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
> I am a novice user of the Altera software and I am stuck with a design that > I run through Quartus 2000.03 and I get an error during compilation: clock > skew is greater then data delay, ciruit will not function. If this is Is this on extra effort? Did you try to assign a clique to a block? Are your setup/hold times set? Check to ensure that the settings in the Assignment Dis-Organizer are configured properly for your design. Also, did you specify a global clock, or did you "specify" the clock? Check that, and if it still gives you grief, let me know. Cheers, XanatosArticle: 22476
I have the following problem trying to run synthesis: FPGA Express Macro Compiler Version 3.1.1.0w Initialize DPM... Checking license... Checking license for Synopsys failed. I received and set up a license.dat file according to the supplied instructions, but apparantly it doesn't work. I am using the license.dat file (verbatium) as e-mailed to me by the online registration. For some reason they emailed me two attachments, one with linefeeds and one without. neither works. Since Xilinx does not offer live support I am unable to contact them, and I am unable to make use of their software. This is very aggravating. None of the problems in their FAQ or database provide a resolution to this problem. Please help. Thanx RobArticle: 22477
> Hi > I have an unusual request: we have a DataIO programmer that can't > program our Altera PROMs. We don't want to invest in an other programmer > so we are looking for someone/a company/anything else who owns an Altera > programmer near us (Courbevoie, France) who could program the PROMs for > us. > Thanks in advance Did you try the Altera distributors (Arrow?) themselves? They can usually pre-program the PROMs for you..... AshokArticle: 22478
Rennie Allen wrote: > > OneStone wrote: > > > > I probably just move in much older circles than you, > > Hmmm, I'm feeling pretty old right now, thanks for making me feel > younger ;-) > > > before the computer > > world went insane with jargon. > > What year, exactly, was that ? Was that the year Grace Hopper coined > the term "bug" ? Seems to me, that any substantial technology will > inevitably require its own terminology. And isn't it interesting trying to keep up with it all, hence the glossary is a great idea. > I have no problem with this, > I > simply want whatever terminology that is developed to be clear, and > concise (btw: I don't consider "bug" to be in the category of clear and > concise, although it is quaint). Bug is so well known it now transcends culture, as in different technical backgrounds. While mutex doesn't as can be seen from other replies. To each his own, my comment is simply that to include omne term, and omit another leaves the glossary incomplete. > > > Mutual exclusion mechanism smacks of the > > same sort of mentality that turned housewifes into domestic engineers, > > and directory enquiries operators into information consultants. Since > > I disagree. MUTual EXclusion is the function of the device. > Personally, > I find functional naming clearer than say, the term "bit flag" (which is > incredibly overloaded, and gives no clue as to the intended function of Flags are not necessarily bits, although the term is most often used in this way. I have used it for so long that I really have never felt the requirement for anything else. > the > device - and in no way conveys the requirement for atomic state change). > > Contrast the rationale behind this, to that behind the term "domestic > engineer". The term "domestic engineer" is more ambiguous than > housewife. > I contend that mutex is *less* ambiguous than "bit flag", since a mutex > can only be one thing, while a "bit flag" could be many things (much as > housewife can mean only one thing, while "domestic engineer" could mean > many things). I agree, sorry, poor comparison, although mutex only conveys meaning if you already fully understand the context, it might not be as meaningful when trying to explain soemthing to a mildly techniocal client. > > > pthreads are a POSIX multithreading standard I have never had cause to > > read about them extensively or use them in any embedded systems. And a > > review in Amazon showed that only the Butenhof book seemed to be really > > popular, the numbers sold on this subject hardly makes the book > > 'popular'. > > Hmmm, Amazon.com, as dearly as they may want us to believe so, is not > the > only source of technical reference material on the planet, hence your > data > (with regards to popularity) may be incomplete. I did check through 2 mmajor national book sellers as well, and one of the university book stores. The only one showing any real activity, as might be expected, was the uni store, even then, from an overall perspective of book sales related to programming it could not be termed exactly popular. > > Certainly, I didn't mean to imply that these books were popular relative > to, > say Stephen King novels :-) > > > Perhaps reasonably popular amongst POSIX programmers, but my > > A reasonable guess would be that every real-time Posix programmer, has a > copy > of at least one pthreads book. > > > HC12, HC16, and even my MCORE systems aren't running POSIX, definitely > > none of my 8 bitters do. I guess my response was too related to comp.arch.embedded, where I read it. I'm not aware of too many embedded POSIX systems. > > Understood. My original point was simply that I don't believe the > Embedded > Systems Glossary was being obscure, simply by including the term mutex; > in > fact, it appears to be a necessary inclusion (since the purpose of a > glossary > is to explain terms). Your point was that the term "bit flag" should be > included, and while this would be nice, I can understand why someone > attempting > to produce a concise glossary would avoid this term (due to the fact it > is > extensively overloaded). But then the glossary is incomplete without it. The author asked for comment, I sent a few comments to him, hopefully in a constructive manner, not suggesting that any term be excluded, in fact the very existence of the glossary has forced me to go and have a look at a couple of items which were known to me, but not in everyday usage. I simply suggest that a glossary of embedded systems programming, which I believe this is, should include as many terms as are in common use. I personally feel that flag (and semaphore) are more common amongst embedded systems programmers, I could of course be wrong, I often am. AlArticle: 22479
PeterS wrote: > > >> I don't like the use of "flag" in such specific terms - to me, a flag is > >> simply an > >> indication of state, with no protection mechanism for setting or > resetting. > > > >The original system of semaphores, for signalling, used flags. A > >semaphore is simply a signal, as is a flag, flags are not necessarily > >binary, originally semaphore, the method of singalling, each with > >different context sensitive meaning. > Them ancient fingers don't work so well at 3am any more. To correct my cock up:- the method of signalling used several flags of different designs, each with different context sensitive meaning. > Hmmm, interesting sentence. Wonder what it means ;-) > > Chains of simple signal flags were also used to "post" entire messages from > the rigging > of sailing ships (like Nelson's "England expects this day..." signal at > Trafalgar), > but there was nothing, I guess to stop anyone else (ie no interlock) putting > up a > message from their own rigging, which is why I dislike the interchangeable > use of > "flag" and "semaphore". The interlock, in Napoleonic times was the yard arm, as dispensed by british maritime law, disobey the signal and suffer for it. I guess that makes Fletcher Christian a 'bug'. Al > > >I believe it originated in the 18th century. > > > >Semaphore noun [U] > > a system of communication using two mechanical arms or > > hand-held flags which are moved into different positions to > > represent different letters, numbers or symbols > > Semaphore was widely used at sea, before the advent of > > electricity. > > > > (figurative) When I lived opposite her we would send > > semaphore signals (=messages without speaking) to each > > other from our bedroom windows. > > > >Al > > I live near to a restored semaphore tower dating from the time of the > Napoleonic wars > which signalled from the Admiralty in central London down a chain of towers > to > Portsmouth and Plymouth.( Just off the A3 outside the M25.) > It has working arms, and an interesting display model of an earlier > system which used six flat, square shutters which could be individually > displayed > upright, face-on, or rotated horizontal. Somewhat OT, I realise, but a > fascinating > monument. Oh yes, a short message took less than 15 minutes from London to > Portsmouth, weather permitting (50 odd miles, 80+ km). Amazing how these ancient systems are faster thean the net can be some days ;@}Article: 22480
Rennie Allen wrote: > > PeterS wrote: > > > > > > >The original system of semaphores, for signalling, used flags. A > > >semaphore is simply a signal, as is a flag, flags are not necessarily > > >binary, originally semaphore, the method of singalling, each with > > >different context sensitive meaning. > > > > Hmmm, interesting sentence. Wonder what it means ;-) Hung the flags upside down. ;@} > > > > Chains of simple signal flags were also used to "post" entire messages from > > the rigging > > of sailing ships (like Nelson's "England expects this day..." signal at > > Trafalgar), > > but there was nothing, I guess to stop anyone else (ie no interlock) putting > > up a > > message from their own rigging, which is why I dislike the interchangeable > > use of > > "flag" and "semaphore". > > I agree absolutely. IMO, clarity is not served by overloading existing > terms > with new meanings. Sometimes, it is simply necessary to create a new > word to > describe a specific concept. > > Rennie But this is exactly one of the arguments used in favour of C++ and other object oriented languages, here overloading is portrayed as a 'good' thing. I don't agree, the original post was not about likes and dislikes, simply completeness of a glossary.Article: 22481
Rennie Allen wrote: > > David Brown wrote: > > > > Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>... > > >On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote: > > > > > >>Result of check: I asked a CS friend (he has a masters degree and is > > >>active in the profession) if he knew or could guess what a mutex is. He > > >>had a vague memory that it is a species of mosquito. > > >> > > >>Jerry > > > > > >Funny, I study CS in the Netherlands and the theory of operation > > >systems (like synchronisation algorithms, multitasking theory) was > > >part of the first half of the bachalors program, dealing with > > >semaphores, mutexes, critical regions, producers/consumers, deadlocks > > >etcetera. Guess it depends on the university where you're studying... > > > > > > > It certainly does. Some universities and courses will cover things like > > Windows programming, which does not really have anything to do with CS, but > > which looks good to PHBs for later employment. > > Are you implying, that mutex is MS-speak ? It isn't (although the > ingrates > do use the term .... there should be a law .... ;-) > > Rennie There is, and finally they've (michael soft) run foul of it, or so it seems. ;@}Article: 22482
OneStone wrote: > > Rennie Allen wrote: > > But this is exactly one of the arguments used in favour of C++ and other > object oriented languages, here overloading is portrayed as a 'good' > thing. Hmmm, I've seen a little too much of this "good" thing in C++ code. Inappropriate semantic overloading, is one of the most abused facilities of C++ IMO. > I don't agree, the original post was not about likes and > dislikes, simply completeness of a glossary. Definately off topic. RennieArticle: 22483
Ray Andraka wrote: > > If you look at the FPGA architectural features as they apply to a computer > design, I think you'd wind up settling on the Xilinx architectures. First, > you'll probably want a fast carry chain so that you can do fast arithmetic > with minimal logic. That rules out pretty much everyone except Xilinx, > Altera and Lucent. I haven't seen any commercial boards with Lucent, so bye > bye Lucent. Xilinx has a capability of using a LUT (a four input look-up > table which is the basis of the FPGA architecture) as a small RAM or shift > register. This makes for very compact register files, pipeline delays, > reordering queues etc. In Altera, each bit of storage chews up another > logic element unless you use the buld memories. Also, Altera's carry logic > structure is not as powerful as the Xilinx structure, which means you will > probably need two levels of logic for real-world arithmetic vs 1 level of > logic in Xilinx. > > If you want it small, you could always go to a bit serial design. A pretty > decent bit serial scientific calculator will fit in an XCS-05 Hey Ray! What about the PC104C31 which uses not only one, but four Lucent FPGA chips? I realize that our price does not fit the $300 limit imposed in this case, but don't say there aren't any commercial boards around that use the Lucent chip! :) Of course our board is not really intended to be an FPGA evaluation board, but rather a DSP board with FPGA capability. But it is so loaded with FPGA capability that anyone needing a lot of FPGA in multiple chips would do well to look at it and just consider the DSP to be a fancy boot loader. In fact the OR3T30 or OR3T55 that can be used on our board would be a very good choice for an attached coprocessor for the TMS320C31 that is on the board. I believe that APS also has a Lucent board much like their APS-X84 board although they may not be marketing it. They sent me a board at one point as a beta test which I never got around to using. > Andy Holt wrote: > > > This is the sort of thing I would expect to be an FAQ, but there doesn't > > seem to be one for this group. > > > > I have been thinking about "playing" with an FPGA both from the view of > > learning about an interesting-looking technology and with the hope of > > constructing an emulation of a '60s mainframe (more about this later). > > > > I am looking for advice on low-cost ways of doing this (Let's say price > > ceiling of about £200 [$300]). It seems that I am going to need two main > > things: > > * A package of software. > > * A prototyping/evaluation board. > > > > Taking the second of these first there seem to be few choices available > > (without paying lots of $$$) - Our board is a lot of money by comparison. The single piece price I have been quoting is between $2000 and $3000 depending on features and speed. But this will be changing to something a little more palatable at the end of the month. And of course the price is much better at quantity. ...snip... > > As for software, the Kanda and Atmel packages come with some, for the > > Xess one I would also have to spend another $100 for the Foundation > > student edition. The PC104C31 comes with a boot monitor program to allow you to program the Flash, boot the FPGAs and perform a lot of testing of the various parts of the board. Full source is provided along with the designs for the FPGAs. A low end package for the Lucent FPGAs is available for $150 very much like the introductory package from Xilinx. This package includes Viewlogic schematic capture and simulation tools. I also believe it includes VHDL synthesis. > > ** so, first question: any known "gotcha's" with the above alternatives? > > [ISTR a recent hint that the Atmel software was weak in one respect - > > it is noticeable that their web site seems to say almost nothing about > > its functionality - and the low cost version of Foundation doesn't > > include VHDL?] > > Are there other reasonable options? > > > > The other main question I have concerns estimating how big an FPGA I > > would need for the mainframe emulation. I assume that the "usable gate" > > counts for all devices tend to be as much marketing as technical > > statements. I have detailed (but only "almost complete") descriptions of > > the logic design for the mainframe that I am interested in (ICT 1905 - > > aka FP6000) and I can be reasonably confident that it has less than 6000 > > gates including FPU ... probably less than 4000 without. Is this likely > > to fit in a "10000 gate" FPGA?, a "20000 gate" one, or whatever? Keep in mind that your gate count also needs to include the IO system for accessing the memory and any peripherals that will be attached. This can easily double the size of the design. Yes, you are right about the gate counts for the FPGAs. But the largest part of the gate inflation comes from the inclusion of memory in the gate counts. They make certain assumptions about the percentage of the memory used. Your design may use more or less of that memory. This will change the used gate count dramatically. -- Rick Collins rick.collins@XYarius.com remove the XY to email me. Arius - A Signal Processing Solutions Company Specializing in DSP and FPGA design Arius 4 King Ave Frederick, MD 21701-3110 301-682-7772 Voice 301-682-7666 FAX Internet URL http://www.arius.comArticle: 22484
"R. T. Finch" schrieb: > I received and set up a license.dat file according to the supplied > instructions, but apparantly it doesn't work. I am using the license.dat > file (verbatium) as e-mailed to me by the online registration. For some > reason they emailed me two attachments, one with linefeeds and one without. > neither works. One of my colleagues had such a problem when he received a license via email. The problem were missing \ (backslash) at the end of the lines in the license. Maybe tjis can solve your problem. regards Joachim -- Joachim Hoch Fraunhofer Institut Physikalische Messtechnik Heidenhofstr. 8 D-79110 Freiburg im Breisgau Tel: +49(0)761/8857-149 Fax: +49(0)761/8857-224Article: 22485
I do not know what problem you have. if you describe 2 or more consecutive registers without reset and only shif funktion, you always get a shift reg, iven if a single FF and a smallae shift reg would be better. This works at all versions of FPGA-Compiler II after 1.1.2000 With best regards E.BlaschekArticle: 22486
Rennie Allen wrote: > > OneStone wrote: > > > > Rennie Allen wrote: > > > > > But this is exactly one of the arguments used in favour of C++ and other > > object oriented languages, here overloading is portrayed as a 'good' > > thing. > > Hmmm, I've seen a little too much of this "good" thing in C++ code. > Inappropriate semantic overloading, is one of the most abused facilities > of C++ IMO. Couldn't agree with you more. > > > I don't agree, the original post was not about likes and > > dislikes, simply completeness of a glossary. > > Definately off topic. > > Rennie But an embedded systems glossary is on-topic, surely? Cheers AlArticle: 22487
Yes, use a "alexander-decoder" for phase recovery either with 3 times oversampling or with 2 consecutive externel delay-lines and an external VCO for clock recovery or an alligner with best regards E.BlaschekArticle: 22488
On Wed, 10 May 2000 07:33:33 +0200, Joachim Hoch <joachim.hoch@ipm.fhg.de> wrote: >"R. T. Finch" schrieb: > >> I received and set up a license.dat file according to the supplied >> instructions, but apparantly it doesn't work. I am using the license.dat >> file (verbatium) as e-mailed to me by the online registration. For some >> reason they emailed me two attachments, one with linefeeds and one without. >> neither works. > >One of my colleagues had such a problem when he received a license via email. >The problem were missing \ (backslash) at the end of the lines in the license. >Maybe tjis can solve your problem. I received two, as well. The files use CR LF or else just LF, I think: MS-DOS or UNIX defaults. Probably their software uses either, so it's more a convenience for those wanting to use TYPE or cat to display the contents. Mine appears to work just fine, though I'm no expert on the subject. One thing to check is the SET variable LM_LICENSE_FILE and make certain that the license file sits in that same directory. In my case, I think that's in fndtn\data. JonArticle: 22489
In article <NQ3S4.67488$VR.1345263@news5.giganews.com>, R. T. Finch <robfinch@cyg.net> writes >I have the following problem trying to run synthesis: > >FPGA Express Macro Compiler >Version 3.1.1.0w > >Initialize DPM... > >Checking license... > >Checking license for Synopsys failed. > > We use this system on our PCs when we setup courses, and it works. We normally put the file in c:\flexlm\xilinx.dat and then add it to the environment variable LM_LICENSE_FILE, e.g. c:\flexlm\xilinx.dat;1700@fred if you have floating licenses on server "fred" as well. As far as I know, if you put it in the foundation installation directory, you probably don't need to set LM_LICENSE_FILE. If you put it anywhere else, you should set LM_LICENSE_FILE regards Alan -- Alan Fitch DOULOS Ltd. Church Hatch, 22 Market Place, Ringwood, BH24 1AW, Hampshire, UK Tel: +44 (0)1425 471 223 Email: alan.fitch@doulos.com Fax: +44 (0)1425 471 573 ** Visit THE WINNING EDGE www.doulos.com **Article: 22490
Rennie Allen wrote in message <391835C4.8ACF6356@computermotion.com>... >OneStone wrote: >> >> I probably just move in much older circles than you, > >Hmmm, I'm feeling pretty old right now, thanks for making me feel >younger ;-) > >> before the computer >> world went insane with jargon. > >What year, exactly, was that ? Was that the year Grace Hopper coined >the term "bug" ? Seems to me, that any substantial technology will >inevitably require its own terminology. I have no problem with this, I >simply want whatever terminology that is developed to be clear, and >concise (btw: I don't consider "bug" to be in the category of clear and >concise, although it is quaint). > The term "bug" goes at least as far back as Thomas Eddison - one of his journals refers to a hardware problem in a design as a "bug".Article: 22491
Rennie Allen wrote in message <391837BA.D0B39CFB@computermotion.com>... >David Brown wrote: >> >> Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>... >> >On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote: >> > >> >>Result of check: I asked a CS friend (he has a masters degree and is >> >>active in the profession) if he knew or could guess what a mutex is. He >> >>had a vague memory that it is a species of mosquito. >> >> >> >>Jerry >> > >> >Funny, I study CS in the Netherlands and the theory of operation >> >systems (like synchronisation algorithms, multitasking theory) was >> >part of the first half of the bachalors program, dealing with >> >semaphores, mutexes, critical regions, producers/consumers, deadlocks >> >etcetera. Guess it depends on the university where you're studying... >> > >> >> It certainly does. Some universities and courses will cover things like >> Windows programming, which does not really have anything to do with CS, but >> which looks good to PHBs for later employment. > >Are you implying, that mutex is MS-speak ? It isn't (although the >ingrates >do use the term .... there should be a law .... ;-) > Not at all. MS might use the term (although only a very small proportion of Windows programmers will ever come across it), but I was implying that you will meet the term more often in courses with a solid theoritic basis, such as Johan seems to be studying in the Netherlands, than in simpler courses covering specific systems (such as Windows) aimed at immediate employment.Article: 22492
On Tue, 9 May 2000 15:20:01 +0200, "David Brown" <david.nospam@westcontrol.com> wrote: >Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>... <snip> >>Funny, I study CS in the Netherlands and the theory of operation >>systems (like synchronisation algorithms, multitasking theory) was >>part of the first half of the bachalors program, dealing with >>semaphores, mutexes, critical regions, producers/consumers, deadlocks >>etcetera. Guess it depends on the university where you're studying... >It certainly does. Some universities and courses will cover things like >Windows programming, which does not really have anything to do with CS, but >which looks good to PHBs for later employment. Others will teach you the >theory of computer systems, and will cover semaphores, etc., in a manner >which can then be applied to any system. Part-time courses and newer >universities tend towards the former, while older, more accademic >universities tend towards the later (I know that's a terrible, >unsubstantiated generalisation, but it is basically true). > >Dutch universities will of course tend towards the theoretical - after all, >a high proportion of the work in synchronisation was done by Dutch >mathematicians such as Djikstra (spelling?). It's Dijkstra (turn the i and j the other way around). We also have Andrew Tanenbaum hanging around here, who has done a lot of research in OSs, being responsible for Minix, the OS that inspired Linus Thorvalds. >>Johan.Article: 22493
On Tue, 9 May 2000 14:06:53 +0100, "PeterS" <peters AT aston desIGNS comMERCIAL> wrote: >Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>... >>systems (like synchronisation algorithms, multitasking theory) was >>part of the first half of the bachalors program, dealing with >>semaphores, mutexes, critical regions, producers/consumers, deadlocks >>etcetera. Guess it depends on the university where you're studying... >> >>Johan. >> >><snip> >> >I studied CS some 20 years ago and also covered semaphores, but don't ever >recall the term mutex, until I came across it a couple of years ago doing >some NT stuff. Is "mutex" purely Microsoft-speak for what everyone else >calls a semaphore? No. As described earlier in this thread, these are two different things, at least in theory. A semaphore can handle multiple instances of a resource. Say you have 10 free buffers for IO, then a semaphore is used to assign them to processes; it holds a value from 0 to 10 to indicate how many are used. A mutex handles one particular instance, for example a specific file, which one proces at a time can use (therefore mutual exclusion). It has little to do with Microsoft. >I don't like the use of "flag" in such specific terms - to me, a flag is >simply an >indication of state, with no protection mechanism for setting or resetting. Agreed. >BTW Didn't Dijkstra first describe semaphores, and wasn't he from the >Netherlands? > (not that its got anything to do with anything, much ;-) Dijkstra (who indeed was from the Netherlands) first described an algorithm for preventing multiple access to the same item, that he compared with a semaphore at a railway; allowing you to pass on or wait for the other train to pass. Johan.Article: 22494
In comp.realtime Michael Barr <mbarr@netrino.com> wrote: : I have put a lot of work into writing the book and creating and : maintaining these online resources. I hope that they will be a : valuable contribution to the community of embedded systems hard- : ware and software designers. When still a student I actually bought and read your book. At least I think it was your book... I had just finished a major hardware project centered around a TMP68303 mcu with the ever popular 68000 cpu kernel. Around it was everything I could come up with, including DRAM, network and SCSI. I was then faced with the task of programming this thing. Of cause, beeing a student, I choosed GCC for the task. (Still does, but that is another history) I had used GCC on Linux before, but now I suddenly had to boot this machine myself, handle interrupt myself, set up DMA myself and - jiiiha - control supervisor/user space myself. Not to mention LibC. There is documentation on GCC, but almost nothing on how to do this hardware fiddling. In retroperspective, that is not a GCC issue. But green as I was, this is where I expected to find it. Well I didn't find it and I turned to your book for help. It helped all right, just not where it hurt the most. Semaphores and message passing is fine per se, but I need to boot that bloody system before I could start thinking about those issues. So my suggestion is that you reconsider what your audience is, or else find a new title for your book. Write a couple of chapters on how to set up the simplest of ROM monitors. Don't be afraid to use assembly - green programmers like me is much more scared of C. Assembly I can see exactly what does, while C is a black box where everything can happen. And boy, was I scared of the dark... Just don't overdo the assembly part - demonstrate the principles. Show intermixed C and assembly listings. Demystify the "crt0.s" issue - after all, all it does is set up a valid C runtime environment. All it needs to do so is some memorymapped ram and some rom to store itself and the main application. And a reset vector of cause... An issue that has linker written all over it. If someone would just write so, damn it! Which leads into the issue of object code sections - .data, .text, .bss. These are not that scary if someone would just point out what parts of the C code goes where and why there will not be any leftover C constructs that does not fit into one of these three sections. This requires a look at cpu addressing modes versus automatic and global C variables. Again - don't be afraid of explaining it in terms of hardware, it is exactly this understanding of the interaction between C code and the resulting hardware instruction issuing I was looking for. I already knew hardware - I wanted to expand this understanding into the abstraction of C/C++ code. It was not until I read up on COFF that I finaly came to peace with what the C compiler does - and more importantly where the *linker* took over. If only someone had told me how object formats works. Placeholder symbols is a natural consept everyone can understand if only told about them instead of beeing patted on the back and told that this is beyond you. Relocation by the linker is then just as natural. And then it becomes just as obvious why "crt0.s" does what it do *and nothing more.* (Yes, it was just as big a problem for me that "crt0.s" looked far to simple - I was looking for problems that was just not there until I finally realised what the the actual problem constitued of.) To conclude - include an example hardware setup in your book - cpu, ram, rom and RS-232. No black boxes including on-chip ROM! Block diagrams plus memory maps are sufficient. I suggest the 68K for cpu. There still is no cleaner and more academicaly correct cpu. And it has got a vector table - an important issue. Then demonstrate how to bring this machine up and running using GCC and Newlib. This ROM monitor should be capable to communicate over RS-232 to a host and possibely know how to download and run ready linked binaries using the kermit filetransfer protocol. This way you have both a terminal emulator and file transfer software on the host. There is a 68K systems book by Clemets that goes a long way towards this goal that I learned a lot more from than your book. I can find the ISBN number if you want. My two cents plus shipment and handling... -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.and.so.forth ******************************************************Article: 22495
Hi all, we are targeting an PCI-based network interface ASIC prototype on VIRTEX1000(E), but I found only very few boards which meet our requirements (PCI_64, VIRTEX1000(E)). One of this is the DN2000k10 from the Dini Group (http://www.dinigroup.com). The problem with this board is, that it is not compatible with the PCI LogicCore from Xilinx and I'm not willing to design a PCI-Interface. Does anyone has experience with this board? Does anyone has an idea for a appropriate board?? Thanks Patrick -- Patrick Schulz (schulz@rumms.uni-mannheim.de, pschulz@ieee.org) University of Mannheim - Dep. of Computer Architecture 68161 Mannheim - GERMANY / http://mufasa.informatik.uni-mannheim.de Phone: +49-621-181-2720 Fax: +49-621-181-2713Article: 22496
I bought a spartan device, Xcs10 pc84. I thought it was a 5V device. But when I use 5V Vcc all pins are high... So, when I use 3.3V Vcc it workes okej. But my question is... When I look at the topp of the device, I it lookes like this.... /------------\ |xilinx |xcs10 |PC84CKN9837 |A2014956A |3C \____________ Is this a 5V device or a 3.3V device? Or is there anything wrong? Thankful for help Björn Lindegern Sent via Deja.com http://www.deja.com/ Before you buy.Article: 22497
Il s'agit d'un message multivolet au format MIME. --------------F69C531C6C324827424F03BB Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Hello, You could have a look to http://www.mvd-fpga.com/ -> products -> tiny prog Greetings, Emmanuel Rick Filipkiewicz a écrit : > Dan wrote: > > > The Data IO programmers cost $1500ish to program Xilinx SPROM. Does anyone > > know of a more reasonably priced one ? > > > > Thanks Dan > > If you're in smallish volume why not consider the reprogrammable 1800 series > devices from Xilinx. They are JTAG programmable. Once you've got the design > stable you could get your SPROM distributor to program one-time ones for you. > > Alternatively avoid SPROMs altogether & use a standard EPROM+small CPLD to > generate the bit stream. --------------F69C531C6C324827424F03BB Content-Type: text/x-vcard; charset=us-ascii; name="emmanuel.lecomte.vcf" Content-Transfer-Encoding: 7bit Content-Description: Carte pour Emmanuel Lecomte Content-Disposition: attachment; filename="emmanuel.lecomte.vcf" begin:vcard n:Lecomte;Emmanuel tel;fax:05 61 06 72 60 tel;work:05 62 13 52 32 x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:emmanuel.lecomte@mvd-fpga.com fn:Mutli Video Designs end:vcard --------------F69C531C6C324827424F03BB--Article: 22498
It is not on extra effort; I don't know about assigning a clique? to a block?; my setup and clock->out's are constrained; I will look into the Assignment Dis-Organizer settings; I specified the clock (not global). I did find one thing, if I do a context help on the error, it tells me I need to add lcells to increase data delay to the 7 problem paths, so next I will try to figure out how to do this. Thanks. Don Xanatos wrote: > > I am a novice user of the Altera software and I am stuck with a design > that > > I run through Quartus 2000.03 and I get an error during compilation: clock > > skew is greater then data delay, ciruit will not function. If this is > > Is this on extra effort? Did you try to assign a clique to a block? Are your > setup/hold times set? Check to ensure that the settings in the Assignment > Dis-Organizer are configured properly for your design. > Also, did you specify a global clock, or did you "specify" the clock? > > Check that, and if it still gives you grief, let me know. > > Cheers, > XanatosArticle: 22499
Rickman <spamgoeshere4@yahoo.com> wrote: > Ray Andraka wrote: > > .... I haven't seen any commercial boards with Lucent, so bye > > bye Lucent. > > Hey Ray! What about the PC104C31 which uses not only one, but four > Lucent FPGA chips? I realize that our price does not fit the $300 limit > imposed in this case, but don't say there aren't any commercial boards > around that use the Lucent chip! :) A German company makes a PCI board with the Lucent FPSC (FPGA with embedded PCI) chip: http://www.morethanip.com/products/or3eval/index.html They even have a Linux driver for it. Lucent also has an evaluation board with the PCI/FPGA chip. At one time they planned to sell it for $250 (it's pretty much just the chip on a board with a few leds, switches, and daughterboard connectors). Now they seem to have some weird plan to loan these boards to customers. -- Don Husby <husby@fnal.gov> http://www-ese.fnal.gov/people/husby Fermi National Accelerator Lab Phone: 630-840-3668 Batavia, IL 60510 Fax: 630-840-5406
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