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
Actually, the Viterbi algorithm is a very good demonstration of the power of an FPGA. Each level in the algorithm requires six adds or subtracts to be performed (four to produce the four possible path metrics, and two to compare the metrics from each node. The compare result's MSB operates a mux to select the survivor for each path. In a DSP processor, each of these has to be done independently and the results stored. In an FPGA, all of these operations can be done in parallel using separate adders and muxes. Note that registers in the FPGA can be used for storage too, so there is no need for external memory. You should have no problem fitting this in even a considerably smaller FPGA (4020 sized ought to do it). I think if you check the Xilinx alliance cores, there may even be a parameterizable viterbi core available. Greg Goslin, formerly of Xilinx did a paper that included a viterbi design a few years ago for designCon (96 I think?) You can probably find that paper on Xilinx's web site as well (do a search on viterbi on the xilinx site: http://www.xilinx.com ). I think his conclusion was that half of an FPGA (4020?) replaced an entire 66MHz DSP processor plus some external memory and outperformed it by a significant margin (x2 or x3) while drawing less power. Again, if your data rates are low enough, this can be implemented as a bit serial Kartheepan, Madasamy wrote: > Thank you for the last two informative emails. I definitely understand the need > to have a different design philosophy when tackling hardware implementations of > signal processing functions.A little off topic, I want to bounce this off you. > I have been thinking of putting Viterbi algorithm for finding MLS onto an > FPGA. Have you come across similar implementations and what is your take on the > feasibility of putting the algorithm handling 32 states (5 memory elements) on > a 100K FPGA?. I would appreciate any input from you, > > Thank you, > Kartheepan, M -- -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: 12326
ems@nospam.riverside-machines.com wrote: > > On Wed, 07 Oct 1998 11:31:23 -0700, Derek Palmer <derekp@synopsys.com> > wrote: > > >Let's make rumor fact! > >FPGA Express 3.0 will have VHDL 93 support. > >Available on the Synopsys web site at the end of Oct. > > great news, and nice to see synopsys here. however, the last > i heard from xilinx was that it would support "some" '93 > features; they weren't specific. can you tell us more? does > this mean (a) a normal synthesisable subset of '93, or > (b) limited support of '93? and what about VSS/DC? > > will the new version be shipped with foundation, or will > it be an (expensive) upgrade? > > evan I called Xilinx last night and asked your last question. I was told that version M1.5 includes FPGA Express 2.1.2 I believe and will later have a patch available to 2.1.3. But the main point was that you won't get VHDL '93 support until M2.0 which will not be shipping until early next year. So if you are a Foundation user, don't hold your breath. -- Rick Collins redsp@XYusa.net remove the XY to email me.Article: 12327
Steven K. Knapp wrote: > > At these "costs", the weak link is probably the design engineer. Hiring him > or her away (or the whole team) at a fraction of the cost would be the easy > solution. :) > > Also, if the devices are _really_ that secure, Actel should offer a big cash > prize to "crack" a secured device. A $250,000 prize should be risk-free for > Actel according to their arguments. Imagine the marketing messages they > could make then. I don't think they will ever do this. It comes under the heading of "It is better to remain silent and appear to be a fool than to speak and remove all doubt". As long as no one cracks it, they can claim this advantage. But if they offer a prize and someone does crack it... they have lost a lot more than the $1/4 mil. -- Rick Collins redsp@XYusa.net remove the XY to email me.Article: 12328
This is a multi-part message in MIME format. --------------C8C12CB4143F1A30C7089B57 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Ido Kleinman wrote: > Hi, > Anyone knows how do I activate VHDL93 compiling ability in MaxPlus2? > > I wrote some code with direct component instanciation that would compile > nice in Aldec, but won't in Maxplus... > > MaxPlus specifies many VHDL93 constructs in its help so I guess the VHDL93 > compiling ability exists. (I have the full software, no "unavailable > features"...) > > -- > > Ido Kleinman. > kleinn@mail.biu.ac.il Hi Ido, You just have to open the compiler window, be sure to have the compiler window active. Choose menu "Interfaces" -> and then i'm not so sure, but i think there must be something like "VHDL-Compiler -Settings" ....there you find a switch for VHDL 93 .... Hope this helps !!! --------------C8C12CB4143F1A30C7089B57 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Davor Lukacic Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Davor Lukacic n: Lukacic;Davor org: Robert BOSCH GmbH adr: Tuebinger Str. 123;;P.O. Box 1342;Reutlingen;;72703;Germany email;internet: dlukacic@eis.k8.rt.bosch.de title: Development of Integrated Circuits tel;work: ++49 7121 35 4430 x-mozilla-cpt: ;0 x-mozilla-html: FALSE end: vcard --------------C8C12CB4143F1A30C7089B57--Article: 12329
Hi: What is the best way to make delay in FPGA? Regards John Huang emmanuel jolly ¼¶¼g©ó¤å³¹ <361d1908.103204@news.hol.fr>... >LCell time depends on Component family and speed grade (10K, 10KA, >10KE, -4, -3, -2, -1). >It is not a good design rule to use LCELL to implement delay. Use it >only if you have no other solution > >Regards > >On Fri, 9 Oct 1998 01:08:15 +0800, "John Huang" ><hungi@tpts4.seed.net.tw> wrote: > >>Hi all: >> >> Does anyone know the delay time of one LCELL, >>if I implement an VHDL statement >> >> U1: LCELL(CLK1, CLK2); >> >> the CLK2 will be delay how many ns? >> >> John Huang >> >> > >--- >Pour me contacter enlever "removethis." de mon adresse Email. >To contact me, remove "removethis." from my Email address.Article: 12330
Sorry i know i am out of topic. This will be the first and once time.Thanks to all other PC-user for your understandling. Bye and great thanks LUXEMBOURG: DEMOCRACY OR POLICE STATE As an italian citizen born in Luxembourg and still married to a luxembourg woman, I am raising questions to myself every day. In January this year my wife left our home.The day after 4 police officers (Gendarmerie Grand-Ducale) entered in my home and picked up my 4 children (Julie,Mélissa,Andy and Laurent) in order to bring them to their mother. This was done without any written protocol, or either without any written aproval from any court.In South-America and Eastern-Europe a few years ago some dictatures was the objective of our medias for these same police activities. I would like to believe that in Luxembourg we still have democracy order and not same destination. Regards JeanArticle: 12331
Have you tried the 74ACT161 - clocks at >100MHz. Synchronous. Don't know about skew. Simon ================================= "Thomas Dölle" <doelle@fb.sony.de> wrote: >Hi, > >I'm looking for high speed (up to 80 MHz input) clock divider chips with > >programmable ratio (1/2, 1/4, 1/8). The outputs should be at CLK and >CLK/X with low skew (100 - 1000ps) and low jitter. The input is driven >by a external clock source. Does anybody know whether there are chips >commercially available? > >Thomas > > > > > > Design Your Own MicroProcessor(tm) http://www.tefbbs.com/spacetime/index.htmArticle: 12332
Hi, I'm looking for high speed (up to 80 MHz input) clock divider chips with programmable ratio (1/2, 1/4, 1/8). The outputs should be at CLK and CLK/X with low skew (100 - 1000ps) and low jitter. The input is driven by a external clock source. Does anybody know whether there are chips commercially available? ThomasArticle: 12333
Error in the URL - revised below: >I have a nifty - I think - tool for designing cores. > >Check out: > Design Your Own MicroProcessor(tm) http://www.tefbbs.com/spacetime/index.htmArticle: 12334
Hi, Open the Compiler window. Go to the "Interfaces" -> "VHDL Netlist Reader Settings" menu. Check VHDL 93 and accpet. Regards, Adam Biniszkiewicz -------------------------------------------- Adam Biniszkiewicz Technical University of Zielona Gora, Poland e-mail: A.Biniszkiewicz@pz.zgora.pl -------------------------------------------- On Fri, 9 Oct 1998 01:38:30 +0200, "Ido Kleinman" <kleinn@mail.biu.ac.il> wrote: >Hi, >Anyone knows how do I activate VHDL93 compiling ability in MaxPlus2? > >I wrote some code with direct component instanciation that would compile >nice in Aldec, but won't in Maxplus... > >MaxPlus specifies many VHDL93 constructs in its help so I guess the VHDL93 >compiling ability exists. (I have the full software, no "unavailable >features"...) > >-- > > > Ido Kleinman. > kleinn@mail.biu.ac.il > > > >Article: 12335
Hi, 1. Open the Compiler window. 2. Go to the "Interfaces" - > "VHDL Netlist Reader Settings" menu. 3. Check VHDL 1993 and accept. Regards, Adam Biniszkiewicz -------------------------------------------- Adam Biniszkiewicz Technical University of Zielona Gora, Poland e-mail: A.Biniszkiewicz@pz.zgora.pl --------------------------------------------Article: 12336
On Thu, 8 Oct 1998 12:11:10 -0700, "Andy Peters" <apeters@noao.edu.NOSPAM> wrote: > >1) Supposedly there's a way to keep revs of source code with implementation >revs but I can't figure out how to do it. > We received our copy of Foundation 1.5 this very morning, and quite out-of-character had a look through the bright yellow 'Read Me First' leaflet inside whilst consuming my second cup of tea of the day. Under Known Issues it says that the revision control mechanism only handles implementation data, and that you have to manually handle your source code/netlists if you want to keep them. I guess this answers your question 1.....it's a crying shame they didn't include source control in the Project Manager, along with implementation revision management. Unless you are scrupulously careful with versions and revisions etc., you can come very unstuck later, as I'm sure most of us have discovered at one time or another! Sorry, no answers yet to your other points....I need to spend some time with it myself first! Graeme Durant HELION Technology Limited Programmable Logic Design Specialists The Granary, Home End, Fulbourn, Cambridge CB1 5BS, UK. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ mailto:gd@heliontech.com http://www.heliontech.com XILINX CERTIFIED GOLD LEVEL DESIGN CONSULTANT ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Article: 12337
To turn on VHDL93 support in MaxPlus2, do the following: 1) Bring up the compiler window 2) Go to the Interfaces-VHDL Netlist Reader menu pulldown 3) In the dialog box, click the VHDL93 button and click OK That's all there is to it ! Ido Kleinman wrote: > Hi, > Anyone knows how do I activate VHDL93 compiling ability in MaxPlus2? > > I wrote some code with direct component instanciation that would compile > nice in Aldec, but won't in Maxplus... > > MaxPlus specifies many VHDL93 constructs in its help so I guess the VHDL93 > compiling ability exists. (I have the full software, no "unavailable > features"...) > > -- > > Ido Kleinman. > kleinn@mail.biu.ac.ilArticle: 12338
Anyone interested in learning about what a truly embedded IrDA infrared software protocol stack is, should check out: http://www.embednet.com This stack is: . modular and written in "clean" C. . small (<10K bytes) . efficient ( Zero copy buffer management ) . able to support speeds from 9.6Kbps to 4Mbps . RTOS ready . simple to port . well tested. . supported on all popular platforms: MIPS, ARM, Motorola, Intel, Hitachi SH, etc ... . (not to forget) inexpensive (<$6k)Article: 12339
Is the Call for Papers for FCCM 99 already out? If so, where can I find it? Thanks, Christof -- ************************************************************************* Christof Paar http://ee.wpi.edu/People/faculty/cxp.html Assistant Professor email: christof@ece.wpi.edu Cryptography Group phone: (508) 831 5061 ECE Department, WPI fax: (508) 831 5491 100 Institute Road Worcester, MA 01609, USA *************************************************************************Article: 12340
Hello, I'm using Active-VHDL now and I like it a lot. Very nice GUI and loaded with features. You can always request a CD or have your computer running overnight... It worth the download. Son Huynh Ido Kleinman wrote in message <6vjffo$nir$1@news.inter.net.il>... >Dear all, > >I am looking for a good VHDL development (compiling, simulation, gate-level >sim/FSM support, Testbench generator) software tool. I need it to have a >comfortable interface to maxplus2 for exporting compiled designs for >synthesis in Altera's devices. >I've been looking around lately and I've evaluated Aldec's "Active-VHDL" and >quite happy with it, but I have a slow internet connection, therefore, >before I start downloading tens of megabytes again, I would like to know if >there are any other compact software tools worth looking at...? > >Anyone got any experience with Aldec's tool? >Are the big ones (Synopsys, Examplar..) worth the investment? > > >-- > > > Ido Kleinman. > kleinn@mail.biu.ac.il > > > >Article: 12341
I got a great note yesterday from David Dye of Xilinx answering my comments. Graeme Durant wrote in message <361df931.81804368@news.force9.net>... re: source code revs w/implementation revs. >I guess this answers your question 1.....it's a crying shame they >didn't include source control in the Project Manager, along with >implementation revision management. Unless you are scrupulously >careful with versions and revisions etc., you can come very unstuck >later, as I'm sure most of us have discovered at one time or another! David says that it's going to be in the "next full release." >Sorry, no answers yet to your other points....I need to spend some >time with it myself first! They're also working w/Synopsys to reduce the amount of timing constraints exported to the .XNF file, and also trying to have one constraints editor. It turns out that the constraints editor needs information that's generated by the Translator, so you have to do a translate, then kill the implementation, the import the necessary files into the constraints editor. The constraints editor is not included in Project Manager because it was decided late in the game to add it. Hey, as long as it's there, I'm happy... There's a patch to the constraints editor up on the Xilinx Web site. -a -- Andy Peters Sr. Electrical Engineer National Optical Astronomy Observatories 950 N Cherry Ave Tucson, AZ 85719 520-318-8191 apeters@noao.eduArticle: 12342
The Call for Papers for FCCM 99 is not out as far as I know (and the program committee usually gets it first). It usually comes out pretty late, and the due date is normally in the 1st week of January. Scott In article <6vl9ea$3fp$2@bigboote.WPI.EDU>, christof@goya.WPI.EDU (Christof Paar) wrote: >Is the Call for Papers for FCCM 99 already out? If so, where can I find it? > >Thanks, > >Christof >Article: 12343
Rickman wrote in message <361D7017.200058C5@yahoo.com>... >I called Xilinx last night and asked your last question. I was told that >version M1.5 includes FPGA Express 2.1.2 I believe and will later have a >patch available to 2.1.3. > >But the main point was that you won't get VHDL '93 support until M2.0 >which will not be shipping until early next year. Won't that be when VHDL 2000 is out? Oh, wait, that's Office 2000. Silly me. -- Andy Peters Sr. Electrical Engineer National Optical Astronomy Observatories 950 N Cherry Ave Tucson, AZ 85719 520-318-8191 apeters@noao.eduArticle: 12344
Christof Paar wrote: > Is the Call for Papers for FCCM 99 already out? If so, where can I find it? > > Thanks, > > Christof > > -- > ************************************************************************* > Christof Paar http://ee.wpi.edu/People/faculty/cxp.html > Assistant Professor email: christof@ece.wpi.edu > Cryptography Group phone: (508) 831 5061 > ECE Department, WPI fax: (508) 831 5491 > 100 Institute Road > Worcester, MA 01609, USA > ************************************************************************* I don't think it's out yet. The fccm.org page (i have a link to it from my page [lab staff]) is not updated yet. regards -martinArticle: 12345
How can I implement this very simple dual port RAM using genmem to embed it on a Flex10K device. Altera memory type "csfifo" is a "cycle-shared" FIFO (what is that?) *************************************************************** module fram2_8 ( clk, write_n, // Write enable write_ptr, // Write Address read_ptr, // Read Address data_in, // Write Data data_out // Read Data . . . always @(read_ptr or reg0 or reg1 or reg2 or reg3 or reg4 or reg5 or reg6 or reg7) casez (read_ptr) 3'b000 : data_out = reg0; 3'b001 : data_out = reg1; 3'b010 : data_out = reg2; 3'b011 : data_out = reg3; 3'b100 : data_out = reg4; 3'b101 : data_out = reg5; 3'b110 : data_out = reg6; 3'b111 : data_out = reg7; endcase // Update RAM on a write pulse always @(posedge clk) if( !write_n) casez (write_ptr) 3'b000 : reg0 <= data_in; 3'b001 : reg1 <= data_in; 3'b010 : reg2 <= data_in; 3'b011 : reg3 <= data_in; 3'b100 : reg4 <= data_in; 3'b101 : reg5 <= data_in; 3'b110 : reg6 <= data_in; 3'b111 : reg7 <= data_in; endcase endmodule *********************************************************************Article: 12346
Hi, I have not heard of any but we had done one way back using a microcontroller. Do let me know if you ultimately land up looking for guys who can program a microcontroller. My company offers such services, we shall identify an appropriate microcontroller, program it and give you some prototypes. If they work you can tell us how many you need. We shall get them for you. Visit our web-site at http://www.spiketech.com - Sri (sri@spiketech.com) Thomas Dölle <doelle@fb.sony.de> wrote in article <361DD1B3.4EFC8575@fb.sony.de>... > Hi, > > I'm looking for high speed (up to 80 MHz input) clock divider chips with > > programmable ratio (1/2, 1/4, 1/8). The outputs should be at CLK and > CLK/X with low skew (100 - 1000ps) and low jitter. The input is driven > by a external clock source. Does anybody know whether there are chips > commercially available? > > Thomas > > > > > > >Article: 12347
One method: Run the signal off-chip into a fast buffer, then through an RC delay, then buffer it again with another fast buffer, and run it back into your chip. It would be nice if the RC part of the delay is large compared to the sum of the clock-->output delay, the pin buffer I/O delays, and the buffer delays. You could consider getting rid of the fast buffer if your R is large. The best method would be to get rid of the delay requirement if at all possible. If it's not possible, your synchronous design is entering the nasty asynchronous world where you have to actually think carefully about what you are doing. John Huang <hungi@tpts4.seed.net.tw> wrote in article <6vkgsh$jok$1@news.seed.net.tw>... > Hi: > > What is the best way to make delay in FPGA? > > Regards > > John Huang > > emmanuel jolly ¼¶¼g©ó¤å³¹ <361d1908.103204@news.hol.fr>... > >LCell time depends on Component family and speed grade (10K, 10KA, > >10KE, -4, -3, -2, -1). > >It is not a good design rule to use LCELL to implement delay. Use it > >only if you have no other solution > > > >Regards > > > >On Fri, 9 Oct 1998 01:08:15 +0800, "John Huang" > ><hungi@tpts4.seed.net.tw> wrote: > > > >>Hi all: > >> > >> Does anyone know the delay time of one LCELL, > >>if I implement an VHDL statement > >> > >> U1: LCELL(CLK1, CLK2); > >> > >> the CLK2 will be delay how many ns? > >> > >> John Huang > >> > >> > > > >--- > >Pour me contacter enlever "removethis." de mon adresse Email. > >To contact me, remove "removethis." from my Email address. > > >Article: 12348
> . RTOS ready You sent this to a Linux group but did not mention if it's available as a Linux module.Article: 12349
I have met someone from a major ASIC house who described to me how firms in Hong Kong have reverse engineered their clients' (digital) ASICs back to a netlist of gates (!!) and done it reportedly for $25k; this cost was for a 10k gate device but they can do much bigger ones since the process is quite straightforward, for a given ASIC vendor. One gets a hex listing of the ROM of an on-chip micro, or a normal masked micro, for a similar amount. This firm offers 8051 cores and they had this done to them. All the above is done purely optically. I don't know if an antifuse is optically visible, but Neocad obviously did not spend $300M reverse engineering Xilinx's bitstream, and all the "secrets" of every FPGA's SRAM/antifuse mapping are in any case revealed when you disassemble their place & route tools. -- Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.
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