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
Garry Allen <garrya@abc.gov.au> wrote: >How can I slag the ispVHDL?Viewlogic combination enough. The lattice >fitter is a good tool. the Viewlogic VHDl tools OTOH were the worst I >have ever used. That bad, eh? This morning, I had a visit from a local representative. She demonstrated the ispVHDL/Viewlogic design flow, and I think a package is pretty complete (although my reference is pretty limited in this area). However, I feel the Viewlogic tools may need better integration. I would prefer a single IDE, not a bunch uncorrelated windows that clutter my desktop. >The minimised code Viewsynthesis produced for one of my designs would >not fit into a 256 macrocell device with less than 5 levels of logic. I >expected around 190 with 2 levels. Using another tool (Synplicity) I got >the types of resiults I expected. Thsi suggests that the minimisation >from viewlogic is around 30-40% worse than a more optimal tool. FWIW >Lattice use Synplicity for their technical support and are happy to sell >it to you at a much reduced price. If you are serious about using the >lattice devices and VHDL, get the Synplicity tool. Hmm...the local Lattice distributor over here recommends the ispVHDL/Viewlogic combo. Will ask about the Synplicity tools. Thanks, Gary. Rene Kellenbach, The NetherlandsArticle: 10701
aaj15@dial.pipex.com (Kim Carter) wrote: >Bear in mind that once you have fixed your pins, getting an acceptable fit >gets increasingly difficult - If you go much over 70% on the 3256's you are >running at rapidly increasing risk of not being able to modify and still get >a fit. If you are pushing for speed as well this is accentuated - If as you >suggested in an earlier post you are planning in the field mods, leave a good >margin for it or you will come unstuck. Good tips! Is the 70% limitation caused by the chip architecture, or by the fitter? What about other software packages - do they all depend on the Lattice fitter? >We were using the 3256 devices - considerably larger, along with a few >1024s. There was a lot of interest in using the 6000 series - at the time >Synario could not handle these in a properly integrated fashion ie one could >neither call the RAM/FIFO sections up as 'symbols' in the schematic editor >nor as macros within Abel, is this still the case? and how are these functions >dealt with in the Viewlogic system? AFAIK, the ispVHDL/Viewlogic supports the 6000 series, but only via schematic entry. The ABEL/Synario package doesn't support the 6000 series, to my knowledge. >It rather depends on what else you are likely to do in future - if you are >likely to move on to FPGAs as opposed to other CPLDs, VHDL will handle this >whereas Abel will not. Maybe - learning VHDL seems like a good investment. I am pretty fluent in ABEL, but don't know anything about VHDL. Any good VHDL tutorials on the net? >There is also the issue of what form the schematics take - Some schematic >editors are basically HDL code generators, so you end up with code which >is entirely in HDL and therefore (within limits) portable, others including >the Synario/Abel arrangement are entirely proprietry and therefore are non >portable. AFAIK the Synario schematic format cannot be exported as anything >else so you can't eg document your PLD by just embedding the schematic in >a word document, only print it. This is a major disadvantage of proprietary formats. A standard system like Viewlogic would be a better choice in this respect. >You should also be aware that some systems enable you to take your fitted design >and feed it back into a schematic editor to give a gate level diagram. The Viewlogic system has this capability. >So rumour (mainly from Lattice) had it, the other manufacturers had problems >with data retention - fine if you want built in obsolescence but not much good >if it was expected to work for 10 years or so. Probably bs**t but if you do >want to use Vantis, might be worth comparing with Lattice. Hmm...should I be worried about this? Some of our products are used in pretty harsh environments (high temperatures and other nasty things). Would Lattice be a safe choice in this respect? Thanks, Kim. Rene Kellenbach The NetherlandsArticle: 10702
In article <357f7caf.0@139.134.5.33>, "John Maher" <jmaher@silicon-systems.com> wrote: > > For all of you VHDL or Verilog users: > > Silicon Systems Solutions produce a product called "ED for Windows - HDL" > (ED4W-HDL) which is a very sophisticated productivity editor, to FAST TRACK > your VHDL/Verilog code development. > > The editor runs under Windows 3.1/95/98/NT > > Features include: > > + Syntax Highlighting > + Automatical Language Template insertion and expansion > + Automatic TESTBENCH template generation > + Dozens of common VHDL models for FPGA development > + Advance Search/Replace across files etc > + Syntax and Template Expansion support for 30+ other languages, including > C. C++, BASIC, PASCAL > > and much more ... > > A FREE evaluation copy is available, at > http://www.silicon-systems.com/prod01.htm > > The evaluation version is restricted to 45 days use, from the time of > installation > > Regards, > > John > Seems your site is password protected. Makes things a little difficult to view. Further, is there a port for UNIX platforms? Not everyone works in a PC environment. Martin Meserve -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 10703
ems@see_sig.com (ems) wrote: >my 2 cents: if you're just moving up from plds, get the starter >kit, and stick with abel. i think you can get support for >everything up to the 1032 (1048?) with the starter kit. this gives >you the minimum-investment solution and the smallest learning curve. >you'll have working devices *much* faster than if you decide to go for >VHDL, and it'll cost you a *lot* less. Agreed, the starters kit seems much easier to use than the ispVHDL/Viewlogic combo. Price isn't that important, but my time is limited. > >schematics: > >why bother? not only do you have to learn an HDL, but >you also have to learn the schematic package, and how to >integrate the HDL with the schematics. You have to cope with bugs >from both environments and support from 2 vendors (except that >one of them isn't supporting you - you try going to viewlogic/ >aldec/etc to tell them that your lattice/xilinx/etc package has >a bug). i personally never use schematics, but i've used synario and >viewlogic, and both are second-rate (but i've seen worse). you often >end up just putting HDL modules on a page with their inputs and >outputs labelled - you might as well just put in a top-level HDL >source instead. I have seen both Synario and Viewlogic in the past few days, and I think the integration of both packages with the Lattice fitter still have a lot of room for improvement. Maybe an HDL-only approach would be wise, just to circumvent quirks. >Abel: > >+ easy to learn, except that the manuals are useless >+ you can buy it separately - you don't need synario >+ functional simulation is very easy; much easier than > using a gate-level simulator such as viewsim (also > second-rate) >- has it reached end-of-life? i think it probably has. > this is not a good time to be paying thousands of > dollars for it. >- annual maintenance (what do you get for this? nothing, > in my case) >- buggy, but you quickly learn what not to do >- very expensive. why buy it if you're only interested > in one vendor, and they do a cheap vendor-locked > version? Agreed, I would not start using Abel from scratch nowadays. However, I have had some exposure to Abel - this would get me started with Synario/Abel in a short period of time. >Lattice starter kit: > >+ much cheaper than abel (i paid 2K UKP for abel; if i'd > waited a couple of months i could have got all the > devices i was using at that time for a couple of > hundred dollars in the starter kit) >+ no maintenance >+ covers most of the devices you'd want to use (i reckon > lattice isn't cost-effective at the 1048 or higher > level, but i think the kit now includes the 1048 > anyway) I already tried the starter kit and I was not impressed. >ispVHDL/Viewlogic: > >haven't used it, but i have used viewlogic with vhdl and >abel for xilinx, and i guess it's much the same. i didn't >like it - see comments about multiple vendors, viewsim, >need for schematics, etc. > >timing: > >i'm not sure why, but i've never felt the need to use >anything other than a static timing analyser for >lattice devices (i've used 1016/24/32/48). the timing >is well-defined and the devices don't encourage you >to do things which you'll need a timing simulator to >verify. on the other hand, i always use a timing >simulator with xilinx. The timing model of Lattice CPLDs is pretty simple - probably simple tools will suffice to verify timing. >VHDL: > >i personally use VHDL for almost everything, but i >would say that - for me - it's not an efficient solution >for devices as small as a 1016 or 1024. > >- *very* hard to learn. if you're faced with a choice > of learning abel or vhdl, and it's important to get > results quickly, then it's a no-brainer - go for > abel. This is the scary part. > it's not just the language, which is bad enough, > but the synthesis issues, the entire > design flow, the differences between vendors, > etc. I have seen a demo of the design flow this morning. It's not straightforward, but I think I can manage this. >+ the primary advantage - simulatability. you > can do far more complex simulations than > you could hope to do with abel or a gate- > level simulator. >- but - you have to buy a separate simulator. > the "cheap" (!) vendor-locked packages include > only a synthesiser, which supports only > a subset of the language. > this can be expensive - i use modelsim, at > about 5K UKP. The ispVHDL/Viewlogic package inludes a simulator (Viewsim), but I haven't tried it yet. Probably gate-level only. >you often hear people saying that they don't use >vhdl because it's too high-level, too far from the >hardware, etc. but you can use whatever subset of >the language you want - you could even write in >a completely structural style, and just wire up >modules in the same way that you would on a >schematic (or in abel, for that matter). >synthesis shouldn't be a problem - if you stick with the >simple stuff it'll synthesise well. Just like moving from C to C++, eh? >and forget the device utilisation arguments. if you're >a software engineer and you synthesise vhdl code, >then you may very well end up with lots of unnecessary >hardware; but, if you know what you want, then there's >no reason that you shouldn't get it. Hey - software engineers should spend some time programming in solder (or solder++) before they start programming in VHDL. That'll teach them to use hardware resources efficiently! >in short - you need a very good reason, and lots of time, >money, and patience, to start down the vhdl route, even >more so if you want to use schematics as well. Pretty discouraging....especially the 'lots of time' part. >on the other hand, you could start with abel and the >starter kit for almost nothing, and have working >hardware quickly. Will give the starter kit another look. Thanks, Evan. Rene Kellenbach The Netherlands.Article: 10704
Tools : MTI VHDL, Leonardo synthesis, Xilinx M1.4.12, Windows NT 4 Target part: large 4000XL Question : I want to do the equivalent of floorplaning. I have a block with some predefined blocks (from the core generator) that. I instanciate this block (with some memory, multipliers, inferred (from the vhdl) adders/subtractors/counters etc.) several times and notice that M1 makes some bad decisions as to where to place things. I would like to tell the tool to keep all logic in one instance in a square relocatable region so that it won't put a multiplier from instance 1 on the other side of the chip from the adder which feeds it in the same instance. Please note that in each coregen block the rloc attribute is used and also Leonardo uses the rloc attribute for each modgen (ie. adder ,counter...) that it produces. Furthermore, the M1 tool does keep the relative locations with in each modgen/coregen part. The problem is I also want to keep relative locations of modegen/coregen parts with in each block instance close. I hope I've been clear on want it is I'm after. Thanks in advance for any and all help. Joshua -- Joshua Schwartz ICC Design 972-8-9287375 Josh@gezernet.co.ilArticle: 10705
Tim Forcer <tmf@ecs.soton.ac.uk.nojunk> wrote: >I got the pDS starter kit several years ago, which was OK but very >limited. Because I wanted to use the 1024, I had to buy "Full pDS", and >all I got extra over the starter kit was support for the full range of >devices. Even after an upgrade, this software is AWFUL. It isn't a >true Windows app, doesn't save configuration, etc. My sentiments exactly. > Since my application was, in >several respects, a trivial one, this wasn't any great hardship. I've >heard several people say the starter Synario is pretty good for a >freebie, but haven't tried it myself. I have tried the Synario/ABEL tutorial - seems pretty straightformard, IMHO. Other people over here tell me it has some quirks, though. >As to the ICs, they are fine, and have delivered excellent performance. >I've got reliable 9ns glitch recognition out of a 90MHz part, which I >think is good going. The only reservation I have, which I pass on >whenever anyone asks about ispLSI, is WATCH YOUR 5V line. With most >logic ICs, the data sheet says max Vcc (or Vdd) is 7V. With 74-type >logic, and with many PLDs I've used, if there's a microsecond-order >supply transient which exceeds 7V, even one of over 10V, the ICs don't >die, and some just carry on working as is nothing had happened. But >ispLSI chips fail short-circuit in an EXTREMELY short time if the rail >exceeds 7.0V. The fix in my case was to use a 1N5908 across the power >pins relatively close to the IC. I also used a tantalum bulk decoupler >to limit the rise rate of 5V transients. If you can guarantee that your >board's power rails will NEVER glitch this way, then you are very >fortunate, and should ignore this problem. I couldn't so protected the >devices explicitly. The failure mode is related to the charge pump >involved in providing 5V-only isp, so is likely to be found in other isp >families. Good tips - will give this some extra attention if we are going to use the Lattice parts. >Support, from distributors and from Lattice, has been patchy. Sometimes >it's been excellent, sometimes poor. No obvious reason for the >variation. Support from the local distributor has been excellent so far. Thanks, Tim. Rene Kellenbach, The Netherlands.Article: 10706
Currently, floorplanning support seems to be a bit light from Xilinx if you're using the M1 tools. However, there are some excellent third-party tools available from Morphologic. You should take a look at their MorphMCFP (multi-chip floorplanner) tool. There is more information available at http://www.morphologic.com/mmcfp.htm. ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Joshua Schwartz wrote in message <357FEA27.8CE67D6F@gezernet.co.il>... >Tools : MTI VHDL, Leonardo synthesis, Xilinx M1.4.12, Windows NT 4 >Target part: large 4000XL > >Question : I want to do the equivalent of floorplaning. I have a >block with some predefined blocks (from the core generator) that. >I instanciate this block (with some memory, multipliers, inferred >(from the vhdl) adders/subtractors/counters etc.) several times >and notice that M1 makes some bad decisions as to where to place >things. I would like to tell the tool to keep all logic in one >instance in a square relocatable region so that it won't put a >multiplier from instance 1 on the other side of the chip from >the adder which feeds it in the same instance. > >Please note that in each coregen block the rloc attribute is used >and also Leonardo uses the rloc attribute for each modgen (ie. adder >,counter...) that it produces. Furthermore, the M1 tool does >keep the relative locations with in each modgen/coregen part. The >problem is I also want to keep relative locations of >modegen/coregen parts with in each block instance close. > >I hope I've been clear on want it is I'm after. Thanks in >advance for any and all help. > >Joshua > > >-- >Joshua Schwartz >ICC Design >972-8-9287375 >Josh@gezernet.co.il > >Article: 10707
Due to Esperan's rapid expansion in the provision of VHDL and Verilog training services in the US and Europe. We are actively seeking to recruit consultants as trainers to out source the delivery of the training courses to, on a week by week basis. Esperan specialize in supplying independent, application orientated VHDL and Verilog based training and have very close relationships with most of the main EDA and FPGA vendors such as Synopsys, Mentor, Viewlogic, Xilinx and Altera our courses are recommended or re-sold by many of them. In addition to the obvious financial benefits our experience with other consultants has shown that delivering the training courses provides an excellent opportunity for the consultant to market their own, and their organizations expertise to potential customers. Primarily we want to to talk with Designers/Consultants who have: # Real world experience of VHDL and/or Verilog based design using synthesis. # Experience in the use of FPGAs would be a distinct advantage. # Good communication skills and the ability to explain complex topics in a clear, concise manner. # Availability for periods of up to one week for between four and ten weeks per year. # A willingness to travel. We own and develop all the training material, but subcontract the delivery to consultants. All the logistics of the courses, such as marketing, software licenses, training materials, presentation materials etc. are managed by Esperan. If you think you have what we are looking for please return by email your answers to the following questions to give me a summary of your experience to date. How many years experience of VHDL/Verilog?= What real design experience have you had?= Does it include Synthesis?= How many ASIC's or FPGA's have you synthesized?= What experience of testbench design have you had?= What design tools have you used?= What would your availability be during the year for training (ie. 6 five day course )?= When would you be available to start?= Would you also send me an up to date resume'. I look forward to discussing this opportunity with you. Regards Simon Moreton Training ManagerArticle: 10708
In article <357FBE69.9F9D85B1@bltinc.com>, Ed McCauley <edmccauley@bltinc.com> writes: >Would turning on the internal pull-up resistor in the 'load' IOB solve >your problem? I'm assuming you're trying to guard against the sensor >going open circuit. No, it is not the usual pullup. The thing acts as a 'Nand' in the old MOS-style. The OBUFT is an open drain, the sensor is the second switch in the chain. When both are active, the 'dot' should be drawn down to logic Low. And there is the problem. When the open-drain is not active, a leakage current somewhere should not bring the 'dot' beneath the High threshold. On the other hand, R must be big enough that when both the sensor has its 20K and the OBUFT connects to ground, the 'dot' is not above the Low threshold. From looking into the handbook I think R=100K might be OK, but I might have forgotten something; so I'd rather hear from somebody who has more experience than me. Regards, Thilo >Thilo Thiessenhusen wrote: >> >> Hello, >> >> I have a configuration like this: >> >> Vdd >> | >> --- >> | | >> | | R >> --- >> _ | _ >> OBUFT _>--------o---------<_ IBUF >> >> What is the maximum resistance that guarantees a valid input level >> at any time? Would 100K be O.K.? >> >> [The actual configuration is a bit more complicated since there is >> a sensor between OBUFT and the dot, which can switch between >> open and about 10-20K]. -- Support the anti-Spam amendment - Join at http://www.cauce.org/Article: 10709
A small well funded Boston area firm is searching for “A” Players for its state of the art compression/decompression and image processing/enhancement technology products. This is a world leader in the development and marketing of image processing chips, controller boards and software tools. There are two positions that are being hotly searched for. The company is willing to pay “A” salary and bonus for “A” players. These are permanent positions. 1. ASIC Designers/ASIC Test Engineers Responsible for specification, architecture, behavioral modeling, design, synthesis, implementation, and simulation of state-of-the-art ASICs. Minimum 3 years IC design; C, algorithm development, image processing/coding and MSEE preferred. We have developed a 133 MHz SIMD DSP, and the world's fastest JPEG @ 70 MHz. 2. Principle Image Processing Engineer In this leadership role, you will design & implement real-time image processing algorithms for color and monochrome printers, copiers, and scanners. Requires a BSEE (MSEE preferred), 5+ years algorithm development experience, RISC/DSP assembly, and C. Project Management and printer experience a plus. 3. Digital Design Engineer Design, implement and debug advanced imaging board-level products. Requires BSEE, 5 + years of digital circuit design experience, familiarity with embedded system design, microcontrollers, high speed logic (>50 MHz), PLD and FPGA design. Experience in system and board level design analysis. Position includes project level responsibility. If you are interested in finding out more regarding these positions, about the challenges, and one of the highest rate of salary - higher than the bay area, please contact me by email. richkol@aol.com Rich Kolikof Winthrop Research Thanks for your time. Rich Kolikof Winthrop Research 503-644-9111Article: 10710
Yes, Speedwave will run "testbenches". Indeed any VHDL simulator should have this capability. There is nothing special about testbenches. The general form of a testbench is an empty entity with the DUT component Instantiated in the architecture. There will then be a process to provide stimulus. See template below. There are also tools available to automatically generate testbenches from timing diagrams such as QuickBench from Chronology. If you require any further information on this please email me directly. Regards Will ------------------------------------------------------------------------------ -- -- Typical testbench skeletons -- ------------------------------------------------------------------------------ -- -- Top-level entity/architecture -- library ieee ; use ieee.std_logic_1164.all ; entity testbench is end testbench; architecture only of testbench is -- -- Paste in any necessary signals and component declarations here. -- Declare DUT component here. -- begin -- -- Paste in the instantiation(s) of the DUT component and any other testbench -- components here. -- -- -- Stimulus thread -- stim : process begin -- -- Paste in your test sequence here -- wait; end process; end only; configuration testbench_config of testbench is for only -- -- Paste in your configuration statements here. -- end for; end testbench_config; In article <6l853l$cee$1@goanna.cs.rmit.edu.au>, Hamish Moffatt <hamish@moffatt.nu> writes >In comp.arch.fpga John Huang <hungi@tpts4.seed.net.tw> wrote: >> I got some questions about Viewlogic/Speedwave, >> I have a design that use VHDL for a FPGA. >> 1. Can I use VHDL testbench in the speedwave, if >> it is YES, how will I do? >> 2. What simulaors support the VHDL testbench? > >Aldec's ActiveVHDL seems to run them nicely. > >> 3. Would you please give me an example for >> write testbench? > >Best to grab a book eg Skahill's VHDL for Programmable Logic, >it has a whole chapter on it. Aldec's also has a thing for generating >test benches for you, I haven't tried it. > > >Hamish -- William White < ------------------------ < < < ---------- Mailto: will@fpga.demon.co.uk | Direct Insight Ltd < < < < > Tel: +44 1280 700262 | | < < < Fax: +44 1280 700577 | --------------------------- < ------------------------------------------Article: 10711
In article <357f7caf.0@139.134.5.33>, "John Maher" <jmaher@silicon-systems.com> wrote: ... > > A FREE evaluation copy is available, at > http://www.silicon-systems.com/prod01.htm > > The evaluation version is restricted to 45 days use, from the time of > installation > > Regards, > > John > > I tryed this link and received a request for a username and password to get into this site! Of course I don't have the username and password. Even trying to go to www.silicon-systems.com gets the same response. Not a very effective way to present a company on the Web. Cheers, Jim Means -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 10712
Apologies to all who tried to access our site, and found a password protected page! The web is now available. -John John Maher wrote in message <357f7caf.0@139.134.5.33>... >For all of you VHDL or Verilog users: > >Silicon Systems Solutions produce a product called "ED for Windows - HDL" >(ED4W-HDL) which is a very sophisticated productivity editor, to FAST TRACK >your VHDL/Verilog code development. > >The editor runs under Windows 3.1/95/98/NT >Article: 10713
Apologies to all who tried to access our site, and found a password protected page! The problem has been resolved by our web administrator, and the web is now available. -John John Maher wrote in message <357f7caf.0@139.134.5.33>... >For all of you VHDL or Verilog users: > >Silicon Systems Solutions produce a product called "ED for Windows - HDL" >(ED4W-HDL) which is a very sophisticated productivity editor, to FAST TRACK >your VHDL/Verilog code development. > >The editor runs under Windows 3.1/95/98/NT >Article: 10714
Joshua Schwartz wrote: > Tools : MTI VHDL, Leonardo synthesis, Xilinx M1.4.12, Windows NT 4 > Target part: large 4000XL > > Question : I want to do the equivalent of floorplaning. Call our Xilinx FAE and beg him/her for help. The floorplanner is coming back! He/She should have access to a pre-released copy.Article: 10715
Dear William White How can I use SpeedWave to do testbenches? In the past, I used ModelSim's V-System to do testbenches. Yet I know SpeedWave requires synthesis before, right? Leslie Yip In article <RePGPJAfYAg1EAfI@fpga.demon.co.uk>, William White <will@fpga.demon.co.uk> wrote: > > Yes, Speedwave will run "testbenches". Indeed any VHDL simulator should have this > capability. There is nothing special about testbenches. The general form of a testbench > is an empty entity with the DUT component Instantiated in the architecture. There will > then be a process to provide stimulus. See template below. There are also tools > available to automatically generate testbenches from timing diagrams such as QuickBench > from Chronology. If you require any further information on this please email me > directly. > > Regards > > Will > > ------------------------------------------------------------------------------ > -- > -- Typical testbench skeletons > -- > ------------------------------------------------------------------------------ > > -- > -- Top-level entity/architecture > -- > library ieee ; > use ieee.std_logic_1164.all ; > > entity testbench is > end testbench; > > architecture only of testbench is > > -- > -- Paste in any necessary signals and component declarations here. > -- Declare DUT component here. > -- > > begin > > -- > -- Paste in the instantiation(s) of the DUT component and any other testbench > -- components here. > -- > -- > > -- Stimulus thread > -- > > stim : process > > begin > > -- > -- Paste in your test sequence here > -- > > wait; > end process; > end only; > > configuration testbench_config of testbench is > for only > > -- > -- Paste in your configuration statements here. > -- > > end for; > end testbench_config; > > In article <6l853l$cee$1@goanna.cs.rmit.edu.au>, Hamish Moffatt <hamish@moffatt.nu> > writes > >In comp.arch.fpga John Huang <hungi@tpts4.seed.net.tw> wrote: > >> I got some questions about Viewlogic/Speedwave, > >> I have a design that use VHDL for a FPGA. > >> 1. Can I use VHDL testbench in the speedwave, if > >> it is YES, how will I do? > >> 2. What simulaors support the VHDL testbench? > > > >Aldec's ActiveVHDL seems to run them nicely. > > > >> 3. Would you please give me an example for > >> write testbench? > > > >Best to grab a book eg Skahill's VHDL for Programmable Logic, > >it has a whole chapter on it. Aldec's also has a thing for generating > >test benches for you, I haven't tried it. > > > > > >Hamish > > -- > William White < > ------------------------ < < < ---------- Mailto: will@fpga.demon.co.uk > | Direct Insight Ltd < < < < > Tel: +44 1280 700262 | > | < < < Fax: +44 1280 700577 | > --------------------------- < ------------------------------------------ > -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 10716
Hi, We're Hunter International, and we are currently looking for qualified DSP based Engineering Professionals from a broad variety of disciplines. Our client is one of, if not THE, best firms you could be associated with in the Northern California area. This beautiful firm, located in the Silicon Valley, is simply at the top of it's class with opportunities and care for it's employees. The benefits are second to none, and the technology is stellar, so you can't go wrong. There is even a fully functioning Health Club built into the premises! We absolutely guarantee that you will be thrilled with this firm if you are qualified! PLEASE note before you read further: This company is a publicly held, incredibly successful firm, BUT, because of certain clients, they do require the ability to secure a Secret Clearance. This means citizenship of you, and your direct family to Mother and Father, if they are still living. Because of this, the candidate pool becomes thin. So if you are qualified, AND can get clearance, you will be received with enthusiasm and given the 'red carpet' treatment, as they need engineers in a big way! In an effort to help with the massive growth they are experiencing, we need to find Engineers who fall into the areas of expertise listed below. The descriptions are intentionally brief, and this is because the amount of available jobs are more than you could read through in a single post. All jobs are new positions, not vacancies. Software Engineers: *Experience developing Real-Time, software intensive projects using Ada, C, C++, in a UNIX, NT, and/or VMS environment. *Requirements definition, preliminary/detailed design code/unit testing and software integration. *Familiarity with OO methodologies, signal processing, telecommunications, distributed systems, GUI builders, reusable software, relational data bases and automated test tools. *Looking for a BS/MS, CS or Math, or equivalent. Experience much more important than degree! DSP Engineers: *Lead, or be a part of, a multi-functional team working on digital/ DSP/communications subsystems utilizing state of the art design/analysis tools/techniques and modern packaging. Responsibilities may include: Participate in architectural and design trades, generate hardware and software specifications, simulate/evaluate DSP algorithm, design digital circuits used in standard and custom interfaces and micro-processor based designs, design high speed digital logic, develop/code DSP firmware, implement custom DSP hardware using PLD/FPGA, and test and integrate the hardware in the subsystem. *Languages: C or C++ in a UNIX environment, with knowledge of Windows NT a plus. Also, knowledge of high speed digital logic, Modem/PLL Datacom products, embedded control firmware, custom DSP hardware implementation, ASIC design and methodology a plus. Hardware Engineer: *Serve as Technical Leader for small to medium sized projects. Coordinate technical efforts of small engineering project teams. *Perform high level system design. Develop algorithms for key communication functions. Define architecture for state of the art communication products. *MSEE with emphasis on Communications theory and DSP. Minimum of 5 years of related experience. Experience with SPW, Xilinx, and Viewlogic design tools. Please also consider that we need Systems Engineers, Intelligence Engineers, a variety of Systems Administrators, and Directors also. The descriptions above are only general! There are multiple openings in every area, and the expertise varies from position to position, so please don't hesitate should you feel there might be a match! Again, this IS a fantastic company! You will not be wasting your time if you decide to respond, and ALL qualified candidates WILL be responded to very quickly. Thank you for making it this far, and we hope to hear from you soon! Contact us at: Hunter International Fax: (815)356-9225, or, cleaner@starnetinc.com Our thanks in advance, Dave Steiger...Article: 10717
Victor Levandovsky said... > On Wed, 10 Jun 1998 17:39:44 -0700, Gary Helbig <ghelbig@slip.net> > wrote: > > >I won't get into the VHDL vs/ Verilog argument. But, AHDL is good > >for Altera-made chips only, which limits its usefulness. > > > >AHDL also has poor documentation, and a few bugs. > > What bug? > > > > >In other words, Verilog, VHDL; pick one. Or both. AHDL, stay away. > > > >My opinions only, > >Gary. I'm missing an intermediate posting on this thread but... AHDL has one very compelling virtue: It is amazingly easy to pick up and use successfully. I've seen Xilinx schematics jockies develop working Altera devices, using AHDL for the first time, in a day. And they are amazed, and encouraged. I've seen the same folks try their hands at VHDL and come away muttering and frustrated. Say what you want about Altera's technology and their design tools, but that is a very compelling argument for a lot of my clients! I haven't seen a VHDL or Verilogs development environment yet that presents such a "shallow" learning curve. If you think there is a comparable product, in terms of ease of attaining moderate proficiency, I'd love to hear of it. **************************************************************** Bob Elkind mailto:eteam@aracnet.com 7118 SW Lee Road part-time fax number:503.357.9001 Gaston, OR 97119 cell:503.709.1985 home:503.359.4903 ****** Video processing, R&D, ASIC, FPGA design consulting *****Article: 10718
Gary Helbig (ghelbig@slip.net) wrote: : AHDL also has poor documentation, and a few bugs. Out of all the help files I ever seen on MS Windows systems, the on line AHDL documentation would have to be amongst the best - lots of info, well cross referenced. Graeme Gill.Article: 10719
Want a free computer? I mean REALLY free--no gimmicks or gags and really easy to do! Don't pass up this superb offer. Here's what you get: -IBM Pentium 233 MMX-classified processor (faster than Intel) -Mini-tower case -7 expansion slots (5 open) -32mb RAM -4.3gb hard drive (Quantum or Cyrix) -3.5 floppy drive -56k flex Fax/Modem -Acer 24X CD-ROM drive -16-bit MMX sound card -Cyrix Logic 5466 full-3D video card (DVD Compatible) -80-watt speakers -L2 meg cache w/ TXPIO motherboard -Digi-View 15" SVGA monitor -Windows95 and over 70 software titles Software includes: Microsoft Office97 Professional, Microsoft Internet Explorer 4.0, Groliers Multimedia Encyclopedia, 3D Key Design Center, 3D Card Design, Video Game Pack, Master Clip 6000 (clipart), Media Wizard Design, Bitware, Zoo Explorer, and much more! And...it's all FREE!!! "What's the catch?" you ask. The catch is not really a catch at all. All you need to do is sign up 6 people for this program under you. That's it! Sign up 6 people for this great program with you as their sponsor and you're done. You never need to do anything else with this program again. You will have 30 days to sign up these 6 people...which should be no problem. If you don't sign up all 6 in time, you will have to begin paying the lease until you have signed up 6 people. If you sign up 6 or more people, the lease will be paid for you and as an added bonus, you will receive $150 for every person that you signed up. Even if you don't get 6 people within 30 days, if you get one person you will have $150, which can be applied to the lease payment. "What happens when I've signed up all 6 people?" you ask. Once you have signed up 6 or more people, you are under no obligation to continue participating in this program. This program is incredible! "How do I get involved?" It's easy. Go to this web site: http://www.camelotenterprises.com/jklukan/ read through the information presented there, and if you're still interested, follow the instructions there. If you have any questions, feel free to e-mail me (this address and the one listed there end up at the same mailbox) and I'll try to answer them. Thank you for your interest! -Jeremy PS: This is not spam. I have posted it to newsgroups that I have found computer-related material in. Please do not criticize me for that. -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Keith Christensen <radiopro@gte.net > 309093 I am a Washington State resident! Spamming me is subject to a $500 (or more) penalty. Complaints WILL be sent to: your provider and the Wa. Attorney General's Office. Be warned that I reserve the right to post any unsolicited messages to public newsgroups as well!Article: 10720
In article <6lq05h$dj7$1@nnrp1.dejanews.com>, leslie.yip@asmpt.com writes >Dear William White > >How can I use SpeedWave to do testbenches? In the past, I used ModelSim's >V-System to do testbenches. Yet I know SpeedWave requires synthesis before, >right? > <Will's testbench guidelines snipped> No. SpeedWave is a Windows-based VHDL simulator supporting full 1076 VHDL and VITAL. So is ModelSim. Therefore you can simulate the design together with its testbench in SpeedWave both before and after implementation of that design. SpeedWave and ModelSim are two directly competing products, one belongs (via Viewlogic Systems) to Synopsys, and the other to Mentor. SpeedWave, Viewsim and VCSi are, respectively, the VHDL, Gate-level and Verilog simulation engines within the Viewlogic Fusion environment. The user sees a single GUI, with Fusion calling whichever simulators are needed during simulation of the various components of the design hierarchy. I hope this helps - I sense your confusion is merely to do with product names! In Will's absence (he's on the beach in Majorca!) -- David Pashley < ------------------------ < < < ---------- Mailto: david@edasource.com | Direct Insight Ltd < < < < > Tel: +44 1280 700262 | | < < < Fax: +44 1280 700577 | --------------------------- < ------------------------------------------Article: 10721
Hi, What are the fastest and biggest FPGA available in samples within next few months? The reason for me asking is that we have a large design that has to run fast, 66MHz. I know this does not sound too hard, but within these 15ns a 64 bit magnitude comparing (data greater or less than a register value) and 180 bit equality comparing has to be done. This plus alot of other logic adds up to over 100K ASIC gates. So the FPGA has to be big. We can accept pipelineing to some extent, but this requires also more resources in registers to hold data. The alternative is of cource ASIC, but our vloumes are low, so FPGA would be great. Has anyone done 64 bit magnitude comparator in an FPGA before? Does anyone have a feel about if this is feasable or not in an FPGA? Any input from you FPGA experts would be nice.Article: 10722
I am an experienced user of FPGAs, doing also conversion to ASIC from FPGAs. Yet, (as the conversion is done in Orbit, AMI, ...) I have lack of knowledge on the ASIC field (such as , but NOT only: the processes during the production, the impact on using gate-array vs. standard-cell and other issues.) I am looking for a recommendation on a tutorial book. Can any one suggest one ? Thank you in advance.Article: 10723
I am tasked with understanding an old design that used the Altera EP1210, which is equivalent to the Intel 5C121. Both of these PLDs are long since discontinued. I have the Jedec files but no logic equations. I have PLDShell versions 4 and 5, both of which appear to be too new to understand the EP1210/5C121. Altera's tech support had a datasheet for the part, but no old software available. Can anyone point me to, or send me, a version of PLDShell old enough to cover this part, or another means of disassembling the Jedec file to recover the logic equations? Thanks in advance, Bob Koblish -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 10724
larsher@online.no wrote: > The reason for me asking is that we have a large design that has to run > fast, 66MHz. I know this does not sound too hard, but within these 15ns > a 64 bit magnitude comparing (data greater or less than a register > value) and 180 bit equality comparing has to be done. This plus alot of > other logic adds up to over 100K ASIC gates. So the FPGA has to be big. > We can accept pipelineing to some extent, but this requires also more > resources in registers to hold data. Lucent does pretty good with wide functions. They claim to have their 3T80 part available now. Adding up the numbers for the -6 Speed grade, it can do a 64-bit arithmetic function in ~13ns. If you include routing delays, this will probably exceed your 15ns requirement. However, it's very easy to pipeline this function. Registers are essentially free. A 2-deep 64-bit pipelined comparator with input and output registers would take 9 PFU (Programmable Funcion Units) which is about 2% of the 3T80. (A single PFU can be an 8-bit comparator with 8-bit register at its input and 1-bit register at its output) My guess is that this could easily run at 100MHz with their -5 speed grade. A 180-bit pipelined equality comparator would also have similar performance. Each PFU can do a 16-bit equality-compare. SLICs (Supplimental Logic and Interconnect Cells) can be used to OR the outputs. My guess is that it would take less than 15 PFU to implement this function and could possibly be done in a single 15ns cycle. It would take 23 PFU if you want to include a 180-bit input register. The 3T80 has 484 PFU arranged in a 22x22 array. Also, if the number you are comparing to can be hard-wired (or firm-wired), the logic becomes much smaller and faster because it can be coded directly into the logic blocks. Then it might be possible to compare as many as 32 bits per PFU (probably limited by routing to ~24). > > The alternative is of cource ASIC, but our vloumes are low, so FPGA > would be great. > > Has anyone done 64 bit magnitude comparator in an FPGA before? > Does anyone have a feel about if this is feasable or not in an FPGA? > Any input from you FPGA experts would be nice. -- Don Husby <husby@fnal.gov> Phone: 630-840-3668 Fermi National Accelerator Lab Fax: 630-840-5406 Batavia, IL 60510
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