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
In article <463h9i$pcl@su102w.ess.harris.com>, tboydsto@su102s.ess.harris.com (Ted Boydston) wrote: >I am getting ready to produce a quick turn ASIC by translating an fpga XNF >netlist to ASIC. One of the vendors we are considering to perform this task is >Chip Express; therefore, I was wondering if anyone has had any experiences they >could relay to me about Chip Express's translation process. > >Specifically: > >* How much time was spent translating your design? >* How many problems did you run into? >* Did you participate in the translation process or did Chip Express do all the work? >* What did they require of from you to perform the translation? > >Since I have personally gone through a translation process, I know that it is riddled >with many pitfalls. This fact makes me wonder if Chip Express's (as well as other >quickturn vendors--like Orbit) brochures oversimplify the issue. > >Any information would be greatly appreciated. > >Thanks :) > I am interested in information on Chip Express as well. If e-mailing a response to this posting, please copy me on the response. We have just completed an FPGA conversion with Orbit Semiconductor. Since they were mentioned, I thought I'd comment briefly. We used VHDL for design entry and mapped a ~2K gate design operating at 2.3 MHz to an Altera FPGA using Synopsys FPGA Compiler. (Yes, I know, not lightening speed or density.) After testing this design in the FPGA, we sent Orbit the following information: 1. EDIF Netlist (Synopsys Output) 2. Print-on-change style simulations from Mentor Graphics Quicksim II 3. Desired Package/Pinout 4. Package Marking Orbit successfully converted this design into their Encore! gate array process with very little interaction from our engineering group. In fact, the level of interaction required was so minimal it bothered us! They completed the conversion and had working parts in our hands in about 8 weeks (just within the upper variance of their commited design schedule). The parts performed 100% to specification. After meeting with their Engineering staff in Sunnyvale last month, I learned that Orbit is much more successful with designs generated from a top-down, synochronous design methodology (as most ASICs are developed), as opposed to schematic-based FPGA designs which tend to be problematic. Orbit's process technology is not the most advanced (1 micron I believe), but they are VERY cost-effective. I would recommend them for small to medium size (up to 45K gates) designs up to moderate speeds. They offer superior value, particularly in low volumes. (We only needed around 2,000 parts, but couldn't use an FPGA for production due to the board layout... we needed a drop-in replacement for an obsolete part.) The nearest ASIC competitor quoted 2.5 times the NRE for this design and wanted a guarantee of 6K parts! I highly recommend Orbit Semiconductor for FPGA conversions... they are very experienced. Hope others find this information useful... ---------------------------------------------------------------------- Nelson W. Willhite phone: (770) 623-7417 AT&T Global Information Solutions FAX: (770) 623-7044 Retail Products & Systems - Atlanta VoicePLUS: 751-7417 2651 Satellite Boulevard Duluth, Georgia 30136 email: Nelson.Willhite@AtlantaGA.ATTGIS.COM ----------------------------------------------------------------------Article: 2151
I would also appreciate that information. Please include me in that response. --- David Hinterberger Eastman Kodak, ASIC Design Group, Digital Technology Center Phone: 716-726-7202 Fax: 716-726-7131 E-Mail: hinter@dtc.kodak.com Mail: 901 Elmgrove Road, Rochester, NY 14653-5211Article: 2152
Hello! I have some problems regarding the XC4025 and I hope someone is able to give me some hints about it. First, the device is very hard to route. Even though it is only filled to 50%, ppr may fail to route it. If it can route it, it takes _hours_ to get is acceptable. My current design fills the device to about one third, and doues route, but I have BIG problems with the timing. To be more specific, the design is a simple signal processor whith four 16-bit data buses, about 12 XBLOX-add_sub, some 16-bit registers and many state machines. I only use one clock (15MHz), but I have an 7.5MHz internally generated clock-enable to about half the registers (total about 900 ff's). I use four timespec statements, from 15M to 15M=66ns, from 15M to 7,5M= 66ns, from 7.5M to 15M=66ns and from 7.5M to 7.5M=133 ns. PPR is able to meet three of the four timespecs, but the one specifying from 15M to 7,5M is increased to 150!! I have tried everything, but nothing helps.. HELP!!Article: 2153
>David Pashley <david@fpga.demon.co.uk> wrote: >" >" >There's a product out there called PLD Pilot, which contains >comprehensive databook info on every PLD and FPGA, as well as >macrocell diagrams etc, and the ability to search by various >criteria. > >I'll try to find contact details... > Check out http://www.reconfig.com/pilot.html on the Web. I've downloaded PLD Pilot demo. It has quite a bit of info. My only caveat is to be wary that it may not include the latest information. -- Steve Knapp Xilinx, Inc.Article: 2154
Hi there, I would like to design some components with my own rounting and these components can be used in VHDL programs. Thus, I have to use Hardmacro. I am using Viewlogic and XACT on Sun workstation. I try the following ways: 1. In Viewlogic Schemtic tool, I can use FMAP (for XC4000). I sucessfully use "expt1076" to convert a schematic file to a VHDL program. But when I want to compile VHDL, it gives me error message. Could anyone point me how to use FMAP in VHDL? 2. I can use XDE to generate a .lca file. My question is how can I make it (the .lca file) a VHDL component so that I can use this "hard macro" component in VHDL programs. Is the above doable in Viewlogic and XACT? Does anyone have better idea? Thank you very much in advance. -John ChengArticle: 2155
I gather FPGAs are used as a nice way of getting around the millions (or too many, at least :) of simple ICs that creep their way into designs. (hello 74ls08 :) However, I've found no simple way that will let me take my design, with glue logic, and implement that in an FPGA such that it's the same design, just with an FPGA instead of many discrete ICs. Someone help me! Also, are there any inexpensive ways to accomplish my task? (I'm a school student - aargh!) ...Duraid MadinaArticle: 2156
A DIRECT WAY TO REACH CHINESE BUSINESS CHINESE INDUSTRIAL & COMMERCIAL GUIDE V95 is a newly-developed software runs on Windows . It is based on the database including 38000 records about main Chinese Industrial & Commercial operations. We have carefully selected these companies, only those who have good business records are qua- lified to join the name list. The database includes companies involved in Agriculture, Forestry, Fishery, Animal Husbandry, Mining, Food, Textile & Clothing, Coal, Power, Petroleum, Steel & Nonferrous Metal, Machinery, Auto- mobile, Locomotive & Rolling Stock, Shipbuilding, Aircraft, Aerospace, Com- puter & Software, Electronic Appliances, Chemical & Household Chemical, Pharmaceutic, Instrument & Meter, Medical Apparatus & Instrument, Tourism & Transport, Postal & Telecommunication Service, Construct, Mass Media, Adver- tisement and other fields. All the information is in English. China is developing its economy at an ultra speed, it's the time now to seek the world's potentially biggest market or to get it's excellent and inexpensive goods. The Guide will surely be a direct way to reach Chinese businesses. The powerful browsing, retrieving and printing function can give you convenience. Each record shows the company's Address, Zip Code, Tel/Fax number, President's Name, Main Products and the SIC (Standard Industrial Code) number. The whole package costs only $1100 (U.S.). This means that each record will cost 2.9 (cent) only and the world-wide EMS or DHL service fee have al- ready been included. Give your order now to get a short-cut expanding your business! Please pay by T/T(Telegraph Transfer) to: Keysoft Co.,ltd. Shanghai THE PEOPLE'S CONSTRUCTION BANK OF CHINA, SHANGHAI BRANCH, XUHUI SUB-BRANCH. Account No.: 5122-20026513158 In order to get the software as soon as possible, please FAX us your name, address, Zip code, phone number and Internet E-mail address if you have, also don't forget to attach the copy of the T/T voucher. We will im- mediately mail the software to you. Thanks & Best Regards! Sincerely Yours: ****************************************************************** Keysoft Co.,ltd., Shanghai, P.R.China Tel: 86-21-64823578 64823579; Fax: 86-21-64823381 E-Mail: keysoft@public.sta.net.cn Add.: 2F, Astronautics Tower, 222 CaoXi Rd., Shanghai 200233 PRC. ******************************************************************Article: 2157
In article <46d214$o46@ping1.ping.be>, Vincent Himpe <Vincent.Himpe@ping.be> wrote: > >Are there any FPGA / EPLD / whatever with low pincounts and lot's of >gates ?. >My deisgns mostly use serial communication and don't need lots of I/O. >So it's pretty silly to put an 132 pin PQFP on your board when you are >only using 12 pins. You might look at ICT's PEEL arrays. The PA7024 (the smallest) has 20 I/Os but 40 registers/latches. It's still not as small as you want since it's in a 24-pin DIP, but it's one of the few small programmable devices I know that doesn't limit the number of latches to the I/O count. ICT's phone is (408) 434-0678. I've never used the PEEL arrays so I don't know how good they are. -- || Dave Van den Bout -- XESS Corp. || || devb@xess.com -- (919) 387-1302 ||Article: 2158
Hi, I need a pld that comes in an 8 pin package and something like 16 ff's and a bunch of colmbinatorial stuff. Pretty much like an iPLD610 or 610 from altera but squeezed into an 8 pin package. Anyone ? PLLC28 is still too large for me. I only have 5 I/O's the rest is buried. Are there any FPGA / EPLD / whatever with low pincounts and lot's of gates ?. My deisgns mostly use serial communication and don't need lots of I/O. So it's pretty silly to put an 132 pin PQFP on your board when you are only using 12 pins. Regards vincentArticle: 2159
David Van den Bout (devb@lys.vnet.net) wrote: : You might look at ICT's PEEL arrays. The PA7024 (the smallest) has : 20 I/Os but 40 registers/latches. It's still not as small as you : want since it's in a 24-pin DIP, but it's one of the few small : programmable devices I know that doesn't limit the number of : latches to the I/O count. ICT's phone is (408) 434-0678. I've never : used the PEEL arrays so I don't know how good they are. Actually their PEEL parts are pretty good; if you can get them when you need them. Since they are fabless, they are at the mercy of their foundries. And since AMI no longer fabs/second sources, you may get stuck with long lead times. -- Gerry Belanger, WA1HOZ wa1hoz@a3bbak.nai.net Newtown, CT g.belanger@ieee.orgArticle: 2160
In article <199510202137.RAA02235@ground.cs.columbia.edu> Fu-Chiung Cheng, cheng@news.cs.columbia.edu writes: > I would like to design some components with my own rounting >and these components can be used in VHDL programs. Thus, I have to >use Hardmacro. I am using Viewlogic and XACT on Sun workstation. > > I try the following ways: > >1. In Viewlogic Schemtic tool, I can use FMAP (for XC4000). I >sucessfully use "expt1076" to convert a schematic file to a VHDL >program. But when I want to compile VHDL, it gives me error message. >Could anyone point me how to use FMAP in VHDL? Perhaps I'm missing something, but in my book FMAPs are not hardmacros. > >2. I can use XDE to generate a .lca file. My question is >how can I make it (the .lca file) a VHDL component so that I can use >this "hard macro" component in VHDL programs. > What ever you want to do, what you need to create is a hierarchical xnf, with the hardmacro (or predefined block or whatever) as a subunit. This tends to imply structural VHDL, but it map also relate to function calls inside a process/procedure. ASYL, which we use, has the notion of a black-box map - where particular operators at least can be mapped to arbitrary xnf. You need something similar. You don't say which version of XACT you use, but they are supposed to be a thing of the past. I've not used their replacements for XACT v5, but we did try something similar with the previous software - it required getting some extra tools for Xilinx to allow the designs from Xde to be made into hardmacros. John _____________________________________________________________ Dr John Forrest Tel: +44-161-200-3315 Dept of Computation Fax: +44-161-200-3321 UMIST E-mail: jf@ap.co.umist.ac.uk MANCHESTER M60 1QD,UK URL: http://www.ap.co.umist.ac.uk/~jfArticle: 2161
This is a multi-part message in MIME format. ---------------------------------110932314213284245191748639348 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii You should try the PCI Special Interest Group (PCI SIG) Home Page: http://www.teleport.com/~pc2/pcisigindex.html There you can subscribe to the Internet mailing list, which serves as a newsgroup for topics around designing PCI devices. Regards Christian Grebe Here are instructions for the mailing list: ---------------------------------110932314213284245191748639348 Content-Transfer-Encoding: 7bit Content-Type: text/plain >From pci-sig-request@znyx.com Tue Sep 26 13:51:05 1995 From: pci-sig-request@znyx.com Date: Tue, 26 Sep 95 05:42:39 -0700 To: grebe@sun2.zfe.siemens.de X-Loop: pci-sig@znyx.com Reply-To: Please.write.a.new.mail.instead.of.replying@FIRST.WORD.archive Content-Id: <rules%pci-sig-request@znyx.com> Subject: archive retrieval: rules Content-Length: 7160 Status: RO X-Lines: 189 File: rules BEGIN------------cut here------------- ============= PCI SIG GROUP Internet Mailing List ===================== Purpose ------- This Internet mailing list is provided to help in the distribution of pertinent information regarding the PCI bus, as specified by the PCI Special Interest Group. The PCI bus has suddenly become one of the most important developments in the computer industry, The topics for discussion may be anything of general interest to members of the PCI-SIG group. The orientation is technical, but may also include non-technical subjects such as marketing if there is an interest. Ideally, we want to promote PCI compatibility between various vendors products. This will benefit everyone in the SIG group. This mailing list is *not* the official information dissemination for the PCI-SIG group. Ideas and information presented on this list are the products of individuals only, and are implicitly provided without warranty. The PCI-SIG documents, such as the specification and updates, that are mailed to SIG members by the PCI-SIG group are the only authoritative source of information on the PCI bus. Another area we wish to avoid is silicon product specific and BIOS implementation issues, such as specifications, bugs, workarounds, etc., unless specific vendors decide to dedicate an FAE to following up on the postings to the list. I expect most vendors will not want to do this, since they have already established mechanisims for dispensing information on their products, and will probably wish to keep closer control over proprietary information. (This is not to say that individual product names will not come up in conversation.) List Owner ---------- This list is being operated by Jochen Roth (jochen@znyx.com), and Alan Deikman (alan@znyx.com) of ZNYX Corporation. If your name server cannot resolve znyx.com, the Internet address assignment is 198.211.96.4. If you are still having difficulty getting service, call Alan Deikman at US 510 249 0800, or FAX at US 510 656 2460. Who May Participate ------------------- This list is open to all parties, although the list owners reserve the right to remove individuals who violate the rules of etiquette, (see below) or whose e-mail addresses generate excessive bounce messages. PCI-SIG membership is not required but is encouraged. Mailing List Addresses ---------------------- The following addresses are used to post subscribe/unsubscribe requests: pci-sig-request@znyx.com for the pic-sig list pci-com-request@znyx.com for the pci-com list pci-bin-request@znyx.com for the pci-biny list pci-all-request@znyx.com to all lists simultaneously The following addresses are the addresses of the mailing list, which are used for posting articles. pci-sig@znyx.com Discussion on PCI spec. and PCI market pci-com@znyx.com PCI Product/Services Commercial Messages pci-bin@znyx.com Binary files of one type or another How To Subscribe ---------------- Send a message to the appropirate -request address(es) listed above. In the "Subject:" line have the word "subscribe" and that is all there is to it. If you do not want binaries or commercial announcements, simply do not send a subscription request to pci-bin-request or pci-com-request. If you send a subscription or unsubscription request to pci-all-request@znyx.com, it will be automatically sent to all three lists. Subscriptions will be accepted starting October 9, 1994. The mail reflector will start reflecting mail on Wednesday, October 13, 1994. How to un-subscribe ------------------- Post a message to the appropriate -request address(es) exactly the same way that you use to subscribe, except use "unsubscribe" in the Subject: line instead of "subscribe." Binary Files ------------ Binary files may be distributed in any of the popular forms by mailing to pci-bin@znyx.com. Remember that large files will be given lower priority by many mail forwarders. Some sites will not be able to receive binary files longer than a certain length. The recommended format is uuencode. Commercial Messages ------------------- A separate list is provided for commerical messages, which may advertise products or services related to PCI. Submissions are sent to pci-com@znyx.com. Since subscribers can shut off receipts of this list, there are no strict guidelines of what may or may not be posted here, although I offer the following items of suggestion/advice: 1. Avoid pre-announcing products. Wait at least until you have your samples ready. Remember, this list is being mailed to your competitors as well as potential customers, and they will use it to figure out how to beat you in the market. 2. Mention suggested retail or low quantity prices only. Legitimate volume purchasers will call you. 3. Refrain from unsubstantiated claims regarding products. For example, statements like "Fastest PCI <blank> board in the world." 4. Provide information on whom to contact for more information. 5. If you are offering consulting services, don't post a full resume. Keep it to a list of areas of expertise. Other than that, happy hunting. PCI-SIG Mailing List Etiquette ------------------------------ The normal Internet rules for etiquette apply. Briefly, this means the following: - No flaming. It is OK to disagree, but do so constructively. If personal insults start appearing, I will have to intercede. Also, slander and libelous slander will be acted upon immediately. - "Proper" language only. I don't think anyone on this list is likely to hide under their desk because they see a four letter word, but I prefer a professional environment and I think most of the people in the PCI-SIG group do as well. - Keep posts short and to the point. If you quote someone from a previous post, don't simply include the entire text, particularly if it is long. Summarize quotes to make your post meaningful and readable. - If English is your primary language, take the time to spell it right. Be tolerant of posters who are doing us a favor by participating when English is not their primary language. - Format posts for no more than 70 characters a line. Some Internet access software tends to send a whole paragraph as a single line. If you see this happen to your post, take the time to figure out how your software works. - Keep signature files under control. Finally, keep in mind that although there are many knowledgable and helpful individuals on this list, they cannot provide free consulting. Many on this list has his or hers own company's products to support as a first priority. There is nothing wrong with posting a question, but there is also no obligation on the part of any one person to provide an answer. Suggestions welcome. Volunteers to maintain a FAQ will be pounced upon. Regards, Alan Deikman %%% EOF %%% END--------------cut here------------- ---------------------------------110932314213284245191748639348--Article: 2162
In article <469jtr$vgg@zipper.zip.com.au>, duraid@zip.com.au (Duraid Madina) wrote: >I gather FPGAs are used as a nice way of getting around the millions (or too >many, at least :) of simple ICs that creep their way into designs. (hello >74ls08 :) However, I've found no simple way that will let me take my design, >with glue logic, and implement that in an FPGA such that it's the same >design, just with an FPGA instead of many discrete ICs. > Generally, FPGAs require synchronous design. Why won't your design (including glue logic) fit into the FPGA? Size problems, or design-style problems (async. vs. sync design). Regards,Article: 2163
Unfortunately, I think you're out of luck. >From what I hear, the 4025 is basically a 4013, with lots more cells, and not lots more routing. So, you're probably going to fill it up pretty quick. I haven't used the latest Xilinx tools; they're supposed to have some sort of floorplanner. You might try using it to force as much of your logic to the outside edges, putting blocks that have lots of internal routing out there. Then, you'll have more room in your center for interblock routing. Of course, you're screwed on timing. But, at least it is more likely to fit. This sounds like a job for ORCA or Altera (here's a great chance for them to come to the rescue of a student - come on, jump in there FPGA salespeople!). I understand they both have some large parts, and they claim to be more routable than Xilinx (though I have no personal experience with the large parts). Let me know if you have any other questions,Article: 2164
unsubscribe -- Sharon OkoliArticle: 2165
In article <469jtr$vgg@zipper.zip.com.au> Duraid Madina, duraid@zip.com.au writes: >I gather FPGAs are used as a nice way of getting around the millions (or too >many, at least :) of simple ICs that creep their way into designs. (hello >74ls08 :) However, I've found no simple way that will let me take my design, >with glue logic, and implement that in an FPGA such that it's the same >design, just with an FPGA instead of many discrete ICs. > > Someone help me! > >Also, are there any inexpensive ways to accomplish my task? (I'm a school >student - aargh!) > Xilinx, and I expect Altera and a few of the others, have equivalent macros so TTL designs can be converted into FPGA. However, only a subset of TTL ICs have equivalents and the timings will be different in general, equivalents to registers are tricky. My suggestion would be to use CPLDs instead. You can get hold of free software for some devices (eg. use old Intel Pldshell for what are now Altera Flexlogic devices, available from Altera). You can get hold of equivalent TTL macros for these too, but I expect it would be easier to restate the logic. You can save yourself some money by making the download cable up yourself - look at the archives of this group to see how. _____________________________________________________________ Dr John Forrest Tel: +44-161-200-3315 Dept of Computation Fax: +44-161-200-3321 UMIST E-mail: jf@ap.co.umist.ac.uk MANCHESTER M60 1QD,UK URL: http://www.ap.co.umist.ac.uk/~jfArticle: 2166
I've been using mach445's for about a year now. Here's some of my experience: A word on programming AMD MACHs with a JTAG cable: Terminate the 4 JTAG lines to the PLD with 470ohm resistors to Vcc. You might want to try to run the return line (TDO) through a 470ohm on the way back to the CPU. As of late '94 this info was not part of the data sheet. Power supply: Use capacitors liberally. Place one 0.1mfd on each side between Vcc and GND. I've had sporadic results programming with the JTAG cable. Maybe my homebrew "programmer" (a ZIF socket for holding an Aries PGA adaptor) is flaky. I have had much better luck programming with the part out of the circuit. Good Luck, -PeteArticle: 2167
This is a multi-part message in MIME format. ---------------------------------4228160711820124532718224029 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Hello Mats, >First, the device is very hard to route. Even though it is only filled to 50%, >ppr may fail to route it. If it can route it, it takes _hours_ to get is >acceptable. This is a known problem with this device. You should try to floorplan your design with 'place instance' and 'place set' statements in the .cst constraints file for ppr. With this statements you can place related logic - using wildcards for instance names - in rectangular areas on the chip, which makes things much easier for ppr. You may consider to buy the Xilinx Floorplanner, a good tool if you want to place parts of the design manually and good for analyzing floorplanning, but there are restrictions though. Mail me if you want to know more about this. Use 'realistic' timing constraints with ppr to dramatically speed up place & route. Otherwise ppr will route forever without meeting your timing. Look at your critical paths and try to figure out why the delay is that big. If there are extensive routing delays, floorplanning may help. If you have question, don't hesitate to mail me. Regards, ---------------------------------4228160711820124532718224029 Content-Transfer-Encoding: 7bit Content-Type: text/plain Christian Grebe Siemens Nixdorf Informationssysteme AG Line of Business High-Performance Printers ASIC's (HLD ST 35) Siemensallee 2 85586 Poing Germany Phone : +49 (8121) 72-4912 Fax : +49 (8121) 72-3173 EMail : grebe@sun2.zfe.siemens.de ---------------------------------4228160711820124532718224029--Article: 2168
So far, I have received many requests for a summary of responses to my original post about Chip Express. Unfortunately, I have received no responses about Chip Express itself. In fact, I have recieved more informaiton about Orbit than Chip Express (1 responses versus 0 responses). I would be willing to cut-n-paste the responses I get and post them, BUT I do not want to become the maintainer of a FAQ for Chip Express. So, please email me only about information concrening Chip Express. That means no "me too" mailings requesting a copy of the responses. I will relay a copy of all the responses I get to the following newsgroups on Monday October 30: comp.arch.fpga comp.cad.synthesis comp.cad.cadence comp.lang.vhdl If I get no responses, I'll post a statement about that too. Thanks :) --- +-------------------------------------------------------------------------+ / TED L. BOYDSTON IV \ + Harris Corporation, GASD * PO Box 94000 * MS 102-4823 * Melbourne, FL 32902 + \ Email: tboydsto@harris.com * Fax: (407) 729-2782 * Voice: (407) 729-7999 / +-------------------------------------------------------------------------+ ====================================================================================== THE ORBIT RESPONSE I RECEIVED ----------------------------- We have just completed an FPGA conversion with Orbit Semiconductor. Since they were mentioned, I thought I'd comment briefly. We used VHDL for design entry and mapped a ~2K gate design operating at 2.3 MHz to an Altera FPGA using Synopsys FPGA Compiler. (Yes, I know, not lightening speed or density.) After testing this design in the FPGA, we sent Orbit the following information: 1. EDIF Netlist (Synopsys Output) 2. Print-on-change style simulations from Mentor Graphics Quicksim II 3. Desired Package/Pinout 4. Package Marking Orbit successfully converted this design into their Encore! gate array process with very little interaction from our engineering group. In fact, the level of interaction required was so minimal it bothered us! They completed the conversion and had working parts in our hands in about 8 weeks (just within the upper variance of their commited design schedule). The parts performed 100% to specification. After meeting with their Engineering staff in Sunnyvale last month, I learned that Orbit is much more successful with designs generated from a top-down, synochronous design methodology (as most ASICs are developed), as opposed to schematic-based FPGA designs which tend to be problematic. Orbit's process technology is not the most advanced (1 micron I believe), but they are VERY cost-effective. I would recommend them for small to medium size (up to 45K gates) designs up to moderate speeds. They offer superior value, particularly in low volumes. (We only needed around 2,000 parts, but couldn't use an FPGA for production due to the board layout... we needed a drop-in replacement for an obsolete part.) The nearest ASIC competitor quoted 2.5 times the NRE for this design and wanted a guarantee of 6K parts! I highly recommend Orbit Semiconductor for FPGA conversions... they are very experienced. Hope others find this information useful... Nelson W. Willhite Nelson.Willhite@AtlantaGA.ATTGIS.COMArticle: 2169
In article vgg@zipper.zip.com.au, duraid@zip.com.au (Duraid Madina) writes: > I gather FPGAs are used as a nice way of getting around the millions (or too > many, at least :) of simple ICs that creep their way into designs. (hello > 74ls08 :) However, I've found no simple way that will let me take my design, > with glue logic, and implement that in an FPGA such that it's the same > design, just with an FPGA instead of many discrete ICs. > > Someone help me! > > Also, are there any inexpensive ways to accomplish my task? (I'm a school > student - aargh!) > > ...Duraid Madina > Many FPGA schematic capture libraries have descrete equivalents so that you can rebuild your schematic with these elements. However, as a student, you should take the long term view. You really ought to stop thinking in terms of discrete ICs, which are quickly becoming obsolete, and think in terms of generic ANDs/ORs, Flip flops, counters, etc. Another way to design is with equations and state machines (Abel, Verilog, VHDL). So use a more employable design methodology, throw your schematic away ;) and design with an FPGA from the start. Put that 74ls08 on your shelf next to the vacuum tubes. -tom -tomArticle: 2170
Vincent.Himpe@ping.be (Vincent Himpe) wrote: >Hi, >I need a pld that comes in an 8 pin package and something like 16 ff's >and a bunch of colmbinatorial stuff. Pretty much like an iPLD610 or >610 from altera but squeezed into an 8 pin package. >Anyone ? PLLC28 is still too large for me. >I only have 5 I/O's the rest is buried. >Are there any FPGA / EPLD / whatever with low pincounts and lot's of >gates ?. >My deisgns mostly use serial communication and don't need lots of I/O. >So it's pretty silly to put an 132 pin PQFP on your board when you are >only using 12 pins. >Regards >vincent I have worked with lots of PLDs and I haven't come across a PLD like that. The smallest PLD I have seen so far is 20pin PLCC, 16v8. But then, I have only used PLDs from major manufactures. Manoj ChaubalArticle: 2171
Hi Xilinx-Users, how can I add TNM attributes to the design without changing the XNF-File before Place&Route with PPR. I use the SYNOPSYS-Design-Flow, therefore I can't add TNM attributes during design entry or within SYNOPSYS. Any ideas or recommendations welcome. Thanks. -------------------------------------------------- Peter Wurbs MAZ Hamburg GmbH pwu@maz-hh.de --------------------------------------------------Article: 2172
fmicale@motown.ge.com (Francesco Micale, X4438) wrote: >You can avoid an invalid bitstream causing any havoc by specifying CRC inclusion >in the bitstream. That will cause the device to abort during programming, and >stay in its quiescent mode. At least it does in XC40xx series devices. >-fm I'm not sure this helps in this case - we are looking at hacking the bitstream to identify the function of various bits. Unless a good CRC is computed for each hacked bitstream (or CRC is disabled), nothing can work, and we can learn nothing. For those with access to it, a safer method might be to use the XACT configuration editor. Say, set up a basic Boolean function in a known CLB location, with all its wiring hand routed. Build a bitstream. Now change one item (location of a route, or the Boolean function itself), and look for the changes in the bitstream. Very tedious, I admit, but hopefully all the CLB sites program identically, so you would only need to attack a limited subset. David R. Brooks <daveb@iinet.net.au> Tel/fax. +61 9 434 4280Article: 2173
We currently have an opening for an electrical engineer with several years of experience in digital logic. The ideal candidate would be familiar with digital video, FPGA development, new DRAM architectures like SDRAM and Rambus, and have an appreciation for the problems associated with high speed design. We value solid skills, innovation, the ability to learn quickly, and flexibility more than we do sheepskins, but this position will require the equivalent of at least a BSEE and several years of solid experience. Good analog skills would be a definite plus. Displaytech manufactures a variety of electro-optic devices based on high speed ferroelectric liquid crystals, including the next generation of high resolution full color miniature displays for head mounted applications, portable computing, and communications. The person hired will work on display controller electronics based largely on FPGAs, CPLDs, and off the shelf devices, but should be prepared for a migration to custom silicon. Displaytech is located in Boulder, CO and is privately held. Although we have been in existence for more than ten years, we are currently in a period of growth more typical of a startup. Please send resumes to resumes@displaytech.com (preferred) or to Human Resources Displaytech, Inc. 2200 Central Ave. Boulder, CO 80301 No phone calls please; I will try to answer email unless I get flooded. -- Rainer M. Malzbender Senior Research Physicist rainer@displaytech.com Displaytech, Inc. 303.449.8933 2200 Central Ave. Boulder, CO 80301Article: 2174
Vincent.Himpe@ping.be (Vincent Himpe) wrote: >I need a pld that comes in an 8 pin package and something like 16 ff's >and a bunch of colmbinatorial stuff. Pretty much like an iPLD610 or >610 from altera but squeezed into an 8 pin package. >Anyone ? PLLC28 is still too large for me. >I only have 5 I/O's the rest is buried. manoj_chaubal@psilongbeach.com (Manoj Chaubal): >The smallest PLD I have seen so far is 20pin PLCC, 16v8. But >then, I have only used PLDs from major manufactures. So what *is* the smallest PLD available today? I have seen a number of fine pitch packages that were tiny, but didn't pay that much attention. --- - Bill Wolf, Raleigh NC - My opinions, NOT my employer's
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