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
Although its response shape is limited, The cascaded integrator comb filter is one you might find useful. See E.B. Hogenauer, “An Economical Class of Digital Filters for Decimation and Interpolation”, IEEE Trans on ACSSP vol ASSP-29 no.2 April 1981 I haven't seen too much on designing FIR filters with power of two coefficients. If you find anything, please post it for the rest of us. What is the complete citation for the H. Samueli paper? flavioas@my-deja.com wrote: > Hi, > > We are looking for some papers on how to > implement multiplierless filter structures in > FPGAs. Papers like "Design of multiplierless > digital data transmission filters with > powers-of-two coefficients" H. Samueli. and others > talking about "CSD coefficients" will be helpfull. > Can any one help? > We are trying to subscribe to this newsgroup, > using outlook express to connect to the server : > nntp-serv.cl.cam.ac.uk, and getting the message: > "you have no permission to talk to this server". > So, How to subscribe? > > Thanks in advance, > > Flávio Andrade > > Sent via Deja.com http://www.deja.com/ > Before you buy. -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randrakaArticle: 18176
FPGA designers needed in Portland, Oregon. Requires 3+ years' experience including recent experience designing FPGAs. Must be proficient with Verilog and VHDL. These positions are full time and require the designer to work onsite at the client's facility in Portland. Please respond with your resume. Principals only.Article: 18177
flavioas@my-deja.com wrote in message <7tcsra$rkh$1@nnrp1.deja.com>... > We are trying to subscribe to this newsgroup, >using outlook express to connect to the server : >nntp-serv.cl.cam.ac.uk, and getting the message: >"you have no permission to talk to this server". >So, How to subscribe? you need to talk to the sysadmin for the server in order to get permission to use it. -- ----------------------------------------- Andy Peters Sr Electrical Engineer National Optical Astronomy Observatories 950 N Cherry Ave Tucson, AZ 85719 apeters (at) noao \dot\ edu "Creation Science" is oxymoronic.Article: 18178
On Tue, 05 Oct 1999 09:21:44 GMT, mush@slip.net (Dave Decker) wrote: >On Tue, 05 Oct 1999 13:01:38 GMT, flavioas@my-deja.com wrote: > >> >> >> Hi, >> >> We are looking for some papers on how to >>implement multiplierless filter structures in >>FPGAs. Papers like "Design of multiplierless >>digital data transmission filters with >>powers-of-two coefficients" H. Samueli. and others >>talking about "CSD coefficients" will be helpfull. >>Can any one help? >> We are trying to subscribe to this newsgroup, >>using outlook express to connect to the server : >>nntp-serv.cl.cam.ac.uk, and getting the message: >>"you have no permission to talk to this server". >>So, How to subscribe? >> >> Thanks in advance, >> >> Flávio Andrade >> >> >>Sent via Deja.com http://www.deja.com/ >>Before you buy. > > >There are some interesting papers at the U of K. on automatic filter >generation using sparce numbers of ones in the coefficients to >minimize the size of the constant multipliers. >try: >ftp://ftp.tisl.ukans.edu/pub/projects/DSP/FPGA/Filters_and_CAD.pdf > >for one. > >Dave Decker Also look at: http://www.ittc.ukans.edu/projects/FPGA/ This page seems to have moved since the last time I tried to find it. Webmaster Doug Herbers was kind enough to point me to it's new puka. Thanks Doug Cheers, Dave DeckerArticle: 18179
This is a multi-part message in MIME format. --------------00806641D7476D06B694E93A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit We require Sr. Design engineers who are willing to work in a high tech company in the Maryland area on a contract basis. The job requires to work on Board design candidate should be proficient in ASIC/ FPGA, Verilog should have worked on Altera. Please provide your contact details also with the resume. Thanks --------------00806641D7476D06B694E93A Content-Type: text/x-vcard; charset=us-ascii; name="rsingh.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Ravi Singh Content-Disposition: attachment; filename="rsingh.vcf" begin:vcard n:Singh;Ravi tel;cell:(703) 582 7819 tel;work:(703) 556 0033 x-mozilla-html:FALSE org:Comptel, Inc. version:2.1 email;internet:rsingh@comptelinc.com title:Director - Business Development adr;quoted-printable:;;2102 C, Gallows Road,=0D=0ASouth Office Tyson Park,=0D=0A;Vienna;VA ;22182;U.S.A fn:Ravi Singh end:vcard --------------00806641D7476D06B694E93A--Article: 18180
Jamil Khaib wrote: > > Hi, > Have you ever imagined that you could build hardware with ZERO design > cost? > Do you want to test your digital designs? > Do you want to have your own CPLD prototyping board? > > You can check our project at > http://www.openip.org/oc/projects/board/board.html > > The design is available for free under the openip license. It is based > on free and only free software tools and design methodology. > > The site contailn links to lot of free EDA tools > > for more information about the openip organization you can visit our > site at > http://www.openip.org > or contact our mailing list at openip@egroups.com or > http://www.egroups.com/group/openip/ on the web > > Please give us your comment on the design itself and the methodology we > are using > > Thanks > Jamil Khatib > khatib@openip.org > OpenIP Organization > http://www.openip.orgArticle: 18181
I will give this a try. Is there some information on how to do this in VHDL somewhere? I haven't seen in the documentation. In case it's not obvious yet this is my first real project with FPGA's and VHDL. Tyrone In article <37F7F87E.186502C3@ids.net>, Ray Andraka <randraka@ids.net> wrote: >Presumably, the reference clock is stable as is the device parameters >that affect the >delay. If the output ratio is constant, then I would expect the delay >to try to stay >at one point until the temperature, voltage or reference change. That >point may be (is >probably) between two of the taps in the delay line, but since the exact >duty cycle is >probably not critical you could set it up with enough filtering >(basically hysterisis) >to keep it from jumping back and forth. Basically, the DLL picks the >tap in a delay >line that is closest to the delay that give the desired output. The >finer the delay >increment the less the size of the jump (which is why I was suggesting >the carry logic >might make a good tapped delay line). > >Alvin E. Toda wrote: > >> On Sun, 3 Oct 1999, Ray Andraka wrote: >> >> > Depending on delays in FPGAs is generally a bad practice, as the >design becomes >> > very sensitive to routing, as well as to process improvements. You >might be able >> > to reliably use the FPGA circuit delays by constructing a delay lock >loop in the >> > FPGA logic. It uses feedback to adjust how many delay elements are >in the signal. >> > You could probably use the carry chain as the delay line, as it has >relatively >> > short and consistent delays for each "tap". You will have to be >careful about the >> > routing off the taps though. I haven't tried this to see how well it >works, but at >> > 25 MHz, you might be able to pull it off. >> >> It would seem that another consideration would be the amount of >> jitter on the clock edge that could be tolerated, since it would >> seems that switching between taps would produce such behavior. >> Perhaps since the clocking is only needed for a special device, >> that specific asynchronous logic be designed so that the sequencing >> of signals is always correct and IAW whatever signals come >> back from the device. >> >> --al toda >> >> ########################################################### >> Alvin E. Toda aet@lava.net >> sr. engineer Phone: 1-808-455-1331 >> 2-Sigma WEB: http://www.lava.net/~aet/2-sigma.html >> 1363-A Hoowali St. >> Pearl City, Hawaii, USA > > > >-- >-Ray Andraka, P.E. >President, the Andraka Consulting Group, Inc. >401/884-7930 Fax 401/884-7950 >email randraka@ids.net >http://users.ids.net/~randraka > > -- -------------- thompson@eecis.udel.edu University of Delaware Tyrone Thompson EE Graduate StudentArticle: 18182
Has anybody manged to get post route simulation to work using Xilinx's Verilog simprim libraries ? As far as I can tell the M1.5i LUT models are broken at least as far as ModelSim is concerned. I've had to fix both these and, to get compatibility with Synplify synthesis , the sync set/reset FF model as well. There are also compatibility problems between the timing checks in the sync. RAM models and the way SDF backannotation is done.Article: 18183
I recommend doing your first FPGA projects in schematics rather than VHDL for the simple reason that schematics don't insulate you from the device architecture. Once you become familiar with what works and doesn't work, then you can move on to trying to get VHDL to produce the circuit and layout you now know is the best. Yes, it can be done with VHDL, but with something like this, I think you'll find that placement is critical to the success. Doing placement within VHDL is not much fun and is not well documented. Tyrone Thompson wrote: > I will give this a try. Is there some information on how to do this in VHDL > somewhere? I haven't seen in the documentation. In case it's not obvious yet > this is my first real project with FPGA's and VHDL. > > Tyrone > > In article <37F7F87E.186502C3@ids.net>, Ray Andraka <randraka@ids.net> wrote: > >Presumably, the reference clock is stable as is the device parameters > >that affect the > >delay. If the output ratio is constant, then I would expect the delay > >to try to stay > >at one point until the temperature, voltage or reference change. That > >point may be (is > >probably) between two of the taps in the delay line, but since the exact > >duty cycle is > >probably not critical you could set it up with enough filtering > >(basically hysterisis) > >to keep it from jumping back and forth. Basically, the DLL picks the > >tap in a delay > >line that is closest to the delay that give the desired output. The > >finer the delay > >increment the less the size of the jump (which is why I was suggesting > >the carry logic > >might make a good tapped delay line). > > > >Alvin E. Toda wrote: > > > >> On Sun, 3 Oct 1999, Ray Andraka wrote: > >> > >> > Depending on delays in FPGAs is generally a bad practice, as the > >design becomes > >> > very sensitive to routing, as well as to process improvements. You > >might be able > >> > to reliably use the FPGA circuit delays by constructing a delay lock > >loop in the > >> > FPGA logic. It uses feedback to adjust how many delay elements are > >in the signal. > >> > You could probably use the carry chain as the delay line, as it has > >relatively > >> > short and consistent delays for each "tap". You will have to be > >careful about the > >> > routing off the taps though. I haven't tried this to see how well it > >works, but at > >> > 25 MHz, you might be able to pull it off. > >> > >> It would seem that another consideration would be the amount of > >> jitter on the clock edge that could be tolerated, since it would > >> seems that switching between taps would produce such behavior. > >> Perhaps since the clocking is only needed for a special device, > >> that specific asynchronous logic be designed so that the sequencing > >> of signals is always correct and IAW whatever signals come > >> back from the device. > >> > >> --al toda > >> > >> ########################################################### > >> Alvin E. Toda aet@lava.net > >> sr. engineer Phone: 1-808-455-1331 > >> 2-Sigma WEB: http://www.lava.net/~aet/2-sigma.html > >> 1363-A Hoowali St. > >> Pearl City, Hawaii, USA > > > > > > > >-- > >-Ray Andraka, P.E. > >President, the Andraka Consulting Group, Inc. > >401/884-7930 Fax 401/884-7950 > >email randraka@ids.net > >http://users.ids.net/~randraka > > > > > > -- > -------------- > thompson@eecis.udel.edu University of Delaware > Tyrone Thompson EE Graduate Student -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randrakaArticle: 18184
We are using an Altera 10K50V and have noticed that, at power up, it will sometimes suck as much as 600mA off the 3.3V supply. Warming the part will make the inrush problem disappear, while cooling the part makes the problem happen every time. Our on-board supply is good for about 500mA and, looking at the ramp up of the 3.3V, it is clearly being limited to a 1.2V level even as the current increases. A more powerful lab supply overcomes the current problem and the current will suddenly drop to normal (about 20mA) allowing the 3.3V to climb to it's full value. Note that the Altera part is the only thing powered by the 3.3V supply and this occurs before any attempt at programming the part. The 3.3V is derived from the on-baord 5V supply through a three-terminal regulator. This problem seems to be worse with some date codes and other date codes don't show the problem at all. I'd be very grateful to hear from anyone who has experienced this problem. -Dave Krueger ---------------------------------------------------------- WEB: http://www.kruegerphoto.com MAIL: dave@kruegerphoto.comArticle: 18185
While you are looking at multiplierless filters, you might also look carefully at the distributed arithmetic approach. I've done some sizeable and fast filters using this technique. Some benchmarks: a 1024 tap FIR filter with 12 bit inputs and 12 bit coefficients will fit in a Xilinx XCV600 (virtex) and will run at 12 MS/S in a -4 part or 15.5 MS/S in a -6 using serial distributed arithmetic. a 32 tap SDA FIR filter with 12 bit coefficients can fit inside a 10x10 CLB (virtex) area. a 32 tap FIR filter with 12 bit inputs and 12 bit coefficients using parallel distributed arithmetic will run at 140 MS/S in a Virtex XCV300-4 or at around 180 MS/S in a -6 speed grade. The advantages of the DA filters are 1) data rate is relatively independent of the number of taps, 2) serial implementations can reduce the size of the hardware considerably for data rates below 10-15 MHz, 3) Conventional filter design programs can be used to generate the coefficients. Dave Decker wrote: > On Tue, 05 Oct 1999 13:01:38 GMT, flavioas@my-deja.com wrote: > > > > > > > Hi, > > > > We are looking for some papers on how to > >implement multiplierless filter structures in > >FPGAs. Papers like "Design of multiplierless > >digital data transmission filters with > >powers-of-two coefficients" H. Samueli. and others > >talking about "CSD coefficients" will be helpfull. > >Can any one help? > > We are trying to subscribe to this newsgroup, > >using outlook express to connect to the server : > >nntp-serv.cl.cam.ac.uk, and getting the message: > >"you have no permission to talk to this server". > >So, How to subscribe? > > > > Thanks in advance, > > > > Flávio Andrade > > > > > >Sent via Deja.com http://www.deja.com/ > >Before you buy. > > There are some interesting papers at the U of K. on automatic filter > generation using sparce numbers of ones in the coefficients to > minimize the size of the constant multipliers. > try: > ftp://ftp.tisl.ukans.edu/pub/projects/DSP/FPGA/Filters_and_CAD.pdf > > for one. > > Dave Decker -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randrakaArticle: 18186
> Has anybody manged to get post route simulation to work using Xilinx's Why do you want to do a post route simulation? Why not just do a functional (unit delay) simulation and then make your static timing when routing the design?Article: 18187
In article <37F8A4B9.8196E368@ids.net>, Ray Andraka <randraka@ids.net> wrote: > Actually, the output can be taken from after any register. The output from > any register is just a copy of the input to the first register delayed by a > number of clocks. > > rseglie@my-deja.com wrote: > > > Tank you for your help > > > > Norm is for "standard" (sometimes I forget my basic english). > > > > I solve the problem on friday. The polynom is ok, I've done a mistake > > about where to get the output. > > > > A few answers to your email : > > About degree : I start from 0 (d0 for Q0) so there's, of course, 43 > > registers. > > The + means xor. > > > > Standards always show serial scrambler so I thought that it's easier to > > talk about them. > > In my FPGA, I transform the "serie" version in 8 bits parallel "version" > > , looking the content of all registers after 8 clock cycles. Of course > > you can use this method for 16 bits version. > > > > My mistake was to take the output after the register but it must be > > taken before ! (when I wrote this message I have any idea and I thought > > the diagram was wrong). > > I test it and it's ok !! > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > -- > -Ray Andraka, P.E. > President, the Andraka Consulting Group, Inc. > 401/884-7930 Fax 401/884-7950 > email randraka@ids.net > http://users.ids.net/~randraka > > I thought too that I can take the output from any register (and it's better to take output after register rather after combinational logic) but it wasn't work ! In fact I think that because scrambling it's only for the payload (48 last bytes of the 53) of the ATM cell (the scrambler is stop during the first 5 bytes), the payload was shift of one bit and the PSA don't pass... I think there's no problem of taking output after registers for scramblers won't stop or for device won't worked with scrambling data and non scrambling at the same time. Greeting. Sent via Deja.com http://www.deja.com/ Before you buy.Article: 18188
Hi! I'm desinging a 32-bit ALU on a FPGA (XC4010) using Foundation Series. It has a 32-bit control word and 32-bit data word. What I'm trying to do is to send these commands via 8-bit parallel port. Could you help me how to design a ciruit that will accept four 8-bit words (the control word), followed by four 8-bit words(input or output data) going the the ALU or from the ALU. This is part of my Honours thesis and I'm running out of time and the solutions I have so far are not very advanced and probably not working. Have you got any material (schematics) that I could use. I would be greatful if you could help me. I'm using Xilinx Foundation Series and implementation is on XC4010E.Article: 18189
For those working with test procedures for FPGAs and fault-tolerant FPGA systems: ***************************************************** 1ST IEEE LATIN-AMERICAN TEST WORKSHOP Marina Palace Hotel, Rio de Janeiro, Brazil March 13 - 15, 2000 ***************************************************** CALL FOR PAPERS The 1st IEEE Latin-American Test Workshop provides a forum for specialists from all over the world, especially from Latin America, to present and discuss various aspects of system, board, and component testing with design, manufacturing and field considerations in mind. The best papers of the workshop will be invited to re-submit to a special issue of the Journal of Electronic Testing: Theory and Applications, published by Kluwer. *** TOPICS OF INTEREST INCLUDE, BUT ARE NOT LIMITED TO: - Analog Mixed Signal Test - Fault-Tolerant Systems Applications - Autom. Test Gen./Fault Simulation - IDDQ Test - Built-In Self-Test - On-Line Testing - Defect & Failure Analysis - Process Characterization, Reliability - Dependability Estimation - Process Control and Measurement - Distributed Fault-Tolerant Systems - Simulation and Design Verification - E-Beam Testing and Thermal Testing - SW Testing/SW Design for Test - Economics of Test - Synthesis for Testability/DFT - Fault Modeling & Diagnosis - System-on-Chip Test - Fault-Tolerance in HW/SW - Yield Learning and Enhancement - Fault-Tolerant Architectures *** PAPER SUBMISSION INFORMATION: To encourage and facilitate informal discussions participation will be limited. Those interested in presenting recent results at the workshop please submit four copies of an extended abstract, one to three pages long, or full length paper to one of the Program Chairs below: Victor Champac Marcelo Soares Lubaszewski Inst. Nacional de Astrofisica, Univ. Federal do Rio Grande do Sul Optica y Electronica (INAOE) Departamento de Engenharia Eletrica Apdo. Postal 51 y 216 Av. Osvaldo Aranha- Sar. Leite, 103 Puebla, 72000 Mexico Porto Alegre, 90035-190 Brazil champac@inaoep.mx luba@iee.ufrgs.br Authors should include the complete address, phone/fax numbers and e-mail address, and designate a contact person and a presenter. The Program Committee also welcome proposals for panels and special topic sessions. Alternatively, it is strongly encouraged Post-Script or PDF electronic submissions via the workshop web page: www.epo.pucrs.br/~la-tttc/latw00.html *** SUBMISSION DEADLINE: November 15th, 1999. *** NOTIFICATION OF ACCEPTANCE: December 30th, 1999. *** SPONSORED BY: IEEE Computer Society Test Technology Technical Council (TTTC) *** IN COOPERATION WITH: Brazilian Computer Society (SBC), Brazilian Microelectronics Society (SBMicro) Catholic University (PUCRS), Federal Brazilian Research Agencies CNPq-CAPES Federal University of RGS (UFRGS), Federal University of Rio de Janeiro (UFRJ) GENERAL CHAIRS: Fabian Vargas, PUCRS - Brazil, vargas@computer.org Yervant Zorian, LogicVision - USA, y.zorian@computer.org PROGRAM CHAIRS: Victor Champac, INAOE - Mexico Marcelo S. Lubaszewski, UFRGS - Brazil PUBLICITY CHAIRS: Bernard Courtois, TIMA/INPG - France Ingrid Jansch-Porto, UFRGS - Brazil LOCAL ARRANGEMENTS CHAIR: Vladimir Castro Alves, UFRJ - Brazil BRAZILIAN COMPUTER SOCIETY LIAISON: Ingrid Jansch-Porto, UFRGS - Brazil BRAZILIAN MICROELECTRONICS SOCIETY LIAISON: Wilhelmus Van Noije, USP - Brazil EUROPE LIAISON: J. Paulo Teixeira, INESC/IST - Portugal ASIA LIAISON: Kozo Kinoshita, Osaka Univ. - Japan NOTRH AMERICA LIAISON: Alex Orailoglu, UC San Diego - USA PROGRAM COMMITTEE: J. Abraham - U. of Texas at Austin, USA V. D. Agrawal - Lucent Tech., USA E. Barros - UFPE, Brazil B. Bennetts - Bennetts Assoc., England E. A. Bezerra - PUCRS, Brazil A. M. Brochi - TELEBRAS, Brazil O. Calvo - UNLP, Argentina H. Castro - UFC, Brazil S. Demidenko - Singapore P., Singapore R. Duarte - UFMG, Brazil G. Espinosa - INAOE, Mexico A. O. Fernandes - UFMG, Brazil J. Figueras - UPC, Spain J. L. Huertas - U. of Sevilla, Spain A. Ivanov - UB Columbia, Canada B. Kaminska - OPMAXX, USA J. A. Martino - FEI/FCA-LSI/USP, Brazil E. Martins - UNICAMP, Brazil M. Marzouki - LIP6 Lab., France P. C. Maxwell - Hewlett Packard, USA J. V. Medina - UDV, Colombia M. Nicolaidis - TIMA/INPG, France A. R. Paula Jr. - UNIVAP, Brazil D. Pradhan - Texas A&M U., USA P. Prinetto - Politec. di Torino, Italy E. Randon - USB, Venezuela M. Renovell - LIRMM, France A. J. L. Ribeiro - TELEBRAS, Brazil J. M. Santos - FEUP, Portugal M. Sarmiento - INAOE, Mexico M. Strum - USP, Brazil L. E. Toledo - UCC, Argentina R. Velazco - TIMA/INPG, France T. W. Williams - Synopsys Inc., USA H. J. Wunderlich - U. Stuttgart, Germany ***************************************************************** ***************************************************************** Fabian Vargas LA-TTTC Chair ___________________________ Catholic University - PUCRS Electrical Engineering Dept. Email: vargas@computer.org Av. Ipiranga, 6681 Phone: +55 51 320.3540 90619-900 Porto Alegre FAX: +55 51 320.3625 BrazilArticle: 18190
Rick Filipkiewicz wrote: > Has anybody manged to get post route simulation to work using Xilinx's > Verilog > simprim libraries ? > > As far as I can tell the M1.5i LUT models are broken at least as far as > ModelSim is concerned. I've had to fix both these and, to get > compatibility with Synplify synthesis , the sync set/reset FF model as > well. There are also compatibility problems between the timing checks in > the sync. RAM models and the way SDF backannotation is done. Hi, Simulator: Verilog-XL 2.6.36 My Synthesis tool: Synplify v5.21 My P&R tool: Design Manager M1.5 I am using $XILINX/verilog/data for postsynthesis simulations and $XILINX/verilog/src/simprims for postlayout (or post-route, according to your terminology) simulations. This is the recommended method by Xilinx. You can find this information at http://support.xilinx.com software manuals. We have not observed any incompatibility between the gate-level and postlayout-level. The only timing violations occur during reset insertion but not any abnormal difference observed between RTL (synthesizable) model and postlayout model. Target is XC40150XV. I think Austin is completely right since the postlayout simulations are aggressively long. For example our minimal 20 ms. postlayout simulation lasts 38 hours on a Sun Ultra-10/256MB/300 MHz UltraSPARC. Static Timing Analyzer TRACE can give practicaly faster results. Despite that we perform postlayout simulations. Utku -- I feel better than James Brown.Article: 18191
You need to match the delays for the scrambled and non-scrambled paths. Don't put the not-to-be scrambled data into the shift register at all (in other words, don't turn off the scrambling by breaking the feedback path, that will break it). It will work if the delays are matched and you don't upset the contents of the shift register while passing non-scrambled data. rseglie@my-deja.com wrote: > In article <37F8A4B9.8196E368@ids.net>, > Ray Andraka <randraka@ids.net> wrote: > > Actually, the output can be taken from after any register. The output > from > > any register is just a copy of the input to the first register delayed > by a > > number of clocks. > > > > rseglie@my-deja.com wrote: > > > > > Tank you for your help > > > > > > Norm is for "standard" (sometimes I forget my basic english). > > > > > > I solve the problem on friday. The polynom is ok, I've done a > mistake > > > about where to get the output. > > > > > > A few answers to your email : > > > About degree : I start from 0 (d0 for Q0) so there's, of course, 43 > > > registers. > > > The + means xor. > > > > > > Standards always show serial scrambler so I thought that it's easier > to > > > talk about them. > > > In my FPGA, I transform the "serie" version in 8 bits parallel > "version" > > > , looking the content of all registers after 8 clock cycles. Of > course > > > you can use this method for 16 bits version. > > > > > > My mistake was to take the output after the register but it must be > > > taken before ! (when I wrote this message I have any idea and I > thought > > > the diagram was wrong). > > > I test it and it's ok !! > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > -- > > -Ray Andraka, P.E. > > President, the Andraka Consulting Group, Inc. > > 401/884-7930 Fax 401/884-7950 > > email randraka@ids.net > > http://users.ids.net/~randraka > > > > > I thought too that I can take the output from any register (and it's > better to take output after register rather after combinational > logic) but it wasn't work ! > In fact I think that because scrambling it's only for the payload (48 > last bytes of the 53) of the ATM cell (the scrambler is stop during the > first 5 bytes), the payload was shift of one bit and the PSA don't > pass... > I think there's no problem of taking output after registers for > scramblers won't stop or for device won't worked with scrambling data > and non scrambling at the same time. > Greeting. > > Sent via Deja.com http://www.deja.com/ > Before you buy. -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randrakaArticle: 18192
Utku Ozcan wrote: > Rick Filipkiewicz wrote: > > > Has anybody manged to get post route simulation to work using Xilinx's > > Verilog > > simprim libraries ? > > > > As far as I can tell the M1.5i LUT models are broken at least as far as > > ModelSim is concerned. I've had to fix both these and, to get > > compatibility with Synplify synthesis , the sync set/reset FF model as > > well. There are also compatibility problems between the timing checks in > > the sync. RAM models and the way SDF backannotation is done. > > Hi, > > Simulator: Verilog-XL 2.6.36 > My Synthesis tool: Synplify v5.21 > My P&R tool: Design Manager M1.5 > > I am using $XILINX/verilog/data for postsynthesis simulations and > $XILINX/verilog/src/simprims for postlayout (or post-route, according to your > terminology) simulations. This is the recommended method by Xilinx. You > can find this information at http://support.xilinx.com software manuals. > We have not observed any incompatibility between the gate-level and > postlayout-level. The only timing violations occur during reset insertion > but not any abnormal difference observed between RTL (synthesizable) > model and postlayout model. Target is XC40150XV. > > I think Austin is completely right since the postlayout simulations are > aggressively long. For example our minimal 20 ms. postlayout simulation > lasts 38 hours on a Sun Ultra-10/256MB/300 MHz UltraSPARC. Static > Timing Analyzer TRACE can give practicaly faster results. Despite that > we perform postlayout simulations. > > Utku > > -- > I feel better than James Brown. Yes in general I agree with this argument BUT the M1.5 s/w was sufficiently buggy both at the place and route and timing analysis levels for the Virtex parts that post route, or at least, post MAP simulation was essential. For example post synthesis simulation would not have found the wonderful MAP bug that silently threw away any inverter after a output FF that was placed in an IOB. This meant that my initial build had all the DRAM control signals inverted. The same bug affected TriFFs and also inverters placed before InFFs, so that the PCI output enables were also upside down. In terms of the simprims I'm not surpised they work for you since the models say something like: // Applicable version Verilog-XL 1.6c.5 (??) but at least the LUT ones do not work on ModelSim [Sorry about the obvious signs of poverty here but $60K for XL was a bit more than we could afford]. An example of the problem lies in the LUT2 simprim statement [approx] assign out = INIT[{a1,a0}] If the LUT is configured as an and gate with INIT = 4'b1000 and a1, a0 = 2'b0x then the output is `x' instead of `0'. What does XL produce under these conditions ? There is another problem caused by incompatibility between the sync set/reset model in UNIVIRTEX libs and the one in the simprims lib. This means that post-synth & post-route simulations can differ. Maybe this has all been fixed up in Synplify 5.1.5a+ & M2.1i. If so it has been done very silently.Article: 18193
Dave, I don't use this part, but I've seen lots of CMOS parts go into SCR latchup mode on powerup if any of their input protection diodes are foreward-biased. This happens if any i/o pin is pulled below ground or above Vcc. 'Above Vcc' can happen if the fpga is being driven by incoming logic inputs that pull up before the fpga supply is up. John ps - nice photos. * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free!Article: 18194
I already have the Xilinx JTAG cable and was wondering if anyone knows if I can use it from MAX+Plus II. I think The BitBlaster Cable is supposed to be a JTAG cable. CC :)Article: 18195
> In terms of the simprims I'm not surpised they work for you since the models say > something like: > > // Applicable version Verilog-XL 1.6c.5 (??) In 2.1i, the Verilog models are tested/sqa against the defined supported simulators at the time. Verilog-XL 2.7.5, Modelsim 5.2, and VCS 5.0. The above comment is no longer listed in the model header files. > but at least the LUT ones do not work on ModelSim [Sorry about the obvious signs > of poverty here but $60K for XL was a bit more than we could afford]. An example > of the problem lies in the LUT2 simprim statement [approx] > > assign out = INIT[{a1,a0}] > > If the LUT is configured as an and gate with INIT = 4'b1000 and a1, a0 = 2'b0x > then the output is `x' instead of `0'. > What does XL produce under these conditions ? This is fixed in 2.1i. All bugs fixed in 2.1i, were documented in a solution record. There was no silence involved. http://www.xilinx.com/techdocs/5270.htm http://www.xilinx.com/techdocs/5262.htm > There is another problem caused by incompatibility between the sync set/reset > model in UNIVIRTEX libs and > the one in the simprims lib. This means that post-synth & post-route simulations > can > differ. What type incompatibilty? -- Paulo //\\\\ | ~ ~ | ( O O ) __________________________________oOOo______( )_____oOOo_______ | . | | / 7\'7 Paulo Dutra (paulo@xilinx.com) | | \ \ ` Xilinx hotline@xilinx.com | | / / 2100 Logic Drive (800) 255-7778 | | \_\/.\ San Jose, California 95124-3450 USA | | Oooo | |________________________________________oooO______( )_________| ( ) ) / \ ( (_/ \_)Article: 18196
Hi, Any one tried to order the HOT II Development System from Virtual Computer Corporation ( www.vcc.com )? How long is need for waiting the PCI core netlist file from them after you sign the LogiCORE "PCI Master and Slave interface" License Agreement and fax it back? I had fax it back to Xilinx/VCC at 27/09/1999 but seem that isn't any response from them up to now (06/10/1999). Is there any problems? Roger Yau rogeryau@casilrd.com.hk CASIL R&D Co., Ltd.Article: 18197
> Yes in general I agree with this argument BUT the M1.5 s/w was sufficiently buggy > both at the place and route and timing analysis levels for the Virtex parts that > post route, or at least, post MAP simulation was essential. I believe you can still do unit delay simulation post route... I understand your (er, or at least the Xilinx) problem ;-)Article: 18198
Xilinx claims that Virtex is compliant with PCI 5V. Virtex is 3.3V IO, without 5V clamping diodes on its IO pins, which is required by PCI Spec. Somebody comment it? Sent via Deja.com http://www.deja.com/ Before you buy.Article: 18199
John, Thanks for your response, but I don't think this is quite the same. The problem persists even when the FPGA is the only part with power applied. Since it's the only device powered by 3.3V (the rest of the board is 5VB), it was a simple matter to power just the FPGA and leave the 5V shut down. Since the FPGA I/O isn't being driven by other circuits, I don't see how it could be a latchup issue. The limiting effect at 1.2V does seem like some kind of "zernering" effect, though. -Dave
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