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
Stuart Clubb wrote: > On 13 Aug 1997 22:48:15 GMT, lass@xilinx.com (Steve Lass) wrote: > > >M1 does support the MAXSKEW constraint which can be used for clock nets. > > Please excuse my lack of knowledge, but does M1 support 3K series, the > origin of this thread from Peter? We are shipping 3K beta this week. Official support for 3K in M1 willbe in November. > Isn't this "preference" from the original NeoCAD engine anyway? > Yes. SteveArticle: 7251
[M.Vorbach] ;-)) Nice try. > > Will someone please send me > the file License.dat for > FPGA Express v1.2? > Mine is corrupt! > >Article: 7252
> Roger, > [M.Vorbach] MCKENZIE has normal (non zero insertion force) sockets: 810 Page Avenue Fremont, California 94538 USA Telephone (510) 651-2700 Fax (510) 651-1020 We use also single pin sockets and mount 503 pieces of them. Supplier is: RS Components Part. No. 519-959 The advantage is, that the heigth of the board is not much higher than by mounting the FPGAs without sockets. > In article <33F334B7.7F9E@net.polyu.edu.hk>, > "Yau Man Wai , Roger" <rogeryau@net.polyu.edu.hk> wrote: > >Hello, > > I ordered a Altera 10K100GC503. I have search for AMP amd > >3M but no such high pin counts PGA socket for this device, does > >anyone know where can I order a 503 pins PGA socket for this device? > >Thank you! > > > >Roger Yau > >R & D Engineer > >Easson Precision Ltd. > >http://www.net.polyu.edu.hk/~rogeryauArticle: 7253
Sometimes I believe that XILINX-fitting is better. But only sometimes and this is an architectural problem. We had problems using 9K-technology, but within 2 weeks we got help from ALTERA (a bug fix). 10K works (up to now) great. I believe the tool is OK and the support too. > can anybody compare this tool with other comparable tools? > > how easy is it to learn?how is the support? > __________________________________________________________________Article: 7254
Hi! I'm having problems using LogiBLOX components as component instances in my VHDL-code. I'm using the LBGUI tool to customize the blocks to match my need. I then try to include these block in my code by inserting the generated VHI-file into my code and completing the port assignments. But when I try to read in the design into Synopsys Design Analyzer the tool does not recognize the component. I either would like to inlude the LogiBLOX like any other library, and then customize the component with attributes (but don't know how), or somehow make Synopsys realize that the implementation of the component is given by the generated netlist in the NGO-file. I've read through the material in the dltext documentation, but it doesen't help me out. Any help greatly appreciated. Joachim StrömbergsonArticle: 7255
Graham Rhodes wrote: > > Tim Conway wrote ... >> We are considering the use of ISP CPLD's Anyone have any amusing >> or helpful anecdotes regarding their use? > >..cut.. >I currently use cypress ISP parts, and have no real problems with >programming them - it all works very nicely. The only minor snag is that >they need a 12V programming voltage. Cypress gets away with this by >creating this in the PC cable. Other devices are nicer and only need >a 5V programming voltage (AMD?) An advantage of requiring external 12V supply for programming is that the internal charge pump circuit is eliminated. That makes the device more robust in terms of effects of Vcc overvolts (in my experience an 8V Vcc spike on a 5V-only programmable part destroys it completely, 10V Vcc spike had no ill effects on a part using 12V programming - both spikes were outside data book ratings of the standard 7V absolute maximum). Downsides are obvious (extra pin, extra net to be routed, have to generate 12V somehow, etc). It depends on the intended product type whether the swings are nicer than the roundabouts. ISP of any form has many plus points which are set out by all manufacturers offering it. Not sure any qualify as "amusing", but they can be truly helpful. I can repeat my anecdote of how to blow up Lattice ispLSI if you like, but that was not amusing either. Tim Forcer tmf@ecs.soton.ac.uk Department of Electronics & Computer Science The University of Southampton, UK The University is not responsible for my opinionsArticle: 7256
On Fri, 15 Aug 1997 12:37:26 -0500, Tim Conway <timc@primafacie.com> wrote: >We are considering the use of ISP CPLD's. Anyone have any amusing >or helpful anecdotes regarding their use? >Thanks much. >-- >Tim Conway >Prima Facie, Inc. >(314) 989-0644 ext. 16 >(314) 989-0654 Fax I don't know if its amusing or not but... If you're using Lattice parts, you have to fully complete and debug your design before you lay out a circuit board. And the signal pinout will be completely arbitrary, not easily circuit board routable. If you think that you can go in and re-assign pinouts or make logic changes after the fact you may be in for a nasty suprise - the Lattice router may no longer be able to fit your design into the part ! I have found this to be the case if the part is over 50% utilized. The Synario software is easy to use, much better than their own PDS. Other than these discrepancies we find the Lattice parts to be useful and have shipped several thousand units of various designs with no manufacturing problems. Steven J. Ackerman, Consultant ACS, Sarasota, FL sja@gte.net http://www.acscontrol.comArticle: 7257
Hi! I need to prototype a board incl. a complex ASIC. Do anyone know about of-the-shelf boards with space for a lot of big FPGAs and perhaps configurable routing? Any other hints? Regards, Bjorn.Article: 7258
I'm not sure about other tools but I've been using this for about a year and have NO compalaints. I'm doing a 10k20 desgn. > >can anybody compare this tool with other comparable tools? > >how easy is it to learn?how is the support? > >thnaks >umesh > >__________________________________________________________________ >Undertake something that is difficult,it will do u good; >Unless u try to do something beyond what u have already mastered >you will never grow > -:Ronald E Osborn >__________________________________________________________________ > Regards, Aaron Quantz \^ ^/ )@ @( +---------------------------oOO--(_)------------------------------------+ + Mgr Software Development, Turret Control Systems + + HR Textron | Phone: (805) 253-5471 + + 25200 W. Rye Canyon Rd. | Fax: (805) 253-5962 + + Valencia, CA USA 91355-1265 | Email: aquantz@ibm.net + + Visit the Textron web site: http://www.textron.com + +-----------------------------------Oooo--oOO---------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_)Article: 7259
Graham Rhodes wrote: <snip> > I currently use cypress ISP parts, and have no real problems with > programming > them - it all works very nicely. The only minor snag is that they need a > 12V programming voltage. Cypress gets away with this by creating this in the > PC cable. Other devices are nicer and only need a 5V programming voltage > (AMD?) > This is a two edged sword - Sure 5V is convienent, BUT I'd feel happier shipping product that was GUARANTEED to never kick into PGM mode by 'accident'. ( see an earlier posting about being careful about PowerUP / ISP plugging order ) The other advantage of 12V, is it allows the ISP pins to be SYSTEM available, the 5V only parts cannot do this. Also a SPI interface / connector for ISP, that is also RUNTIME accessible, has to be appealing. On FLASH PROM devices, which are runtime written, 5V makes sense, but surely not on CPLD ? Maybe the ideal solution is BOTH voltage options, with one die, but factory selected to 5V for RunTime PGM customers, and 12V for factory/safe PGM. - jim -- ======= Manufacturers of Serious Design Tools for uC and PLD ========= = PLD Applications libraries available, for i2c, SPI, SPL = for more info, Email : DesignTools@xtra.co.nz Subject : c51ToolsArticle: 7260
Thanks for the info. But what the *@#$ are you saying? I had assumed 89c2051s are 8051 object compatible. Is "i2c" the serial protocol used by "24Cxx" parts? Is "SPI" "Serial Programming Interface", for programming Xilinxes? (a wheel I didn't need to re-invent?) In that case, i2c -> SPI makes sense, but it's the verification that's rough. What would I use an "SPI <-> SPI data pump" for? Have I slept through an important but quiet revolution? jim granville wrote: > > > Have look at the 89C2051 ( 20 pins ) and the new SPI DataFLASH from > > > ATMEL. > > About how much does a low-end (slow, OTP) 89C2051 cost? > Here they are between $3-4US, in small quantities. For this you get > DIP2O or SO20, and 2K FLASH with 128 Bytes RAM, 24MHz Clock, 2.7-6V Vcc. > They will run your 80C52 SPI code unchanged. > ( the 89C1051 is cheaper again, with 1K FLASH, 64 Bytes RAM, and an ISP > version is comming ... ) > > i2c LIB will cost appx 105 bytes, and SPI is appx 60, so you can do a > 24Cxx <-> SPI or SPI <-> SPI data pump with a lot of room to spare.Article: 7261
There is a basic difference between the Xilinx and Altera daisy-chain structure: In the Xilinx case, all configuration data is passed through the daisy chain, while ( I think ) in the Altera case it is an ENABLE signal that is passed through the daisy chain. Xilinx devices count every rising edge of CCLK in their internal length counters ( "every " means each and every one ! ), and they all pass the preamble through the daisychain. After the preamble, the head of the daisy chain consumes data until it is full, then passes the incoming data through to the next device, which consumesit until it is full, etc. It might be quite a challenge to combine these bitstreams, and I ask: Why do it? Why not run them in parallel ? Peter Alfke Xilinx ApplicationsArticle: 7262
APTIX makes such a board. In article <33F9B003.1381@emw.ericsson.se> Bjorn Sihlbom <emwbs@emw.ericsson.se> writes: >Hi! > >I need to prototype a board incl. a complex ASIC. >Do anyone know about of-the-shelf boards with space for a lot of big >FPGAs and perhaps configurable routing? > >Any other hints? > >Regards, > >Bjorn.Article: 7263
Dirk, What exactly do you mean by the "same" configuration lines? You should be able to connect the "DONE" pins together in both the Altera and Xilinx parts so that the two set of devices wake up together. Jose -- Jose A. Paredes IBM RS/6000 Division Office: 512-838-3855 Fax: 512-838-1852 josep@austin.ibm.comArticle: 7264
> Can anyone come up with a plausible explanation of how this could be legit? > Better yet, given that license.dat files are usually associated with a > physical key (and the serial number it contains), or as in M1, the serial > number of the C drive, can anyone come up with a way that sending your > license.dat file to this person could be anything other than pointless. Better yet, that the license.dat file is a text files! A text file could still be corrupted....but it shouldn't be that hard to repair...and they usually send it to you in e-mail, or on a disk...so....where's the original? AustinArticle: 7265
The PCI bus spec guarantees 7ns setup and 0ns hold. In order to meet the PCI spec, the 4k IOB flip flops have to be set to delay...which claims to require a 7ns setup, and 0ns hold. Ok, all that seems fine...but when I generate a .vsm file and simulate this using ViewSim and a command file that gives 7ns setup and 0ns hold, I get setup violations..claiming setup is 6.2, 10.2 is required. Interestingly enough, it's not the 7ns setup that is causing the problem, it's the 0ns hold. When I change the 0ns hold to 11, it works fine. Now...my first guess is something is wrong with the timing model that Xilinx provides. I have not gone into the .vsm file to see what it is that is wrong, but I am about to. What my question is, has anyone simulated a Xilinx PCI design using ViewSim and using 7ns setup and 0ns hold....successfully? Thanks, Austin Franklin darkroom@ix.netcom.com To reply to this post, remove the number from the reply address.Article: 7266
Umesh Nair wrote: > > can anybody compare this tool with other comparable tools? > > how easy is it to learn?how is the support? > > thnaks > umesh > > __________________________________________________________________ > Undertake something that is difficult,it will do u good; > Unless u try to do something beyond what u have already mastered > you will never grow > -:Ronald E Osborn > __________________________________________________________________ Hi Umesh, I have used the software over a few years now, and it's really easy to work with it. I prefer the Altera sw instead of some sw from Xilinx for example. Regards Gerhard Vogt -- ------------------------------------------------------------------------------------------------------------------ Gerhard Vogt | Newbridge Networks Corporation E-mail: Gerhard.Vogt@newbridge.com | Suite 301, 8555 Baxter Place Phone: (604) 421-2643 Ext. 5718 | Burnaby, BC Direct: (604) 415-1400-5718 | Burnaby, BC Fax: (604) 421-2644 | Canada, V5A 1V7 PGP Key available ---------------------------------------------------------------------------------------------------------------------Article: 7267
Bill Ewing wrote: > > jim granville wrote: <snip> > > Have look at the 89C2051 ( 20 pins ) and the new SPI DataFLASH from > > ATMEL. > > This has 4MBit Serial FLASH, at under ByteWIDE prices. > > Yes, I've seen the recent discussion. I often need a few I/O pins, but > I guess you can get that through the Xilinx, once it's operational. > About how much does a low-end (slow, OTP) 89C2051 cost? I have a pending > (i.e., overdue) project I think it might fit into. Here they are between $3-4US, in small quantities. For this you get DIP2O or SO20, and 2K FLASH with 128 Bytes RAM, 24MHz Clock, 2.7-6V Vcc. They will run your 80C52 SPI code unchanged. ( the 89C1051 is cheaper again, with 1K FLASH, 64 Bytes RAM, and an ISP version is comming ... ) i2c LIB will cost appx 105 bytes, and SPI is appx 60, so you can do a 24Cxx <-> SPI or SPI <-> SPI data pump with a lot of room to spare. - jim -- ======= Manufacturers of Serious Design Tools for uC and PLD ========= = Optimising Modula-2 Structured Text compilers for ALL 80X51 variants = Reusable object modules, for i2c, SPI and SPL bus interfaces = Safe, Readable & Fast code - Step up from Assembler and C = Emulators / Programmers for ATMEL 89C1051, 2051, 89C51 89S8252 89C55 = *NEW* Bondout ICE for 89C51/89C52/89C55 = for more info, Email : DesignTools@xtra.co.nz Subject : c51ToolsArticle: 7268
Exjobbare Joachim Strombergson wrote: > > I'm having problems using LogiBLOX components as component instances in > my VHDL-code. I'm using the LBGUI tool to customize the blocks to match > my need. I then try to include these block in my code by inserting the > generated VHI-file into my code and completing the port assignments. But This is the problem. DO NOT insert the code into your code. The VHDL output is really just a simulatable model, NOT a synthesizable model. You should treat the logiblox as a hierarchical piece of the code and instantiate the module as a lower hierarchy. > when I try to read in the design into Synopsys Design Analyzer the tool > does not recognize the component. Correct, because it is not synthesizeable. You just used LBGUI to optimize the block, don't get Synopsys to have its way with it. The NGO output of the logiblox IS the synthesized code. In your hierarchical design, you need a synthesis black_box or similar don't touch attribute to get synopsys to compile the design correctly but not synthesis the LogiBlox component. > > I either would like to inlude the LogiBLOX like any other library, and > then customize the component with attributes (but don't know how), or > somehow make Synopsys realize that the implementation of the component > is given by the generated netlist in the NGO-file. > LogiBlox does customize the component and you do include the component like any other library. I suggest you look in the documentation for examples of including logiblox components. (I'd give it a shot but I'm a verilog guy). Tim. -- You better be doing something so that in the future you can look back on "the good old days" My opinions != Nortel's opinion;Article: 7269
jim granville wrote: > > Steven J. Ackerman wrote: > > > <snip> > > If you're using Lattice parts, you have to fully complete and debug > > your design before you lay out a circuit board. And the signal pinout > > will be completely arbitrary, not easily circuit board routable. If > > you think that you can go in and re-assign pinouts or make logic > > changes after the fact you may be in for a nasty suprise - the Lattice > > router may no longer be able to fit your design into the part ! I have > > found this to be the case if the part is over 50% utilized. > > Yikes! > We have just completed a design where we routed the PCB FIRST ! > ( using ATMEL ATF1500 PLDs, 100% & 87% packed ) Yes, but doesn't that series have a fully populated switch matrix. It's a very nice feature to be able to lay out your PCB before you do the logic. However, correct me if I'm wrong, aren't you limited to 32 macrocells and a 15ns Tpd with the Atmel parts? > This 'my fitter knows best' philosophy is too common, but with pressure, > things will > get better :-) True, very true. Both with the fitters and with the parts themselves. A Vantis marketing guy told me that they have a new generation of CPLD's coming that are designed based on the fitter, as opposed to designing the fitter to work with the PLD. An interesting concept. -- Tim Conway Prima Facie, Inc. (314) 989-0644 ext. 16 (314) 989-0654 FaxArticle: 7270
Bill Ewing wrote: > > Thanks for the info. But what the *@#$ are you saying? I had assumed > 89c2051s are 8051 object compatible. Correct. I mentioned this because not all tiny C51's were 100% compatible. Fortunately the 2051 is just a C51 in 20 Pin Box. > Is "i2c" the serial protocol used by "24Cxx" parts? Yes. i2c is the Philips Inter IC bus, also called 'two wire' bus by those paranoid about acknowledging Philips :-) Many suppliers make 24C16/32/64/128/256 K Bit Serial EEPROMS/FLASH. They are probably cheaper than the more niche 17xxx FPGA series. > Is "SPI" "Serial Programming Interface", for programming Xilinxes? (a wheel I didn't need to re-invent?) Serial Peripheral Interface, is a simpler BUS than i2c, and is very similar to JTAG - HW level. Typically you have a long shift register :: CLK MISO - Master In, Slave Out MOSI - Master Out, Slave IN CS - Some form of ChipSelect. > In that case, i2c -> SPI makes sense, but it's the verification that's rough. > What would I use an "SPI <-> SPI data pump" for? Have I slept through an > important but quiet revolution? Here I used SPI to mean CLK.DATI.DATAO Shifter BUS The DataFLASH devices I mentioned are serial ( SPI R/W interface, SO28 package ), and the XILINX and other FPGAs are also Serial, but with different Start.Block.Stop requirements, so you would use a SPI-SPI pump to handle these differences, tho mostly the data stream would be simply duplicated. - jimArticle: 7271
jim granville <Jim.Granville@xtra.co.nz> wrote: >> If you're using Lattice parts, you have to fully complete and debug >> your design before you lay out a circuit board. And the signal pinout >> will be completely arbitrary, >I shudder to think how a 'PLD determined' design would have performed >:-( >This 'my fitter knows best' philosophy is too common, but with pressure, >things will get better :-) It isn't a case of fitter knows best it is a case of CPLD architectural restrictions and all CPLDs have them to a greater or lesser degree. If you lock down pins to guide a fitter towards a possible solution then you are making life easy for it. If you lock down pins towards an impossible solution then you make it work hard and end up with something that will probably break with a small design change. Lattice parts are a little quirky, having a more complex set of small restirctions than Xilinx 9500 or Altera 7000 for example. When you begin to understand the architectural restrictions you begin to understand if and how you can guide the fitter. Lattice parts have a GRP (global routing pool) which can route any signal anywhere but unfortunately with some undocumented limitations, when you lock down pins and the fitter can't fit (which sounds like the problem the previous poster was getting) it is because of these GLB restrictions. Being undocumented it is pretty hard to know what to do about it. Cheers Terry...Article: 7272
Steven J. Ackerman <sja@gte.net> wrote in article <5tc6sr$6l9$1@gte2.gte.net>... > If you're using Lattice parts, you have to fully complete and debug > your design before you lay out a circuit board. And the signal pinout > will be completely arbitrary, not easily circuit board routable. If > you think that you can go in and re-assign pinouts or make logic > changes after the fact you may be in for a nasty suprise - the Lattice > router may no longer be able to fit your design into the part ! Cypress likes to pride themselves on the ability to use completely arbitrary pin assignments when they're trying to convince you to use their stuff over Lattice's. In general I think they're correct, but one thing they don't mention is that even their pin assignments start becoming non-arbitrary when you start utilizing over about 75% of the device. Better than Lattice, but definitely not as cool as they'd have you believe. With FPGAs the problem is that you may not be able to get a signal over to the pin you want it _fast_ enough, right? ---Joel KolstadArticle: 7273
Try vcc.com. We bought one on a isa card; the software support for general purpose win95 use was sad but it works and I am now integrating it into win95 succesfully ( after much trial and error). OTOH they are activily supporting multi FPGA systems for something like "virtual" computing or some such. Enjoy RayArticle: 7274
DON'T BE FOOLED BY IMITATIONS There is only one Pam & Tommy Sex Tape and it is available here for a limited time only. Others may claim to have it for sale but this is the only Hard Core Copyrighten version in existence. See them bare their soles and bodies in public places, but observe closely, a very private blend of love and sex that they thought the public would never see. Visit: http://www.pamsex.com http://www.pamsex.com
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