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
In article <EJB0yA.CDD@iglou.com>, Hul Tytus <htytus@iglou1.iglou.com> writes > > Anyone know of a SCSI-1 host adapter for the PC ISA buss for which a >listing of the control & data ports and their functions is available? This >would seem a simple quest but the opposite appears true. I too had difficulty finding information on SCSI adaptors. I had a look at the Linux source listings which do contain the kind of information you need but are not really structured in a way that makes it easy to get at. Maybe you could see if the Adaptec 1505 or 1510 have drivers within Linux. In the end I used the ASPI interface as my unit ran under DOS and that has helped with multi sourcing 'cos every adaptor you can buy has an ASPI driver. Whether its an appropriate copurse for you is obviously another matter. Good luck anyway and I'd be interested in anything you did find. Regards -- Steve Goodwin... De-spamming active, remove any _DSPM from addressArticle: 8026
Who has some electronic materials about the lastest development of the Digital's Alpha AXP? Please give replies to my own email box. Thanks!Article: 8027
Ray Andraka wrote: > > Reetinder P. S. Sidhu wrote: > > > > Hi all > > > > Could anyone provide info on design techniques used to > > implement arithmetic division on FPGAs (both LUT and sea-of-gates)? > > > > Thanks in advance. > > Division is not trivial as you've probably realized by now. ..... Would you like to try a new, simple, high-radix approach? There's no charge for non-profit use. I came up with new, comparator-based high-radix division algorithms, which can be read at http://math.isu.edu/~vkantabu/radix8.pdf If anyone implements any of these algorithms, I'd love to know the delay & area data.Article: 8028
Ok, as great as the internet is with information, I couldn't find any online documentation, (or PDFs) that would answer me with basic questions to start me off on learning about this great technology. Are FPGAs re-porgrammable ? If so is there a limit to the number programming ? Where does the word "field" come from in FPGA ? I've heard of people using programmable logic devices for building hardware that has the capability to be reconfigured at any time. Would this be FPGA ? Does this implementation have a name ? Obviously I am a beginner to programmable logic. Over the last few weeks of looking at whets around I see its a huge and growing industry. For some reason FPGAs seem to be most popular over other technologies. AdamArticle: 8029
I`m looking for a converter which may transform *.pds or *.jed -files (PLDasm/ ALTERA`s PLDShell Plus) to *.tdf - files (MAX+PLUS II) vic@alpha.podol.khmelnitskiy.ua -- Victor Levandovsky Technological University of Podolia Khmelnitsky, UkraineArticle: 8030
Little offer here for people who do not have access to the Altera program but would like to use one for a project. I have a proto type board that has an Altera FLEX 10K10 on it. This is a Programmable logic device with 134 input/output pins and 10,000 logic gates. I have designed the PCB to use the surface mount version of this chip and the board is 2.5" by 4". The board brings all the I/O pins to wire wrap connectors on a 0.1" grid. The board has a socket to plug the 10K10's programming PROM. (this configures the device). What I am offering is the chip/board fully wired. You supply the schematic I will program the PROM. The way the programming software works is that you must have a key.(The software is very expensive) This chip is very powerful, just about any logic you could think of you can put into this chip. Just about any type of 74 series chip for example. I will provide the boards/ chips at my cost. I will program the chips to your specification for ~$30. (Note this is not design just entering the schematic and a basic test. If you want more we can talk about it.) I feel that I can provide the board and chip for about $55. If a change in your schematic is required I will do the change and send you a new EPROM for ~$20. This is a for sale type post but I think that anyone can see it's not a money making venture to say the least. I just want the experience and some of you could benefit as well. If you don't see it that well please forgive the intrusion. Anyway if this is of any use to you please email me at kkibbe@golden.comArticle: 8031
C A L L F O R P A P E R S THE SIXTH ANNUAL IEEE SYMPOSIUM ON FIELD PROGRAMMABLE CUSTOM COMPUTING MACHINES Napa, California April 15-17, 1998 http://www.fccm.org PURPOSE: To bring together researchers to present recent work in the use of reconfigurable logic as computing elements. This symposium will focus primarily on the current opportunities and problems in this new and evolving technology for computing. Contributions are solicited on all aspects of custom computing, including but not limited to: Architecture of reconfigurable computing devices and systems, including coprocessors, attached processors, and hybrids. Languages, compilation techniques, tools, and environments for programming and run time support; Application domains; Prototyping for architecture emulation. SUBMISSIONS: Authors are invited to send submissions for either full length papers (10 page maximum) or extended abstracts (2 page maximum) for posters by January 5, 1998, to Jeffrey Arnold. Notification of acceptance will be sent in early March. Final papers will be due on the first day of the Symposium. The proceedings will be published following the Symposium. Authors are encouraged to submit PostScript, Microsoft Word, or FrameMaker manuscripts by FTP. For instruction on electronic submission, please see the Web page or contact Jeffrey Arnold (jmarnold@znet.com). SPONSORSHIP: The IEEE Computer Society and the Technical Committee on Computer Architecture. CO-CHAIRS: Kenneth L. Pocek Intel Mail Stop RN6-18 2200 Mission College Boulevard Santa Clara, California 95052 Voice: 408-765-6705 Fax: 408-765-5165 kenneth_pocek@ccm11.sc.intel.com Jeffrey M. Arnold 10686 Mira Lago Terrace San Diego, CA 92131 Voice: 619-547-9257 Fax: 619-547-9010 jmarnold@znet.com PROGRAM COMMITTEE: Peter Athanas, Virginia Tech. Donald Bouldin, University of Tennessee, Knoxville Duncan Buell, Center for Computing Sciences Michael Butts, Quickturn Design Systems, Inc. Steve Casselman, Virtual Computer Corp. Pak Chan, Univ. California, Santa Cruz Apostolos Dollas, Technical Univ. of Crete Scott Hauck, Northwestern Univ. Brad Hutchings, Brigham Young Univ. Tom Kean, Xilinx, Inc. (U.K). Phil Kuekes, HP Labs. Wayne Luk, Imperial College John McHenry, NSA Robert Parker, Institute for Information Sciences Herman Schmit, Carnegie Mellon University Mark Shand, Digital Equipment (Paris) Satnam Singh, Univ. of Glasgow Stephen Smith, Altera Corp. -- Jeffrey M. Arnold jma@super.org or jmarnold@znet.com 10686 Mira Lago Terrace Tel: 619-547-9257 San Diego, CA 92131 Fax: 619-547-9010 USAArticle: 8032
>Protel has not yet discovered what hierarchy means. Multiple instantiations >of a single library part means ridiculous workarounds. Until that's fixed, >I won't touch it for FPGA work. Can you give more detail? What happens if one does it? Of course, this should work. Peter. Return address is invalid to help stop junk mail. E-mail replies to z80@digiXYZserve.com but remove the XYZ.Article: 8033
It's time to get YOUR free pager. It's as simple as calling the number below!! NO YEAR CONTRACTS, all service is month-to-month. YOUR CHOICE OF COLORS!! These are NEW MOTOROLA PAGERS. NO GIMMICKS, JUST CALL THE NUMBER BELOW AND GIVE THEM THE FOLLOWING ACCESS #. CALL: (800) 784-6452 ACCESS#: 4095038 Ask how YOU can get paid to give away free pagers!!!!Article: 8034
Adam Seychell wrote: > > Ok, as great as the internet is with information, I couldn't find any > online documentation, (or PDFs) that would answer me with basic > questions to start me off on learning about this great technology. > > Are FPGAs re-porgrammable ? > > If so is there a limit to the number programming ? > > Where does the word "field" come from in FPGA ? > The answers are Yes, No, As in "in the field". Field Programmable Gate Arrays can be reprogrammed, in the field (ie where your end user is), a large number of times (there may be a limit but I don't know it). Some models can even be reprogrammed while running. alan ----------------------------------------------------------------------- Alan Donovan, FORE Audio & Video, 14 Regent Street, Cambridge CB2 1DB Tel: +44 1223 518325 adonovan@eu.fore.com http://www.fore.com -----------------------------------------------------------------------Article: 8035
On Mon, 10 Nov 1997 10:03:11 +0000, Alan Donovan <adonovan@eu.fore.com> wrote: >Adam Seychell wrote: >> >> Ok, as great as the internet is with information, I couldn't find any >> online documentation, (or PDFs) that would answer me with basic >> questions to start me off on learning about this great technology. >> >> Are FPGAs re-porgrammable ? >> >> If so is there a limit to the number programming ? >> >> Where does the word "field" come from in FPGA ? >> > >The answers are Yes, No, As in "in the field". > >Field Programmable Gate Arrays can be reprogrammed, in the field (ie >where your end user is), a large number of times (there may be a limit >but I don't know it). Some models can even be reprogrammed while >running. > >alan > >----------------------------------------------------------------------- > Alan Donovan, FORE Audio & Video, 14 Regent Street, Cambridge CB2 1DB > Tel: +44 1223 518325 adonovan@eu.fore.com http://www.fore.com >----------------------------------------------------------------------- Not all FPGAs are re-programmable, e.g. ACTEL; which use antifuse technology rather than SRAM. This means that they are not reprogrammable "in the field", although they are still called FPGAs. This can have its advantages where reprogrammability is not a requirement (i.e. cheaper/more reliable/instant operation from boot-up). The number of re-programming cycles depends on the device. It can be anything from a couple of hundred to tens of thousands of times. The best way to get basic information is to get some data books from the manufacturers, and have a look at them. If your familiar with PLDs, then FPGAs shouldn't be much of a problem. DerekArticle: 8036
Austin Franklin wrote: > > Why not just declare ifoobar (for internal foobar) and foobar... > so you only need to set foobar to 1 in state 3 > > Marc Heuler <marc@aargh.mayn.de> wrote in article > <TbclB*xte@aargh.mayn.de>... > > I'm defining a state machine in ABEL HDL. > > > > Some nodes are LOW only during 1 state of 16.... > > This is very tiring and invites to forget it once... Another suggestion would be to leave the assertion of foobar out of the state machine section entirely, and put it in the equations section. Thus: equations foobar = (StateMachineRegs != 3); This causes foobar to be 1 whenever the state machine is in any state other than 3. With no driving logic, the output will go to zero. -TBBArticle: 8037
On Mon, 10 Nov 1997 10:03:11 +0000, Alan Donovan <adonovan@eu.fore.com> wrote: >Adam Seychell wrote: >> >> Ok, as great as the internet is with information, I couldn't find any >> online documentation, (or PDFs) that would answer me with basic >> questions to start me off on learning about this great technology. >> >> Are FPGAs re-porgrammable ? >> >> If so is there a limit to the number programming ? >> >> Where does the word "field" come from in FPGA ? >> > >The answers are Yes, No, As in "in the field". > The answers are "It depends." Some FPGAs are re-programmable, some are not -- i.e., Actel's. Some are RAM-based, so have "inifinite" re-programmability. Some are based on Flash or EEPROM-technology based, so can be programmed "lots and lots"--probably more than anyone would care. Jason >Field Programmable Gate Arrays can be reprogrammed, in the field (ie >where your end user is), a large number of times (there may be a limit >but I don't know it). Some models can even be reprogrammed while >running. > >alan > >----------------------------------------------------------------------- > Alan Donovan, FORE Audio & Video, 14 Regent Street, Cambridge CB2 1DB > Tel: +44 1223 518325 adonovan@eu.fore.com http://www.fore.com >-----------------------------------------------------------------------Article: 8038
Adam Seychell wrote in message <3465CE15.E8ADFCA8@cybec.com.au>... >Ok, as great as the internet is with information, I couldn't find any >online documentation, (or PDFs) that would answer me with basic >questions to start me off on learning about this great technology. > > Are FPGAs re-porgrammable ? Some FPGAs are re-programmable, namely those built using SRAM technology and those using FLASH memory technology. There is another technology called anti-fuse that is used in some FPGA devices. It can only be programmed once but has some advantages for design security. See http://www.optimagic.com/faq.html#Memory_Technology for additional information. > > If so is there a limit to the number programming ? For SRAM-based parts, there is no limit to the number of programming cycles as there is no know wear-out mechanism for SRAM technology. FLASH-based devices do have a maximum rating which varies from one vendor to another. > > Where does the word "field" come from in FPGA ? Good question. I think it hails back to the days of 'field programmable logic arrays'. My guess is that 'field' means that it can be programmed in the field and does not have to be programmed in the factory. > >I've heard of people using programmable logic devices for building >hardware that has the capability to be reconfigured at any time. Would >this be FPGA ? Most probably they are referring to SRAM-based FPGAs. They may also be referring to FLASH- or EEPROM-based FPGAs or CPLD devices. >Does this implementation have a name ? Some more generic names include 'reconfigurable computing' (RC), configurable computing, reconfigurable hardware. If you are interested in this technology, take a look at the following pages: http://www.netcom.com/~optmagic/reconfigure http://www.optimagic.com/research.html http://www.optimagic.com/boards.html > >Obviously I am a beginner to programmable logic. Over the last few >weeks of looking at whets around I see its a huge and growing >industry. For some reason FPGAs seem to be most popular over other >technologies. > >Adam > > You may be interested in our site as it contains a fairly comprehensive set of links related to programmable logic. You can find The Programmable Logic Jump Station at http://www.optimagic.com. -- Steve KnappArticle: 8039
Alan Donovan wrote: > > Adam Seychell wrote: > > > > Ok, as great as the internet is with information, I couldn't find any > > online documentation, (or PDFs) that would answer me with basic > > questions to start me off on learning about this great technology. > > > > Are FPGAs re-porgrammable ? > > > > If so is there a limit to the number programming ? > > > > Where does the word "field" come from in FPGA ? > > > > The answers are Yes, No, As in "in the field". > > Field Programmable Gate Arrays can be reprogrammed, in the field (ie > where your end user is), a large number of times (there may be a limit > but I don't know it). Some models can even be reprogrammed while > running. > > alan > More accurately depends, depends, like he said. SRAM based FPGAs (Xilinx 3K,4K,5K,6K;Atmel 6K,40K;Altera 10K etc) are infinitely reprogrammable and must be reloaded after power is cycled. In these cases the FPGA configuration is stored in volatile registers int he device. Some of these devices also accept partial reprogramming so that a portion of the circuit remains while the rest is replaced. Other devices use different technologies for storing the device configuration. Actel and Quicklogic use antifuses which can be programmed only once. While these are a throwaway device if the programming needs to be changed, there is also no need for programming circuitry on your board. Still others use EEPROM technology which allow the device to be reprogrammed. In this case the program is non-volatile, but there is typically a limit on the number of program / erase cycles that can be done. -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: 8040
> Reetinder P. S. Sidhu wrote: > > > > Hi all > > > > Could anyone provide info on design techniques used to > > implement arithmetic division on FPGAs (both LUT and sea-of-gates)? > > > > Thanks in advance. > > > > Reetinder Sidhu Simple long division, as taught in grade school, can be implemented in Xilinx 4000 series FPGAs in a simple manner. The high speed carry logic of the part is configured to perform the subtractions at each stage, and the carry out is used to control multiplexors which provide either the difference (if positive), or the subtrahend(?) (if difference is negative). The muxes are configured into the LUTs associated w/ the carry logic. The operations can be pipelined for maximum speed. This may not be the most area-efficient way to go, but is probably the easiest to design and get running. Don Husby has posted on this previously, you may want to search the archives.Article: 8041
Dear all, a while ago an FPGA designer told me that Xilinx does not guarantee any minimum skew between clock signals from different clock sources. This means that if I have two clock signals ck_1, ck_2 with a duty cycle of 50%, frequency f_1, f_2 = 2*f_1 and little skew (say 1-2 ns) at the FPGA input pins, there is no gurantee that the skew remains the same due to inpredictable delays in the internal clock distribution network. For this reason I can't exchange data between different clock domains by simply connecting the output of a register clocked by ck_1 to, say, the input of a register clocked by ck_2. I did not find any guidelines for dealing with this problem, so I would like to know how these isses are usually handled by the experts. thanks in advance -Arrigo BenedettiArticle: 8042
You might be interested in the following Adobe Acrobat document available on the Xilinx web site. I believe that it directly addresses the problem that you posed. Trouble-Free Switching Between Clocks http://www.xilinx.com/xcell/xl24/xl24_20.pdf ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Arrigo Benedetti wrote in message ... >Dear all, > >a while ago an FPGA designer told me that Xilinx does not guarantee >any minimum skew between clock signals from different clock sources. >This means that if I have two clock signals ck_1, ck_2 with a duty >cycle of 50%, frequency f_1, f_2 = 2*f_1 and little skew (say 1-2 ns) >at the FPGA input pins, there is no gurantee that the skew remains the >same due to inpredictable delays in the internal clock distribution >network. For this reason I can't exchange data between different clock >domains by simply connecting the output of a register clocked by ck_1 >to, say, the input of a register clocked by ck_2. I did not find any >guidelines for dealing with this problem, so I would like to know how >these isses are usually handled by the experts. > >thanks in advance > >-Arrigo BenedettiArticle: 8043
Jack Following is from Xilinx's online documentation in the M1 Design tools. It is an example and definition of a UCF(User Constraint File). Gus Anderson Field Applications Engineer Nu Horizons 305 710 6894 CPLD Attributes Used in a User Constraint File (UCF) Copyright © 1984÷1997 ALDEC, Inc. · INST inst_name BUFG=CLK|OE|SR; on an IBUF symbol. or NET net_name BUFG=CLK|OE|SR; on an input pad net. (Assigns global buffers to input pins in top-level design.) · INST inst_name COLLAPSE; on a logic symbol. or NET net_name COLLAPSE; on a logic net. (Forces a logic mode to be collapsed into all of its fanouts.) · INST inst_name FAST | SLOW; on an OBUF, OBUFE or OBUFT type symbol. or NET net_name FAST | SLOW; on an output pad net. (Selects output slew rate for output pins in top-level design.) · INST inst_name INIT=R|S; on a flip-flop or other registered macro symbol. or NET net_name INIT=R|S; on the output net of a flip-flop or other registered macro symbol. (Defines initial state of registers.) · INST inst_name KEEP; on a logic symbol. or NET net_name KEEP; on a logic net. (Preserves internal nodes during design implementation.) · NET net_name LOC=Ppin_number|BGA_pin_name; [eg. NET ABC LOC=P23; or NET XYZ LOC=AC17;] on any pad net. (Assigns a device pin number to an input or output pad of the design.) · INST inst_name LOC=FBnn; [eg. INST U3 LOC=FB3;] on a logic symbol. or NET net_name LOC=FBnn; on a logic net or an output pad net. (Assigns an internal Function Block number to logic driving an internal node or device output; typically used only to control XC9500 local feedback routing.) · INST inst_name NOREDUCE; on a logic symbol. or NET net_name NOREDUCE; on a logic net. (Preserves redundant logic terms during design implementation.) · NET net_name OFFSET=IN:delay:BEFORE:clock_net_name; on an input pad net. (Defines an input pad setup time constraint before the active edge of the named clock input pad.) · NET net_name OFFSET=OUT:delay:AFTER:clock_net_name; on an output pad net. (Defines a clock-pad-to-output-pad propagation delay timing constraint.) · INST inst_name PWR_MODE=LOW|STD; on a logic symbol. or NET net_name PWR_MODE=LOW|STD; on a logic net. (Selects macrocell power mode for nodes or outputs in CPLD designs.) · INST inst_name TNM=group_name; [eg. INST FLOP1 TNM=MY_GROUP1;] on any flip-flop symbol. or NET net_name TNM=group_name; on any logic net or pad net. (Tags signals with a Timing-group Name to be used in a Timespec.) · TIMESPEC TSnn=FROM:source_group:TO:dest_group:delay; [eg. TIMESPEC TS02=FROM:PADS(IBUS*):TO:MY_GROUP1:16;]. (Defines a point-to-point timing constraint.) · TIMESPEC TSnn=PERIOD:clock_group:delay; [eg. TIMESPEC TS10=PERIOD:CLK1_GRP:12;]. (Defines a clock period timing constraint.) Note: The Foundation Project Manager automatically creates an empty template UCF file with the same name as your project and the fitter automatically reads it. You can add constraints to the UCF file by double-clicking its file name in the Project Manager’s window. Copyright © 1984÷1997 ALDEC, Inc.Article: 8044
Hi - "Steven K. Knapp" <sknapp@optimagic.com> wrote: >You might be interested in the following Adobe Acrobat document available on >the Xilinx web site. I believe that it directly addresses the problem that >you posed. >Trouble-Free Switching Between Clocks >http://www.xilinx.com/xcell/xl24/xl24_20.pdf I think that the circuit you've cited addresses a different question, namely, how to safely multiplex between two clocks. The original poster was asking how to safely transfer data between logic domains, one of whose clocks is an integral multiple of the other, when the skew between the two clock trees inside the FPGA isn't well specified. Good question. I can suggest several techniques: - use only the N*f1 clock. In the domain where you want to clock at f1, use an enable that's active every N clocks. With only one clock tree, skew shouldn't be a problem. Of course, power dissipation may increase. This is my favorite approach. - Ensure through your design that there's some minimum amount of logic between the registers connecting the two domains. Unfortunately, FPGA vendors do not specify minimum delays, and those minimum delays get smaller all the time, as parts are migrated to faster and faster processes. But, you may be able to rationalize that the clock trees, and thus the skews, are getting smaller and smaller, too. You'd have to be a sounder sleeper than I am, though. - Avoid using clock edges that are close to one another. If, for example, you are trying to transfer data from a 10MHz domain to a 20MHz domain, and the 10 and 20MHz rising edges are close to one another, consider clocking the data into the 20MHz domain on the falling edge of 20MHz, then reclocking it into another set of registers on the 20MHz rising edge. This works, but with some drawbacks. First, it chews up registers. Second, you have less than half a 20MHz clock in which to transfer data between domains, because you're penalized by the potential skew between the two clocks, and by the potential logic HIGH pulse shrinkage of the 20MHz clock (which usually isn't well specified, either). But it works, if you're careful. - Ignore the whole issue and hope for the best. In my experience this is the most unsatisfactory solution but the also the most popular. [Aside: I had a chance to look at the article referred to, titled "Trouble-Free Switching Between Clocks," and it looks pretty scary to me. Given a pair of clocks with an arbitrary phase relationship and a clock select that's not synchronized to at least one of them, the circuit appears to be Metastability Central. Can someone explain how this is supposed to work? Or is "Trouble-Free" a typo for "Free Trouble?" Or maybe I'm missing the point, which has happened before.] Bob Perlman ------ "Never wrestle with a pig: you both get dirty and the pig likes it." To send me e-mail: my login ID is bobperl, my ISP is best.com.Article: 8045
Check at http://www.associatedpro.com/aps for some great FPGA informatioon and deals on low cost eda software. Adam Seychell wrote: > Ok, as great as the internet is with information, I couldn't find any > online documentation, (or PDFs) that would answer me with basic > questions to start me off on learning about this great technology. > > Are FPGAs re-porgrammable ? > > If so is there a limit to the number programming ? > > Where does the word "field" come from in FPGA ? > > I've heard of people using programmable logic devices for building > hardware that has the capability to be reconfigured at any time. Would > this be FPGA ? > Does this implementation have a name ? > > Obviously I am a beginner to programmable logic. Over the last few > weeks of looking at whets around I see its a huge and growing > industry. For some reason FPGAs seem to be most popular over other > technologies. > > AdamArticle: 8046
Philip Freidin comments on Bob Perlman's post about switching clocks: In article <648phq$g6o$1@nntp2.ba.best.com> nospam@notime.nohow (Bob Perlman) writes: >Hi - > >"Steven K. Knapp" <sknapp@optimagic.com> wrote: > >>You might be interested in the following Adobe Acrobat document available on >>the Xilinx web site. I believe that it directly addresses the problem that >>you posed. > >>Trouble-Free Switching Between Clocks >>http://www.xilinx.com/xcell/xl24/xl24_20.pdf > >I think that the circuit you've cited addresses a different question, >namely, how to safely multiplex between two clocks. The original >poster was asking how to safely transfer data between logic domains, >one of whose clocks is an integral multiple of the other, when the >skew between the two clock trees inside the FPGA isn't well specified. >Good question. > >I can suggest several techniques: > Bob suggests several techniques, and comments on design practice in general. >[Aside: I had a chance to look at the article referred to, titled >"Trouble-Free Switching Between Clocks," and it looks pretty scary to >me. Given a pair of clocks with an arbitrary phase relationship and a >clock select that's not synchronized to at least one of them, the >circuit appears to be Metastability Central. Can someone explain how >this is supposed to work? Or is "Trouble-Free" a typo for "Free >Trouble?" Or maybe I'm missing the point, which has happened before.] >Bob Perlman Bob has not missed the point. The Xilinx article is indeed JUNK, and should not be used for switching clocks, as it is indeed an example of a design that can go metastable, and the results will be yeuchy. For those of you who care, the article is titled: "Trouble-Free Switching Between Clocks" and was in the XCELL-24 magazine, Q1 1997 If you have the Xilinx AppLINX CD, there is a copy on it too. Here is the scenario: Assume that the select line is high, and clock B is therefore enabled. Lets say that clock B is at 10MHz, and clock A is at 51.44MHz, with no phase relationship. At the time of clock B falling, the select line goes low, violating the setup/hold requirement of the lower flip flop. The flip flop goes metastable and its output transitions low, and then returns high, after maybe 10nS. Assuming that select remains low, it will transition more permanently to a low level on the following falling edge of clock B, 90nS later. While it was low for 10nS though, the gate feeding the upper flip flop was supplying a high, and a falling edge of clock A came along. The upper flip flop would now be set also, and the following high on clock A will be passed to the ouput, and will merge in an uncertain way with clock B. Real Junk. Here's another failure mode: Clock B at 66MHz (15.151nS cycle time). Setup changes async to clock B, and the flip flop goes metastable. It goes low for 10nS, then returns high. When it returns high, we are 2.424 nS into the clock high time of B, and the and gate for B into the or gate is re-enabled. The result is a runt clock high pulse of 5.151 nS. Obviously, for all possible metastability recovery times, there is a range of runt widths from 0nS to just less than the correct width. Here's another failure mode: Select changes, disabling B, but the arrival of the B flip flop signal to enable the A flip flop is such that the async relationship with the A clock causes the A flip flop to go metastable, with the above runt clock problem as described above, but on the A side. Here's three more failure modes: Swap A and B in all of the above scenarios. Bob did not miss the point, there was none. The design is JUNK. Philip Freidin. (I really hate when applications information from a vendor is just patently wrong)Article: 8047
Victor Levandovsky wrote: > I`m looking for a converter > which may transform *.pds or *.jed -files (PLDasm/ ALTERA`s PLDShell Plus) > to *.tdf - files (MAX+PLUS II) > > vic@alpha.podol.khmelnitskiy.ua > -- > Victor Levandovsky > Technological University of Podolia > Khmelnitsky, Ukraine Altera provides (free of charge) a 'pla-to-tdf' program that will convert the pla output of Abel to Altera's tdf format. I don't have a copy handy, but I'm sure if you contact Altera, they'll point you to one. Might be on their website as well. Good luck. -johnArticle: 8048
In article <3465CE15.E8ADFCA8@cybec.com.au>, Adam Seychell <aseychell@cybec.com.au> wrote: > > Ok, as great as the internet is with information, I couldn't find any > online documentation, (or PDFs) that would answer me with basic > questions to start me off on learning about this great technology. > > Are FPGAs re-porgrammable ? > > If so is there a limit to the number programming ? > > Where does the word "field" come from in FPGA ? > > I've heard of people using programmable logic devices for building > hardware that has the capability to be reconfigured at any time. Would > this be FPGA ? > Does this implementation have a name ? > > Obviously I am a beginner to programmable logic. Over the last few > weeks of looking at whets around I see its a huge and growing > industry. For some reason FPGAs seem to be most popular over other > technologies. > > Adam Adam Try the Xilinx web site (www.xilinx.com) they have a lot of info on FPGAs. Deb -------------------==== Posted via Deja News ====----------------------- http://www.dejanews.com/ Search, Read, Post to UsenetArticle: 8049
In article <01bceb68$efb9f940$fd4f27ca@jack.hinet.net>, "Jack Huang" <jacktch@tpts1.seed.net.tw> wrote: >Hello, > >I'm now using Fundation design entry and M1implementation tools on XILINX >XC4000 series. Where can I find any document talking about the keywords or >parameters >I can use in the UCF file? > >Jack > I went through the same thing, so I'm not being sarcastic: Put the CD that says 'docs' in, install the browser software, and look through the table of contents. It's pretty easy to spot there. Well, I have the Alliance stuff. I imagine the docs are in much the same format though.
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