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
1. I guess the real answer is what will happen when you are caught? 2. If the company already has the software then there wouldn't be a copyright violation. Otherwise for them to have a copy of the software would be a copyright violation -- That is the criminal part of the equation. For them that is. 3. If you are charging for the use of the license then you would probably have to move to Borneo and put on a sweaty ape suit to hide from the S Police. chipfactory@hotmail.com wrote: > Hi Folks: > When we read the documentation for Synopsys we found out that > a license server can also be somewhere in the internet. Now we > came up with the idea to connect our Synopsys/Cadence/Modeltech > license server to the internet and share our licenses with others. > During the night here in Europe our licenses are not used. > Would there someone be interested and is this legal? > > Best regards > Jack > > email: chipfactory@hotmail.com -- Tim DavisArticle: 17301
bill morris wrote: > Has anybody tried to implement a frequency multiplier in XC4k? I need to > implement a frequency multiplier by 2^n (n=1,2,3,4). I know VIRTEX contain > DLL's which can implement that but unfortunately I have to use XC4kE. > > Please help! Bill, I drew a circuit that might do the trick: There is an internal oscillator consisting of a 2-input NAND gate driving a number of non-inverters feeding back to one input. The other input is driven from the Q of a Control flip-flop clocked by your incoming clock, its D is permanently High. You use the internal oscillator to clock a counter that is one bit too long ( e.g. 4 bits for divide by 8 ) The MSB of this counter drives the asynchronous Reset input of the Control flip-flop. The Qbar output of the Control flip-flop is an active High asynchronous Reset input to all counter bits, except its an asynchronous PRESET for the LSB. That should do it. Incoming clock edge sets Control flip-flop, thus starting oscillator and raking away the asynchronous preset to a count value of 1. When counter reaches 2-to-the-n, the MSB clears the CONTROL flip-flop, which in turn presets the counter to a 1 and holds it there until the next rising edge of your main ( slow) clock. Let me know when you got any further with this. Regards PeterArticle: 17302
My two cents: Ignoring possible export restrictions (due to encryption, "national security," etc.), you've paid for x number of license tokens. Why should the vendor care where they're run? As long as the number of simultaneous users doesn't exceed the number of licenses you've paid for, what's the difference? I have the (floating) FlexLM license daemon running on my NT box at work. I can dial into our network from home. My home machine has a static IP address, and I can running the Xilinx tools and FPGA Express without any problems. All of the design files are on my home machine; the only network traffic is what's needed to deal with the license. While on the topic of licensing, what about the following: for one seat, node-locked licenses are generally about half the price of a floating license. Say you have a Unix network, and you install the license server and the tools on a machine called gromit. Give any potential users a login on gromit. To use the tools, all one has to do is rlogin to gromit and do your work. (Cross-mount drives and make it even easier to handle files across the network.) So, if you have a handful of people who need access to the software and they don't require full-time access, purchasing one node-locked license and doing the rlogin seems like one way of saving some cash. 'course, ya can't do that trick on NT. And, of course, the Unix licenses are generally about double the price of the NT licenses, so... -- ----------------------------------------- Andy Peters Sr Electrical Engineer National Optical Astronomy Observatories 950 N Cherry Ave Tucson, AZ 85719 apeters (at) noao.edu "You want partial credit? You build bridge, bridge falls down - no partial credit." -- Dr A. Chang, professor of Mechanical Engineering at Stevens Institute of Technology chipfactory@hotmail.com wrote in message <37964ef8.605363216@news.maltanet.net>... >Hi Folks: >When we read the documentation for Synopsys we found out that >a license server can also be somewhere in the internet. Now we >came up with the idea to connect our Synopsys/Cadence/Modeltech >license server to the internet and share our licenses with others. >During the night here in Europe our licenses are not used. >Would there someone be interested and is this legal? > >Best regards >Jack > >email: chipfactory@hotmail.com >Article: 17303
R. Mark Gogolewski <gogo@netcom.com> wrote in message news:7n00fp$g5u@dfw-ixnews11.ix.netcom.com... [snip] > Let's say that the license agreements didn't specifically > have wording to make sure that this usage is restricted. > IMO, I would think that everyone in this industry - designers > and vendors - would agree that software licenses were > never priced with this type of usage in mind. > > Does it _have_ to strictly be in violation of a license > agreement before it feels like stealing? Let's be > realistic, if everyone did this, vendors would have to > either technically do something to stop it, or drastically > change the price model. > > Thoughts? [snip] I don't think that 'feels like stealing' has anything to do with it. A license is a binding agreement between a buyer and seller. If the license says that they can do this, then by all means they should do it. If the license says they can't do it, then shey should not do it. If the license does not anticipate this possibility, then they should probably spend a couple of hundred bucks to get an attorney's opinion on the agreement and go from there. As far as being a good idea or not, I think that some of the EDA tool companies have decided that this is a good idea. I was down at the DAC conference in New Orleans last month, and there was a company preparing to sell 'timeshare licenses' over the internet. Apparently, they were working with Synopsys (and others) on the idea. If you need synthesize or simulate on Synopsys, and don't want to spend the $30K for a license, then you could use theirs on a 'per use' basis. I think it's a great idea. -- Wade D. Peterson Silicore Corporation 3525 E. 27th St. No. 301, Minneapolis, MN USA 55406 TEL: (612) 722-3815, FAX: (612) 722-5841 URL: http://www.silicore.net/ E-MAIL: peter299@maroon.tc.umn.eduArticle: 17304
Hello, I have just purchased the Xilinx Foundation Series 1.5 Student Edition, a XS40 and a XS95 board from Xess Corp. Unfortunately, the license software dosn’t seem to work and Xilinx now claims that they don’t support the student edition. I was hoping that someone here knows a solution to the problem. Everything seems to work except the VHDL synthesis. When attempting to synthesize a VHDL entry I get: Pcm :Synopsys server initialization Dpm :Invalid Host(-9,57) Pcm :Cannot find a valid license for Synopsys synthesis I have spent several hours attempting to get this to work. Initially I used Drive number 0B87-7DBF with the Xilinx license generator which is the id for the D drive where the Foundation Series software resides (as reported by Vol C: in DOS.) This did not work, so I manually changed the license.dat file so the id was 1878-08CD which is the id the master drive, C:. This did not work. Then I downloaded Hostid.exe from Synopsys which gave the host id as f8d709c10000. This also did not work. The Autoexec.bat dose contain: C:\flexlm\license.dat. Any insight to the problem would be appreciated. Thanks in advance -NickArticle: 17305
Last time I checked, VHDL wasn't included in the student edition, of course that was a while ago now. Nicholas Brown wrote: > Hello, > I have just purchased the Xilinx Foundation Series 1.5 Student > Edition, a XS40 and a XS95 board from Xess Corp. Unfortunately, the > license software dosn’t seem to work and Xilinx now claims that they > don’t support the student edition. I was hoping that someone here knows > a solution to the problem. > Everything seems to work except the VHDL synthesis. When attempting > to synthesize a VHDL entry I get: > -- -Ray Andraka, P.E. President, the Andraka Consulting Group, Inc. 401/884-7930 Fax 401/884-7950 email randraka@ids.net http://users.ids.net/~randrakaArticle: 17306
In article <7n0mr4$l76$1@news1.tc.umn.edu>, Wade D. Peterson <peter299@maroon.tc.umn.edu> wrote: >R. Mark Gogolewski <gogo@netcom.com> wrote in message >news:7n00fp$g5u@dfw-ixnews11.ix.netcom.com... >[snip] >> Let's say that the license agreements didn't specifically >> have wording to make sure that this usage is restricted. >> IMO, I would think that everyone in this industry - designers >> and vendors - would agree that software licenses were >> never priced with this type of usage in mind. >> >> Does it _have_ to strictly be in violation of a license >> agreement before it feels like stealing? Let's be >> realistic, if everyone did this, vendors would have to >> either technically do something to stop it, or drastically >> change the price model. >> >> Thoughts? >[snip] > >I don't think that 'feels like stealing' has anything to do with it. A >license is a binding agreement between a buyer and seller. If the license >says that they can do this, then by all means they should do it. If the >license says they can't do it, then shey should not do it. If the license >does not anticipate this possibility, then they should probably spend a >couple of hundred bucks to get an attorney's opinion on the agreement and >go from there. > >As far as being a good idea or not, I think that some of the EDA tool >companies have decided that this is a good idea. I was down at the DAC >conference in New Orleans last month, and there was a company preparing >to sell 'timeshare licenses' over the internet. Apparently, they were >working with Synopsys (and others) on the idea. If you need synthesize >or simulate on Synopsys, and don't want to spend the $30K for a license, >then you could use theirs on a 'per use' basis. I think it's a great idea. Good comments - and I agree. I would add, however, that in this last paragraph, the scenario is that this type of licensing is being _discussed and negotiated_ - rather than simply being implemented just because the license agreement didn't anticipate it. There has been a smattering of discussion both on the newsgroups and on ESNUG about "getting around" the specifics of licensing. I believe that there should be some consideration of the intent of the agreement even if the specific wording doesn't cover it. While both the vendors and the designers can hire lawyers until the cost of doing business raises prices through the roof, when unforseen contingencies arise, the success or failure of the resolution boils down to the relationship between the vendor and the client. Just my $0.02, Mark P.S. I think the idea of specifically licensing software this way on a pure usage model has a heck of a lot of merit. My comments began from my perception that there is a portion of the audience at large that looks for ways to slip around the rules as much as they can...Article: 17307
Does anyone know what happened to Esprimo: www.vhdl.com. I have their VHDL editor but trying to find out about upgrades etc. is fruitless as they don't reply to any email.Article: 17308
You and everyone else who knows the port number! This is a known (potential) security breach for flexLM. Whilst you can control which external IP addresses can access the internal license server the increase in site maintenance may not make it worth the effort. In article <37964ef8.605363216@news.maltanet.net>, chipfactory@hotmail.com writes >Hi Folks: >When we read the documentation for Synopsys we found out that >a license server can also be somewhere in the internet. Now we >came up with the idea to connect our Synopsys/Cadence/Modeltech >license server to the internet and share our licenses with others. >During the night here in Europe our licenses are not used. >Would there someone be interested and is this legal? > >Best regards >Jack > >email: chipfactory@hotmail.com >Article: 17309
It currently claims to be capable of VHDL. The first instance of VHDL starts on page 61 of the manual, "Entering the Design Using the VHDL Hardware Description Language.". Ray Andraka wrote: > Last time I checked, VHDL wasn't included in the student edition, of course > that was a while ago now. > > Nicholas Brown wrote: > > > Hello, > > I have just purchased the Xilinx Foundation Series 1.5 Student > > Edition, a XS40 and a XS95 board from Xess Corp. Unfortunately, the > > license software dosn’t seem to work and Xilinx now claims that they > > don’t support the student edition. I was hoping that someone here knows > > a solution to the problem. > > Everything seems to work except the VHDL synthesis. When attempting > > to synthesize a VHDL entry I get: > > > > -- > -Ray Andraka, P.E. > President, the Andraka Consulting Group, Inc. > 401/884-7930 Fax 401/884-7950 > email randraka@ids.net > http://users.ids.net/~randrakaArticle: 17310
Correction to the schematic. There is an invertor from the counter to the first stage nand gate. Evan evansamuel@earthlink.netArticle: 17311
http://www.cstn.com A staffing option that does not involve recruiters or headhunters!!!!!!! Regards, BlakeArticle: 17312
You should have Synopsys FPGA Express as your synthesis tool for VHDL which is provided in the 1.5 Student Edition. Do not edit the license.dat file as it's a special file and isn't that easy to do (there will be a checksum at the botttom of the file to stop people just changing the configuration file whenever they wanted to). I haven't any experience with 2 drive configurations, but would assume that the volume-id would have to be on the primary boot drive (C)- check on the Xilinx support site for this as they should have info on the Flex-LM licensing issue- do you have a network card in your PC?- if you have then host to this as it is a better option since you don't hit a problem when your hard drive crashes. Other thing to check is the the 'set' in your autoexec has worked properly, got to a command prompt and type: set which will echo your configuration settings. Hope this helps and feel free to e-mail if you can't get it running. CraigArticle: 17313
Thanks for the drawing. Close but no cigar! Eliminate all the nets driven by the two Q outputs. Then make the Control flip-flop's Q drive the NAND gate and, through an inverter, the active High counter Reset. So when the Control flip-flop gets set, it starts the oscillator and allows the counter to count. The MSB of the counter, through an inverter, must drive the active Low Reset of the control flip-flop. So when the counter reaches two-to-the-n, it immediately resets the control flip-flop, which stops the oscillator and also resets the counter. There is a benevolent race condition between the MSB of the counter and the Control flip-flop. If there are more questions, we can continue off-line. The whole newsgroup may not be interested... Peter Alfke ================================================= Earthlink News wrote: > Correction to the schematic. There is an invertor from the counter to the > first stage nand gate. > > Evan > > evansamuel@earthlink.net > > [Image]Article: 17314
We are currently investigating upgrading of our platform hardware. We have a team of 3 designers each with an out-dated PC, these are networked with two somewhat obsolete Spark Workstations. We are running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to Quartus planned). With the good price performance ratio of PC platforms and the ease of use of NT, we are considdering migrating our toolset to high end NT machines. What the heck, we need new PC's anyway. What does concern me however is the lack of multi-user support and performance levels. I would hate to overlook any important issues which should be considdered in making this decision, so if you have any input please mail it to me / post it to the group. ...how beneficial is a Multi-CPU system for these tools ? Regards, Mark Kinsley mkinsley @ xs4all.nlArticle: 17315
I think I may know what your problem is. The Foundation series software wants to be node locked, using either your hard disk serial number or your ethernet address. The problem is that the software doesn't recognize hard disk serial numbers that start with a zero "0". I have 1.5i and have been on temp licenses for the last 8 months. I'm told Xilinx is working the problem, but no date for a fix yet. Good luck.Article: 17316
In article <379f7c9e.221323198@news.rdc2.occa.home.com>, David Rogoff <drogoff@home.com> wrote: >Cadence and Synopsys (and probably others) have terms in their license >agreements and prohibit license servers from being more than a specified >distance from the machine requesting the license. This makes it against the >rules to share licenses between offices of the same company if they are in >different cities! I don't think that there is anything that can technically >stop you from doing it, but you are breaking a legal contract. Not to belabor the point, but the question of whether, in general, the rather amusingly restrictive license agreements issued by most commercial SW vendors constitute "legal contracts" is something that has yet to be definitively decided in US courts (don't know about the rest of the world). I'm not picking on Cadence/Synopsys/Mentor, most of these license agreements are pretty bad, and many of the PC-oriented licenses are absurdly bad. I particularly enjoy the ones that restrict me to a "single backup copy" of the software-- I take a backup of that SW every bleeping week, every month I set one of those aside for two years, then keep one of those per year for three more years (five years total), something like 35 backup copies total. :-) But most customers behave ethically, and *nobody* wants to take one of these agreements to court, least of all the vendors. -- Jay Lessert Portland, Oregon USA jdl@teleport.comArticle: 17317
First and foremost, you should demand that your tools be supported by Linux. Forcing you to go to NT is a huge step back in usability (as you've noticed). Without this valuable feedback, those companies will just keep shoveling support at NT which is a clearly inferior platform for development of this sort. I suppose it all depends on how your designers use the CAD tools. Do they log in from remote and run large scripts on them (like many do)? Do they run scripts to do format conversions (like many do)? If so, then your designers may be happier with the newer Solaris boxes. Solaris machines have come down in price considerably in the past year or two. Without support for Linux, I would not even consider a PC for serious CAD work. Cheers, Jake In <3794B84E.49BF@xs4all.nl> Mark Kinsley <mkinsley@xs4all.nl> writes: >We are currently investigating upgrading of our platform hardware. We >have a team of 3 designers each with an out-dated PC, these are >networked with two somewhat obsolete Spark Workstations. We are >running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to >Quartus planned). > >With the good price performance ratio of PC platforms and the ease of >use of NT, we are considdering migrating our toolset to high end NT >machines. What the heck, we need new PC's anyway. What does concern me >however is the lack of multi-user support and performance levels. > >I would hate to overlook any important issues which should be >considdered in making this decision, so if you have any input please >mail it to me / post it to the group. > >...how beneficial is a Multi-CPU system for these tools ? > > >Regards, >Mark Kinsley >mkinsley @ xs4all.nl -- janovetz@uiuc.edu | Once you have flown, you will walk the earth with University of Illinois | your eyes turned skyward, for there you have been, | there you long to return. -- da Vinci PP-ASEL | http://www.ews.uiuc.edu/~janovetz/index.htmlArticle: 17318
<!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> You can do most of your development on Linux. ModelSim is not available for Linux as far as I know, although you should ask them, but several other simulators are. I'm using Finsim + Undertow which has a bundle price that is similar to ModelSim's, and I've been happy with it so far.VCS and Avant's simulator are also available on Linux. As far as I know there are no synthesis tools on Linux yet so you might what to get one Solaris box to put on your network. Linux and Solaris work together naturally. What you don't want to do is use NT, if you're coming from a Unix environment you will find that NT is hopelessly inadequate. <p>Josh <br> <br> <br> <br> <p>Mark Kinsley wrote: <blockquote TYPE=CITE>We are currently investigating upgrading of our platform hardware. We <br>have a team of 3 designers each with an out-dated PC, these are <br>networked with two somewhat obsolete Spark Workstations. We are <br>running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to <br>Quartus planned). <p>With the good price performance ratio of PC platforms and the ease of <br>use of NT, we are considdering migrating our toolset to high end NT <br>machines. What the heck, we need new PC's anyway. What does concern me <br>however is the lack of multi-user support and performance levels. <p>I would hate to overlook any important issues which should be <br>considdered in making this decision, so if you have any input please <br>mail it to me / post it to the group. <p>...how beneficial is a Multi-CPU system for these tools ? <p>Regards, <br>Mark Kinsley <br>mkinsley @ xs4all.nl</blockquote> </html>Article: 17319
Thanks for the input, I really like Linux and the concept of an open OS, but as far as our design team goes, we're quite happy with whatever makes doing our design work the easiest, and let's face it, Linux is unfortunately a few steps behind MS in this area. For uneducated hardware designers, system admin is a touch easier (and more confortable) under NT. I do agree that pressure should be placed on the software vendors to support Linux so that as Linux gets better, all the tools will be available. ? What exactly do you define as "a huge step back in useability". ? What format conversions are you referring to ? Currently our designers have Win 95 PC's on which the design capture is performed (using Codewright & Ease VHDL). The VHDL output files are simulated on either the PC or the Spark (we have licenses for both). For synthesis, the VHDL files are FTP'd accross to the Spark and synthesized by running a script file through Leonardo. The EDIF netlist output is then placed and routed on either a PC or Spark (we have one license for both). I have seen that we can run similar scripts on a PC platform. What are the genuine merits of a Solaris solution over an NT solution ?.... 1. Useability: ? Here i say NT wins ? Explain why i should change me view if you disagree. 2. Cost/Performance Ratio: ? 3. Multi-User: Solaris wins hands-down 4. Tool support: Seems about the same for NT & Solaris (rumour has it Quartus initially only supports PCs) 5. Software Licensing: Solaris sometimes more expensive 6. Multi-CPU support: Slightly better under Solaris, but this is more a software dependent issue, and is only relevant with multiple users (thus Solaris) 7. Anything else i should be looking at.. Just thinking aloud, and ready to listen... Mark Jake Janovetz wrote: > > First and foremost, you should demand that your tools be > supported by Linux. Forcing you to go to NT is a huge step > back in usability (as you've noticed). Without this valuable > feedback, those companies will just keep shoveling support at > NT which is a clearly inferior platform for development of this > sort. > > I suppose it all depends on how your designers use the CAD tools. > Do they log in from remote and run large scripts on them (like many > do)? Do they run scripts to do format conversions (like many do)? > If so, then your designers may be happier with the newer Solaris > boxes. Solaris machines have come down in price considerably in > the past year or two. Without support for Linux, I would not even > consider a PC for serious CAD work. > > Cheers, > Jake >Article: 17320
asic & eda today [i believe i got the name right] a while back did a detailed study of this question, mostly for the asic flow. they did a head to head comparison of nt vs. unix and addressed linux in depth. an independent lab did it. also considered was the software packages that you had to get and their cost. of course, the #1 thing is the software, really, and what you want to run. me? i have linux, '95, NT, and unix boxes, although i do almost all my work on nt or '95. just not enough s/w for linux to make it very useful for design and compatibility for office work isn't adequate on linux [we've run our own independent tests]. it seems that there have been some new releases of sim s/w so linux is something to watch. i don't want to have to maintain the linux box, don't want to invest the time, and '95 and NT are easy to maintain. for asic, you can get all the tools on unix, so if that's your target probably the way to go, as they are getting cheaper. for fpga's, either unix or windoze would work, although i prefer the pc for that, limiting the amount of different computers i have to deal with [office type stuff has to be on pc for compatibility with the rest of the world]. if you want fully remote operation, then unix/linux is nice with good telnet. i've heard running X over the network from home doesn't hack it very well. but if you're running command line and scripts and no graphical output that'll work. there are remote s/w add-ons for PC which i've heard good things about but i haven't tried them. if you run remote, you might also be able to run local on home pc and check out license over the net. that might work fine, too. for performance, the study showed that there was no big difference between pc/nt and unix, with NT perhaps a bit ahead. with the new cpu releases, NT will probably be a bunch ahead. the big thing that affects performance, from the study and my own observations is the size of memory. get lots of it. even if you don't get it all now, make sure your motherboard supports at least 1 gigabyte. sometimes the synthesis algorithms just get into a mode where they want lots of ram, and then they die if they have to hit the hard disk. good luck, rk ---------------------------------------- Mark Kinsley wrote: > We are currently investigating upgrading of our platform hardware. We > have a team of 3 designers each with an out-dated PC, these are > networked with two somewhat obsolete Spark Workstations. We are > running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to > Quartus planned). > > With the good price performance ratio of PC platforms and the ease of > use of NT, we are considdering migrating our toolset to high end NT > machines. What the heck, we need new PC's anyway. What does concern me > however is the lack of multi-user support and performance levels. > > I would hate to overlook any important issues which should be > considdered in making this decision, so if you have any input please > mail it to me / post it to the group. > > ...how beneficial is a Multi-CPU system for these tools ? > > Regards, > Mark Kinsley > mkinsley @ xs4all.nlArticle: 17321
In article <379518D6.27E1@xs4all.nl>, Mark Kinsley <mkinsley@xs4all.nl> wrote: >Thanks for the input, > >I really like Linux and the concept of an open OS, but as far as our >design team goes, we're quite happy with whatever makes doing our design >work the easiest, ... >What are the genuine merits of a Solaris solution over an NT solution >?.... > >1. Useability: ? Here i say NT wins ? Explain why i should change me >view if you disagree. >2. Cost/Performance Ratio: ? >3. Multi-User: Solaris wins hands-down >4. Tool support: Seems about the same for NT & Solaris (rumour has it >Quartus initially only supports PCs) >5. Software Licensing: Solaris sometimes more expensive >6. Multi-CPU support: Slightly better under Solaris, but this is more a >software dependent issue, and is only relevant with multiple users (thus >Solaris) >7. Anything else i should be looking at.. 7 It is much easier to deal with remote access to "server" machines (using X windows, etc.) using Solaris/Linux... 8. Stability/Reliability... Solaris/Linux Ed Hepler Dr. Edward L. Hepler President, Adjunct Professor, VLSI Concepts, Inc. Villanova University Graduate Courses: VLSI and System ECE-8440 System Design and Modeling Architecture, Design, ECE-8445 Advanced Computer Architecture and CAD ECE-8460 VLSI Design email: hepler@vlsi-concepts.com or elh@ece.vill.edu voice: (610) 408-9121 fax: (610) 408-9121 www: http://www.vlsi-concepts.com Read: I John 5:10-13Article: 17322
Mark Kinsley <mkinsley@xs4all.nl> writes: > With the good price performance ratio of PC platforms and the ease of > use of NT, we are considdering migrating our toolset to high end NT > machines. What the heck, we need new PC's anyway. What does concern me > however is the lack of multi-user support and performance levels. First, demand Linux tools from the vendors (they'll ignore you but do it anyway). Then, if you choose unix over NT and you are willing to switch tools, you can find simulators for Linux but no synthesis or P&R. If you simulate a lot relative to synthesis and P&R time, you might be better off with a Solaris box running the latter and Linux boxes running the former. This way you stay in a productive environment while still keep the cost low (and you can keep demanding Linux versions :-). If you go NT, you may get back to that "ease of use" clause later ... Regards, Zoltan -- +------------------------------------------------------------------+ | ** To reach me write to zoltan in the domain of bendor com au ** | +--------------------------------+---------------------------------+ | Zoltan Kocsi | I don't believe in miracles | | Bendor Research Pty. Ltd. | but I rely on them. | +--------------------------------+---------------------------------+Article: 17323
Mark Kinsley wrote: > > Thanks for the input, > > I really like Linux and the concept of an open OS, but as far as our > design team goes, we're quite happy with whatever makes doing our design > work the easiest, and let's face it, Linux is unfortunately a few steps > behind MS in this area. For uneducated hardware designers, system admin > is a touch easier (and more confortable) under NT. I do agree that > pressure should be placed on the software vendors to support Linux so > that as Linux gets better, all the tools will be available. ? What > exactly do you define as "a huge step back in useability". ? What > format conversions are you referring to ? Depends all on how much automation you do. Attempting to automate things under NT *sucks*. I really believe in automating as much of the process as possible; That way, some user (usually me) can't screw it up or forget a step. We do post-layout simulation with an EDIF netlist, and to jump back and forth between the unix ws (for everything but the Maxplus2 fitter) and the PC just to fit the FPGA is very annoying. What I like to be able to do is make some change to the VHDL, and kick off the entire process again *without* having to be there. It's so much more efficient, esp. if the design is large. What I define is the easiest is to make a change to the source and then walk away until I have simulation results available. System admin is easier under NT? May be just me, but I don't think so. Each NT box has to have the software installed under it, vs 1 for the unix box. I can set up another linux box to add to our group a heck of a lot faster than I could ever set up another NT box. All the software's installed on one NFS server box. Makes upgrades so much easier, too. I spend as much time maintaining one PC as I do the entire unix cluster, if not more on the one PC. At least in our setup, Backup of the NT boxes is a pain. We don't have any good way of doing it. And, unfortunately since we run Veribest for the board layout, we're dependent on an NT box to serve out ODBC data for it. And, as expected, NT ends up messing up every week or so and the machine has to be rebooted for one reason or another. Then every user in the dept has to shut down to avoid problems while the machine is rebooted. We also have to keep the PCB files on those machines to make performance acceptable. Makes a real problem when the box acts up. > Currently our designers have Win 95 PC's on which the design capture is > performed (using Codewright & Ease VHDL). The VHDL output files are > simulated on either the PC or the Spark (we have licenses for both). > For synthesis, the VHDL files are FTP'd accross to the Spark and > synthesized by running a script file through Leonardo. The EDIF netlist > output is then placed and routed on either a PC or Spark (we have one > license for both). I have seen that we can run similar scripts on a PC > platform. Yes, but can you script all of the stages together on NT? I think not. We also have written a lot of wrapper tools to make test benches for us, create stimulus macros, edit incompatibilites between tools, extract info from report files... etc. This works so much better under Unix. BTW, why aren't you putting the VHDL files on a network drive? Then you wouldn't have to ftp them. > What are the genuine merits of a Solaris solution over an NT solution > ?.... > > 1. Useability: ? Here i say NT wins ? Explain why i should change me > view if you disagree. Depends on your user :). If you want batch capability or would rather type than click a lot, NT doesn't cut it. The newer GUIs on Linux are a wash, if not better than NT's. > 2. Cost/Performance Ratio: ? I think on the same hardware, Solaris will beat out NT. I know linux does over 95. Exemplar was slower by 2x under 95 than under linux. > 3. Multi-User: Solaris wins hands-down Quite. > 4. Tool support: Seems about the same for NT & Solaris (rumour has it > Quartus initially only supports PCs) Pretty much a wash in my experience. > 5. Software Licensing: Solaris sometimes more expensive How many dongles can you put off the back of the PC? Sometimes they interact with each other, too. Floating licences are wonderful, esp if you end up having to go to the factory floor to debug something and need to make a change. > 6. Multi-CPU support: Slightly better under Solaris, but this is more a > software dependent issue, and is only relevant with multiple users (thus > Solaris) No, some programs (albeit very few) can run with multiple CPUS... Don't remember which ones, though. I think some simulators.... > 7. Anything else i should be looking at.. 8. Reliability/Uptime. NT Loses by a huge margin. > Just thinking aloud, and ready to listen... > > Mark For me, automating the design process as much as practical is what makes it the easiest. That leaves me time (and brain cells) to concentrate on designing. Repetitive steps should be done by the computer, not by me telling it to do the same exact steps each time. I'd rather just type make and go get a soda than sit, click on one pgm, click several times, click again, start another program, click some more..... It's so much more efficient. Gives me more time to think about the design. JerryArticle: 17324
Hello, I installed the Xilinx WebPack, there free CPLD Fitter environment. The documentations says it should include a chip viewer for the XC9500 family, but I did not see it. Does anybody know more about? Best regards Klaus -- Klaus Falser Durst Phototechnik AG I-39042 Brixen Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.
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