... |> 2) given the problems that others have highlighted for me, and given that |> the target audience is probably not highly familiar with FPGAs, I dont |> think I want to support large-complex designs. How do the following |> constraints strike you (as examples) I can route for myself, but if I have to use your service, here are my 0.01$: |> A) only 3020, 3030, 3042, 4002/3/4/5/6 That's OK, someone who needs an 4025 is certainly no hobbyist :-) |> B) no more than 70% full The number of used CLBs has no great meaning for routing time. PPR has a very funny behavior if you run several routings on the same(!) design, sometimes it takes twice as long (the resulting .lca file works as fine as the others, but not better...) |> C) Only 1 clock net, all FFs must use this net, via a global buf I can't imagine that FPGAs for hobbyists are just fully synchronuos state machines with no asynchronous tricks. If you started with TLLs and GALs it's difficult to get away from these behaviors, and sometimes you simply need this tricks to get the design running. |> D) No async reset/presets Ditto, if you design an PCI interface, you won't need this, but if you design some 'integrated' glue logic you can't avoid it (sometimes...). |> E) No locked pinouts If you just accept designs without locked pinouts, forget it... If one detects an error in the design running in 'real world' and has to reroute, he is forced to change his PCB again... That's not much fun :-( |> F) Clock rates no faster than (pick a number: 1MHz, 5MHz, 10MHz..) 10-15MHz are low enough to avoid trouble (setup/hold etc.), if you are using the 3000A-series. 3) It seems that the PD tools MUST include simulation, and that it should be a pre-requisite before a design is submitted. I certainly thoroughly simulate all my designs before I get near the place-n-route tools. I don't know, wether an hobbyist can simulate the design without forgeting an important bug. Especially if timing simulation (after the routing) shows that the design won't run properly :-( or if it even works in timing simulation but then you have trouble with the evils of high frequency, clock jitter, signal ringing and anything else that cause so much fun in designing ;-) PS: Use the fastest PC you can get. I've worked til last week with a 486SX/33MHz and PPR needs about 60min to get it routed. Now I have an Pentium/90MHz and it's done in 5-6min! -- Bye Georg Acher +--------------------------------------------------------------+ | Georg Acher, acher@informatik.tu-muenchen.de | | "Oh no, not again !" The bowl of petunias | +--------------------------------------------------------------+Article: 1472
Does anybody know if Lattice Semiconductors has got a WWW page somewhere ? Geir Olav Berg g-o-berg@dicron.co.ukArticle: 1473
Beser, We are wondering where the IOBUF is and how to use it ? Should we instantiate it each time we need or it can be auto instantiated (inserted) by the Synopsys FPGA Compiler ? By the way, Since we use the Synopsys SGE as our interface editor, and if we must instantiate it each time we need, is there any IOBUF.sym that we can use ? Thanks your answer ! Have a nice day. -- Feng-Chen Chang - (035) 712712-56668 - jean@pds.cis.nctu.edu.tw National Chiao Tung University - Computer and Infomatiom Science. Dept. Hsinchu, Taiwan 30050, Republic of China Parallel and Distribute LaboratoryArticle: 1474
Don Husby (husby@fnal.gov) wrote: : Here we see the difference between the way things are really done, and the : imaginary world that marketing would like to see. I have never done a : Xilinx design that didn't require a lot of hand mapping, placing, and : routing. It's nice that they have all these automatic tools that work well : with slow, simple designs... Rather a sweeping generalisation, this. I have never done a Xilinx design which *did* require any hand mapping, placing or routing. This includes 3195As with over 94% of CLBs used and clock speeds of 27 MHz. With the advent of the 'Neocad' place-and-route software the need for hand tweaking should be reduced even further. Richard.
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