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
Hi, beeing a bloody newcomer in the FPGA area, can anyone tell me how much effort it is to learn & use both PCI and PCMCIA (once they are available) FPGA cards in a software package, provided I don't have to write the FPGA parts myself (``only'' loading the algorithms and interfacing them is required). * It probably includes learning how FPGA works in general * Writing or using FPGA glue code (to download the software, and interfacing the algorithms running on the FPGA) Maybe someone who has faced the same problem some time ago can share his experience. I wonder wether we're talking about days, months, years... Best Regards, Gerd Rausch -- voice: +49-2407-575-353 email: Gerd.Rausch@ericsson.comArticle: 18451
I observe some signals keep unknown value in some simulation steps. When I program an FPGA where some signals keep unknown values in some executions, does it harm the device? I have heard that unknown value corresponds to an excessive current from GND or VDD. Utku -- I feel better than James Brown.Article: 18452
Hi, I am trying to simulate a very simple circuit with MAX+plus II: 74161 counter, 10 MHz clock, clear input driven high (i.e., inactive), all other inputs floating (defaults to: count inputs grounded, load input inactive, count enables active). The count outputs count properly but, the RCO output indicates an undefined state at all count values. It should be low except when the count value is fifteen. Oddly enough, when the RCO is gated through an AND gate, the AND gate's output shows the undefined state during the time when the RCO should be high. I've selected a FLEX 6K device. Any ideas what I'm doing wrong? TIA, PatArticle: 18453
Andreas Doering wrote: > Hi, > more by accident I found the program XDL in my Alliance distribution. > XDL is a program that converts a XILINX design between ncd > (binary undocumented internal) and XDL an ASCII format. > Beside of two work around answers in the data base I have not yet found > any hint or documentation on this. > I think this is a great thing, because it is much easier > finding things, you can use grep/wc/perl what you like > to check things. > Without xdl the only way back to a processable text file is pre-map with > ngd2vhdl/ngd2edif and the like. > Only timing information can added. > I think that such open interfaces are a big win, > especially for reconfigurable computing because access > to very low end features are possible without re-engineered > bit-stream manipulation. > (Of course, meanwhile there is also JavaBits). > Andreas > > ----------------------------------------------------------------- > Andreas C. Doering > Medical University Luebeck, Germany > Home: http://www.iti.mu-luebeck.de/~doering > quiz, papers, VHDL, music > "The fear of the LORD is the beginning of ... science" (Proverbs 1.7) > ---------------------------------------------------------------- I sent this request in to Xilinx a few months ago and got the answer that the docs will be included in the 2.1i release. Unpacking my 2.1i and there they were in $(XILINX)\userware\doc. They are in HTML format and nowhere near as good as the main tool descriptions but just about useable. Looking at the xdl-ucs-ext file it appears that there are a lot of undocumented ``hidden flags''. One real omission is that XDL cannot convert the .ngd type files produced by ngdbuild and ngdanno.Article: 18454
In article <01bf1cf6$834c2000$207079c0@drt1>, Austin Franklin <austin@darkr00m.com> wrote: > > >Allan Herriman <allan.herriman.hates.spam@fujitsu.com.au> wrote in article ><380fdc19.6028487@newshost.fujitsu.com.au>... >> On Thu, 21 Oct 1999 17:55:14 -0400, Ray Andraka <randraka@ids.net> >> wrote: >> >> [snip] >> >Someone asked offline what >> >constitutes "up" on the chips. A good question, so I'll repeat the >answer here. >> >Up is what shows as up when you open the floorplanner. IIRC, the quad >flatpacks >> >have pin one on the top edge. The best way to make sure you get it >right is to >> >open the floorplanner for the chip and package you are using and look at >the pin >> >numbers - they are printed on the floorplan screen. > >Is there a way to do this without having to create a blank design for it to >'read' in? Yes. Fliptronics ChipView creates floorplanning templates with all pin numbers shown, and it can even print them so you can use it. No input file required. Works for All XC4000 and Spartan products. >OR better yet, why they don't have either an Adobe Acrobat (preferred) or >PostScript printable and legible document for each package/die combination. > It gets VERY tedious making these for the new Virtex chips.... Use ChipView, send to a postscript printer, set "print to a file" Philip Freidin Fliptronics You know, I wouldn't have to post these blatant commercials, if you all bought a copy.Article: 18455
Kai Troester wrote: > the first thing you should do is a static timing analysis. static timing > analysis means every possible path (pad to ff, ff to pad and ff to ff) > will be checked. if you are using the M1 Foundation Tool of Xilinx than > you have a timing analyzer. the timing analyzer uses your timing > constraints and reports which paths fails the constraints. you should > have timing constraints for every clock and for the delay time at the > pads. if you have multiple clocks in your design you should design the > crossing from one clock to an other very carefully. > > this are the general things i can suggest you. > > -- > If you have access to an HDL simulator it might be worth trying some post Place&Route simulation. This has helped me find missing constraints before. It can be a bit tedious to set up and work through a few bugs in the Xilinx ``gate-level'' model generation but once you are there you can play with different clock speeds and see if you can trigger set/hold violations.Article: 18456
Hello all, I'm having the following problem: In a new product I have to program a Xilinx CPLD (XC95216) with Boundary Scan. As we have the Parallel Cable and Xilinx Foundation Series available, we want to program with this tool. As long as we 'isolate' the CPLD, and make a one-element chain, we are capable of programming it (so the tool is ok, as well as the CPLD). However, if we define a chain with more elements (the normal case will be 7, maybe 9), then we always get a basut error for everything we want to do with CPLD (programming, ID-check, erase, etc). We tried to make the chain already in two ways: a. Starting from the one-element-chain that is capable of programming, and adding elements by loading their BSDL-files b. Initializing the chain with the programmer itself (Seven elements are found, but just the CPLD is recognized), loading the jedec-file for the CPLD, and linking BSDL-files to the others. If I'm not completelly wrong, the only thing the programmer needs to know is how to put the other elements in bypass, as the other functions don't matter... Did anyone succeed in making a chain of several elements yet (of which some aren't Xilinx-parts), and did this need some tricks ??? I give you an example of the errors I get (here a chain with two elements is defined, first device is the CPLD XC95216, which is linked with jedec-file top_cpld ; the second element is a Motorola 68360 QUICC-µP, linked with bsdl-file quicc240.bsd) >JTAG Programmer Started 1999/10/25 15:39:49 >Loading Boundary-Scan Description Language (BSDL) file >'C:/fndtn/data/xc95216.bsd'.....completed successfully. >Loading Boundary-Scan Description Language (BSDL) file >'c:\DIR1\quicc240.bsd'.....completed successfully. >Checking boundary-scan chain integrity...done. >Verifying device positions in boundary-scan chain... >Instance 'top_cpld(Device1)' at position '1'...verified.ERROR:basut - >Boundary scan chain has >been improperly specified. Please check your >configuration and re-enter the boundary-scan chain information. >Boundary-scan chain validated unsuccessfully. >ERROR:basut - : The boundary-scan chain has not been declared correctly. > Verify the syntax and correctness of the device BSDL files, correct the > files, reset the cable and retry this command. (BTW, I already posted an article simular to this one in May, but we didn't succeed then, even with the hints given ; and due to priorities it didn't matter for a while, but now it's there again...) TIA, Alain CloetArticle: 18457
Hello, I have used a more compact instance naming, and 'Map' worked fine. It seems it is a string length overflow. I do not know if it is possible to overcome this limitation since my designs might be even bigger in the future. Khaled.Article: 18458
Khaled BENKRID wrote: > Hello, > > I have used a more compact instance naming, and 'Map' worked fine. It seems > it is a string length overflow. I do not know if it is possible to overcome > this limitation since my designs might be even bigger in the future. > > Khaled. Khaled, if it is possible, generate the instance name suffixes in hexadecimal numbers. Like "inst_1/inst_11/inst_111/inst_1111/inst_11112/inst_111121/inst_1111211/...." might be "inst_0/inst_1/inst_3/inst_F/inst_1F/.." This will reduce the instance name length by at least four. Utku -- I feel better than James Brown.Article: 18459
in addition, drop out "inst_"... why don't you use just a "i" for instance? -- I feel better than James Brown.Article: 18460
This is a multi-part message in MIME format. --------------F5CD52424994BBFBA72F0792 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Given the number of people who ask for a way to get started with programmable logic, we put together a tutorial on CPLD design using Xilinx's free WebPACK suite of tools. You can find it at http://www.xess.com/webpack.pdf The tutorial shows how to get and install the WebPACK software, and it shows how to use the software to do a simple combinational logic design and a hierarchical, sequential design using VHDL. It covers all the phases from design entry, logic synthesis, fitting, bitstream generation, and downloading into the CPLD. This tutorial will not make you an expert FPGA/CPLD designer and it is specific to Xilinx's tools, but it will get you started with a minimum investment of time and money. --------------F5CD52424994BBFBA72F0792 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 --------------F5CD52424994BBFBA72F0792--Article: 18461
Child Your posting got me interested in the delta sigma DAC of XAPP154 so I read it over. It looks to me that the ideas are basically sound and should work well as shown. I think the important points are to follow the RC filter with a high impedance load and to use their oversampling recommendations. It occurred to that two FPGA outputs could be used generating the complements of the same bit stream and then you would have a differential signal to low pass filter. The two pins would form a pair that have a constant current and this could reduce ground bounce related distortion. My guess is that their single ended solution would work great however. Pete Dudley Child K.L. Sun <u8713501@cc.nctu.edu.tw> wrote in message news:38131eb8.344767491@netnews.nctu.edu.tw... > Dear Guys, > > I am thinking of using Xilinx FPGA to implement a DAC. > From the provided application notes XAPP154 (9/1999), an RC > LPF has to be added. Could any one tell me if such a DAC can > performs as well as an ordinary DAC (e.g. Rise/Fall Time, > speed...)? > > Thankx. > > ChildArticle: 18462
On Mon, 25 Oct 1999 17:55:08 -0600, "peter dudley" <padudle@worldnet.att.net> wrote: >Child > >Your posting got me interested in the delta sigma DAC of XAPP154 so I read >it over. It looks to me that the ideas are basically sound and should work >well as shown. I think the important points are to follow the RC filter with >a high impedance load and to use their oversampling recommendations. I read it over too, but I wouldn't say that *all* of the ideas are basically sound. "For example, the 16-bit audio DACs in a CD system would require a clock frequency of 2.9 GHz for full resolution of the highest frequencies. In practice, a much lower clock frequency is used." Audio converters generally aren't 1st order DSMs, so these figures don't apply. Regards, Allan. >It occurred to that two FPGA outputs could be used generating the >complements of the same bit stream and then you would have a differential >signal to low pass filter. The two pins would form a pair that have a >constant current and this could reduce ground bounce related distortion. > >My guess is that their single ended solution would work great however. > > Pete Dudley > >Child K.L. Sun <u8713501@cc.nctu.edu.tw> wrote in message >news:38131eb8.344767491@netnews.nctu.edu.tw... >> Dear Guys, >> >> I am thinking of using Xilinx FPGA to implement a DAC. >> From the provided application notes XAPP154 (9/1999), an RC >> LPF has to be added. Could any one tell me if such a DAC can >> performs as well as an ordinary DAC (e.g. Rise/Fall Time, >> speed...)? >> >> Thankx. >> >> ChildArticle: 18463
I have been working on pinning out a Virtex in a 432 BGA....so, for resources, I was using the pin diagram shown in the Xilinx data book....along with the Floor Planner and the FPGA Editor. Well, I noticed a major discrepancy. The data book showed pin C30 in the upper right, and both the Floor Planner, and the FPGA Editor showed C30 in the upper LEFT! What turns out to be the 'discrepancy' is....the die is UPSIDE DOWN in the BGA, and they don't reflect that in either the FPGA Editor or the Floor Planner...so...the left edge in the Floor Planner and the FPGA Editor is actually the RIGHT edge of the physical package. The problem this can create, is if I was to use just the Floor Planner as my guide for physical pinning of the device, I would end up with pins on the wrong side of the device which would create a PCB routing mess. The data flow of the 4k series used to work best going from left to right...because of the horizontal routing between CLBs...and the carry chain ran bottom to top...but it appears that in the Virtex parts (at least according to the physical layout viewed in FPGA Editor), there is no real advantage to data flow going left to right, or right to left, since all the inputs to the CLBs are on the bottom, and outputs are on the top. The Floor Planner shows a DIFFERENT story, though....it has all the elements looking like they flow from left to right (though, they are really from right to left in the BGA because the die is mounted upside down...). I have chosen to believe the FPGA Editor...unless someone else knows otherwise... Anyone have any notes on data flow in the Virtex parts that I missed (like something is better going from left to right or right to left)? The carry chain appears to only run up...so I'll keep that in mind while doing my floor planning, and bit ordering. Comments?Article: 18464
Have you tried both of the chips/software from these two companies? What's the difference between them? ChildArticle: 18465
On Tue, 26 Oct 1999 01:51:57 GMT, allan.herriman.hates.spam@fujitsu.com.au (Allan Herriman) wrote: >"For example, the 16-bit audio DACs in a CD system >would require a clock frequency of 2.9 GHz for full resolution of the >highest frequencies. In practice, a much lower clock frequency is >used." > >Audio converters generally aren't 1st order DSMs, so these figures >don't apply. > >Regards, >Allan. > For my applications, I just wanna turn a demultiplexed(1:4) bit stream (of bit rate 155Mbps) into 16-PAM signals. So I want to know if such a fpga DAC structure can perform as well as a tranditional DAC does. Child :~]Article: 18466
Will Smith Totally Lost It AT http://www.hardcore.tmfweb.nl/Article: 18467
Hi, Atmel / Kanda Systems have just produced a new low cost FPGA Starter Kit which is available to purchase online. More information can be found at: http://www.kanda.com There is also an FPGA book which can be bought separately (but is included in the kit). Hope that helps. Sent via Deja.com http://www.deja.com/ Before you buy.Article: 18468
Andreas Doering wrote: > more by accident I found the program XDL in my Alliance distribution. > XDL is a program that converts a XILINX design between ncd > (binary undocumented internal) and XDL an ASCII format. > Beside of two work around answers in the data base I have not yet found > any hint or documentation on this. > I think this is a great thing, because it is much easier > finding things, you can use grep/wc/perl what you like > to check things. > Without xdl the only way back to a processable text file is pre-map with > ngd2vhdl/ngd2edif and the like. > Only timing information can added. > I think that such open interfaces are a big win, > especially for reconfigurable computing because access > to very low end features are possible without re-engineered > bit-stream manipulation. > (Of course, meanwhile there is also JavaBits). > Andreas Hi, Does this tool (XDL) actually convert the binary format to something legible? And does it work both ways? What phase of the design trajectory is this ncd format? Can you also use XDL to interface between say tech mapping and placement? And what is JavaBits? Joni (feeling stupid) -- Department of Electronics and Information Systems, Faculty of Applied Sciences, University of Ghent, St. Pietersnieuwstraat 41, B-9000 Ghent, Belgium 99999 work phone: +32-9-264.34.09 9 o - 9 email: jdambre@elis.rug.ac.be 9 | 9 \_/Article: 18469
On Tue, 26 Oct 1999 06:18:50 GMT, u8713501@cc.nctu.edu.tw (Child K.L. Sun) wrote: >On Tue, 26 Oct 1999 01:51:57 GMT, >allan.herriman.hates.spam@fujitsu.com.au (Allan Herriman) wrote: > >>"For example, the 16-bit audio DACs in a CD system >>would require a clock frequency of 2.9 GHz for full resolution of the >>highest frequencies. In practice, a much lower clock frequency is >>used." >> >>Audio converters generally aren't 1st order DSMs, so these figures >>don't apply. >> >>Regards, >>Allan. >> > >For my applications, I just wanna turn a demultiplexed(1:4) bit stream >(of bit rate 155Mbps) into 16-PAM signals. So I want to know if such >a fpga DAC structure can perform as well as a tranditional DAC does. In this case you would *not* want to use a DSM DAC, as you wouldn't be able to get the (over)sampling rate high enough. Allan.Article: 18470
hi, I have some question regarding the PC system that run the Xilinx Foundation series. Will a multi-processor pc speed up the place and route process or memory is more important? Can the Xilinx foundation software benefit from multi-processor environment? ( o/s most likely will have to be NT) from my own observation during PNR ( which take up the most time ), CPU Utilitization is 100 % and memory usage is < 256 Mbytes. No much swapping is occuring as the HardDisk LED is not flashing. The design is about 1000 CLB , going to about 3000 ~ 3500 when the design is complete clock = 40 MHz. o/s = win 98 ram = 196 Mbyte swapdisk = 256 Mbyte CPU = PII 400 Because we are looking into a upgrade of the PC system that run the Xilinx Foundation s/w. Ram : 512 Mbyte or 1 GByte of system ram Processor : PIII 600 , single or multi-processors. Any recommendation?? or where can I find the information ??? Thanks alot Sheau PyngArticle: 18471
"Haneef D. Mohammed" wrote: > > > Are you planning on a Linux release? How about source code -- that would > > really be no-strings-attached. :-) > > Linux release? Maybe, maybe soon (no promises). > Source-code? Maybe not :-) Linux would be very good ! My main OS is linux. If I have to do some fpga designing, I have to switch to bill's toys. So I am glad about any linux EDA tool that will be available. MatthiasArticle: 18472
Hi, What is the difference between the BlockRAM of a VIRTEX and the BlockRAM of VIRTEX-E? In the Features of VIRTEX-E (Advance Product Specification version 1.0) XILINX say that this Family has True Dual Read/Write Port Block RAM capability. What is the mean of True? thanks in advance! Regards, Rafael Gadea Departamento Ingenieria Electronica Universidad Politecnica de Valencia SpainArticle: 18473
Does anyone have any suggestions on how I can create a power on reset type signal so that I can initialise signals/ variables to a set of default values within a Virtex? The requirement is to ensure that on boot up a set of commands are issued to an external display to initialise it. Either VHDL or schematic entry methods are suitable, although VHDL preferable many thanks DarylArticle: 18474
Hi Xilinx 4000 parts have a power-on reset capability, I suppose the Virtex ones have it too. You have to instantiate the startup component: COMPONENT startup PORT(gsr : IN std_logic); END COMPONENT; ... startup_inst : startup PORT MAP(gsr => rst); -- where rst is your reset signal ... (you might have to tell your synthesis tool not to remove this component with no output) This will automatically set or reset your design on power on. Daryl Bradley wrote: > > Does anyone have any suggestions on how I can create a power on reset type > signal so that I can initialise signals/ variables to a set of default > values within a Virtex? > > The requirement is to ensure that on boot up a set of commands are issued to > an external display to initialise it. > > Either VHDL or schematic entry methods are suitable, although VHDL > preferable > > many thanks > > Daryl -- Nicolas MATRINGE DotCom S.A. Conception electronique 16 rue du Moulin des Bruyeres Tel 00 33 1 46 67 51 11 92400 COURBEVOIE Fax 00 33 1 46 67 51 01 FRANCE
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