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
"Pawe³ J. Rajda" wrote: > > Does anyone has an idea how to quickly count number of 1's (or 0's) > in a word (i.e. 8 or 24 bits). I have to implement this as a part of > algorithm in FPGA. I would: Assign the word vector to a variable Loop the length of the variable incrementing an unsigned count for '1' Assign the count to an unsigned signal. -Mike TreselerArticle: 20226
In article <s9dng6smer2121@corp.supernews.com>, Larry Eisner <leisner@bryceusa.com> wrote: >Does anyone know of a PCI core which is the public domain. My specific needs >are for a simplified target (non-master) version only. All the licensed >designs I have found are full blown PCI, with associated full blown license >fees. > Larry, Altera Corp provides a target-only 32/33 core and variants thereof. I don't know the price, but it might be worth looking into. www.opencores.org is working on one. http://www.tkt.cs.tut.fi/~havu/ has some VHDL source, but looking at the config register area there are some holes. -steenArticle: 20227
Use a carry save (also known as 3 to 2 counters) adder tree. Josh "Pawe³ J. Rajda" wrote: > > Does anyone has an idea how to quiclky count number of 1's (or 0's) > in a word (i.e. 8 or 24 bits). I have to implement this as a part of > algorithm in FPGA. > > -- > Regards, > Pawel J. Rajda > > ----------------------------------------------------------------------------- > > Pawel J. Rajda, MSc. E.E. mail: pjrajda@uci.agh.edu.pl > Dept. of Electronic Engineering www: > http://galaxy.uci.agh.edu.pl/~pjrajda > AGH Technical University tel: (+48-12) 617 3980 > Al. Mickiewicza 30 fax: (+48-12) 633 2398 > 30-059 Cracow, POLAND > -----------------------------------------------------------------------------Article: 20228
Seeing thiis whole thread is about PLL's etc. I have a simple question? Given a Manchester encoded input at 125Mhz. Can one use a Zilinx PLL, DPLL or whatever to generate a phase sync'd clock at 125Mhz to extract the data? Thanks Anthony Don Husby wrote in message <8746qh$spc$1@info3.fnal.gov>... >Ray Andraka <randraka@ids.net> wrote: >> I hadn't considered a VCO made as a ring oscillator. > >Take a look at the Lucent 3T parts. Their clock synchronizer >can be used in either DLL or PLL mode. It's implemented as a >multi-tap voltage-controlled delay line. > > >-- >Don Husby <husby@fnal.gov> http://www-ese.fnal.gov/people/husby >Fermi National Accelerator Lab Phone: 630-840-3668 >Batavia, IL 60510 Fax: 630-840-5406Article: 20229
F1.4 is two major revisions old. Apparently you didn't buy the yearly maintenance, as that gets you the upgrades when they come out. It also buys you the lukewarm line support, which compared to others is actually pretty good (I should know, I seem to find all the bugs in the tools). Keith Wootten wrote: > Hi > > I've been using Foundation F1.4 for a while, using XC5215 and Spartan > XCS40. I want to change to the 3.3V XCS40XL part, but my software won't > support this part. > > Apparently, there is no upgrade path for F1.4 and I'll have to *buy* > some new software. To cope with both the XC5215 and the XCS40XL parts, > I'll need to spend over GBP1000 - yes, one kilopound. I already spent > over GBP2000 for the F1.4 stuff, and I'm not a volume user. ^^^^^^ you might say you spent a ton on it :-) (a US ton is 2000 pounds) > > > Why do they do this? Surely the small company user has *some* value? > The support from the dealer was poor and the promised training sessions > never materialised once the money was paid, so cost of support is no > justification. > > Can anyone recommend a UK dealer who is not a shark? > > Cheers > -- > Keith Wootten -- -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: 20230
The 'best' implementation for FPGAs depends on the amount of time you have to do it and how many bits you are counting. For relatively small numbers of bits, use the 4LUTs to create partial sums of the individual bits. the 4 LUT outputs will be weighted. YOu combine like weighted outputs until you are down to 2 of every weight, then add those vectors together in a conventional adder. That's a merged tree implementation. Note that most merged tree work in the literature deals with gates with a fanin and fanout of 2. The FPGA LUTs generally have a fan-in of 4, so for the most efficient merged tree, you need to modify the approach slightly. For larger numbers of bits, you can take the odd bits, shift them down by one and add them to the even bits. That will get you N/2 values of 0,1 or 2, but you get to use the carry chain to cut down on the real-estate. Then take every other two bit output, shift it down two bits and add to the even sets. Keep repeating that till you have one set. If you have a clock that is n times faster than the data rate (not likely in your case based on your problem statement), you can use a shift register and counter (but that was obvious. huh?) "Pawe³ J. Rajda" wrote: > Does anyone has an idea how to quiclky count number of 1's (or 0's) > in a word (i.e. 8 or 24 bits). I have to implement this as a part of > algorithm in FPGA. > > -- > Regards, > Pawel J. Rajda > > ----------------------------------------------------------------------------- > > Pawel J. Rajda, MSc. E.E. mail: pjrajda@uci.agh.edu.pl > Dept. of Electronic Engineering www: > http://galaxy.uci.agh.edu.pl/~pjrajda > AGH Technical University tel: (+48-12) 617 3980 > Al. Mickiewicza 30 fax: (+48-12) 633 2398 > 30-059 Cracow, POLAND > ----------------------------------------------------------------------------- -- -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: 20231
This is a multi-part message in MIME format. --------------0ED228255F832848066B58B3 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Traditionally: You could start with a 2 bit 1-counter, which is called a half-adder. ___ A---| H |--Carry B---|_A_|--Sum Next, use two of those and an or gate to build a 3 bit 1-counter, called a full-adder. ___ ___ |-----------| O | A---| H |--| ___ |--|_R_|---S1 B---|_A_|---| H |---| C-----------|_A_|--------------S0 Next, take two full adders to add up six bits, plus two more full adders to add up the output from the first two full adders plus another bit, and you have a 7 bit counter: ___ A--| F | B--| A |----------| C--|___|-----| | ___ | |-------| F | ___ |- | -----------| A |---S2 D--| F | | | ___ |--|___|---S1 E--| A |--| |--| F | | F--|___|--------| A |--| G---------------|___|-------------S0 Next, A pair of seven-bit counters, and three more full adders to add up the output from the seven-bit counters plus another bit, and you have a 15-bit 1-counter. Next, ... For FPGAs: If it is going into FPGAs, which commonly use 4-LUTs as basic logic element, you might start at the bottom with a 4 bit 1-counter, use two of those feeding into an (optimum for your FPGA) two input, 3-bit adder to make an 8 bit 1-counter. Generally: The end stages of bit counting algorithms are usually adders, which sum up the results of different sections of the word being counted. So optimising bit counting is usually the same as optimizing adders. Look into adder optimizations. "Pawe³ J. Rajda" wrote: > > Does anyone has an idea how to quiclky count number of 1's (or 0's) > in a word (i.e. 8 or 24 bits). I have to implement this as a part of > algorithm in FPGA. > > -- > Regards, > Pawel J. Rajda > > ----------------------------------------------------------------------------- > > Pawel J. Rajda, MSc. E.E. mail: pjrajda@uci.agh.edu.pl > Dept. of Electronic Engineering www: > http://galaxy.uci.agh.edu.pl/~pjrajda > AGH Technical University tel: (+48-12) 617 3980 > Al. Mickiewicza 30 fax: (+48-12) 633 2398 > 30-059 Cracow, POLAND > ----------------------------------------------------------------------------- --------------0ED228255F832848066B58B3 Content-Type: text/x-vcard; charset=us-ascii; name="jsmith.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for John L. Smith Content-Disposition: attachment; filename="jsmith.vcf" begin:vcard n:Smith;John L. tel;work:858-320-4102 x-mozilla-html:FALSE url:http://www.visicom.com org:Visicom;Imaging Products adr:;;10052 Mesa Ridge Court;San Diego;CA;92121;USA version:2.1 email;internet:jsmith@visicom.com title:Principal Engineer x-mozilla-cpt:;27888 fn:John end:vcard --------------0ED228255F832848066B58B3--Article: 20232
Hello Xue Zhong, also notice that Lattice will be soon releasing their FPGA Godfather family in Q1 - 2000. www.latticesemi.com regards, Michel WMU "#XUE ZHONG#" <P144850500@ntu.edu.sg> wrote in message news:0CF260C495FED111A6610000F866308D09C8C10E@mail3.ntu.edu.sg... > Hi folks, > > In face, I am familiar with Lattice's ispPLD chip 1000, 2000, 3000 > series: > Such as 1016, 2032, etc. including there structures, and programming. > > How about the difference between FPGA and the same. Where can I find > (website) > the information about FPGA, and learn it quickly, I need to develop > circurts with FPGA > in the near future. > > Thanks. >Article: 20233
Something these FPGA companies (who I know monitor this list) should make a point of remembering that those of use that work at small companies now, won't ALWAYS be working at small companies (either because our small companies go big time or we go to another larger company). I, for one, will remember which companies treated me well and which ones didn't. -- Keith F. Jasinski, Jr. kfjasins@execpc.com Keith Wootten <Keith@wootten.demon.co.uk> wrote in message news:Fjn8RCAvgul4EwTT@wootten.demon.co.uk... > Hi > > I've been using Foundation F1.4 for a while, using XC5215 and Spartan > XCS40. I want to change to the 3.3V XCS40XL part, but my software won't > support this part. > > Apparently, there is no upgrade path for F1.4 and I'll have to *buy* > some new software. To cope with both the XC5215 and the XCS40XL parts, > I'll need to spend over GBP1000 - yes, one kilopound. I already spent > over GBP2000 for the F1.4 stuff, and I'm not a volume user. > > Why do they do this? Surely the small company user has *some* value? > The support from the dealer was poor and the promised training sessions > never materialised once the money was paid, so cost of support is no > justification. > > Can anyone recommend a UK dealer who is not a shark? > > Cheers > -- > Keith WoottenArticle: 20234
Hello: I have read XAPP158 (Powering Virtex FPGAs http://www.xilinx.com/xapp/xapp158.pdf) and found the following recommendations: VCCINT -> Guideline ------------------- 0.1 uF -> One per VCC 47 uF -> Four per device (XCV50 - XCV300) 470 uF -> One per device 470uF!!! In my experience, this seems excessive, even after reading the disclaimer in Answers record #777 (http://support.xilinx.com/techdocs/777.htm)... I was planning on following the guidelines sans 470uF cap. It's a low power design in a XCV100 running at mostly 4 to 32 MHz... Can others comment on power related issues in Virtex parts, particularly related to decoupling and startup??? Thanks... JohnArticle: 20235
Please help to give any info for the best verilog book for self-study. T.I.AArticle: 20236
Keith, It cannot possibly cost you that much to upgrade your tools, a new Foundation Base Express package costs $495 USD. DS-FND-BSX-PC is the P/N that you want. Keith Jasinski, Jr. wrote in message ... >Something these FPGA companies (who I know monitor this list) should make a >point of remembering that those of use that work at small companies now, >won't ALWAYS be working at small companies (either because our small >companies go big time or we go to another larger company). I, for one, will >remember which companies treated me well and which ones didn't. > >-- >Keith F. Jasinski, Jr. >kfjasins@execpc.com >Keith Wootten <Keith@wootten.demon.co.uk> wrote in message >news:Fjn8RCAvgul4EwTT@wootten.demon.co.uk... >> Hi >> >> I've been using Foundation F1.4 for a while, using XC5215 and Spartan >> XCS40. I want to change to the 3.3V XCS40XL part, but my software won't >> support this part. >> >> Apparently, there is no upgrade path for F1.4 and I'll have to *buy* >> some new software. To cope with both the XC5215 and the XCS40XL parts, >> I'll need to spend over GBP1000 - yes, one kilopound. I already spent >> over GBP2000 for the F1.4 stuff, and I'm not a volume user. >> >> Why do they do this? Surely the small company user has *some* value? >> The support from the dealer was poor and the promised training sessions >> never materialised once the money was paid, so cost of support is no >> justification. >> >> Can anyone recommend a UK dealer who is not a shark? >> >> Cheers >> -- >> Keith Wootten > >Article: 20237
Dave Vanden Bout <devb@xess.com> wrote in message news:389589F9.27B3C708@xess.com... > Our software for testing, programming, and debugging the XS40 Board runs under Win95/98/NT in both a DOS window and using the GUI. You can read more about that at http://www.xess.com/gxstools-v3_0.pdf. Cool, that's really great. Now if only the XS40 had a way to get more data back into the PC...Article: 20238
I have the Xilinx Student Edition Foundation 1.5, which includes a VHDL and Verilog compiler. I want to drive this from the command line, using a makefile (eventually using wine). The fndmake package from Xess has the command line params for the hitop, plus6, ndgbuild commands, which is fine, but for the compiler includes Tcl code for fe_shell, which is not in the student edition. The 'command' is probably something like dpmcomp. Thanks in advance, JamieArticle: 20239
Back in the days when fpga's were too small to be full masters with dma etc Xilinx gave it all away. I nearly used it recently and it looked very good. ftp://ftp.xilinx.com/pub/applications/pci Sent via Deja.com http://www.deja.com/ Before you buy.Article: 20240
On Sun, 30 Jan 2000 20:06:15 -0500, "B. Joshua Rosen" <bjrosen@polybus.com> wrote: >I've just put up a HowTo page for running Xilinx place and route tools under Linux. > >http://www.polybus.com/xilinx_on_linux.html Very nice work, but what about the overhead of running the windows emulator? Is this significant? Thanks EvanArticle: 20241
smithers12@my-deja.com wrote: > I was wondering if anyone has had any experience with the various > Virtex-based prototyping boards out on the market (e.g., Avnet, VCC, > etc.). Any recommendations would be appreciated. Thanks. > We've just bought in a Virtex developement board from VCC. I've not had much chance to use it yet but we've powered it up and downloaded a design (via Multilinx Cable) and it seemed to work fine. I know that Nallatech (www.nallatech.com) also produce a board, I think it's a PCI type board. Cheers DavidArticle: 20242
http://freecore.com has now been fully restored. Rune BaeverrudArticle: 20243
My comments did not refer to any particular manufacturer. -- Keith F. Jasinski, Jr. kfjasins@execpc.com Kirk Saban <ksaban@home.com> wrote in message news:F0Ol4.32585$up4.565141@news1.rdc1.ab.home.com... > Keith, > > It cannot possibly cost you that much to upgrade your tools, a new > Foundation Base Express package costs $495 USD. > > DS-FND-BSX-PC is the P/N that you want. > > > > Keith Jasinski, Jr. wrote in message ... > >Something these FPGA companies (who I know monitor this list) should make a > >point of remembering that those of use that work at small companies now, > >won't ALWAYS be working at small companies (either because our small > >companies go big time or we go to another larger company). I, for one, > will > >remember which companies treated me well and which ones didn't. > > > >-- > >Keith F. Jasinski, Jr. > >kfjasins@execpc.com > >Keith Wootten <Keith@wootten.demon.co.uk> wrote in message > >news:Fjn8RCAvgul4EwTT@wootten.demon.co.uk... > >> Hi > >> > >> I've been using Foundation F1.4 for a while, using XC5215 and Spartan > >> XCS40. I want to change to the 3.3V XCS40XL part, but my software won't > >> support this part. > >> > >> Apparently, there is no upgrade path for F1.4 and I'll have to *buy* > >> some new software. To cope with both the XC5215 and the XCS40XL parts, > >> I'll need to spend over GBP1000 - yes, one kilopound. I already spent > >> over GBP2000 for the F1.4 stuff, and I'm not a volume user. > >> > >> Why do they do this? Surely the small company user has *some* value? > >> The support from the dealer was poor and the promised training sessions > >> never materialised once the money was paid, so cost of support is no > >> justification. > >> > >> Can anyone recommend a UK dealer who is not a shark? > >> > >> Cheers > >> -- > >> Keith Wootten > > > > > >Article: 20244
In comp.emulators.ms-windows.wine eml@riverside-machines.com.NOSPAM wrote: : On Sun, 30 Jan 2000 20:06:15 -0500, "B. Joshua Rosen" : <bjrosen@polybus.com> wrote: :>I've just put up a HowTo page for running Xilinx place and route tools under Linux. :> :>http://www.polybus.com/xilinx_on_linux.html : Very nice work, but what about the overhead of running the windows : emulator? Is this significant? W_ine I_s N_ot an E_mulator. Wine is a reimplementation of the Windows Api with a loader for window programs. So the code in your programm runs as fast as in Window, only calls from your program to the windows API are treated different. They may be slower, but must not. Bye -- Uwe Bonnes bon@elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------Article: 20245
This is a multi-part message in MIME format. --------------25755396617441B1751D0F3B Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Joel Kolstad wrote: > Dave Vanden Bout <devb@xess.com> wrote in message > news:389589F9.27B3C708@xess.com... > > Our software for testing, programming, and debugging the XS40 Board runs > under Win95/98/NT in both a DOS window and using the GUI. You can read more > about that at http://www.xess.com/gxstools-v3_0.pdf. > > Cool, that's really great. Now if only the XS40 had a way to get more data > back into the PC... Data goes from the XS40 to the PC through the five parallel port status lines. You can use another six data lines if you have a bidirectional parallel port. That's not enough for many applications, but the XS40 is not intended to operate as a coprocessor to the PC. For that, you need to select one of the many fine PCI-based boards listed at the optimagic site. --------------25755396617441B1751D0F3B Content-Type: text/x-vcard; charset=us-ascii; name="devb.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Dave Vanden Bout Content-Disposition: attachment; filename="devb.vcf" begin:vcard n:Vanden Bout;David tel;fax:(919) 387-1302 tel;work:(919) 387-0076 x-mozilla-html:FALSE url:http://www.xess.com org:XESS Corp. adr:;;2608 Sweetgum Drive;Apex;NC;27502;USA version:2.1 email;internet:devb@xess.com title:FPGA Product Manager x-mozilla-cpt:;28560 fn:Dave Vanden Bout end:vcard --------------25755396617441B1751D0F3B--Article: 20246
This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------ =_NextPart_000_01BF6DF3.79EA22F4 Content-Type: text/plain Joel Kolstad wrote: > Dave Vanden Bout <devb@xess.com> wrote in message > news:389589F9.27B3C708@xess.com... > > Our software for testing, programming, and debugging the XS40 Board runs > under Win95/98/NT in both a DOS window and using the GUI. You can read more > about that at http://www.xess.com/gxstools-v3_0.pdf. > > Cool, that's really great. Now if only the XS40 had a way to get more data > back into the PC... Data goes from the XS40 to the PC through the five parallel port status lines. You can use another six data lines if you have a bidirectional parallel port. That's not enough for many applications, but the XS40 is not intended to operate as a coprocessor to the PC. For that, you need to select one of the many fine PCI-based boards listed at the optimagic site. ------ =_NextPart_000_01BF6DF3.79EA22F4 Content-Type: text/x-vcard; name="devb.vcf" Content-Disposition: attachment; filename="devb.vcf" Content-Description: Card for Dave Vanden Bout begin:vcard n:Vanden Bout;David tel;fax:(919) 387-1302 tel;work:(919) 387-0076 x-mozilla-html:FALSE url:http://www.xess.com org:XESS Corp. adr:;;2608 Sweetgum Drive;Apex;NC;27502;USA version:2.1 email;internet:devb@xess.com title:FPGA Product Manager x-mozilla-cpt:;28560 fn:Dave Vanden Bout end:vcard ------ =_NextPart_000_01BF6DF3.79EA22F4--Article: 20247
Hi. Using Xilinx Foundation series 2.1i, and I want to build a program with tree states to XC9536 in the programming language Abel via State Editor. In state number one outport q1 activates and all the others are deactivated. In state no two outport q2 activate and q1, q2 is activated and in the last state is only q3 activated. When I program XC9536 with the code in the end of this message, all outports (q1, q2, q3) are activated. What's wrong? Thankful for help Björn Lindegren " " File: E:\PROGRAM\XFS2.1\ACTIVE\PROJECTS\95360130\delay.abl " created: 02/02/00 16:31:30 " from: 'E:\PROGRAM\XFS2.1\ACTIVE\PROJECTS\95360130\delay.asf' " by: fsm2hdl - version: 2.0.1.53 " module delay Title 'delay' Declarations "clocks CLK PIN 43; "input ports "output ports q1 PIN 22; q2 PIN 21; q3 PIN 13; "******** SYMBOLIC state machine: Sreg0 ****** Sreg0 STATE_REGISTER; S1, S2, S3 STATE; "diagram ACTIONS "************* state machine: Sreg0 ************* Equations " clock signals definitions Sreg0.clk = CLK; State_diagram Sreg0 State S1: q1=1; q2=0; q3=0; x=x+1; IF ('x<1000') THEN S1 ELSE IF ('x=1000') THEN S2; State S2: q1=0; q2=1; q3=0; x=x+1; IF ('x<2000') THEN S2 ELSE IF ('x=2000') THEN S3; State S3: q1=0; q2=0; q3=1; x=x+1; IF ('x<3000') THEN S3 ELSE IF ('x=3000') THEN S1 WITH x=1; ENDWITH; " end of state machine - Sreg0 end delayArticle: 20248
Hi people, I'm looking for some informations about FPGA, DPGA and TSFPGA. If someone could help me, mail to: anoriaki@comp.ufscar.br thanks, Nori. Sent via Deja.com http://www.deja.com/ Before you buy.Article: 20249
"Keith Jasinski, Jr." <jasinski@mortara.com> wrote: >Something these FPGA companies (who I know monitor this list) should make a >point of remembering that those of use that work at small companies now, >won't ALWAYS be working at small companies (either because our small >companies go big time or we go to another larger company). I, for one, will >remember which companies treated me well and which ones didn't. Me too. For my own part as a one man business I think Xilinx are generally better than average. The support is generally good and (running Foundation Base Express) I think the software is good value for money. I didn't see the start of the thread but there are said to be ways of getting discounts on software from Xilinx, for example by taking part in their consultant programs, or dropping heavy hints about the volumes of the final sales of the design. One of the problems that the companies who provide poor support for small businesses face, is that they do not know who the final customer might be. For example my current design will ultimately be used by a major UK telecoms operator. And on the next project I'll happily use Xilinx but there will be a good deal more soul-searching regarding TI DSPs. :( Richard ------------Richard Dungan------------- Radix Electronic Designs, Orpington, UK Spamtrapped: Remove the XY ---------------------------------------
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