Site Home Archive Home FAQ Home How to search the Archive How to Navigate the Archive
Compare FPGA features and resources
Threads starting:
Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
I have an old (out of maintenance) Viewlogic schematic based Xilinx system, and I have the choice to either bring it back up to date (I won't bother with Viewlogic VHDL after reading the 100% negative comments on this list!) or else switching to the new Foundation stuff. I've been looking at the VHDL vs schematics thread and not seen any comments on the Xilinx Foundation s/w in general and its VHDL in particular. Maybe I missed it, but it seems that people use either Viewlogic (schematic or VHDL) or Synopsis. Does anyone use Foundation? Is it any good? Any advice would be welcome. I definitely want to keep schematic entry, but I'd like to add VHDL to try it (and maybe switch to VHDL for designs where it has advantages) and it seems that I'd need Foundation to do it. thanks for your posts (or emails). Ian. -- Ian Lazarus Nuclear Physics Support Group, CCLRC, Daresbury Laboratory email: I.H.Lazarus@dl.ac.ukArticle: 4676
A 45 day evaluation licence for SETANTA ED, a VHDL aware editor is available at : http://www.vhdl.com.au Features include: + Keyword colorising + no file length limit + Automatic Testbench Generation (Beta version) + Automatic model insertion + Dozens of free VHDL models including counters, adders + Support for C/C++, JAVA, Perl, and many more languages Try it out, lets us know what you think. -John Maher Technical Director VHDL System Solutions P/L email: jmaher@vhdl.com.auArticle: 4677
Hi all, I have come across a lot of examples of processor architecture using reconfigurable FPGAs. But has anyone heard of reconfigurable FPGAs getting used in switching and routing applications? Links to stuff on the web and other info is welcome. Thanking in advance, MakarandArticle: 4678
Thanks for taking the time to read this message! This is about a lady that gave all she could to everyone, including me. Her name is Wendy. When I was young she became my foster mother after my real mother had abandon me. Never having any kids of her own she became a foster parent. That I got to be in her care was probably the best thing that had ever happened to me. She had other foster children and even ended up taking care of many of the neighborhood kids. Wendy is a lady that just never gave up and always gave more than anyone could imagine. I no longer think of my real mother as mother but rather Wendy as my only true mother. I know that I'm not the only one that thinks that way either. It wasn't just kids that she helped it included the elderly, the disabled and the just down and out. What it takes to qualify for a saint is something I never found out, but I know it can't be more than the things I saw this lady do for anybody. Even people that had wronged her could find themselves looking up at her while she helped them when they were down. I'm 26 now and many of the things I'm telling you I never really realized until now. Since I was a kid she has helped me with everything from recovering from my mistakes to just being there to say "You can do it... It'll be okay." I've never been able to make a good attempt at trying to repay her for all she's done for me and I'm not alone. Many people just have taken for granted the things she's done for them. I know by now you're asking yourself why I've posted this letter. Well, to be honest and straight with you it's because I would like your help. If there's one thing that I've learned from Wendy it's that, "We are our neighbors keepers." You know the golden rule, "Do onto others as you would have them do onto you." Anyway now Wendy is an older lady that is disabled and can't do as much as she used to. With the disability sometimes expenses can be too much. I now live in Arkansas and can't be of much help. With no real family I'd like to do something to try to repay all she done for me and many others. That's were you come in. See if you can think back to someone that helped you no matter how big or small. Maybe it's someone you can't even remember or have lost touch with. For that moment picture Wendy and say, "I'd like to repay that kindness." I wish you would sit down and write a small note saying "Thanks for being the generous person that you have been!" If you could include something to help out. Maybe it's a five dollar bill, a book of stamps, even a dollar. Oh and would you also tell her the Steffin sends his love. Her birthday is next month on the 27th and I'm hoping I can make this own a very memorable one. Thanks for reading and if you choose to help God pless you. Her address is Wendy Mcghee 9636 Elm Taylor, Mi 48180Article: 4679
I found how to assign the pin number . Here is a sample for how to assign pin number . -- library IEEE; use IEEE.std_logic_1164.all; library METAMOR ; use METAMOR.attributes.all ; entity uh6 is port ( PXCLKN : in std_logic ; ); attribute pinnum of PXCLKN : signal is "p160" ; end uh6; -- CongArticle: 4680
vn5s-cng@asahi-net.or.jp wrote: > > I found how to assign the pin number . Hi, what synthesis tools are you using? I've seen this referred to before, but are attributes supported at all under the Viewlogic tools? Cheers, SteveArticle: 4681
Can anybody tell me if there is a low-cost programmer for Cypress CPLD and pASIC380 family? Their VHDL synthesis software is amazingly cheap (32 Pounds for Warp2 Version4.0 + VHDL synthesis book). The Cypress databook describes the Impulse 3 programmer which is based on an OEM version of the DATA I/O ChipLab programmer. Those of you who know this programmer also know is far from low-cost. Thanks for any comments, Hans. email: ees1ht@ee.surrey.ac.ukArticle: 4682
Hi, I am having duty of to make a algorithm which is able to access the abundant memory with minimum address pin lines. Perhaps that memory has about several Tera bytes capacity. Ofcouse this is a hypothesis. But my boss require me it. Probably it could be carried out with only new mathematical approach. With group theory? Any, any hints would be greatly appreciated. Thanks in adavnce.Article: 4683
John Maher <jmaher@vhdl.com.au> wrote: : A 45 day evaluation licence for SETANTA ED, a VHDL aware editor : is available at : : http://www.vhdl.com.au : Features include: : + Keyword colorising : + no file length limit : + Automatic Testbench Generation (Beta version) : + Automatic model insertion : + Dozens of free VHDL models including counters, adders : + Support for C/C++, JAVA, Perl, and many more languages : Try it out, lets us know what you think. Does it run on Unix/Linux? -- Uwe Bonnes bon@elektron.ikp.physik.th-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------Article: 4684
TukryopKim <acsapark@public.bta.net.cn> wrote: >Hi, >I am having duty of to make a algorithm which is able to access the >abundant memory with minimum address pin lines. Perhaps that memory has >about several Tera bytes capacity. Ofcouse this is a hypothesis. But my >boss require me it. >Probably it could be carried out with only new mathematical approach. >With group theory? If you know in advance how much memory you can have, you should be able = to address it with a single pin by sending the address serially instead of = in parallel. If you don't know how much memory you have, you might have to = use something like a stop bit (a signal 1.5 or 2 times the length of your normal signal) to mark the end of the address. ------------------------------------- Michael A. Quinlan mikeq@primenet.com http://www.primenet.com/~mikeq "If it doesn't fit, you must acquit!" -------------------------------------Article: 4685
Kevin Horton (khorton@iquest.net) wrote: : I've been doing a little project on the Altera EPX780LC84-15. I have it : working great (visit: http://www.iquest.net/~khorton/bankzill.htm for : details) but I'm having problems finding the programming spec. for this part. : The goal is to have a micro (80C85) program it on power-up so that changes : and additions are simple to do without replacing the FPGA. : I've done several 'net searches and have come up empty-handed. I visited the : Altera web page, and it's like the part has never existed; there are only : data sheets for their latest offerings. : So, if anyone can tell me the basics on how to get the data into the part's : SRAM cells, I'd be appriciative. BTW, I used the 'pengen' program to convert : the JEDEC file into the bit-stream required by the part; I just need to know : how to play with the TCK, TDI, TDO (output probably not required for : programming), and TMS lines. I've been using this series of parts since they were made by Intel (and the programming s/w was FREE - what an amazing idea to try and make money by selling parts!). If you look in the Altera PLSshell data book you will find that the binary output from PENGEN (or JED2JTAG) interleaves the TMS and TDI bits. What we do is convert the binary to an S3 file and blow it into an EPROM. The board has a little state machine in a 22V10 that gets bytes from the EPROM and serialises the bit pairs, even bits on TDI and odd bits on TMS. After each bit pair is driven the state machine toggles TCLK and every 4 pairs it increments the EPROM address. Note that we don't use the Intel hex format output of JED2JTAG since the first 4 bytes are size info. It would be nice to use a serial EEPROM but I don't know of a cheap one that has the 32KByte capacity needed for 4 daisychained devices. _________________________________________________________________________ Dr. Richard Filipkiewicz phone: +44 171 700 3301 Algorithmics Ltd. fax: +44 171 700 3400 3 Drayton Park email: rick@algor.co.uk London N5 1NU EnglandArticle: 4686
What are you doing? There are some reconfigurable cross-point switches from I-CUBE which do routing of digital signals. This may be what you need. I do also know of someone who uses a XILINX as an address comparitor. They download a different bit stream (which is dynamically calculated) to catch different addresses. The had to reverse-engineer the XILINX device bitstream though. -- Gareth Baron. Email: gareth@trsys.demon.co.ukArticle: 4687
Ian Lazarus wrote: > > I have an old (out of maintenance) Viewlogic schematic based > Xilinx system, and I have the choice to either bring it back up > to date (I won't bother with Viewlogic VHDL after reading the > 100% negative comments on this list!) or else switching to the > new Foundation stuff. > > I've been looking at the VHDL vs schematics thread and not seen > any comments on the Xilinx Foundation s/w in general and its VHDL > in particular. Maybe I missed it, but it seems that people use > either Viewlogic (schematic or VHDL) or Synopsis. Does anyone > use Foundation? Is it any good? Any advice would be welcome. > I definitely want to keep schematic entry, but I'd like to > add VHDL to try it (and maybe switch to VHDL for designs where > it has advantages) and it seems that I'd need Foundation to do it. > > thanks for your posts (or emails). > > Ian. > > -- > Ian Lazarus > Nuclear Physics Support Group, CCLRC, Daresbury Laboratory > email: I.H.Lazarus@dl.ac.uk Ian I bought the top of the line Foundation series with VHDL, XABEL etc a few weeks ago. I am surprised that a leading player in the fpga industry is selling such poor software. The Xilinx written stuff seems to be ok but they buy in the schematic tool and simulator from Aldec and the VHDL sythesiser from Metamor. I am only just getting into VHDL and so don't feel able to comment on the quality of the tool although the editor is VHDL aware and so highlights to code nicely. The big problem is the schematic editor and simulator. There are so many things wrong with it I don't know were to start. One immediate problem you will have is with busses particulary if you use the XBLOX parameterised modules. You seem unable to pass XBLOX busses through the hierarchy, the must be converted to normal busses first, pass through the symbol and then converted back to XBLOX at the next level. The XBLOX bus will not pass correctly to the simulator. There are more problems when you start to change the signals the signals that pass through the hierarchy. The software tries (an often fails) to be clever in ensuring that the signals passing from one level to the next match on the symbol. This would be ok if it worked but I seem to keep loosing the changes I just made. I could go an and on but I am at work and am not being payed to moan about a tool I bought. My recommendation, for what it is worth, is to go for the cheapest tool without VHDL or XBLOX and see for yourself. If you don't like it at least you haven't spent that much. Mike Forster Perkin Elmer Ltd UK These are my own views and not those of Perkin Elmer.Article: 4688
TukryopKim <acsapark@public.bta.net.cn> wrote: >I am having duty of to make a algorithm which is able to access the >abundant memory with minimum address pin lines. Perhaps that memory has >about several Tera bytes capacity. Ofcouse this is a hypothesis. But my >boss require me it. Does your boss know anything about binary arithmetic? >Probably it could be carried out with only new mathematical approach. >With group theory? Probably not. >Any, any hints would be greatly appreciated. Use big word sizes and wide busses. 256 bits or 1024 bits per word (or more?). Disks call them "block addresses". Same idea.Article: 4689
mak@cromp.ernet.in wrote: > > Hi all, > I have come across a lot of examples of processor architecture using > reconfigurable FPGAs. But has anyone heard of reconfigurable FPGAs > getting used in switching and routing applications? Links to stuff > on the web and other info is welcome. > Thanking in advance, > Makarand Talk to Bo Varga at Giga-Ops. I believe they are working on a telecom application using reconfigurable computing. -Ray Andraka, P.E. Chairman, the Andraka Consulting Group 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://www.ids.net/~randrakaArticle: 4690
Is anybody out there using the LP6-MPU programmer under NT 4.0 (or any verision for that matter). I installed the card and have tried all 16 :~( IO Addresses to no avail. The configuration menu refuses to accept my hardware... HELP!!! A direct email of *anything* would be appreciated. Thanks, Eric <ohms@vt.edu>Article: 4691
Windows 95, Visual Foxpro, MS Office 7.0, Visual Basic, Visual C++, Borland C++, Lotus cc:Mail Release 6, Lotus SmartSuite 97, ...... only costs US$40 for all? CorelDraw 7.0! only costs US$40? Hong Kong is the paradise in shopping. You can buy anything you want by an reasonable price. Today, by this chance, we would like to introduce to those who are interest in computing. In Hong Kong, there are five famous spots inclues: Golden Center - Sham Shui Po ; Mongkok Computer Center - Mongkok ; Sino Center - Mongkok ; 298 Computer Center - Wan Chai and Tsuen Fung Center -Tsuen Wan . In which, you can always find the most update warez includes all PC Warez, PC games, CD-Title, Shareware, Video-CD and Playstation game, etc. For Example, Inside a Installer (one CD only, Product No.: IN96112) , it contains MS Visual Foxpro 5.0, MS Windows 95 OEM Service Release, MS Money97, Lotus SmartSuite 97, Lotus cc:Mail Release 6, Symantec Cafe 1.5, Norton Your Eyes Only, Norton Smart Doctor 1.0, Quartdeck SpeedyROM 1.1, KPT BrayceAlpha 2.0, Visual dBASE Professional 4.4a, MacroMedia Animator Designer, Solitaire Antics, Almost Reality, CakeWalk Express 3.01, ClockMan95, Demo-It!2.0, MicroLogic EMAGIC, EXTRA! Version 6.2, MacroMedia Icon Designer, Infinite Disk, InfoSpy Pro, Fractal Deign Ray Dream Studio 4.1, Janna Conract Manager, Kurzweil Voice release 2.0, PC Maclan verion 6.0, NetWizard Plus Version 3.1, Organic Art, Microsoft Edition, PhotoWorks, RandoMedia, Real Oschestra, Reflection 4, RichWin, SoundForge 4.0, SignLab Pro+, Spectrum Pro CD Player, Starfish Internet Sidekick, TeleMagic, TurboCAD 3.0, MS-Visual SourceSafe 5.0, WinBye 32, Xing 3.02 Release, Zydeco Management Desktop. Good news to you today, that is, we would like to introduce all wares in VERY attractive price to you. Not only to provide over 50 warez for you to choose, but also to introduce the most suitable warez to you. We would provide a very attractive price not more than US$50. The more you buy, the more money you can save. Right? Hurry up! Don't miss this chance! Just ask what you want to know OR just decide and place your order to us. If you want some more details , you can go http://www.hkstar.com/~dogz now. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ DogZ Software CenterArticle: 4692
I've just finished a Xilinx 4025E design that PPR required 99% CLB's to place/route. Utilization of CLB resources is a bit on the high side (80%+ FG's, 50%+ DFF's). I've looked at the placed design with FPLAN and PPR appears to have done an OK job given the speed requirements. There are, however, many CLB's where only a few internal resources are used. There are also many split nets (nets routed through CLB's) reported. I'd like to make an informed assessment of how much more logic I can *ADD* to the design. So far, I've got the following to work with: (1) The standard CLB resource utilization stats from XACT. (2) A view of the entire placed design from FPLAN. (3) A grayscale view of routing congestion distribution from FPLAN. (4) An idea of how much more logic I'd like to add. Does anyone have any suggestions on other data I can use ? How would you go about making such an assessment ? Any feedback is greatly appreciated. -- _______________________________________________________________________ Lance Gin "off the keyboard Delco Systems - GM Hughes Electronics over the bridge, OFC: 805.961.7567 FAX: 805.961.7739 through the gateway, C43LYG@dso.hac.com nothing but NET!" _______________________________________________________________________Article: 4693
Lance Gin <c43lyg@dso.hac.com> wrote: >I've just finished a Xilinx 4025E design that PPR required 99% CLB's to >place/route. Utilization of CLB resources is a bit on the high side >(80%+ FG's, 50%+ DFF's). I've looked at the placed design with FPLAN >and PPR appears to have done an OK job given the speed requirements. >There are, however, many CLB's where only a few internal resources are >used. There are also many split nets (nets routed through CLB's) reported. >I'd like to make an informed assessment of how much more logic I can >*ADD* to the design. So far, I've got the following to work with: >(1) The standard CLB resource utilization stats from XACT. >(2) A view of the entire placed design from FPLAN. >(3) A grayscale view of routing congestion distribution from FPLAN. >(4) An idea of how much more logic I'd like to add. >Does anyone have any suggestions on other data I can use ? >How would you go about making such an assessment ? >Any feedback is greatly appreciated. I've been working with Xilinx FPGA's fro several years, though I don't have the latest software (I run Xact v4.2 fro financial reasons). A rule of thumb I have developed over the years that seems to serve me well is to load a device to no more than 70% utilization when doing the initila design. You will ALWAYS find something you need to change/add as the design is brought up, and this will tend to fill the device. I would be scared to death of your device that was 99% utilized for two reasons. 1. You have no room to move if you need to add something later. 2. As the device fill up, the most premium routing channels get filled. Somewhere in that design is a signal (or more) that got routed around Robin Hood's barn to get to a CLB next door. These can be the ones that get you. I assume that you specified the max delays for some critical signals from your post. The ones to watch are the ones you DIDN'T specify. They probably aren't AS critical as the ines you did specify, but if the delays get pushed waaaaay out, they may become critical. I know I didn;t give you any hard and fast answer, but I hope this helps some.Article: 4694
Mike Treseler (tres@tc.fluke.COM) wrote: : [1] Treseler, Michael "Designing State Machine Controllers Using : Programmable Logic" Prentice Hall, 1992, p89. : [2] Ibid, p82-83. This is a good book, by the way. I have it, I read it, I learn from it. CharlieArticle: 4695
============================================================================= Call for Papers 1997 International Symposium on Physical Design April 14-16, 1997 Napa Valley, California Sponsored by the ACM SIGDA in cooperation with IEEE Circuits and Systems Society The International Symposium on Physical Design provides a forum to exchange ideas and promote research on critical areas related to the physical design of VLSI systems. All aspects of physical design, from interactions with behavior- and logic-level synthesis, to back-end performance analysis and verification, are within the scope of the Symposium. Target domains include semi-custom and full-custom IC, MCM and FPGA based systems. The Symposium is an outgrowth of the ACM/SIGDA Physical Design Workshop. Following its five predecessors, the symposium will highlight key new directions and leading-edge theoretical and experimental contributions to the field. Accepted papers will be published by ACM Press in the Symposium proceedings. Topics of interest include but are not limited to: 1. Management of design data and constraints 2. Interactions with behavior-level synthesis flows 3. Interactions with logic-level (re-)synthesis flows 4. Analysis and management of power dissipation 5. Techniques for high-performance design 6. Floorplanning and building-block assembly 7. Estimation and point-tool modeling 8. Partitioning, placement and routing 9. Special structures for clock, power, or test 10. Compaction and layout verification 11. Performance analysis and physical verification 12. Physical design for manufacturability and yield 13. Mixed-signal and system-level issues. IMPORTANT DATES: Submission deadline: December 20, 1996 Acceptance notification: February 1, 1997 Camera-ready (6 page limit) due: March 1, 1997 SUBMISSION OF PAPERS: Authors should submit full-length, original, unpublished papers (maximum 20 pages double spaced) along with an abstract of at most 200 words and contact author information (name, street/mailing address, telephone/fax, e-mail). Electronic submission via uuencoded e-mail is encouraged (single postscript file, formatted for 8 1/2" x 11" paper, compressed with Unix "compress" or "gzip''). Email to: ispd97@ece.nwu.edu Alternatively, send ten (10) copies of the paper to: Prof. Majid Sarrafzadeh Technical Program Chair, ISPD-97 Dept. of ECE, Northwestern University 2145 Sheridan Road, Evanston, IL 60208 USA Tel 847-491-7378 / Fax 847-467-4144 SYMPOSIUM INFORMATION: To obtain information regarding the Symposium or to be added to the Symposium mailing list, please send e-mail to ispd97@cs.virginia.edu. Information can also be found on the ISPD-97 web page: http://www.cs.virginia.edu/~ispd97/ SYMPOSIUM ORGANIZATION: General Chair: A. B. Kahng (UCLA and Cadence) Past Chair: G. Robins (Virginia) Steering Committee: J. Cohoon (Virginia), S. Dasgupta (Sematech), S. M. Kang (Illinois), B. Preas (Xerox PARC) Program Chair: M. Sarrafzadeh (Northwestern) Keynote Address: T. C. Hu (UC San Diego) & E. S. Kuh (UC Berkeley) Special Address: R. Camposano (Synopsys) Publicity Chair: M. J. Alexander (Washington State) Local Arrangements Chair: J. Lillis (UC Berkeley) Technical Program Committee: C. K. Cheng (UC San Diego) W. W.-M. Dai (UC Santa Cruz) J. Frankle (Xilinx) D. D. Hill (Synopsys) M. A. B. Jackson (Motorola) J. A. G. Jess (Eindhoven) Y.-L. Lin (Tsing Hua) C. L. Liu (Illinois) M. Marek-Sadowska (UC Santa Barbara) M. Sarrafzadeh (Northwestern) C. Sechen (Washington) K. Takamizawa (NEC) M. Wiesel (Intel) D. F. Wong (Texas-Austin) E. Yoffa (IBM) =============================================================================Article: 4696
Your problem description is not sufficient. A little more info would help! However... Lloyd Miller wrote: > > TukryopKim <acsapark@public.bta.net.cn> wrote: > > >I am having duty of to make a algorithm which is able to access the > >abundant memory with minimum address pin lines. Perhaps that memory has > >about several Tera bytes capacity. Ofcouse this is a hypothesis. But my > >boss require me it. > 2 ** 10 -> 1KB 2 ** 20 -> 1MB 2 ** 30 -> 1GB 2 ** 40 -> 1TB etc So you need a few more bits than 40 for "several" terabytes! > Does your boss know anything about binary arithmetic? > > >Probably it could be carried out with only new mathematical approach. > >With group theory? > Is your memory 'linear' or 'holey'. If linear, than a simple linear page table with a length register will do the trick. If the memory is holey (ie. non-contiguous virtually), then a tree structured page table is most efficient for very large memories. The Alpha VM implementation is a 3-level tree with 43 (8TB) address lines implemented. There are also a few more tricks used to save on translation buffer entries such as the ability to group contiguous virtual/physical pages into 'super' (can't remember the term in the Alpha Reference Manual) pages of up to 4MB each (ie. 512 normal 8KB pages mapped by 1 TB entry). > Probably not. > > >Any, any hints would be greatly appreciated. > > Use big word sizes and wide busses. 256 bits or 1024 bits per word (or > more?). Disks call them "block addresses". Same idea. Disk blocks and memory pages are really much the same think when it comes to addressability. The width of the bus really only affects the speed of access/transfer, not the addressability. Eg. Ethernet uses a 48-bit address (256TB) which is somewhat of an overkill for most LANs. TonyArticle: 4697
timolmst@cyberramp.net wrote: > Lance Gin <c43lyg@dso.hac.com> wrote: > >I've just finished a Xilinx 4025E design that PPR required 99% CLB's to > >place/route. Utilization of CLB resources is a bit on the high side > >(80%+ FG's, 50%+ DFF's). I've looked at the placed design with FPLAN > >and PPR appears to have done an OK job given the speed requirements. > >There are, however, many CLB's where only a few internal resources are > >used. There are also many split nets (nets routed through CLB's) reported. > > >I'd like to make an informed assessment of how much more logic I can > >*ADD* to the design. > I've been working with Xilinx FPGA's fro several years, though I don't > have the latest software (I run Xact v4.2 fro financial reasons). A > rule of thumb I have developed over the years that seems to serve me > well is to load a device to no more than 70% utilization when doing > the initila design. You will ALWAYS find something you need to > change/add as the design is brought up, and this will tend to fill the > device. > I would be scared to death of your device that was 99% utilized for > two reasons.... I'm working on a design that requires 92% of the FFs in a 4010D and it is routed usually without problems. It has LOTs of counters and pipeline FFs to get to speed, so it's flipflop bound ( 40 MHz in a -6 LCA, cannot get a faster chip because some units are already in the field.) I think that PPR spreads out the logic over as many blocks as it seems usable if it is given the space; when i look at my chip in XACT i _never_ see a free block. I feel i still could add a few bits worth of a LFSR counter or shift registers, but adding a single long line decoder or a multiplexer with lots of inputs or anything else that affects the layout in a non-local way will break the routing for sure. (Note that i don't use a single long line decoder and i have LOTs of spare function generators.) _Q_: Does anybody know the LOC= constraint to move a FlipFlop to an unbonded I/O block? I don't care what block it may end up in, but i do not want to see it in a CLB or in a usable pin. Gerhard (still experimenting with Netscape 3, hope it works.) -- Gerhard Hoffmann ^ I must insist on that second 'n'. on the air: dk4xp in the air: D-1441Article: 4698
Mark, Depending on how you wrote your VHDL code, Synopsys will automatically optimize a flip-flop into the IOB. How did you implement the GSR? If you have a reset signal connected to the GSR on the startup block and have a reset signal in your VHDL code, Synopsys thinks you want an asynchronous reset on the FF but the FF's in the IOB don't have an asynchronous reset so Synopsys must put that FF in a CLB. Synopsys doesn't understand the GSR and how it works. When it sees a reset in the VHDL code, it attaches the reset line to the reset pin on the FF. XNFPREP knows that you are using the GSR so it strips the global reset line that goes to the asynchronous reset pin o the FF's. You should not have to use any special commands in the Synopsys script file. If all else fails, you can instantiate the IOB in your code. Kate Meilicke Xilinx FAEArticle: 4699
Mark Sandstrom <Mark.Sandstrom@martis.fi> wrote: >Has anyone managed to utilize Xilinx 4000e IOB FFs through Synopsys? >I'm using FPGA Compiler v.3.4b and Xact 5.2.1. It seems like no matter >what I try, all my I/O registers are mapped into the CLBs. I'm >describing the I/O registers in the VHDL description in a normal >D-flip-flop style. The flip-flops are controlled by the global reset for >which I use the GSR net. >I tell Synopsys: >set_register_type -exact -flip_flop OFD_F find (design, stm4_out_reg) >Performing set_register_type on design 'stm4_out_reg'. >1 >but the flip_flops synthesize into 'FDC's. >Thanks for any information that could help! >Mark I just recently ran into the same problem. Synopsys is not the problem, because it turns everything into generic registers. It is XBLOX that will move registers into IOBs. There are a number of restrictions that will prevent XBLOX from moving registers into an IOB, listed in the XBLOX documentation somewhere. I ended up hand-instantiating the output registers that I wanted, and that did the trick. Robert
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