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
Another, possibly easier way to do what you want is to use the CLBMAP symbol in your Xilinx schematic. This will allow you to specify the lookup tables in the XC3000 CLB's directly. You might want to look at the XC4000E family - this CLB has 3 RAM and 1 ROM mode where the ROM can be initialised at the schematic stage as a lookup table. See your disty or the Xilinx website for info.... -Andrew MetcalfeArticle: 3551
johnson edward eric wrote: > > I am currently working on a project that will require the usage of a large > number of CLB's available in an XC3042pc84 Xilinx Chip. Each CLB has a > truth table represented by a group of about 10-15 AND and OR gates. > XNFMAP generally groups all of these logic gates in CLBs in the way that > I desire it too, but occasionally it splits up these logic gates into two > CLBs when I want them all to be associated with the same CLB. Does anyone > know if there is a way that I can assign a group of logic gates to a > particuler CLB without assigning the CLB to reside in a specific location > in the Xilinx chip? > i'm a fairly new xc4k user with little experience with xc3k, but i believe there is a primitive in xc3k called a CLBMAP which you can use in conjunction with your schematics (i assume you're not using an HDL) to control logic partitioning. check out p3-207 of the "XACT Libraries Guide" for more info. the equivalent capabilities in xc4k are called FMAP's and HMAP's. alternatives might be to use floorplanner or CST files, but this would involve placement which you wanted to avoid. i'm not sure if CLBMAP's would be less work though. good luck, -- _______________________________________________________________________ Lance Gin "off the keyboard Delco Systems - GM Hughes Electronics over the bridge, OFC: 805.961.7567 FAX: 805.961.7739 through the gateway, C43LYG@dso.hac.com nothing but NET!" _______________________________________________________________________Article: 3552
Dear Sir, Please note that I am the manufacturer’s representative therefore should you see or need anything please contact me as I currently represent about 160 different lines. If you like and would like to represent the manufactures goods then simple contact myself details supplied below and I would be more than honoured to put you in direct contact with the manufacturer, this way there is one less middle man and higher profits to be made all round. A sample of just a few products can be found on my web page which is still and continuously under construction, so if you don't see anything please simple Email me and I will certainly obtain it at the best available prices. Hope this introduction between ourselves is only the beginning of a fruitful relationship. Looking forward to hear from yourself soon. Best regards Mark Goldberg -- GoldMar Holdings TEL:(61-9)249-6162 FAX:(61-9)248-9464 GoldMar@wantree.com.au http://cybernet.net.au/goldmar.htmlArticle: 3553
Hello out there! (((( Caution: I'm a SW type, posing this HW problem! ;-)))) We have been using Xilinx XC1765 serial PROMs for a while. We have bought a bunch of Atmel's AT17C65 serial EEPROM so we wouldn't have to keep buying the Xilinx chips, as the Atmel chips are EEPROMs. We bought them thinking they would be a form fit replacement for the Xilinx chips, because they had the same pin-outs, specs, etc. Problem: whem we make a MCS file from the Xilinx tools, it will burn fine in the Atmel chips, but will not run, at all! ARE WE MISSING A STEP? We are thinking that the Xilinx software prepares this file in a certain way so that only Xilinx chips can be used. Whaddya think? Dean Nelson Loral Aerospace Las Vegas, NvArticle: 3554
On 20 Jun 1996, DeanNelson wrote: > Problem: whem we make a MCS file from the Xilinx tools, it will burn fine > in the > Atmel chips, but will not run, at all! ARE WE MISSING A STEP? No, in my opinion there aren't any differences regarding the bitstreams of the PROMs. Remember: a serial 1765-PROM is a parallel PROM with a build-in address counter and a parallel-to-serial converter - and you can connect a parallel PROM like a C256 to a XILINX with the same burned-in bitstream (XILINX-FPGAs offer many more modes of configuration as serial PROMs). But what I think is you forgot to program the RESET polarity of the AT17C65 correctly. Your PROM programmer must offer you to program this polarity! Try to change it. E-regards --------------------------- Rainer Scharnow (amigo@bintec.de) BinTec Commmunications GmbH ---------------------------Article: 3555
DeanNelson (deannelson@aol.com) wrote: : Hello out there! : (((( Caution: I'm a SW type, posing this HW problem! ;-)))) : We have been using Xilinx XC1765 serial PROMs for a while. We have bought : a bunch of Atmel's AT17C65 serial EEPROM so we wouldn't have to keep : buying [..snip..] : Problem: whem we make a MCS file from the Xilinx tools, it will burn fine : in the : Atmel chips, but will not run, at all! ARE WE MISSING A STEP? No, you are not missing a step. The same thing happened to me. The problem is the ATMEL chips, some of them just don't work. ATMEL recognizes the issue and after badgering will provide devices with appropriate date codes. Just return them to the distributor. Alain ArnaudArticle: 3556
DeanNelson wrote: > We have been using Xilinx XC1765 serial PROMs for a while. We have bought > a bunch of Atmel's AT17C65 serial EEPROM so we wouldn't have to keep > Problem: whem we make a MCS file from the Xilinx tools, it will burn fine > in the Atmel chips, but will not run, at all! ARE WE MISSING A STEP? Check that you have the right OE/RESET polarity, as I believe some Atmel chips has been shipped with the wrong default polarity! Please give me feedback if this turns out to be the problem. -- Rune Baeverrud <r@acte.no> Field Application Engineer |\ _,,,---,,_ ScandComp, ACTE NC Norway AS ZZzz /,`.-'`' -. ;-;;,_ P.O. Box 190, N-2020 Skedsmokorset, Norway |,4- ) )-,_..;\ ( `'-' Tel: +47 6389 8969 Fax: +47 6389 8979 '---''(_/--' `-'\_)Article: 3557
I have used the Atmel programmables successfully. However, I burn them via DATAIO 3900, using ATMEL's algorithm. Originally, I tried with my Xilinx box and never saw a DONE/PROGRAM pin go High. The first samples from Atmel were bad, however that was months ago.Article: 3558
The 1765 has programmable OE/RESET polarity. We have seen the same problem and it depends on the programmer you are using...(Data I/O Chiplab and Link in our case fail). Atmel has stepped back and rather than trying to convince the programmer vendors to change the algorithms they have changed their silicon. The new rev should be available about now and is distinguishable by date code. Ask your Atmel rep/FAE... Kevin Steele ksteele@silcom.com kevins@motnengr.com Motion Engineering, Inc. Santa Barbara CA ---------------------------------------------------------------- Dean Nelson wrote: > > We have been using Xilinx XC1765 serial PROMs for a while. We have bought > > a bunch of Atmel's AT17C65 serial EEPROM so we wouldn't have to keep.... > > Problem: whem we make a MCS file from the Xilinx tools, it will burn fine > > in the Atmel chips, but will not run, at all! ARE WE MISSING A STEP? > > DeanNelson > REPLY > Check that you have the right OE/RESET polarity, as I believe some Atmel chips > has been shipped with the wrong default polarity! Please give me feedback if > this turns out to be the problem. > Rune Baeverrud <r@acte.no>Article: 3559
I tried to run xmake from Mentors design-manager to place and route my design. I used the switches x-Blox Optimization, Map-then-merge and Make Palced & routed Design. The Utilization of CLB's was like this: Preliminary estimate of device utilization for part 4013MQ240: ------------------------------------------------------------------------- 14% utilization of I/O pins. ( 26 of 192) 81% utilization of CLB FG function generators. (934 of 1152) 16% utilization of CLB H function generators. (185 of 576) 51% utilization of CLB flip-flops. (582 of 1152 17% utilization of bus resources. ( 16 of 96) -------------------------------------------------------------- The router wasn't able to route teh whole design on this chip ( ONLY 97 % ) Ist there a way to do it with another tool or works this tool perfect? How can I improve my design in a way that i can root it ? Is it normal to get rooting problem with a utilisation of 80% ? Would it be possible to root it on a bigger device (4015)? Remark: There is no 4015 in my partlist and as a student I have no chance to get better (bigger,newer..) libaries. Witch placing policy should I use (already tried Map-then-merge and Merge than Map, but no difference.) I hope somebody can help me, otherwise I can't enjoy my holydays ;-). Thanx in advance -- Manfred Aigner alias maigner@sbox.tu-graz.ac.atArticle: 3560
I have a problem with a current fpga design I am doing. In the design I have a 24 stage shift register. When I do a functional simulation everything works fine, however when I program the chip and try the real thing, my shiftreg fails to work correctly. I had a close look at the gate level simulation and found that the clock for the first register leaded the clock for the second register by 24ns. This gives me, ineffect, a 23 bit shiftreg, as Q0 and Q1 always have the same data. The clock by the way is 2MHz. It looks to me, that the plug and play fitter with Maxplus2 has put the first SR next to the input pin, routed the clock to this register first and then kindly put an lcell in series with the clock line to the next register while routing Q to D directly! The design is in a fairly full(90%) 8636 device. Can anyone suggest how I can control the fitting so that the registers are grouped the way I want them? The clock is internally generated and I have already used the global inputs :( As I only get to this news group once or twice a week, could any replies kindly be CCed to tclark@aesprodata.com.au Thanks TonyArticle: 3561
What would be the key ingredients for most cost effective-, and productivity enhancing mix of (Pentium platform-) FPGA CAE Design Tools ??? (This question asked by a South African company with half-a-dozen Engineers full-time engaged with FPGA design work, but on a tight budget for software tools!) We need to make best possible decisions for upgrade of all our software tools relating to FPGA design work, including the CAD PCB tools. We're looking at getting: 1) a Generic front-end CAD editor, for both PCB and FPGA schematic capture, 2) a mix of Xilinx and Altera tools, 3) 3rd party Simulators and Synthesis products. 4) 3rd party PCB auto-router We think we should purchase VIEWLOGIC front-end tools, which provide us with a common hierarchical schematic design tool for both PCB's and FPGA's, and shall probably then have a couple've XILINX FOUNDATION sites (VIEWLOGIC schematics can be imported directly), but are also convinced that we should be looking at getting some ALTERA sites, and 3rd party VHDL simulator and possibly additional Synthesis tools as well. (Both FOUNDATION and MAXPLUS shall provide us with basic VHDL synthesis and netlist timing simulation.) Can anybody out there advise, before we break the bank? Thanks in advance. PETER FENN.Article: 3562
Over the last few days there have been numerous postings to comp.arch.fpga with regard to Atmel's EEPROM family of FPGA configuration memories (the configurators). I wish to clarify a few points and offer a contact point for any further questions. Atmel's Configurators do work with all known SRAM based FPGAs. We have found that due to a 'percieved' bug in our competitors s/w that some customers have had difficulty when loading from a MCS file created in their s/w. This problem manifests itself in the FPGA taking approx. 20 secs to load from the serial memory. The problem is not with Atmel's serial memories, but with the competitors s/w. I can supply a work arround to anyone who is interested. Atmel chips are shipped with reset polarity high. Most customers have to change reset polarity to low. The AT17Cxxx devices support programmable reset polarity. A broad range of industry standard programmers support the Atmel EEPROM. The reset polarity change is done in one of two different ways on those programmers. If you would like more information on this contact me at the e-mail address below. Atmel EEPROM are not supported on the Xilinx programmer (!) as some customers have tried !! Atmel does offer a $250 programmer for the devices (ATDH2200) that also support ISP of the devices - see Atmels home page for more details. If anyone would like more details on the in-system-(re)programmable AT17Cxxx serial configuration EEPROMs or any other Atmel products please contact me at the e-mail addresses below. Regards, Martin. ------------------------------------------------------------------------- | Martin Mason | Snr FPGA/17Cxxx Applications Engineer | | Atmel Corp. | (email - me) martin@atmel.com | | 2325, Orchard Parkway | (email - support) fpga@atmel.com | | San Jose | (Fax) + (408) 436 4300 | | CA 95131 USA | (Tele) + (408) 436 4178 | ------------------------------------------------------------------------- |Need Atmel Lit. stuff ? WWW.ATMEL.COM or FAX-on-DEMAND 1-800-29-ATMEL | -------------------------------------------------------------------------Article: 3563
johnson edward eric wrote: > > I am currently working on a project that will require the usage of a large > number of CLB's available in an XC3042pc84 Xilinx Chip. Each CLB has a > truth table represented by a group of about 10-15 AND and OR gates. > XNFMAP generally groups all of these logic gates in CLBs in the way that > I desire it too, but occasionally it splits up these logic gates into two > CLBs when I want them all to be associated with the same CLB. Does anyone > know if there is a way that I can assign a group of logic gates to a > particuler CLB without assigning the CLB to reside in a specific location > in the Xilinx chip? > > Thanks > Ed You can with Viewlogic schematic tools use the FMAP, HMAP attribute. Using attributes, you can group logic into portions of a CLB, or group locic relative to a CLB or group of CLB's, or fix the logic down at an exact location. RobArticle: 3564
Rob Froyd wrote: > You can with Viewlogic schematic tools use the FMAP, HMAP attribute. > Using attributes, you can group logic into portions of a CLB, or group > locic relative to a CLB or group of CLB's, or fix the logic down at an > exact location. i believe fmap's/hmap's are to be used with xc4k only. for xc3k, clbmap is the equivalent. -- _______________________________________________________________________ Lance Gin "off the keyboard Delco Systems - GM Hughes Electronics over the bridge, OFC: 805.961.7567 FAX: 805.961.7739 through the gateway, C43LYG@dso.hac.com nothing but NET!" _______________________________________________________________________Article: 3565
* Atmel FPGA Apps * (martin@atmel.com) wrote: : Atmel's Configurators do work with all known SRAM based FPGAs. : We have found that due to a 'percieved' bug in our competitors s/w that : some customers have had difficulty when loading from a MCS file created : in their s/w. This problem manifests itself in the FPGA taking : approx. 20 secs to load from the serial memory. The problem is not with : Atmel's serial memories, but with the competitors s/w. I can supply a work : arround to anyone who is interested. [..snip..] Could you be more specific about this "perceived" bug, otherwise we can dismiss your post as marketing hype.Article: 3566
Manfred, Xilinx has a Floorplanner that will help with you routing. If you floorplan your data flow it usually helps with the routing. Talk to your Xilinx Rep about getting this tool. As far as Bus Resources vs Muxes it depends on how many signals you are muxing. More than 4, I usually suggest using the Bus Resources. You can place the TBUFS using the Floorplanner and then the router doesn't have any problems routing these signals. If you have a reset signal, use the GSR - Global Reset Line. This is a dedicated resource so it doesn't use the normal routing resources. Use the STARTUP symbol to access the GSR. I think your best bet is to use the floorplanner! Kate MeilickeArticle: 3567
In article <1996Jun24.103014.18313@super.org>, * Atmel FPGA Apps * <martin@atmel.com> wrote: > We have found that due to a 'percieved' bug in our competitors s/w that > some customers have had difficulty when loading from a MCS file created > in their s/w. This problem manifests itself in the FPGA taking > approx. 20 secs to load from the serial memory. The problem is not with > Atmel's serial memories, but with the competitors s/w. I can supply a work > arround to anyone who is interested. > Well, well. Here is company X shipping millions of one-time programmable serial PROM to thousands of customers, with hardly a problem. ( I have learned not to say "never", but here it is really close ). Then comes company A and offers an interesting alternative, an electrically erasable version of such a serial PROM. Some designers like that. Some of these designers then have problems with company A's PROM and ventilate their frustration in comp.arch.fpga. The friendly answer coming from company A is: "..due to a perceived bug in company X's software..." C'mon, if you want to retrofit, you better be 100% compatible. If you are not, it's a cheap excuse to put the blame onto the originator. Peter Alfke, Xilinx ApplicationsArticle: 3568
Geez guys, lighten up. I've used both parts without problems, including cases where I've used the X brand ROM to configure the A brand FPGA and vice-versa. Both are fine parts. Just make sure you set the reset/oe polarity the right way for your application. That's really no different than many other parts out there is it? Those of you using a programmer made for one specific part can't reasonably expect it to program a different part type. Even among some of the popular PALs and ROMs made by several manufacturers there are a variety of combinations of programming algorithms/voltages required. There are reasonably priced programmers out there that will handle a wide variety of parts. An example is the Needham Electronics EMP-20 which I think sells for less than $500 and programs both of these parts as well as most micros, ROMs and PLDs. -Ray Andraka Chairman, the Andraka Consulting Group 401/884-7930 FAX 401/884-7950 mailto:randraka@ids.net http://www.ids.net/~randraka/ The Andraka Consulting Group is a digital hardware design firm specializing in high performance FPGA designs. Services include complete design, development, simulation, and integration of these devices and the surrounding circuits. We also evaluate,troubleshoot, and improve existing designs. Please call or write for a free brochure.Article: 3569
Has anybody created a tristate-based connection network using Xilinx XC3100A devices? Has anyone had any experience of using the tristates in these devices? What sort of delay times are we looking at? Thanks, Richard Meacham Contract ResearcherArticle: 3570
>Just make sure you set the reset/oe >polarity the right way for your application. You are probably right on the mark. The trouble is that some programmers which "support" a given device don't support user config for what that device's manufacturer regards as *user-configurable* features, and the only way is to edit the .hex or .jed file. I found that my Data I/O Chiplab is in this category. It has an even more annoying general shortcoming: it does not allow the user electronic signature (on PLDs) to be specified. Peter.Article: 3571
Dear net, I need to interface a Motorola MC68332 controller to a PCI VGA card. I am looking at various PCI bridge chips, but they seem to offer too much for my needs. I saw that Actel and Xilinx (and perhaps others?) support the PCI bus with some of their chips, and even have drop-in PCI designs (Xilinx LogiCore). Are Actel and Xilinx the most likely candidates for this? Any others? How good is their PCI support, and how good are these designs like LogiCore? Thanks, Ben -- Analogic Engineering BV Phone: +31 (0)23 5621623 Parellaan 44 Fax: +31 (0)23 5629349 2132WS Hoofddorp Email: ben@stuyts.nl The Netherlands (NeXT/MIME Mail OK)Article: 3572
Can anybody tell me whether Foundation VHDL is supported on WINDOWS95? I would be glad if anybody can tell me a good web site about Foundation VHDL Tools.. Thanks -Apurva Brahmbhatt Uni. of Massachusetts, AmherstArticle: 3573
In article <31D104E6.41C67EA6@sheffield.ac.uk>, Richard Meacham <R.Meacham@sheffield.ac.uk> wrote: > Has anybody created a tristate-based connection network using Xilinx > XC3100A devices? Has anyone had any experience of using the tristates > in these devices? What sort of delay times are we looking at? The individual device output turn-on and turn-off times are all in the data sheet( page 2-183 in the data book ).They assume 50 pF load. You can get a feel for the active output drive capability from page 9-23 of the current ( and older ) Xilinx data book. Slew-rate limited outputs have the same dc characteristics as non-slew rate limited outputs. It's only the transition that is longer. To really answer the question, one needs to know something about topology and speed requirements of your network. Peter Alfke, Xilinx ApplicationsArticle: 3574
Ben Stuyts (benst@stuyts.nl) wrote: : Dear net, : I need to interface a Motorola MC68332 controller to a PCI VGA card. I am <stuff deleted> : Are Actel and Xilinx the most likely candidates for this? Any others? The Xilinx family best suited to PCI designs (the XC4000E series) are roughly twice the price of something like the Spanner/Q-Span chips from Newbridge Components (they may now be called Tundra??) which are designed specifically for the Motorola MC68k family. The main advantage of the FPGA PCI bus interface designs is the ability to add application specific logic to the FPGA and hence reduce chip count. Hope this helps, Erik.
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