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
IJ.McCrum@ulst.ac.uk (I McCrum) wrote: >Xilinx XC3000A are pretty low power. For 2MHz clock rates and 3.3V operation, the 3000L series is a better choice and runs at 40% the power of the 3000A. The 3000L's powerdown mode is specified at 10uA but is actually about 1uA. If you could afford to loose the state of your counters during your 75% off time, you could save a lot of power using power down mode of say an XC3064L . Cheers, Dave Decker Dave Decker Diablo Research Co. LLC Please use only one 'h' in mush. I'm trying to reduce the spam. "Animals . . . are not brethren they are not underlings; they are other nations, caught with ourselves in the net of life and time, fellow prisoners of the splendor and travail of the earth." Henry Beston - The Outermost HouseArticle: 9626
In article <01bd5913$d85769e0$3a9602c3@default>, Ian Stevenson <ianstevenson@no-junk.saqnet.co.uk> writes >Is there a FAQ for this group? I would be grateful if anyone can point me >in the right direction. >Cheers, >Ian > I'm afraid there's no official FAQ. No doubt one or two people who've produced informative websites will post their addresses. However, in nearly five years in this group, I've never seen anyone criticised for asking questions, so feel free to ask away. DavidArticle: 9627
On Thu, 26 Mar 1998 16:15:25 -0800, Greg Prior <eng@digital-reflection.com> wrote: >Hi all- > >Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am >looking for a better editor. I would like to try out a share one with >the capability of multiple levels of 'undo'. Also it should be VHDL >aware, and highlite keywords, etc. Can anyone point me to something? > >Thanks, Greg > gnu emacs. it's free, does everything, is infinitely extensible, runs on (almost) anything, and is infinitely better than any other editor i've ever used. you can get current info from: http://www.cs.washington.edu/homes/voelker/ntemacs.html vhdl extensions from: http://www.geocities.com/SiliconValley/Peaks/8287/ by the way, the editor that came with M1.3 had a nasty habit of deregistering any other editor that was registered for ".vhd" files (in win95). it did this automatically every time it was run. does anyone know how to fix this? evan (ems@nospam.riverside-machines.com)Article: 9628
Hi every body... Looking for interface program (written in VHDL, or C) between ISA card and Pc. (to down load bitstream and control the operation via menu, send data, recieve result say for 4x4 serial multiplier) Thanks in advance....Article: 9629
In article <3519CE06.178C9C91@writeme.com>, Rick Collins <redsp@writeme.com> writes <suggestions of how to break Xilinx and Viewlogic security devices> How can you justify promoting the removal of security devices? Surely the resulting illegal copying of software (the use to which many would inevitably put this information) is against the interest of both the paying customers and those who make a living from providing and supporting the software? For me, and I would guess many other engineers in the EDA business who come here, the very existence of this thread is a sad occasion for the group. David PashleyArticle: 9630
David Pashley wrote: > For me, and I would guess many other engineers in the EDA business who > come here, the very existence of this thread is a sad occasion for the > group. There speaks a man who's not had a Xilinx dongle choose to stop working on a tight-timescale project. I pay for my software. I'd like to be able to use it. I'd like to be able to come back to an old project and not have to fight old versions of dongle drivers. I'd like to use all the software I have paid for at the same time, without frantic dongle swapping. Things are getting a lot better, but dongles are still a pain. Perhaps the fact that this thread exists is as much a commentary on the penalty that dongles inflict on legitimate users as it is an indictment of the morals of all EDA professionals. Problem is, I don't know of a better solution, but that doesn't stop me wanting one. The fragility of dongles is currently unacceptable. Locking software to a particular machine is worse, especially if it's keyed to a disk ID, and positively encourages workarounds. Net licensing is yet worse. I simply can't imagine a secure, unhackable system that accurately enforces the license conditions. A question:- what is the position of software vendors towards replacing dead dongles for users who are no longer under maintenance? This doesn't seem to be made clear on shrink-wrap licenses. Strange, that... All just my opinion, as someone who does a lot of work on-site using a portable, running up to 7 dongles, all legitimate, like, I suspect, the vast majority of tool users. For the record, neither I nor my company condone defrauding software companies or their support operations, regardless of my grumbles about dongles. SteveArticle: 9631
Bryan Williams (bk&dontyabespamminme&willi@smart.net) wrote: : Funny, I was digging through the Xilinx AllianceCORE products on their : website, I was dumbfounded seeing a T1 framer there. Trick is, it's a : _single_ framer, has NO elastic stores (according to the block diagram : in the datasheet at least), and it takes 1236 or so CLBs, meaning it : won't fit in anything smaller than a XC4036EX and apparently uses all : the CLB's in a 4036. I checked the price of the device it is tailored : to, it's about $500 in single units, add to that the $6000 cost of the : core design. The problem with T1 framing is that the frame bits are well distributed through the multiframe (not 'bunched' in a single place as they are in European equivalents). It is possible to sync to these bits in very few gates using a serial search, but this can take 2 or 3 seconds for one of the T1 formats. Alternatively I guess you need a large store to hold a whole multiframe, allowing the frame to be found much quicker. This is perhaps what is inside the Xilinx. If anybody has any neat ways of quickly synchronising to T1 ESF framing in a small FPGA the I'm all ears (...eyes?). Nick.Article: 9632
Good for Xilinx! They didn't even want to make me (little me) a measly one-month loaner to at least evaluate their software. I absolutely had to fork out a few $k if I wanted to see how my designs would fit in their part. I now have the Foundation kit limited to a useless 8k gates and even to use that, I have to go through bureaucratic processes to get authorisation codes (for 8k gates!). When will these FPGA vendors finaly step out (like Actel) and provide free software so users can easily compare between different FPGA vendors (or maybe they are afraid of this?). And guess what? Now that I am finaly ready to ship some protos, which vendor will I use? Actel, and with their new MX series devices I can make money in small production quantities and ship them as stand-alone ASICs.! Sam Falaki David Pashley wrote: > In article <3519CE06.178C9C91@writeme.com>, Rick Collins > <redsp@writeme.com> writes > > <suggestions of how to break Xilinx and Viewlogic security devices> > > How can you justify promoting the removal of security devices? Surely > the resulting illegal copying of software (the use to which many would > inevitably put this information) is against the interest of both the > paying customers and those who make a living from providing and > supporting the software? > > For me, and I would guess many other engineers in the EDA business who > come here, the very existence of this thread is a sad occasion for the > group. > > David PashleyArticle: 9633
> <suggestions of how to break Xilinx and Viewlogic security devices> > > How can you justify promoting the removal of security devices? You can't be serious? This has to be either a troll or from someone very inexperienced in hardware keys. > Surely > the resulting illegal copying of software (the use to which many would > inevitably put this information) is against the interest of both the > paying customers and those who make a living from providing and > supporting the software? I have many reasons why I don't want the key. One is I use a notebook, and carrying around the key (and 6 others) is ridiculous and risky. If I loose the key (which they are easy to loose) or someone steals it while I'm out of the office (much easier to steal than the computer it self, and it can't be locked, the computer can) then I'm SOL. No one is promoting making illegal copies, or illegally using the software. That is something you read into this. Anyway, it has been proven beyond the shadow of a doubt that companies are NOT 'deprived' of their income by pirated software, as %99 of pirated software is used by someone who would not have bought it in the first place, also a good number of them, if they are using the software regularly, eventually do buy it. It's the best marketing tool the software industry has to offer. > For me, and I would guess many other engineers in the EDA business who > come here, the very existence of this thread is a sad occasion for the > group. What's sadder to me is you assume people are going to use the key-removals to promote illegal use of the software, when that is NOT the case at all. Austin Franklin darkroom@ix.netcom.comArticle: 9634
In article <01bd598e$c1f66360$6491d9ce@drt1>, Austin Franklin <dar4kroom@ix.netcom.com> writes >> <suggestions of how to break Xilinx and Viewlogic security devices> >> >> How can you justify promoting the removal of security devices? > >You can't be serious? This has to be either a troll or from someone very >inexperienced in hardware keys. ROFL > >> Surely >> the resulting illegal copying of software (the use to which many would >> inevitably put this information) is against the interest of both the >> paying customers and those who make a living from providing and >> supporting the software? > >I have many reasons why I don't want the key. One is I use a notebook, and >carrying around the key (and 6 others) is ridiculous and risky. If I loose >the key (which they are easy to loose) or someone steals it while I'm out >of the office (much easier to steal than the computer it self, and it can't >be locked, the computer can) then I'm SOL. > Fair enough. Hardware keys are a real pain. >No one is promoting making illegal copies, or illegally using the software. > That is something you read into this. I never suggested that anyone posting here is promoting making illegal copies. Read my words again. I am insulted by your implication. What I did say, and if I restate then perhaps it will aid your comprehension, is that if the information is published here, in a public forum read by thousands, some people will inevitably put it to this use, whatever the intent of the publisher. > >Anyway, it has been proven beyond the shadow of a doubt that companies are >NOT 'deprived' of their income by pirated software, as %99 of pirated >software is used by someone who would not have bought it in the first >place, also a good number of them, if they are using the software >regularly, eventually do buy it. It's the best marketing tool the software >industry has to offer. Nonsense. If that's true, tell me why nearly every EDA company continues to protect its software in this way? Don't tell me! they're all wrong, and you are right? In over 12 years in the EDA industry, I've seen a number of cases where logic such as that you present has been used to justify the removal of a hardware lock. In every case, revenue dropped sharply without any corresponding evidence of a decrease in product usage. That's why the locks are there - experience shows they are needed. I've never known a user to intentionally "pirate" the software, but the absence of a lock nevertheless allows users to "stretch" the meaning of the licence agreement, or their co-workers to unknowingly break it. You've got to forgive people under pressure if they don't think too hard about how the software got onto the particular machine. > >> For me, and I would guess many other engineers in the EDA business who >> come here, the very existence of this thread is a sad occasion for the >> group. > >What's sadder to me is you assume people are going to use the key-removals >to promote illegal use of the software, when that is NOT the case at all. There you go again. Where did I talk about "promoting illegal use"? To my mind, anyone who is clever enough to remove the lock is almost certainly also wise enough to not pirate the software. Posting the instructions for how to do this in a newsgroup is, however, a very different matter. -- David PashleyArticle: 9635
Nonsense. Dongles are a pain in the a**e. To provide "security" for the software which I regularly use (which is a lot less than some people use) I would have at least SIX dongles in the back of my PC. Some of these belong to programs "no longer supported", and if such a dongle breaks, the investment in that program goes down the plughole. Don't forget - one has a right to use a piece of s/w without a time limit. My investment in such "no longer supported" software is well over $10k, and when Xilinx stop supporting the XACT6 dongle this figure would have increased by another $4k. And this is before we move on to other topics, like all the devices which don't work through a dongle. EPROM emulators, EPROM programmers, even other dongles. Regarding illegal copying, I would challenge anyone to do something useful with XACT6 (or any other similar software) without having the manuals. I have no idea who did that XACT6 crack but two things I am sure about: I am grateful for it, and it must have been damn difficult. >How can you justify promoting the removal of security devices? Surely >the resulting illegal copying of software (the use to which many would >inevitably put this information) is against the interest of both the >paying customers and those who make a living from providing and >supporting the software? > >For me, and I would guess many other engineers in the EDA business who >come here, the very existence of this thread is a sad occasion for the >group. Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.Article: 9636
Hi, Does anyone know about implementing random number generator in FPGA? Thank you very much, Peixin ZhongArticle: 9637
Greg Prior wrote: > > Hi all- > > Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am > looking for a better editor. I would like to try out a share one with > the capability of multiple levels of 'undo'. Also it should be VHDL > aware, and highlite keywords, etc. Can anyone point me to something? > > Thanks, Greg vim (VI iMproved)! it's free, does everything, is infinitely extensible, runs on (almost) anything, and is infinitely better than any other editor i've ever used. you can get current info from: http://www.vim.org/ (new) version 5 supports syntax highligthing ...Article: 9638
Here are a few VHDL editor options that I came accross. Some are shareware, some not. In no particular order: PeakVHDL -------- http://www.acc-eda.com/demo.htm Not a full-featured editor (with macros and the like), but includes a built-in editor with syntax highlighting, a VHDL hierarchy browser and VHDL wizards, including a test bench template generator. Part of a demo download but can also be requested on CD-ROM. WinEDIT (30-day trial) ---------------------- http://www.windowware.com/winware/winedit.html Supports VHDL syntax highlighting. ED4W ---- http://www.silicon-systems.com/prod01.htm ED4W-HDL is a Window95/NT productivity Editor, tailored specifically for the VHDL/Verilog design engineer. + colorised keywords + automatic testbench generation + automatic template expansion + dozens of free models which can be automatically inserted at cursor + Error tracking for MODEL SIM (VSYSTEM from Model Tech) VHDL Editors (from VHDL FAQ) ---------------------------- http://www.vhdl.org/comp.lang.vhdl/FAQ3.html#sstuff_editors Other VHDL editor options from the VHDL Frequently-Asked Questions page Sledgehammer ------------ http://www.premia.com Sledgehammer is an add-on for CodeWright and is an excellent VHDL editor for Win95 and NT4.0. The VHDL add-on was created by VHDL Technology Group (I think), whose web site is at http://www.vhdl.com. However, they only have a useless marketing message posted on their site. You may also find some of the other freely-downloadable software available from The Programmable Logic Jump Station of interest. See http://www.optimagic.com/lowcost.html ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Greg Prior wrote in message <351AEF9D.CB458CB0@digital-reflection.com>... >Hi all- > >Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am >looking for a better editor. I would like to try out a share one with >the capability of multiple levels of 'undo'. Also it should be VHDL >aware, and highlite keywords, etc. Can anyone point me to something? > >Thanks, Greg >Article: 9639
Most of the programmable logic companies are finally waking up to some of these problems. Most offer either free or low-cost versions of their software. It's true that these versions may not support their entire family, but it at least gives you a good look at their software capabilities. You can find a list to completely legitimate, vendor approved, free downloads and CD-ROMs at http://www.optimagic.com/lowcost.html#DeviceVendors. ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Sam Falaki wrote in message <351BB39F.1F2D8B32@nospam.videotron.ca>... >Good for Xilinx! > >They didn't even want to make me (little me) a measly >one-month loaner to at least evaluate their software. >I absolutely had to fork out a few $k if I wanted to >see how my designs would fit in their part. I now have >the Foundation kit limited to a useless 8k gates and even to >use that, I have to go through bureaucratic processes to >get authorisation codes (for 8k gates!). When will these >FPGA vendors finaly step out (like Actel) and provide >free software so users can easily compare between different >FPGA vendors (or maybe they are afraid of this?). And >guess what? Now that I am finaly ready to ship some protos, >which vendor will I use? Actel, and with their new MX series >devices I can make money in small production quantities and >ship them as stand-alone ASICs.! > >Sam Falaki > >David Pashley wrote: > >> In article <3519CE06.178C9C91@writeme.com>, Rick Collins >> <redsp@writeme.com> writes >> >> <suggestions of how to break Xilinx and Viewlogic security devices> >> >> How can you justify promoting the removal of security devices? Surely >> the resulting illegal copying of software (the use to which many would >> inevitably put this information) is against the interest of both the >> paying customers and those who make a living from providing and >> supporting the software? >> >> For me, and I would guess many other engineers in the EDA business who >> come here, the very existence of this thread is a sad occasion for the >> group. >> >> David Pashley > > >Article: 9640
Lattice has a CPLD the 6192 that offers a true DPRAM or FIFO or SRAM with all the control logic and signals already designed into the block. It also has a register file and 192 macrocells for logic. EDArticle: 9641
In article <351eca21.14313231@news.netcomuk.co.uk>, Peter <z80@ds2.com> writes >Nonsense. > >Dongles are a pain in the a**e. To provide "security" for the software >which I regularly use (which is a lot less than some people use) I >would have at least SIX dongles in the back of my PC. > >Some of these belong to programs "no longer supported", and if such a >dongle breaks, the investment in that program goes down the plughole. >Don't forget - one has a right to use a piece of s/w without a time >limit. My investment in such "no longer supported" software is well >over $10k, and when Xilinx stop supporting the XACT6 dongle this >figure would have increased by another $4k. I don't have a problem with that. I was following up on what Don Husby wrote: "Viewlogic please!! "As someone mentioned earlier, the Viewlogic licensing scheme is a pain "in the "ass, and often makes it impossible to use their software even with a "valid license. And that's what my "nonsense" as you call it was referring to. Does it sound like he's looking to remove the protection from a "no longer supported" product? > >And this is before we move on to other topics, like all the devices >which don't work through a dongle. EPROM emulators, EPROM programmers, >even other dongles. > >Regarding illegal copying, I would challenge anyone to do something >useful with XACT6 (or any other similar software) without having the >manuals. > What about someone who works next to someone who has the manuals? -- DavidArticle: 9642
--------------64D67E47690DCE32A0A2C0B1 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Another UNIX editor I have found very useful is NEDIT. This can be downloaded from ftp://ftp.fnal.gov/pub/nedit. They seem to have it compiled for most flavors of UNIX there but I believe the source code exists as well for any that are not covered. It is much easier to use than EMACs (very mouse-based) but still has alot of capibility. It has the ability to highlight keywords from several languages such as: C C++ Java Javascript Fortran Pascal Perl TCL Awk CSH Makefile HTML Verilog VHDL As well as others I am not so familiar with. There are also alot of familiar UNIX commands (wc, sort, make, etc.) which can be invoked from pull-down menus. Hope some of you out there find it useful. -- Brian ems wrote: > On Thu, 26 Mar 1998 16:15:25 -0800, Greg Prior > <eng@digital-reflection.com> wrote: > > >Hi all- > > > >Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am > >looking for a better editor. I would like to try out a share one with > >the capability of multiple levels of 'undo'. Also it should be VHDL > >aware, and highlite keywords, etc. Can anyone point me to something? > > > >Thanks, Greg > > > > gnu emacs. it's free, does everything, is infinitely extensible, runs > on (almost) anything, and is infinitely better than any other editor > i've ever used. you can get current info from: > > http://www.cs.washington.edu/homes/voelker/ntemacs.html > > vhdl extensions from: > > http://www.geocities.com/SiliconValley/Peaks/8287/ > > by the way, the editor that came with M1.3 had a nasty habit of > deregistering any other editor that was registered for ".vhd" files > (in win95). it did this automatically every time it was run. does > anyone know how to fix this? > > evan (ems@nospam.riverside-machines.com) -- ------------------------------------------------------------------- / 7\'7 Brian Philofsky (brian.philofsky@xilinx.com) \ \ ` Xilinx Applications Engineer hotline@xilinx.com / / 2100 Logic Drive 1-800-255-7778 \_\/.\ San Jose, California 95124-3450 1-408-879-5199 ------------------------------------------------------------------- --------------64D67E47690DCE32A0A2C0B1 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <HTML> <P>Another UNIX editor I have found very useful is NEDIT. This can be downloaded from <A HREF="ftp://ftp.fnal.gov/pub/nedit">ftp://ftp.fnal.gov/pub/nedit</A>. They seem to have it compiled for most flavors of UNIX there but I believe the source code exists as well for any that are not covered. It is much easier to use than EMACs (very mouse-based) but still has alot of capibility. It has the ability to highlight keywords from several languages such as: <P>C <BR>C++ <BR>Java <BR>Javascript <BR>Fortran <BR>Pascal <BR>Perl <BR>TCL <BR>Awk <BR>CSH <BR>Makefile <BR>HTML <BR>Verilog <BR>VHDL <P>As well as others I am not so familiar with. There are also alot of familiar UNIX commands (wc, sort, make, etc.) which can be invoked from pull-down menus. <P>Hope some of you out there find it useful. <P>-- Brian <BR> <P>ems wrote: <BLOCKQUOTE TYPE=CITE>On Thu, 26 Mar 1998 16:15:25 -0800, Greg Prior <BR><eng@digital-reflection.com> wrote: <P>>Hi all- <BR>> <BR>>Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am <BR>>looking for a better editor. I would like to try out a share one with <BR>>the capability of multiple levels of 'undo'. Also it should be VHDL <BR>>aware, and highlite keywords, etc. Can anyone point me to something? <BR>> <BR>>Thanks, Greg <BR>> <P>gnu emacs. it's free, does everything, is infinitely extensible, runs <BR>on (almost) anything, and is infinitely better than any other editor <BR>i've ever used. you can get current info from: <P><A HREF="http://www.cs.washington.edu/homes/voelker/ntemacs.html">http://www.cs.washington.edu/homes/voelker/ntemacs.html</A> <P>vhdl extensions from: <P><A HREF="http://www.geocities.com/SiliconValley/Peaks/8287/">http://www.geocities.com/SiliconValley/Peaks/8287/</A> <P>by the way, the editor that came with M1.3 had a nasty habit of <BR>deregistering any other editor that was registered for ".vhd" files <BR>(in win95). it did this automatically every time it was run. does <BR>anyone know how to fix this? <P>evan (ems@nospam.riverside-machines.com)</BLOCKQUOTE> <PRE>-- ------------------------------------------------------------------- / 7\'7 Brian Philofsky (brian.philofsky@xilinx.com) \ \ ` Xilinx Applications Engineer hotline@xilinx.com / / 2100 Logic Drive 1-800-255-7778 \_\/.\ San Jose, California 95124-3450 1-408-879-5199 -------------------------------------------------------------------</PRE> </HTML> --------------64D67E47690DCE32A0A2C0B1--Article: 9643
Peixin Zhong <pzhong@ee.princeton.edu> writes: >Hi, >Does anyone know about implementing random number generator in FPGA? >Thank you very much, >Peixin Zhong LFSR, linear feedback shift register. You build a shift register and the input is the XOR of some combination of the output bits. The number of bits and which are XORed determines how random it is, so they should be selected with care. How many numbers, how many bits each, how fast? -- glenArticle: 9644
Greg Prior wrote: > Due to a lack of an 'undo' function in the Xilinx Alliance tools, I am > looking for a better editor. I would like to try out a share one with > the capability of multiple levels of 'undo'. Also it should be VHDL > aware, and highlite keywords, etc. Can anyone point me to something? Try emacs. The 95/NT package is available at: http://www.cs.washington.edu/homes/voelker/ntemacs.html While the unix versions can be found at: ftp://prep.ai.mit.edu/pub/gnu/ and add to that vhdl-mode available at: http://www.geocities.com/SiliconValley/Peaks/8287/ This provides: - Highlighting of keywords, comments, built-in types, and strings - Indentation based on versatile syntax analysis (by Rod Whitby) - Electrification (automatic template generation) for most VHDL constructs - Word completion (dynamic abbreviations) - Menu containing all VHDL mode commands - Index menu (jump index to main units and blocks in a file) - Source file menu (menu of all source files in current directory) - Source file compilation (syntax analysis) - Postscript printing with fontification - Lower and upper case keywords - Easy customization - Works under GNU Emacs and XEmacs -- Regards Glenn Eng glenn.eng@nortel.com (remove "delete." in header to reply")Article: 9645
Check out our X84 board download manual at http://www.associatedpro.com/aps check uder support / downloads / X84 manual and labs. Also Azeddien I got your requests for information at our email box, but every time I tried to respond, your email address comes back as invalid. Feel free to call or email me again. Richard Azeddien Sllame wrote: > Hi every body... > Looking for interface program (written in VHDL, or C) between ISA > card and Pc. (to down load bitstream and control the operation via > menu, send data, recieve result say for 4x4 serial multiplier) > > Thanks in advance.... -- __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ Richard Schwarz, President EDA & Engineering Tools Associated Professional Systems (APS) http://www.associatedpro.com 3003 Latrobe Court richard@associatedpro.com Abingdon, Maryland 21009 Phone: 410.569.5897 Fax:410.661.2760 __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/Article: 9646
Peixin Zhong wrote: > > Hi, > > Does anyone know about implementing random number generator in FPGA? > > Thank you very much, > > Peixin Zhong For a uniform distribution without any concern over frequency content, a simple linear feedback shift register will do the trick. This is simply a shift register whose input is the one bit sum of selected delay taps off the register (exclusive or). THe sequence will repeat after some time...if the taps are selected for a maximal length sequence, the sequence will repeat after (2^n)-1 clocks, where n is the length of the shift register. Be aware that each bit from the shift register is a time delayed copy of the first bit. This means to get a better pseudorandom sequence, the shift direction should be so the first shift tap is the msb. You can look at the xilinx app note for details on the design and for a tabulation of the feedback taps for maximal length sequences for any reasonable sequence length. Since each bit is a time delayed copy of the first bit, there is some correlation between successive samples. This can be reduced by separating the bits you read the number from by varying numbers of internal register taps. There will still be a spectral coloring in the random numbers you take off of it however. For no spectral coloring, you can use independent LFSRs for each bit (either very long with different seeds or different lengths that are relative primes). If you need some other distribution, there are ways to get there. The gaussian or normal distribution can be had by summing several samples from a uniform generator. If you use the simple lfsr, there will be some coloring to the gauss distribution. Using a bunch of 1 bit lfsrs or summing many subsequent outputs from one will yeild the binomial approximation to the gaussian distribution that is spectrally pure. For an arbitrary distribution, you can use a cumulative density function of the desired distribution and a uniform generator. Since the CDF of a uniform distribution is proportional to the distribution, the generator can be treated as the CDF of the uniform distribution. Use that as the key to look in the reverse CDF of the target function. You'll need many bits to get an accurate representation, so the table gets large. This can be dealt with using by doing a binary search on the CDF table using the uniform dist as the key. I've done both the gaussian and an arbitrary random generator in FPGAs. The gaussian produces spectrally clean 7 bit gaussian RVs at 40 MHZ sample rate, the arbitrary produces a new 7 bit RV every 7 clocks, also at 40 MHz. These are used in physical system modelling. -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/~randraka The Andraka Consulting Group is a digital hardware design firm specializing in high performance FPGA designs for digital signal processing, computing and control applications.Article: 9647
>There you go again. Where did I talk about "promoting illegal use"? Here's your text: How can you justify promoting the removal of security devices? Surely the resulting illegal copying of software (the use to which many would inevitably put this information) is against the interest of both the paying customers and those who make a living from providing and supporting the software? Peter. Return address is invalid to help stop junk mail. E-mail replies to zX80@digiYserve.com but remove the X and the Y.Article: 9648
Thanks to Ray Andraka and other people who answered my question. The newsgroup has always being very helpful. Thank you all very much, PeixinArticle: 9649
Caspar Steineke wrote: > Hi Scott, > > Xilinx' XC6216 is probably what you are looking for. Besides > supporting > rapid partial reconfiguration it is the only (correct me if I'm wrong) > FPGA > with a documented bitstream format. The evaluation board from VCC > looks very > promising. We tried to order one last year but after a dozen E-mails > and > prospective lead times of several months we turned back to the 4000 > series. > If you are more successful, please let me know. > I'm still very excited about the 6200 series and with an improved > uP-interface, more CLB's and a lower price tag those chips could be > excellent reconfigurable co-processors for the mainstream. > > Caspar > Last year when our HOTWorks product came out every time I did a run it would be sold out before I could get done manufacturing them. Many people said "What a great problem to have!" But I must disagree. The longest delay was because we did a very large run for our distributor. The good news is the board is now in stock at Marshall's http://www.marshall.com/dynamic/html/mfrs/vcc/pages/index.htm There was another little delay while I redesigned the backend of of the Xilinx reference design. The original PCI backend design only allowed the Xilinx PCI core to transfer data at 7 MegaBytes sec. Knowing the the LogiCore PCI could transfer data at the full PCI spec I redesigned the backend design (they were only letting the core do single transfers) to let it be so. Our other product is the Fat Hot works (the above is target only) which has the hardware to do initiator transactions (XC4020-1). It comes with the XC6264 and will be shipping in a week or two. The good thing about the HOTWorks package? It has all the hardware AND all the software you need for only $995. Also the original reference performed very poorly on PCI transfers and had some mechanical errors. Keeping that in mind we desided we will replace anyone's board who bought the original reference design board from us for free. -- Steve Casselman, President Virtual Computer Corporation http://www.vcc.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