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
WORKSHOPS '98 - VHDL for Power Users October 26 - 28, 1998 Adam's Mark Hotel, Orlando Florida For further info and registration: http://www.vhdl.org/viuf/conf/conf.html Advance registrations due October 9. INVITATION "It is time for a change." That is the message that has come through loud and clear from you, the user community. VHDL has come of age and now has a mature following. Isn't it about time to give some dedicated focus to the issues facing experienced VHDL users? With this in mind, I cordially invite you to take advantage of a unique opportunity - to participate in workshops and tutorials geared to moderate and advanced VHDL users and developers. VIUF is replacing the usual Fall Conference with a program that we hope will accomplish two goals: 1.Provide a more extensive forum for open dialog; and 2.Provide information on current industry trends in VHDL thinking and usage. Workshops '98 will attempt to promote an active rather than passive environment. Workshop topics were selected based upon input solicited from the user community; and workshop attendance will be limited to facilitate better dialog. We intend to capture the key points and arguments from each workshop. This information will then be passed on to relevant working groups and standards bodies within the IEEE/DASC. Your voice will be heard by those who are actively evolving VHDL and its companion standards. Preceding the workshops, there will be a day dedicated to tutorials for both novice and experienced VHDL users. For variety, the Program Committee has also embedded a few tutorials within the workshop schedule as well as a special session to showcase papers from the academic community. It is a great pleasure to announce the return of the VIUF International Workshop on Behavioral Modeling and Simulation (BMAS). This valuable forum will address the emerging needs associated with analog and mixed-signal extensions to VHDL and will be co-located with Workshops '98. In closing, I again extend to you my personal invitation to participate in what I believe will be an intriguing, informative and significant dialog. Are you up to the challenge? I hope to see you there. Yvonne T. Ryan - Event ChairArticle: 12101
Sri Saripalle wrote: > > Hello Lothar, > > We offer the following services and have expertise in-house > (a)ASIC/FPGA design, verification & implementation services. > (b) EDA Software solutions > (c) Cell & Library Development > > We have 90 employees world-wide and 50 engineers in our US Design center > and the rest in our engineering center in India. > Please let me know if there are any opportunities we could address and > bring in our expertise in this area. > > Thanks > > - Sri Almost everyone reading this message is a designer who is doing the same type of work. Why are you telling us? -- Rick Collins redsp@XYusa.net remove the XY to email me.Article: 12102
Tim O wrote: > >I am working on an I2C controller to be placed in an FPGA and I will > >appreciate any valuable information. > > > > If you're using Altera devices there's an I2C module on the freecore > webpage that might do what you need. I'll look up the url if you're > interested. (I haven't used this lpm myself, I've only seen it -- it > is well documented though). > > Tim. Could you post that url please! ...AndrewArticle: 12103
saju@wipinfo.soft.net wrote: > Any good reference material on "Metastability". > > Thanx, > > Saju > > -------------------------------------------------------------------- > Posted using Reference.COM http://WWW.Reference.COM > FREE Usenet and Mailing list archive, directory and clipping service > -------------------------------------------------------------------- A couple on the Xilinx site which also have some general discussion and further references: http://www.xilinx.com/apps/xapp.htm#xapp077 http://www.xilinx.com/apps/xapp.htm#xapp094 ...AndrewArticle: 12104
The Freecore URL is : http://193.215.128.3/FREECORE/ Derek >Hello everyone! > >I am working on an I2C controller to be placed in an FPGA and I will >appreciate any valuable information. > >Tank you! >Ovi >-- > >************************************************************ >Ovidiu Lupas >TimTeh Electronics Ltd. > >e-mail : ovilup@hotmail.com >home e-mail : ovilup@mail.dnttm.ro phone : 40-56-121951 >work e-mail : lupas@timteh.dnttm.ro phone/fax : 40-56-198943 >************************************************************Article: 12105
I'm using Xilinx M1 (Syntesys FPGA Express). I've a shift register and every output must drive some inputs. I want to avoid a slow slew rate. Is there an explicit possibility to use Xilinx TBUF? Bye Tom!Article: 12106
Article: 12107
Hi friend, I am faced with the same problem too, since i am also targeting the same XC4028expg299-3 device. I am also faced with the problem of getting the post_synthesis simulation to work using the time_sim.sdf amd time_sim.vhd files that are generated, while i am using the FLOW ENGINE in the design manager of the XILINX. I troed using the same test bench file which i had earlier used for functional simulation, but i can't get any trace/signal on the VHDL simulator using the VHDLDBX &. I just get a single all zero trace labelled TOC. Am i missing something? I would really appreciate, if u could outline the steps that are involved in successfully getting through the post synthesis timing simulation. Thanks, Mohsin Riaz Faculty Of Engineering, Memorial University Of Newfoundland, St. John's,Newfoundland, A1B3X5,Canada. email:mohsin@engr.mun.ca Web:www.engr.mun.ca/~mohsin On Mon, 28 Sep 1998, Gareth Baron wrote: > Are you sure you are providing all the power and gnd pins correctly ? > > Are you exceeding the total current specifications ? Ie the current drawn > by the VCC pins is less than the datasheet specifications ? What about the > Dynamic current consumption when you are switching a load of outputs ? > > Just some thoughts for you. > > > Leprechaun wrote in message <6ue1f8$cpd@ustsu10.ust.hk>... > >Hi all, > > > >I am using Xilinx XC4028XL-1-PG299 to implement a circuit. The utilization > >is 100 % (all 1024 CLBs are used). After I download the circuit, I use a > >IMS Tester to test the chip. > >I found that for some output pin assignment, my circuit works well, but > >for another, it just gives wrong result or just no result (all outputs > >stay zero). > > > >I've done post-sim using the .sdf and .vhd files generated by M1 in > synopsys > >and the it works well (not any glitch problems ) and the timing > >requirement for my circuit is not tight at all. > > > >does anyone have the same experience? I've already spent 2 weeks time in > >debugging this problem but don't have any idea up till now..... > > > >Thanks a lot. > > > >Rgds, > > > >Oliver > > > > > > Mohsin Riaz Box#59, Faculty Of Engineering, Memorial University Of Newfoundland, St. John's,Newfoundland, A1B3X5,Canada. email:mohsin@engr.mun.ca Web:www.engr.mun.ca/~mohsin Tel# res:(709)576-8414 off:(709)737-8809 lab:(709)737-3583Article: 12108
Hi all, I have an application where we use a handful of standard CMOS IC's, a 4028 BCD to decimal, and 3 x 4071 quad OR gates, to generate 10 outputs from a 4 input BCD switch. Speed is not important, as it controls input selection of a consumer audio product. Any ideas on a simple cost effective programmable device that could replace these parts? Something that can implement a simple 12 gate or so equivalent lookup table? It seems that most of the devices out there today are much more complex, therefore too pricey, for what we need. I would like to use a programmable device of some kind, as the marketing people drive me nuts with their requests for feature changes/additions. Every time this happens, the gate routing has to be changed, which means modifying the PCB layout........ Thanks, Lukas LouwArticle: 12109
Hello, Thank you very much for your help and url. I had looked up and it is a good starting point. My controller should be a little more complex, but I think I can handle it. At least this I should do, er ? :) Ovi.Article: 12110
Because you have 10 outputs your circuit won't fit in a 16V8 (only 8 outs). The next part I would consider would be a 22V10 (10 outs). Both these parts come in multiple flavours from multiple vendors. You can even get some ISP parts if you want. Lukas Louw wrote in message <6uqs9b$2io@bgtnsc03.worldnet.att.net>... >Hi all, > >I have an application where we use a handful of standard CMOS IC's, a 4028 >BCD to decimal, and 3 x 4071 quad OR gates, to generate 10 outputs from a 4 >input BCD switch. Speed is not important, as it controls input selection of >a consumer audio product. > >Any ideas on a simple cost effective programmable device that could replace >these parts? Something that can implement a simple 12 gate or so equivalent >lookup table? It seems that most of the devices out there today are much >more complex, therefore too pricey, for what we need. > >I would like to use a programmable device of some kind, as the marketing >people drive me nuts with their requests for feature changes/additions. >Every time this happens, the gate routing has to be changed, which means >modifying the PCB layout........ > >Thanks, >Lukas Louw > >Article: 12111
Lukas Louw wrote: > I have an application where we use a handful of standard CMOS IC's, a 4028 > BCD to decimal, and 3 x 4071 quad OR gates, to generate 10 outputs from a 4 > input BCD switch. Speed is not important, as it controls input selection of > a consumer audio product. > Any ideas on a simple cost effective programmable device that could replace > these parts? Something that can implement a simple 12 gate or so equivalent > lookup table? It seems that most of the devices out there today are much > more complex, therefore too pricey, for what we need. > I would like to use a programmable device of some kind, as the marketing > people drive me nuts with their requests for feature changes/additions. > Every time this happens, the gate routing has to be changed, which means > modifying the PCB layout........ Go for it! To get 10 outputs the cheapest you can do is pretty certainly one of the 22V10 family - 10 outputs, more inputs than you need. Far more speed than you require, but of course that's no problem. If you're making consumer kit then your volumes are probably high enough to get a good deal on the part. True product-life cost of the 22V10s should be a stack cheaper than all the PCB real-estate, PCB design respins, assembly hassle and inflexibility that you have right now. Plenty of free software around (PALASM from AMD, demo versions of CUPL from Logical Devices, etc, etc) for something as simple as this. And what's more you get enough logic to implement extra stuff like latches if you need them. In-system-programmable parts (Lattice) probably too expensive to suit your application, so you will need a programmer - but don't buy anything expensive because in a high volume application you should get your distributor to program the parts for you - tell 'em how many thousands of pieces you need, and they'll crawl all over you offering programming services. You won't look back. Jonathan Bromley --Article: 12112
Sounds like it might be poor power supply characteristics. Are all the power and grounds connected Are the outputs in slow slew mode (default) Are you following the data book rules about number of switching outputs per power/ground pair How far away are the decoupling capacitors? Is the chip in a wirewrap socket? if so how long are the power and ground wires, where are the decoupling capacitors. Philip. In article <6ue1f8$cpd@ustsu10.ust.hk> cpegfa@uxmail.ust.hk (Leprechaun) writes: >Hi all, > >I am using Xilinx XC4028XL-1-PG299 to implement a circuit. The utilization >is 100 % (all 1024 CLBs are used). After I download the circuit, I use a >IMS Tester to test the chip. >I found that for some output pin assignment, my circuit works well, but >for another, it just gives wrong result or just no result (all outputs >stay zero). > >I've done post-sim using the .sdf and .vhd files generated by M1 in synopsys >and the it works well (not any glitch problems ) and the timing >requirement for my circuit is not tight at all. > >does anyone have the same experience? I've already spent 2 weeks time in >debugging this problem but don't have any idea up till now..... > >Thanks a lot. > >Rgds, > >Oliver >Article: 12113
Thomas Reinemann <thomas.reinemann@mb.uni-magdeburg.de> wrote in article <3610C2C7.B19A3B8A@mb.uni-magdeburg.de>... > I'm using Xilinx M1 (Syntesys FPGA Express). > > I've a shift register and every output must drive some inputs. I want to > > avoid a slow slew rate. Is there an explicit possibility to use Xilinx > TBUF? > You're being vague. Do you mean that every output of the shift register must drive some inputs to other *chips* or just inputs to other modules inside your chip? -- Andy Peters Sr. Electrical Engineer National Optical Astronomy Observatories apeters@noao.edu.NOSPAMArticle: 12114
We are looking for: designers, programmers, engineers: We provide working visa if needed. 3 years exp. required Permanent positions only Pls. send your TEXT only resume as an EMAIL if you are interested 1. Standard ASIC designers: Arizona, US. 2. EDA Tool Software Development Arizona, US. Candidates should possess strong software development skills (C++, configuration management tools), experience developing EDA tools is preferred. Knowledge in Verilog, VHDL and synthesis are also preferred. The main products of this group are model generators for EDA tools and VLSI EDA tools such as Delay Calculators, netlist screeners. 3. Design System Development I currently have two openings in this group. These people should be familiar with using EDA tools (simulation, static timing, synthesis, design-for-test, etc.), and preferably have a good understanding of the modeling issues in those tools and how to integrate EDA tools into the design environment. The group combines commercial EDA tools, VLSI tools and libraries for VLSI technology into a design system based on design flows and interoperability of the tools in that environment. 4. Bay Area: ANALOG cicuit design. We can send a simple analog circuit for you to take a look, analyze it and apply for the position. Gary Lang Gary N. Lang Vice President of ACD,Inc. E-mail: garynlang@aol.com http://www.acdcon.com/Article: 12115
Does anyone know how to report delay of internal signals with Altera's Maxplus2 Timing Analyzer? I can only get the delay from one I/O pin to another I/O pin. Any input would be appreciated.Article: 12116
By internal signals, I assume you are talking about register to register delays. If you are only getting i/o to i/o delays, then you have the analysis menu set to delay matrix, which if I remember right is the default. To get a listing of register to register delays: Select the timing analyzer On the analysis menu, select registered performance On the options menu, select time restrictions In the time restrictions dialog box, set either the number of paths per clock or the clocks less than frequency radio buttons and text boxes with appropriate values Run the analysis Push the list paths button You'll get a listing paths sorted by longest path first Also, if you assign any timing constraints, they are included in the *.rpt file with the assigned and realized time values for each of the setup/hold, delay matrix, and registered performance groups. Giang Thach Nguyen wrote: > Does anyone know how to report delay of internal signals with Altera's Maxplus2 Timing Analyzer? I can only get the delay from one I/O pin to another I/O pin. Any input would be appreciated. -- -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: 12117
You can find a list of newgroup pointers at http://www.optimagic.com/newsgroups.html. Also, you can try searching for the appropriate article at www.dejanews.com. ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Victor Levandovsky wrote in message <1103_907072299@p-331.podol.khmelnitskiy.ua>...Article: 12118
This Bit-Serial approach is very clever Jan, but if we were to pass the 16 data values sequentially into the FPGA as 2 8-Bit parallel values per clock over a 16-Bit bus instead of 16 Serial lines (along with a sync line), I think we can compute the max in 10 CLBs. (assuming packing the subtractor, mux and registers in the same CLBs) /=====REG==\ | [MAX] | [MAX] \==>|a o|==/ [hi_byte ]===>|a o|==REG=>|b__| [lo_byte ]===>|b__| Result every 8 clocks, (you have to force the max value to be loaded on clock 0 by mucking with the carry). The problem is the speed. Current FPGAs do an 8-Bit ripple subtract in 6 to 8 ns, but you have to distribute the carry back to the mux in the same clock which will take another 3-4 ns. You can pipeline the delay away, but that will cost another 4 CLBs. (You can actually hide 2 of these registers in the carry CLBs, but you will also need to delay sync with two registers). slow=12ns, 10 CLBs fast= 8ns, 14 CLBs Cheers, Brad Jan Gray wrote: > > I wrote in message <6u8s8g$l5s$1@news-1.news.gte.net>... > >Algorithm analysis 101 tells us that to determine the largest item of n > >items requires at least n-1 2-input "greater than" comparisons. ... > > SEGUE > > Algorithm analysis 101 also teaches us to carefully choose a model (e.g. > elements and a 2-element comparison function) and a goal (e.g. minimize no. > of comparisons to determine the largest element) to represent the problem. > Only then can we compare algorithms or study optimal lower bounds on > algorithms. > > For example, using a comparison based model, it can be shown that any > sorting algorithm on n elements must perform at least ceil lg n! > comparisons. But there are other sorting algorithms which do not perform > any elementwise comparisons. They assume a different model with different > operations on the elements. Consider radix sort, in which we distribute the > n elements into r buckets, according to increasingly significant digits in > their base r representation, and then regather them and repeat. (Remember > the old IBM punch card sorters?) > > So, as radix sort is to a comparison-based sort, is there an analogous > "radix max" to our comparison-based max? > > "PARALLEL BITWISE MAX" > > Yes. Here's a 'digit'al method for finding the max of n k-bit words. > > We scan the n inputs as n serial bit streams, all clocked together, msbs > first. One bit stream is the maximum if it is never observed to be less > than some other bit stream. For any two bit streams A and B, > "A < B" iff ~msb(A)&msb(B) | (msb(A)==msb(B) & "lsbs(A) < lsbs(B)"). > > Design: we keep n candidate-for-max state bits, one for each bit stream. > All are initialized to 1, as each stream is initially a candidate to be the > maximum. A candidate stream is still a candidate if it has a current 1 > input bit. A stream with a current 0 input bit loses its candidacy if there > is some other remaining candidate stream that has a current 1 bit. After > all input bits have been clocked past, the remaining candidate is the > maximum bit stream. If duplicate input bit streams are possible, there can > be more than one remaining candidate, each of which corresponds to the same > maximum value. > > // pidgin netlist generator source: > > input reset; // 1 during the reset cycle > input stream[n]; // current bit of each of the n input streams > reg cand[n]; // 1 if the i'th stream is still a candidate for maximum. > net some1; // 1 if some remaining candidate stream input bit is currently > 1 > output answer; // result, the index of max input stream > > some1 = cand[0]&stream[0] | cand[1]&stream[1] | ... | > cand[n-1]&stream[n-1]; > for (i = 0; i < n; i++) > cand[i] := reset | cand[i]&(stream[i] | ~stream[i]&~some1); > > If we know the input words are never two alike, then bitcount(cand)=1 and we > can use > answer = encode(cand); > otherwise, > answer = priority_encode(cand); > > Implemented in an XC4000 FPGA, for n=16, requires approximately > 4.5 CLBs for some1 > 8 CLBs for cand[16] > 12.5 CLBs for a 16-to-4 priority encoder > ---- > ~25 CLBs, result every 9 clocks for 8-bit input data, at perhaps 10 > ns/clock. > > Radix 4 (2 bits/clock) is also possible but probably unwieldly and slow. > > The nice thing about this approach is it is readily scalable to many inputs; > the slowest part of the design would be the 'some1' or-reduction circuit, > and even with 36 input streams this is only three CLB delays and mostly > local interconnect. > > Jan GrayArticle: 12119
Brad Taylor wrote in message <3611D95A.527FC010@emf.net>... > /=====REG==\ > | [MAX] | > [MAX] \==>|a o|==/ > [hi_byte ]===>|a o|==REG=>|b__| > [lo_byte ]===>|b__| I like it! Certainly nicer than the 4-way hybrid in my first posting. To recap the thread and add yet another approach, if you have n inputs each m bits long, some choices are :- 1. simple m/2+1 CLB max accumulator in ~n clocks 2. ~nm/2 CLB max-mux tree in 1 clock 3. hybrid which takes ~nm/2k CLBs in k clocks 4. "parallel bit serial max" in ~n CLBs in ~m clocks 5. "serial bit serial max" in ~lg m + m/16 CLBs in m*n clocks About 5: use a bit serial max state machine together with an m-bit FIFO implemented using dual port RAM, to store the "current max". Operate by streaming in all the input words, serially, one bit at a time, into the max machine. Each m clocks it bit serially emits the max of all inputs so far. Jan GrayArticle: 12120
APS (Richard Schwarz) wrote: > > ------------------------------------------------------ > APS Newsletter Interim Announcements > ------------------------------------------------------ > > The next APS EDA Newsletter is about to be released. In this issue we > will cover: ... well, but: > > For those who wish to view the APS FPGA tools and Solutions now, the > following websites apply: .. following many, many links to different pages of APS. Hallo Richard, as you know, this is a non commercial newsgroup. cp. the threat "Too much advertising in this news group?" in july <irony> btw: WWW supports multiple hops. Therefore one link to your WWW server is sufficient. Users can branch out by using a second link at no extra charge. </irony> Markus WannemacherArticle: 12121
Hello I have Foundation 1.4 and have developed a board based on Xilinx XC95108 to be programmed in system. Foundation JTAG programming tools work only on Windows 95 or NT so I cannot program the device on DOS. I would not like to oblige my production personnel to install the Windows based program (fairly complicated and tricky process) and some test tools we have work fine only on real mode DOS so they would have to reboot their computer to switch between programming and testing. Older versions of Foundation software support DOS (Eztag) but it does not recognize some commands in the version 1.4's jedec file. Does somebody have suggestions or a software to control the parallel programming cable through DOS? Thanks in advance for your helpArticle: 12122
Check out www.Viewlogic.com. Great tools (Fusion/SpeedWave for VHDL simulation, FPGA-Express for synthesis, etc.) integrated under the Intelliflow manager makes a very easy-to-use package from Viewlogic, the leader in FPGA design solutions. John Huang wrote: > I want to buy a FPGA tool, do you recommand which > one is better, how about Accolade and Aldec? > > John Huang -- *-------------------------------------------------------* * John Willoughby ジヨン ウイロビイ * * System Simulation Mktg office: 508-303-5238 * * Viewlogic Systems mobile: 508-254-9608 * * 293 Boston Post Rd West fax: 508-460-7826 * * Marlboro, MA 01752 email: jww@viewlogic.com * * * * "Well done is better than well said" - Ben Franklin * *-------------------------------------------------------*Article: 12123
Jan Gray wrote in message <6usptm$d1$1@news-2.news.gte.net>... >5. "serial bit serial max" in ~lg m + m/16 CLBs in m*n clocks > >About 5: use a bit serial max state machine together with an m-bit FIFO >implemented using dual port RAM, to store the "current max"... Oops, we don't need an m-bit FIFO, but rather a simple m-bit shift register. Looking to http://www.xilinx.com/xapp/xapp052.pdf for inspiration, the m-bit SR can be implemented in just 2 + m/32 CLBs so I should rather have written: 5. "serial bit serial max" in ~4 + m/32 CLBs in m*n clocks Jan GrayArticle: 12124
We have a Verilog RTL design that needs FPGA for demonstration Gate count - less than 10K (ASIC gates) Speed - 33Mhz Interface - PCI Package - 208-pin PQFP Device - Non-SRAM based FPGA (no EPROM on board) We need conversion, place & route and programming service. Anybody who is interested can send me e-mail at kychan@hintcorp.com Thanks K. Y. Chan HiNT Corp.
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