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
rk: : >: >don't know why people think you can't do good work with non unix/linux : >: >systems phil: : Perhaps some people can't understand how you can get any work done at all : on Win95 (and to a lesser extend with NT). : : Win95 and NT lock you into a graphical paradigm for tools that is : difficult to automate. Let's say I'd like to run several simulations over : night and keep track of potential failures in each. This is easy in : Unix/Linux. Even if a tool has a GUI, it always has a way to execute from : the command line - this makes it very easy to put together a perl script : to automate the overnight simulation runs. In the Win world, most apps : don't have a way to execute from the command line - so you've got to stay : around all night and push the buttons if you want to batch up several : runs. rk: don't know about all of my tools (since not the biggest issue haven't looked into it, just run them off of the home pc, keep phone line open), but most of them do have alternate interfaces via the command line; the gooey is just a shell. and running them via the internet is trivial. rk: : >does microsoft office run on linux? need that to communicate these days. phil: : Who needs it? There are other ways to 'communicate'. On Linux there is : StarOffice (can read and write M$Word doc format), Applix (can also read : and write M$Word format as well as others), WordPerfect, LyX (a GUI : frontend for LaTeX - great for equations, MSWord can't touch it). rk: if it can read and write the latest formats (word, excel, and powerpoint are the standards) then that's fine. and it should accept objects from all other programs too. are these programs available for win '95? if so, then there's an alternative that i think many would be willing to try. rk: : >does aol run on linux? very popular in america. phil: : AOL is a joke. Who needs it, get a real ISP ;-) rk: the people who use it, very popular application, easy to use. got lots of users and the discussion was that people use win and not linux cause they don't want to re-install os. but, there are apps that people want to run. and aol is very popular with many professionals, including those that are involved in cae. cheap, and can dial in from anywhere. real isp is a long distance call. expensive for many. and much harder to use for the typical pc user. sorry if aol is such a joke, perhaps those 10 million people are all idiots. or perhaps that just want to use the internet easily and are less concious of what the 'real' professionals think. rk: : >visual basic run on linux? phil: : Again, who needs it? Especially with the existence of the following on rk: people with a lot of exisiting vb code. why should the switch to linux and rewrite code? why should they make their life harder? why should they learn new tools? phil: : Linux: : : Perl (and Perl/TK) : TCL/Tk : Python (and many different graphical extentions) : Eiffle : Scheme/Tk : Java : ... others too numerous to mention... : : Given the availability of these languages, who in their right mind would : us VizBasic rk: personally i'm not a visual basic user. but there are a lot of people who are. gets the job done for them and they say it's very easy to use and do good thinks with. guess they're all crazy. i'll tell them all immediately! ;) p.s. one of the gals i work with is doing some java stuff, very good programmer. if she could, she'd switch to vb or delphi, but they won't let her. much easier to use (or is it the learning curve?). rk: : >national instruments, do they supply ieee-488 drivers for linux? phil: : I donno... but I'll bet that someone has made a driver for Linux somewhere : (just may not be National Inst) rk: if it doesn't run *ALL* of the national cards i gotta shut my lab down. and so do a lot of others. or get new cards, new driver, and then right an interface module so i don't have to revise about 8 years of code, developed my multiple people. *that's* more important than the os, which i don't particularly care about. rk: : >labview run on linux? : >matlab run on linux? phil: : I believe the answer is yes here (but not too sure). If not there are : several free equivilents that are available (Octave comes to mind). : Mathmatica runs on Linux. rk: oops, good point, left off that last one. again, are they code compatible and humanoid interface compatible with labview and matlab? lot of files and training time. and training and converstion time, for a typical professional, runs $50 - $100 / hour. it's expensive to change. it's inertia. and that does figure into decisions, not just technical merit. rk: : >orcad run on linux? phil: : Now to address the EDA apps together: : : No, most of them don't run under Linux and that's the point of this : thread. Many of us in the design community would like to see some of : these apps on Linux. The fact that they are not here now does not point to : some technical deficiency in the Linux OS. It merely points to the fact : that a large, well financed company with monopolistic tendencies has been : able to convice the EDA companies that its OS will ultimately rule the : universe. The OSs from Redmond are not in the position they are in : because of any technical merit, but because of the marketing muscle (and : strong arm tactics) of Micro$oft. rk: well, i agree with a lot of that <rk quickly checks his pulse>. many of the cae tools run on unix and not windoze, so that part doesn't hold up. and i'd guess that it's an easier port from unix to linux. again, a problem with the argument. and, as we've discussed, there are certain designs which won't fit the pc model but need something more. doesn't mean that you can't do a lot of nt or linux, it's just different niche. now w.r.t. a certain company, who's head got his *ss hauled before congress this week (live right outside d.c.), yes, they have used marketing muscle and strong arm tactics to get where they are, not technical merit (is this the olympics?). it's interesting, when win 3.1 finally was abandoned by s/w developers, how many developed for '95 and how many for nt. perhaps i'm a conspiracy theorist! ;) rk: : >viewlogic run on linux? : >actel designer run on linux? : >xilinx tools run on linux (pete)? : >altera tools run on linux? : >orca tools run on linux (stu)? : >synplicity run on linux? : >examplar run on linux? i think this is a yes, do they still support? : : Answer was yes here, not too sure now. phil: : >aldec run on linux? : >does it network directly with win 95/nt machines with drag and drop? : >delphi run on linux? : See the languages discussion above. rk: yup, but here's the key point. let's say one is developing on both windoze now and unix. windoze for small to medium jobs, say fpgas. and unix for medium sized asics. and win '95 is about to be replaced in a years time with the new version. so, it's time to think, perhaps, about upgrading. and when you get new machines, what do you put on it? obviously, on a design team, you want similar configurations for all team members. do you go to linux, for the sake of argument, technical merit? do you stay at '95 since it does the job and wait things out? do you go to nt 5.0, run existing apps (eda and non-eda) and bet that that's where the action will be? do you go to straight unix/sun configurations? note: for windows choices, keep unix/sun for asics. *that* is the question, as if the chips and boards and boxes don't flow out, $ doesn't flow in. we're all in business of one sort. from what i've read here, from people all over the world, it doesn't look like linux will be a good bet in the 6-12 months time. not from performance, technical merit, or anything like that. from s/w availability. rk: : >sigma plot run on linux? phil: : gnuplot (and its X frontend). Several plotting programs are available : (most free). rk: is it compatible? a lot of people use this package (like the wife!) so i gotta be compatible. again, file conversion/training time. this case, mostly training time. <snip a bunch of apps - apparently there are some equivalents if you're resourceful and can find them, eliminating some conversion hassles> rk: : >can i play the cd-rom interactive disks for the kids on linux? phil: : What's this got to do with getting your design done? rk: oh, we use the 'puter at home for lots of stuff. business (cae) and other stuff. don't want to lose capability if i upgrade. same to other stuff below. rk <who gotta proof read better> : >for engineering. don't know answers, perhaps you can feel us in. phil: : Well, I'm not that kind of guy. ;-) rk: good, me neither. just sometimes types too fast. rk: : >like i say, "it's the s/w, stupid." [paraphrased from the pres.] who : >cares about the os or cpu. it's the apps. phil: : Yeah, and its the stupid software too ;-) But seriously, the OS does have : an impact in how you get your work done. If the GUI gets in the way (as : it does with WinXX) it can be a hinderance. rk: well, you have showed some good alternatives and answered some questions. thanks! and you did better than the guy who doesn't want to move his hand all the way over to the arrow keys. yup, some programs, too much clicking. even the 'explorer.' i do a lot of my navigating with the keyboard, much faster. sometimes better with mouse. a lot of the cae you can run from the command line and batch files. that's something that's easy for eda vendors to change for many apps. but, at day job, they run many of the same apps on both unix/sun and win/pc. and that keeps training time/costs reasonable and shields you, to a certain extent, from the os. if the mainstrain cae apps aren't there, it will be hard to switch to linux. chicken and egg. and will vendors translate sun --> win or sun --> linux or none? and on the other side will they go from pc --> linux or pc --> unix or none? it'll be interesting to see! -- -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) --------------------------------------------------------------Article: 9376
Mark Willey wrote: > This conversation has evolved(?) beyond the scope of any of the forums that > it's posted to. Please move to a more appropriate venue. I would suggest > some advocacy groups.... which I never read. :) Indeed. Unfortunately, this discussion, and any discussion like it, has no home to go to. Perhaps comp.cad.synthesis is closest, but this isn't a comp. discussion, really. comp.arch.fpga never seems to get used for its intended purpose, so perhaps that's the correct place for toolset issues (for those of us hacking fpgas, at least), and, despite appearances, this isn't really an OS war, certainly from where I'm standing... If I knew how to set followups in this godforsaken newsreader, I'd point at comp.cad.synthesis. If anyone else would like to steer us over there, that'd be great. Cheers, SteveArticle: 9377
try signal processing technologies, there are others, too. spt can take you up to 1 GHz. -- -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) -------------------------------------------------------------- Lucian Fogoros <l.fogoros@popmail.csuohio.edu> wrote in article <35019C53.1421@popmail.csuohio.edu>... : We would like to know where we could get 100 MHz A/D's? :Article: 9378
Hi : Are you check Oberon S3 Native, In my opinion one of the best OS today Take one second to check it at http://inf.ethz.ch Walter. rk <rich.katz@gsfc.nasa.gov.NOSPAM> escribió en artículo <01bd493d$15a1cd40$1e83accf@homepc>... > misha: > : There is a significant advantage of Linux over Windows concerning > portables. > : I was very surprised when I touched the processor inside my desktop when > : it was running Linux. It was COLD. So was the power regulator. Under > : Windows95 the processor is always hot, even when OSR2 System Monitor > shows 1% > : CPU utilisation for an hour. Same about NT. > : > : Of course, when CPU-intensive application runs under Linux, the processor > : warms up. But for typing text on the portable Linux should give much > longer > : battery life. > > interesting. i haven't seen any of the rags that report on portables and > battery life consider this. it would make a good benchmark since a lot of > people do x-country traveling. now, the portables do have a lot of power > savings features and do a bunch of stuff to extend battery life so the > difference on the desktop pc may not equal difference on a lapdog. but it > would be interesting to see, for a portable, a comparison for running a > simple text editor between: > > a. win '95 > > b. win nt > > c. linux > > d. unix/pc > > e. unix/risc (pick hp, sun, ibm, etc.) > > f. mac > > may the best win! > > -------------------------------------------------------------- > rk > > "there's nothing like real data to screw up a great theory" > - me (modified from original, slightly more colorful version) > -------------------------------------------------------------- >Article: 9379
wen: : ;: Sorry, I can't do that. The machine I sometimes need to use would have : :: so many processors, disks, and memory and powersupply that it would take : ;: a forklift to carry. I also routinely use manage jobs that runs on a : :: farm of workstations and I need to do it remotely and service them at : ;: odd hours of the day. rk: : :for an application like that, at the high end, you need the high end : ;horsepower. that's not a good spot for NT. but that has nothing to do : :with suitability of nt for eda, as peter pointed out. it only has to do : ;with the suitability of nt for a particular class of job. don't think : :there's any disagreement there. and i've bought some big machines for big : ;jobs (and spent big $). wen: : But that is the direction of EDA future. Larger chips more simulation : and analysis, and more difficult place and route. I already have jobs : that takes 24 hours per random seed on the fastest pc available. rk: a bit confused, thought you were working on super-duper-need-a-forlift-'puter. there's a large amount of design work going on in the < 100 k gates and under arena. with a lot of designs < 25 k gates. these are being increasingly filled with fpgas. and a lot of people are currently designing these with (dare i say) pc's and win and some still with dos, as we've seen here. as the pc hardware grows in performance, and it seems to increase almost every week (and memory is $3/megabyte!), pc developer's are setting their heights higher doing larger designs. and they'll be a whole generation of chip designers using these tools. just like we got hit with a whole generation of programmers who used c and unix in school and we know the impact of that. now, we have previously heard stories about some fpgas taking ~ 24 hours for p&r while their competitors take an order of magnitude or more less. part of this is the fpga architecture. part of this is the software. i think the fpga market will be shaken out some if they require 24 hour p&r. people will switch to devices that don't. now, for the really big jobs, those that require a farm to run, the really mega-jobs at the extreme end of performance. no doubt that current win technology is no place for that. but, as an aside, perhaps you could comment on technologies such as quickturns, where they use boards full of fpgas (and now custom chips, iirc) to emulate the functionality of the asic (pentium 'simulated' this way) and permit running at a sizeable fraction of actual device speed, in-circuit. will this be preferable to farms of computers, multi-processors, or other simulation techniques? will formal equivalence checking of final netlists to original vhdl be accepted for signoff with only a limited amount of gate level simulation? will the use of rapid prototyping (say chip express 1 day turn laser programmable gate array) replace a lot of the gate level simulations? =============================== wen: : ;: The problem is not with NT, the problem is with how Microsoft is having : :: an effect on how EDA programs were written. There is absolutely no reason : :: why EDA vendors couldn't compile their program on NT but link them with : ;: X-window libraries so that the NT machine would be useful as a remote : :: compute engine within a network of unix workstations. NT still has a : ;: long way to go to match Unix for networking and user interface. rk: : :yup, microsoft monopoly, which is not good [see my earlier post on : ;competition and open standards]. and, i believe, microsoft does care about : :you; well about your $, anyways. for some networking features, the : ;microsoft stuff is pretty good (and win 3.11 was horrible); on others it's : :a pain and not up to unix or another of other os'es. their user interface : ;is not bad, very mac-like, and it's easy to write your own programs with : :nice, easy to use and understand humanoid interfaces. having programmed : ;both, i'd go with the win '95/nt interface. wen: :Humanoid interface? You should know by now that engineers are more :machine than human. :-) rk: no comment. :-) -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) --------------------------------------------------------------Article: 9380
steve: : As for basing my choice of synthesis / simulation tools on which : platform rund Word / VB / whatever, heck, that's what the sub-$K PC / : portable was put on this earth for. Don't want to be polluting my 'real' : machine with cycle-hungry dancing paperclips now, do I? rk: for small businesspeople, running two 'puters is not really that good of an option. unless you'd like to make an investment! for the stupid paperclip, geeze, how'd that get past any review cycle at all. steve: : Of cource, I could roll out the panacea that is java, and we can all run : the same tools. After all, cycles are cheap *8-) rk: yeah, right, sure. ;)Article: 9381
How to build a programming cable for the FLASH370i devices from CYPRESS (PC version)? Is there an application note? ThanksArticle: 9382
Russell May wrote: > > I am looking for ideas why version 8.1 of Altera's MaxPlus II > has a delay of about 0.5 to 1 second each time a new window is > selected, either automatically by the program or by clicking > on a different window. Does anyone else experience this? > > I have been using version 8.1 at home on my Pentium 233 MMX > system, which has a 64 Mb of EDO memory and a large fast disk drive, > and version 8.0 at work on a classic Pentium 133 system, which > has 32 Mb of FPM memory and a slower disk drive. My home system > runs the Altera compiler nearly twice as fast as the one at work, > but has these annoying and continual pauses. The system at work > does not have the pauses. > > Altera tech support has been no help, only insisting that I must not > have sufficient memory or something else must be running. The pauses > happen as soon as the program starts, and I have not compiled anything > which takes over about 9 Mb of working memory. Nothing else (except > clock) is visibly running. My system at work is connected to a 3COM > network, the one at home is not connected to a network (except thru > DUN to my Internet provider). Many times, I've seen unexplained delays in programs due to an invalid path on the PATH tree. For example, if the CD-ROM is on the PATH tree, and no CD is present, there could be unusual delays in your program. Especially if the program you're having a problem with is after the invalid path location in the PATH tree. Hope this helps, pmygArticle: 9383
In article <01bd49d4$198ce380$3a80accf@homepc>, rk <stellare@erols.comNOSPAM> wrote: >rk: >: >does microsoft office run on linux? need that to communicate these >days. > >thomas: >: There is StarOffice, that understand the Microsoft "standards" and is >free >: for personnal use. Works pretty well. Really looks like Office. > >rk: >ok, but does it understand for reading and writing *all* of the latest file >formats, not just the standards to be 'office compatible.' one can argue >whether we should be using office, pagemaker, tex, latex, or whatever. >but, micro$oft has one, and to communicate effectively with others and do >joint writing, it *is* the standard. also, i can take the output of my cae How can anything be a serious standard that changes every two years? Office 95 can't read the latest Office 97 formats either. On the otherhand, things like TeX and Postscript don't change nearly so often. But of course M$ needs to constantly 'update' these formats to 'encourage' you to go out and buy the latest Office every two years - someday, maybe, people will figure out this scam. >rk: >: >visual basic run on linux? >rk: >this is important as there is a lot of investment in visual basic code, >amongst engineers. not me (delphi fan), but i know a lot of people who do >it. very important application. There was a Vestigial Basic clone that ran on Linux - it allowed you to run VB byte codes. But again, there are so many superior languages available on Linux - why would anyone in their right mind use Vestigial Basic? >rk: >: >national instruments, do they supply ieee-488 drivers for linux? > >thomas: >: I know a physicist who drives some experiments with a Linux-PC and IEEE >: cards. I'll have to ask him about this version. > >rk: >the national cards, at least in america, are most probably the most >popular, they are fairly 'standard,' and give us code portability from dos >them and for us users. and yes, machines that i use for cae i also use for >laboratory use. The drivers may not be supplied by Nat Inst, but someone else may have developed some drivers on their own. >thomas: >: I don't have time to comment upon the others. Some (especially Xilinx >tools) >: do not have a Linux version, which is explanable by market analysis but >: not technical reasons. >rk: >what succeeds in the market is not determined solely by technical >considerations, which is the point of this whole argument. we can discuss >os/2 vs. win 3.1. microchannel vs. isa vs. eisa. c vs. pascal. etc. etc. >etc. Sure, but the point that the Linux folks are trying to make is that Linux is enjoying INCREASING success in the market. The number of Linux users is increasing at a high rate. Of the other technologies you mention: OS/2: Killed primarily by IBM's incompetance. Linux is not controlled by any one entity, the source code is available everywhere - it can't be killed by one company's incompetance. Win3.1: Killed by Micro$oft. See OS/2 above. MicroChannel,ISA, EISA: Stages in hardware evolution. C vs Pascal: Maybe Unix had something to do with the popularity of C? The point here is that Linux is different from all of those other technologies you mention. The number of Linux users worldwide is on the increase, and the numbers don't seem to be peaking out anytime soon. At this point, conservative estimates place the number of Linux users worldwide at 5Million. Other estimates range as high as 15 Million. It is quite possible that there are 7million. >thomas: >: The whole point is that if all vendors made Linux >: versions, everybody would be happier: the vendors who would have an >: extended market for cheap (good software should be easy to port) and >: the clients, who could benefit from the power and stability of Linux. >thomas: >: Oh well, maybe Bill Gates would not be so happy (niark niark niark). > >rk: >don't underestimate that <fill in choice word here>. i'm sure he'd find a >way to make more billions off of it, selling more upgrades, convertors, >bleah, bleah, bleah. You're invited to get off the treadmill of upgrades, rk. Save yourself $thousands$. >he'd use it to avoid the justice department and any >attempt to break his company up for being a monopoly. look what he's doing >to netscape. first he got them to give away their product for free. now >he has them giving away their source code for free. what do they give away Netscape's move to release source code is brilliant. And it is one that M$ cannot follow; its too much against their mind$et. By releasing source code, netscape has ensured that their browser will never go away, even if Netscape dies as a company. They've also gained thousands of new developers for free. >rk: >performance and interface is secondary. like i said, you can have the best >os, the nicest interface <one where the truly lazy don't have to move their >hands over to the arrow keys, a windows complaint i have recently heard> >and have a worthless machine. >if the applications aren't there, no work gets done. go ask ibm about >os/2. ask apollo about their os? well, you can ask some apollo users, >still running the machines, keeping them for their applications, locked in. > And in the open source world, if the apps aren't there we write them ourselves. We're getting quite a lot of them nowadays. In the meantime, I don't see Linux going away for lack of apps. philArticle: 9384
APS has al arge selection of FPGA boards/routers and fronte end tools including the XILINX Foundation Series with Synopsis FPGA express, Lucent routers and test boards, and VHDL simulators, as well as ATMEL's new 20K kits. Check out these and other tools at : http://www.associatedpro.com/aps -- __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ Richard Schwarz, President EDA & Engineering Tools Associated Professional Systems (APS) http://www.associatedpro.com 3003 Latrobe Court richard@associatedpro.com Abingdon, Maryland 21009 Phone: 410.569.5897 Fax:410.661.2760 __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/Article: 9385
rk: : >: >does microsoft office run on linux? need that to communicate these : >days. thomas: : >: There is StarOffice, that understand the Microsoft "standards" and is : >free : >: for personnal use. Works pretty well. Really looks like Office. rk: : >ok, but does it understand for reading and writing *all* of the latest file : >formats, not just the standards to be 'office compatible.' one can argue : >whether we should be using office, pagemaker, tex, latex, or whatever. : >but, micro$oft has one, and to communicate effectively with others and do : >joint writing, it *is* the standard. also, i can take the output of my cae phil: How can anything be a serious standard that changes every two years? : Office 95 can't read the latest Office 97 formats either. On the : otherhand, things like TeX and Postscript don't change nearly so often. : But of course M$ needs to constantly 'update' these formats to 'encourage' : you to go out and buy the latest Office every two years - someday, maybe, : people will figure out this scam. rk: er, everyone's already have figured out this scam. and we're stuck. heeeeeellllpppp!!!!! and, of course, a microshafty can say, "well, office 97 can write office 95/6.0 file formats." and they'd be correct. and everytime one down converts (and i'm not sure if the file i sent you via email to try was down converted or not, i'll have to check) you get one giant bloat. <insert expletive here>. they did do good ONCE, office 95 used same file formats as word 6.0. but they figured out their mistake and quickly corrected it, since there was little reason for anyone to update. and, i'll give you a hint, the paperclip ain't making it! and bill g. was downtown this week, chatting about this, and how it's good for the world. on the other hand, having used tex, latex, etc., ms word is easier to use. and when i worked in unix joint, tex was required, i fought tooth and nail, easier to format straight ascii with my turbo pascal code editor on my little 286 pc at home. and it is rather standard. and it is often required. now, if you read back, say, like 2,000 posts or so ago, i was on the proverbial soapbox about open standards for eda to enhance competition. same here. the update-thing sucks, gotta do it too, too often. and at day job, take the word update and multiply by 4. gets expensive. personally, i was pretty happy with word 2.0. rk: : >: >visual basic run on linux? rk: : >this is important as there is a lot of investment in visual basic code, : >amongst engineers. not me (delphi fan), but i know a lot of people who do : >it. very important application. phil: : There was a Vestigial Basic clone that ran on Linux - it allowed you to : run VB byte codes. But again, there are so many superior languages : available on Linux - why would anyone in their right mind use Vestigial : Basic? rk: personally, i don't program in it. but a lot of people do. and they can do it quickly. lot of support for it, esp. in instrumentation, can develop good, custom, easy to use apps quickly and cheaply with a lot of functionality. for some example, i'd suggest you look at the national instruments catalog. have some rather amazing stuff there. for performance, code legacy, sanity, etc. reasons, we're using the delphi (pascal) compiler. very, very nice. and, of course, part of microshafts great world domination strategy is to program the world in basic, bill gates' favorite language. that's why they're doing away with macros for all of their apps and sticking in vb. bleah! rk: : >: >national instruments, do they supply ieee-488 drivers for linux? : >thomas: : >: I know a physicist who drives some experiments with a Linux-PC and IEEE : >: cards. I'll have to ask him about this version. rk: : >the national cards, at least in america, are most probably the most : >popular, they are fairly 'standard,' and give us code portability from dos : >them and for us users. and yes, machines that i use for cae i also use for : >laboratory use. phil: : The drivers may not be supplied by Nat Inst, but someone else may have : developed some drivers on their own. rk: for exercise purposes, and to keep an open mind, i checked out the national instruments catalog and their www site for the latest info. they had mac, unix, dos, win 3.1, win nt, win '95, exec 8, no linux. right now i gotta run dos/win 95(or nt) and unix. linux would make a third. now, if the unix apps would all migrate to linux, *that* would be good. maintaining skills and systems for three environments is undesirable, to say the least. thomas: : >: I don't have time to comment upon the others. Some (especially Xilinx tools) : >: do not have a Linux version, which is explanable by market analysis but : >: not technical reasons. : >rk: : >what succeeds in the market is not determined solely by technical : >considerations, which is the point of this whole argument. we can discuss : >os/2 vs. win 3.1. microchannel vs. isa vs. eisa. c vs. pascal. etc. etc. : >etc. phil: : Sure, but the point that the Linux folks are trying to make is that Linux : is enjoying INCREASING success in the market. The number of Linux users : is increasing at a high rate. Of the other technologies you mention: : OS/2: Killed primarily by IBM's incompetance. Linux is not controlled by : any one entity, the source code is available everywhere - it can't be : killed by one company's incompetance. : Win3.1: Killed by Micro$oft. See OS/2 above. : MicroChannel,ISA, EISA: Stages in hardware evolution. : C vs Pascal: Maybe Unix had something to do with the popularity of C? : : The point here is that Linux is different from all of those other : technologies you mention. The number of Linux users worldwide is on the : increase, and the numbers don't seem to be peaking out anytime soon. : At this point, conservative estimates place the number of Linux users : worldwide at 5Million. Other estimates range as high as 15 Million. : It is quite possible that there are 7million. rk: 7 million is a large number, worldwide. very impressive. and if it can do the job with little work, is as easy and cost-effective as the other system, then it's gonna be a winner. *IF* the conversion costs aren't too high, which is the point of above. there is inertia and for good reasons - people have invested a lot of time and $ in what they have and don't change easily. and one guy at work keeps on bugging me to upgrade some dos apps i'm running; real-time programs that have to work correctly everytime. can't justify, even to myself, the time to upgrade it to 95/nt. rather move forward. c vs. pascal? another thread, another time, that one's *ALWAYS* fun. just like the unix <--> dos ones from a decade ago, where the unix guys said messy-dos is dead, after 386 becomes popular, unix will be everything. same arguments as we've seen here, for the most part. big argument for unix in the '80s was that they knew where to put the <crtl> key in unix, dos machines couldn't even get that right, slowed down the entire machine! now they complain about arrow keys. what a difference a decade makes! for linux to suceed in eda (and in an attempt to keep this thread somewhat relevant), it needs something to make people want to use it. better is the enemy of good enough and if people are reasonably happy with what they got and know that it will get their job done (i.e., unix, mac, pc/win), why change to an unknown, NOT be able to deliver on time. when you don't deliver, they never say, "ah, but he used linux" (forth, or smalltalk, or nueral nets, or fuzzy logic, or expert systems, or tagauchi method, or next (jobs machine) or whatever the trendy technology of the day is. they say, "phil <expletive> up, playing with that crap." and that's not good. i know people who spend all their time messing with tools, never have time to design the damn logic! anyways, the thought of having to run *three* systems, is, well, not nice. now for linux to succeed, imho, it can do the following: 1. be a unix on a pc. be able to run most of the major unix apps. perhaps a bit slower, but be able to go to the synopsis store, pick up your copy, and run. or 2. be a better pc. minimal hassles and conversion costs for people to dump the guys from the northwest. don't disrupt the world. look at the hassles in just converting from win 3.1 to 95. people went nuts. ok, you might say, they were mostly aol, vb idiots. and you may be right. but we're talking what, 100,000,000 machines or so? *that's* a big market. world-wide there is, iifc, 100,000 eda people? really quite a small market. so, it is my conclusion, for linux to succeed, they gotta hit critical mass in applications so one of the two cases above is met. for general use, they gotta run *all* of the apps. like 95 does, mostly, in the pc world. or for unix. why give up a sun for something that can't run all of the s/w? and price of the unix boxes is falling, towards pc prices. while pc performance moves up, closing in on workstations. same holds true for the eda world. gotta have the flow. only having some of them on the machine will make it little more than a pain in the butt. let's say they get 10% of the eda users. and let's say my 100,000 eda users is the correct number. and let's say that the eda community, on pc's, is to move, and we're to have low-cost s/w (currently there is a lot of that, check out rich, programmable logic jump statoin, xilinx, actel, quicklogic, bleah, bleah bleah - notice viewlogic isn't on this list;). anyways, take the 10,000 converts, and 10% is a big number, and say they spend $1,000 for eda tools. $10,000,000. big deal. @ $100,000/man-year, that's 100 man-years of effort. and that's not much, considering the variety of tools that are necessary for eda, with all of the vendors, libraries, manuals to be written, machines to run them and check them out on, app notes to right, bleah, bleah, bleah. and let's say the users pay maintenance, $250.00 per year. that's 25%, *higher* than industry average, from what i've seen. another $2,500,000 or 25 man-years of effort. bfd. so, if these assumptions are correct, gotta raise prices. but, for eda, fpga users, for example, are getting back end tools real cheap or even free (actel, quicklogic, etc.). and getting free vhdl compilers (not the best, but they work, and *outstanding* macro generators). and viewlogic maintenance, i think it's $500.00/year, time to re-up anyways, and synopsys/viewlogic said no linux, no customer demand. and orcad is cheap on pc. ==> touch switch for pc users. on the unix side, expensive tools, cost of machine negligible to the s/w bill and the nre for the super-big asics. perhaps tough sell, esp. if all the apps aren't ported, then they have to run two machines. not three, unix guys love tex! anyways, from all that i've seen here over the last week, i think it will be tough, either in eda or general use, for linux to take over. don't think that it's because that's the os that comes on the machine, i think it will be s/w availability and the desire of a lot of people to go with what they know will do the job, even if it's a bit more $ and technically inferior. in electronics, my output is chips and boards and boxes (and some paper too!) and the tools are, to a large extent, just a means to an end. and no way would i jeopardize output for a 'better' tool that isn't well supported. chicken and egg, yes, but i think that will be the issue. unless there's something made available that is way better and won't cause you to screw up. thomas: : >: The whole point is that if all vendors made Linux : >: versions, everybody would be happier: the vendors who would have an : >: extended market for cheap (good software should be easy to port) and : >: the clients, who could benefit from the power and stability of Linux. thomas: : >: Oh well, maybe Bill Gates would not be so happy (niark niark niark). : >rk: : >don't underestimate that <fill in choice word here>. i'm sure he'd find a : >way to make more billions off of it, selling more upgrades, convertors, : >bleah, bleah, bleah. phil: : You're invited to get off the treadmill of upgrades, rk. Save yourself : $thousands$. rk: love to, if it can do the job. see above. rk: : >he'd use it to avoid the justice department and any : >attempt to break his company up for being a monopoly. look what he's doing : >to netscape. first he got them to give away their product for free. now : >he has them giving away their source code for free. what do they give away phil: : Netscape's move to release source code is brilliant. And it is one that : M$ cannot follow; its too much against their mind$et. By releasing source : code, netscape has ensured that their browser will never go away, even if : Netscape dies as a company. They've also gained thousands of new : developers for free. rk: yup, but a lousy way to run a business. not a good move if you need to convince those funding guys to fork over $ for your next startup. again, the key is public domain standards. ibm opening up the pc vs. apple closing the mac. pc development was competitive as hell and look where they are. and the technically superior mac (my favorite, btw, but could never afford one), needs handout from bill gates just to keep the doors open. going back a computer generation, the open apple ii got people working on it. it's when a single man can control what lots of people do and behave because he controls, in secret, developments that can affect the world, we're in trouble. (is this iraq or seattle? ;) so, for eda, i'd like to see more open standards, not less. and they're brawling about that now, right, just like in os. and if standards are open, idiots like us will write utilities, people will form startups, competition and greed will kick in, and capitalism will work. rk: : >performance and interface is secondary. like i said, you can have the best : >os, the nicest interface <one where the truly lazy don't have to move their : >hands over to the arrow keys, a windows complaint i have recently heard> : >and have a worthless machine. : >if the applications aren't there, no work gets done. go ask ibm about : >os/2. ask apollo about their os? well, you can ask some apollo users, : >still running the machines, keeping them for their applications, locked in. phil: : And in the open source world, if the apps aren't there we write them : ourselves. We're getting quite a lot of them nowadays. In the meantime, I : don't see Linux going away for lack of apps. rk: sounds like the unix argument in the '80s. unix is obviously not going away. and it's much improved over what it was. but the open source world hasn't caused it to take off the way the pc's have. with all of the great unix programmers, really smart guys and all, they missed the big apps that caused pc's to take off: visicalc[iirc]. and games. and easy to setup and use. and cheap. back in '87, got an inexpensive '286 (which has slowly morphed into the one i'm typing on now), could program anything (yes, wrote some of my own analog and digital simulators), could develop pals, run abel, spice, do documentation, useful machine. never could afford it for unix. and unix was harder to program. did it professionally for a while so i'm speaking from experience. and any idiot could design their own add-in card for the pc. from general use card to a laboratory instrument. which leads to my great big b*tch about pc's: they're getting rid of the isa bus!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!! now *that* move will cost me tons!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! yeah, yeah, yeah, pci is so much better technically. but it will cost me a fortune in time, effort, and $ to convert so i can stay at the same place i am now! [got a lot of semi-custom and home-designed cards, as well as a pile of other cards that need replacing. i've started to prepare for this but it will be a disaster financially - run about 12 'puters total]. aaaarrrrgggghhhh!!!! have a nice weekend, -- -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) --------------------------------------------------------------Article: 9386
In a previous article "rk" <stellare@erols.com.NOSPAM> writes: : ;wen: :: ;: Sorry, I can't do that. The machine I sometimes need to use would ;have :: :: so many processors, disks, and memory and powersupply that it would ;take :: ;: a forklift to carry. I also routinely use manage jobs that runs on a ;: :: farm of workstations and I need to do it remotely and service them at :: ;: odd hours of the day. ; :wen: ;: But that is the direction of EDA future. Larger chips more simulation :: and analysis, and more difficult place and route. I already have jobs ;: that takes 24 hours per random seed on the fastest pc available. : ;rk: :a bit confused, thought you were working on ;super-duper-need-a-forlift-'puter. I said that is how long it takes. I didn't say that is what I use to do actual productive work. I als said that it is for one random seed, and I would have used several hundred in each job, and that is still not complete to my liking. ;there's a large amount of design work going on in the < 100 k gates and :under arena. with a lot of designs < 25 k gates. these are being ;increasingly filled with fpgas. and a lot of people are currently :designing these with (dare i say) pc's and win and some still with dos, as ;we've seen here. as the pc hardware grows in performance, and it seems to :increase almost every week (and memory is $3/megabyte!), pc developer's are ;setting their heights higher doing larger designs. and they'll be a whole :generation of chip designers using these tools. just like we got hit with ;a whole generation of programmers who used c and unix in school and we know :the impact of that. When pc's get faster, so would unix running on pc's. It is not an issue of hardware. The pc will never scale as fast as the demand of computing power. While speed grows much less than linear compared to the physical measures of the processor (area, gates, channel length, etc), the computing demand grows much faster. There are a lot of things that would tend to grow with n*n or n*(log n) if not managed carefully. What is wrong with EDA tools running on NT is that it is short sighted and because of poor network services, it precludes effective scaling of computational power by aggregating a network of computing engines. :now, for the really big jobs, those that require a farm to run, the really ;mega-jobs at the extreme end of performance. no doubt that current win :technology is no place for that. but, as an aside, perhaps you could ;comment on technologies such as quickturns, where they use boards full of :fpgas (and now custom chips, iirc) to emulate the functionality of the asic ;(pentium 'simulated' this way) and permit running at a sizeable fraction of :actual device speed, in-circuit. Witness the long list of pentium bugs since the processor was released as evidence that insufficient verification cycles, even with the aid of quickturns, has been devoted to check the design. : will this be preferable to farms of computers, multi-processors, or other ;simulation techniques? : ; will formal equivalence checking of final netlists to original vhdl be :accepted for signoff with only a limited amount of gate level simulation? ; : will the use of rapid prototyping (say chip express 1 day turn laser ;programmable gate array) replace a lot of the gate level simulations? I am afraid none of that will be good enough. It is going to take a group of really sharp programmers many years in an effort that is not driven by the short sighted market forces to come up with a real solution that scales. And only multicomputers of some sort can provide that scaling of computing power required.Article: 9387
phil: : Win95 and NT lock you into a graphical paradigm for tools that is : difficult to automate. Let's say I'd like to run several simulations over : night and keep track of potential failures in each. This is easy in : Unix/Linux. Even if a tool has a GUI, it always has a way to execute from : the command line - this makes it very easy to put together a perl script : to automate the overnight simulation runs. In the Win world, most apps : don't have a way to execute from the command line - so you've got to stay : around all night and push the buttons if you want to batch up several : runs. rk: there is a difference in the win and unix world, since in the win world you can carry your cycles with you or perhaps have a home pc (my case). now we have discussed this before with myself and one or two others that some tools are command line driven too, as you mention for unix. i know that the actel tools for fpga, which i did initially in dos and now in win, had command line i/f's and ascii configuration files. this makes it suitable for remote operation with very simple software over the internet: i.e., ftp, drag and drop over the internet, and a small program to stay resident and start other programs. i just finished looking at the simulator and related tools (viewlogic) and see that these tools, which i normally operate via the gooey, also have command line forms, so that is not an issue here. perhaps this is something, as our designs get bigger and more complex, win users should watch for and put in their $0.02 to the vendors to enable this sort of operation. can't speak, obviously, for all apps but did take a closer look at this one. now, however, i'm using a new tool, statecad, for graphical entry, vhdl output, using the gooey. dunno how good these graphic tools will work over a slow phone line and perhaps the cycle here for win and x (although one respondent said that there a program to speed up x) starts to break down. btw, statecad has a built in simulator. so enough there to keep you busy on your home pc. or ftp in the generated vhdl and continue from there. cheers! -- -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) --------------------------------------------------------------Article: 9388
Gavin Melville wrote: > On Thu, 05 Mar 1998 14:45:00 GMT, k.rozniak@XXX.ien.gda.pl (Krzysztof > Rozniak) wrote: > > >Thank you for reply, Gavin and Peter. > > > >I managed to find the reason of VL malfunction. The problem was with > >too long path. I had path about 180 chars long. None of old programs I > >was using complained about it. VL was the first one. I had to shorten > >path (now ~65 chars). Since then all works perfectly. Probably VL has > >a workspace for path limited to 128 chars (DOS <=6.22 limit IIRC). > >Hope it helps someone. > > Oh -- THAT version. I had not read the original post very carefully. > It is actually a little more subtle than that -- not only must the > path be less than 127 characters long, but ALL the directories in that > path must exist. > -- > Gavin Melville > gavin@cypher.co.nz I remember dealing with this once. I was told that the limitation was that the comamnd line which included this path had to fit 128 characters. So the Path had to be somewhat less than that. I seem to recall a practical limit of about 90 characters. Rick CollinsArticle: 9389
Peter wrote: > I have no extra info beyond what you can work out by looking at the > files. However, do please post your final results - ESPECIALLY if you > can work out how the magic number is calculated. > > This magic number is used to stop things like e.g. using a Xilinx/LCA > version of Viewdraw to open a sch or lib file which was created in the > unrestricted version. > > I believe that unless you can generate the magic number, you will not > be able to generate valid Viewlogic files. > > The sch file starts with something like > > V 50 > K 257863394400 FRED > Y 0 > D 0 0 1653 1169 > > Where the 2nd line, after the "K", has the magic number. This, I > believe, is computed with an algorithm which takes in the filename > ("FRED" above), a text string identifying the company selling the > restricted version (e.g. "Xilinx") and some other stuff I don't know. > > I doubt Viewlogic will tell you, since they went to a lot of trouble > to prevent the restricted versions (sold for much less than their own > full version - $30k at one time) being used to design normal stuff, > with general libraries like LSTTL. > But you don't need to know how it is calculated if you use the same filename and the same key number. I have done this with success. The key number does not depend on anything else in the file. Rick CollinsArticle: 9390
Wen-King Su <wen-king@myri.com> wrote in article <6dvk0l$mcp@neptune.myri.com>... : In a previous article "rk" <stellare@erols.com.NOSPAM> writes: : : : ;wen: : :: ;: Sorry, I can't do that. The machine I sometimes need to use would : ;have : :: :: so many processors, disks, and memory and powersupply that it would : ;take : :: ;: a forklift to carry. I also routinely use manage jobs that runs on a : ;: :: farm of workstations and I need to do it remotely and service them at : :: ;: odd hours of the day. : ; : :wen: : ;: But that is the direction of EDA future. Larger chips more simulation : :: and analysis, and more difficult place and route. I already have jobs : ;: that takes 24 hours per random seed on the fastest pc available. : : : ;rk: : :a bit confused, thought you were working on : ;super-duper-need-a-forlift-'puter. : : I said that is how long it takes. I didn't say that is what I use to : do actual productive work. I als said that it is for one random seed, : and I would have used several hundred in each job, and that is still : not complete to my liking. : : ;there's a large amount of design work going on in the < 100 k gates and : :under arena. with a lot of designs < 25 k gates. these are being : ;increasingly filled with fpgas. and a lot of people are currently : :designing these with (dare i say) pc's and win and some still with dos, as : ;we've seen here. as the pc hardware grows in performance, and it seems to : :increase almost every week (and memory is $3/megabyte!), pc developer's are : ;setting their heights higher doing larger designs. and they'll be a whole : :generation of chip designers using these tools. just like we got hit with : ;a whole generation of programmers who used c and unix in school and we know : :the impact of that. : : When pc's get faster, so would unix running on pc's. It is not an issue : of hardware. The pc will never scale as fast as the demand of computing : power. While speed grows much less than linear compared to the physical : measures of the processor (area, gates, channel length, etc), the computing : demand grows much faster. There are a lot of things that would tend to : grow with n*n or n*(log n) if not managed carefully. What is wrong with : EDA tools running on NT is that it is short sighted and because of poor : network services, it precludes effective scaling of computational power : by aggregating a network of computing engines. rk: i agree, for the really large jobs. and as i said, before, doesn't seem to be a place for nt. and that's why i invest in unix workstations, too. now, for fpga jobs, for instance, with the improvement in pc technology, and moving from dos -> 95, and incresing the size and complexity of the jobs, i find time spent on machine running is decreasing, over all, and i work more efficiently. two separate problems, two solutions. while nt would be shortsighted since it "precludes scaling", it is also efficient at the lower end of design size, where a large volume of designs are currently being done. i know that doing my "lower end" designs on unix box would be more expensive. and with some here advocating same price of tools for all platforms, that would drive development costs up through the roof, as compared to what i use for dos/win designs. to get back on thread, a little bit, what will linux mean? unix scalability with unix tools at unix prices? the synopsys guys said same price, unix and nt. my guess would be that the people who pay that $ and use that capability are already using it on unix and there would not be a great reason to switch to nt and they will have a very big infrastructure. some will move but i would predict not the majority. and the amount of win people wanting unix software with the limits of nt are not going to fork over that kind of money. will win apps port to linux, a unix-type environment for the scalability and charge win-type prices? it seems from the discussion here (and at work) the unix die-hards wouldn't go for that, they'd want their unix tools for many reasons, an important one not having to retrain, write new scripts, etc. and the win people would have to maintain two machines for the near term. and idiots like me, would have to work with three environments. aaaarrrrgggghhhh!!!!! rk: : :now, for the really big jobs, those that require a farm to run, the really : ;mega-jobs at the extreme end of performance. no doubt that current win : :technology is no place for that. but, as an aside, perhaps you could : ;comment on technologies such as quickturns, where they use boards full of : :fpgas (and now custom chips, iirc) to emulate the functionality of the asic : ;(pentium 'simulated' this way) and permit running at a sizeable fraction of : :actual device speed, in-circuit. wen: : Witness the long list of pentium bugs since the processor was released : as evidence that insufficient verification cycles, even with the aid of : quickturns, has been devoted to check the design. rk: verification for the mega-chips is a real issue. and historically, the list of processor bugs, even for simple ones like the 8086, has been, well, non-zero. and, iirc, the 386 multiply bug was a result of the test consisting of a bunch of random vectors thrown at the chip, not the most thorough. can't remember the details of the pentium fdiv bug, other than they tried to cover it up! scary problem, really, and these chips are getting into more and more 'critical' applications. rk: : : will this be preferable to farms of computers, multi-processors, or other : ;simulation techniques? : : : ; will formal equivalence checking of final netlists to original vhdl be : :accepted for signoff with only a limited amount of gate level simulation? : ; : : will the use of rapid prototyping (say chip express 1 day turn laser : ;programmable gate array) replace a lot of the gate level simulations? wen: : I am afraid none of that will be good enough. It is going to take a group : of really sharp programmers many years in an effort that is not driven : by the short sighted market forces to come up with a real solution that : scales. rk: and please don't forget the test/verification engineers. tough (and thankless) job. wen: : And only multicomputers of some sort can provide that scaling : of computing power required. rk: not sure i understand/agree here. if one can run logic simulations (not timing, different story) on a multicomputer (pick your favorite type) to scale simulation performance, how would this logically differ from running on a quick-turn or rapid asic prototype? thanks for the info and discussion, -- -------------------------------------------------------------- rk "there's nothing like real data to screw up a great theory" - me (modified from original, slightly more colorful version) --------------------------------------------------------------Article: 9391
I was wondering if anyone had successfully installed the Atmel IDS 5.0 (the free version - http://www.atmel.com/fpga_software.html). I get the following message after trying to run it: "A primitive has failed". Also, there is WorkView Office on the same CD, and for its installation a floppy disk is required. I didn't get any disk in the package though. Anyone had luck with that? Thanks, Ivan.Article: 9392
>But you don't need to know how it is calculated if you use the same filename >and the same key number. I have done this with success. The key number does >not depend on anything else in the file. OK, others have suggested this too. Therefore, can anyone post a few magic numbers generated by the unrestricted version of VL? That would cover all options. What about libraries? One of the objectives, I assume, of this scheme was to prevent owners of the restricted versions from using them for any normal work, e.g. using 74HC libraries from VL. Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.Article: 9393
I am new to this field (FPGA). I want to know how to represent the floating point on FPGA? How floating point operations are carried out in FPGA? Please answer to these questions my e-mailaddress is satish_me@hotmail.com (M.E.Satish Kumar) India. -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreadingArticle: 9394
Allan Redenbaugh wrote: > > Given the following code for a control register where a single bit has to > have > a seperate async reset : > (target device = Xilinx 4ke) > > process ( strobe, reset, clear_bit0) > > begin > if reset = '0' then > cntl_reg <= DEFAULT_REG; > elsif clear_bit0 = '1' then > cntl_reg(0) <= DEFAULT_REG(0); > elsif strobe'event and strobe = '1' then > cntl_reg <= data; > end if; > end process; There is nothing wrong with your code. I tried it with Synopsys and it didn't complain. > > I assign reset to GSR so I expect clear_bit0 to drive the reset line of a > dff > on bit 0 and the reset of the register bit to only have the GSR reset. > I don't understand what you say, but what you have coded will work as follows: - if reset is asserted '0', then all FFs get DEFAULT_REG value - if reset is asserted '1' and clear_bit0 is '1', then only bit 0 will get DEFAULT_REG(0). - else if there is a positive edge on strobe, sample data. > My synthesis tool (Leonardo) says that since I have nothing defined for the > upper bits under the clear_bit0 condition it defaults to a preset which is > not what I inteded. > > I have pulled out bit 0 into its own process and everythings happy, I just > don't > understand why this method doesn't work. > > Any ideas? > > Thanks, > Allan Redenbaugh > Raytheon Systems Kind regards, Jan -- =================================================================== Jan Zegers === Easics === General Manager === VHDL-based ASIC design services === NEW Tel: +32-16-395 601 =================================== NEW Fax: +32-16-395 619 Interleuvenlaan 86, B-3001 Leuven, BELGIUM mailto:janz@easics.be http://www.easics.comArticle: 9395
Ivan wrote in article <35039a7a.1493447@news.sprint.ca>... > > I was wondering if anyone had successfully installed the Atmel >IDS 5.0 (the free version - http://www.atmel.com/fpga_software.html). >I get the following message after trying to run it: "A primitive has >failed". Please make sure you have the latest patch from : http://www.atmel.com/atmel/products/prod180.htm Martin.Article: 9396
Bigwheel...you suck comp.arch.fpga too easy ya gotta change em both.Article: 9397
In a previous article "rk" <stellare@erols.com.NOSPAM> writes: :wen: ;: And only multicomputers of some sort can provide that :scaling ;: of computing power required. : ;rk: :not sure i understand/agree here. if one can run logic simulations (not ;timing, different story) on a multicomputer (pick your favorite type) to :scale simulation performance, how would this logically differ from running ;on a quick-turn or rapid asic prototype? : ;thanks for the info and discussion, With those, you are limited by what the physical test environment can check. If you verify your PCI design by plugging a protptype into the PCI bus of a machine you happen to have, for example, you will get a very inadequate test coverage.Article: 9398
satish_me@hotmail.com wrote: > I want to know how to represent the floating point on FPGA? > How floating point operations are carried out in FPGA? I would be interested in any real life (64 bits) documented implementation of fp arithmetic (possibly IEEE standard) on FPGAs. -- Marc Daumas - Charge de recherches au CNRS (LIP - ENS de Lyon) mailto:Marc.Daumas@ENS-Lyon.Fr - http://www.ens-lyon.fr/~daumas ENS de Lyon - 46, allee d'Italie - 69364 Lyon Cedex 07 - FRANCE Phone: (+33) 4 72 72 83 52 - Fax: (+33) 4 72 72 80 80Article: 9399
This is a multi-part message in MIME format. --------------8AD7F76A08015AD1ABADAF78 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Allan Redenbaugh wrote: > > Given the following code for a control register where a single bit has to > have > a seperate async reset : > (target device = Xilinx 4ke) > > process ( strobe, reset, clear_bit0) > > begin > if reset = '0' then > cntl_reg <= DEFAULT_REG; > elsif clear_bit0 = '1' then > cntl_reg(0) <= DEFAULT_REG(0); > elsif strobe'event and strobe = '1' then > cntl_reg <= data; > end if; > end process; > > I assign reset to GSR so I expect clear_bit0 to drive the reset line of a > dff > on bit 0 and the reset of the register bit to only have the GSR reset. > > My synthesis tool (Leonardo) says that since I have nothing defined for the > upper bits under the clear_bit0 condition it defaults to a preset which is > not what I inteded. > > I have pulled out bit 0 into its own process and everythings happy, I just > don't > understand why this method doesn't work. > > Any ideas? > > Thanks, > Allan Redenbaugh > Raytheon Systems Try this: process ( strobe, reset, clear_bit0) begin if reset = '0' then -- reset cntl_reg <= DEFAULT_REG; -- default if (clear_bit0 = '1') then cntl_reg <= DEFAULT_REG_ALT; -- alternate default end if; elsif strobe'event and strobe = '1' then cntl_reg <= data; end if; end process; -- overwriting should be ok since evaluation is at the end of the process --------------8AD7F76A08015AD1ABADAF78 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Mike Treseler Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Mike Treseler n: Treseler;Mike org: Fluke Networks Division adr;dom: 6920 Seaway Blvd;;;Everett WA;;98203; email;internet: tres@tc.fluke.com title: Staff Engineer tel;work: 425.356.5409 tel;fax: 425.356.5043 x-mozilla-cpt: tc.fluke.com;2 x-mozilla-html: TRUE version: 2.1 end: vcard --------------8AD7F76A08015AD1ABADAF78--
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