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
Graeme Durant wrote: > It is possible to use > the inherent volatility of sram-based fpgas as an asset to security. > If you programme the design into the device in the factory, battery > back it up to keep it in-situ for normal use, then fit tamper sensors > on the casework of your product (microswitches, tilt switches > whatever). It's actually simpler than that:Find an "SRAM-based" FPGA with low standby current consumption (XC3000 in Powerdown have the lowest, but XC4000XL are not so bad ) and provide sufficient battery capacity to maintain configuration during any ac power-outs. Set the bit that diables Readback, and you can be sure that nobody can reverse-engineer your code. It would require keeping the device powered up, de-capping the (plastic) package, and using a very sophisticated instrument to detect on-chip voltage levels at undocumented locations. No way ! See the Xilinx data book, page 13-43 "Design Security by Hiding the Configuration Bitstream". Contact me, if you have additional questions Peter Alfke, Xilinx ApplicationsArticle: 11401
Simon Ramirez <s_ramirez@email.msn.com> wrote in article <ugxDRW0w9GA.329@upnetnews05>... > To Everyone Who Responded: > Thanks to everyone who responded to my plea for information about FPGA > PCI cores information. I received some good and varied information about > these cores and what to do with them and without them. > The information I received makes me think that my silicon vendor choice > will be critical and the design effort will be moderately difficult. > When I make my decision about which vendor to select, do my design and > gain some experience with the design, I will post another message to let > everyone know how it went. I would like to see more information posted > about FPGA PCI designs as I have found out that FPGA PCI designs are about > to start taking off more and more. Yes, Simon, let us know what you choose and how it goes. It's looking like I'll have to do a PCI design Real Soon Now and I do want to know what happens with your design. -andy -- Andy Peters Sr. Electrical Engineer National Optical Astronomy Observatories apeters@noao.edu.NOSPAMArticle: 11402
jim granville wrote: > <snip of some fantastic stuff :-) > > > Very interesting - can those with experience in Radiation Enginnering, bring a more > terestial level to this, and quantify the difference in Radiation levels > - on the earths surface ? > - in Jumbo Jets ? > - and in Orbit ? for ground stuff, it is starting to be a bit of concern, for extremely large memories, for example. there was a symposium on this last year, you might want to contact michele.gates@gsfc.nasa.gov for info (can't remember if there's one 'l' or two in her name, i always get it wrong).perhaps they put some proceedings together. and recently one of the trade rags had an article by an ibm engineer studying this for large computers. obviously, the higher you go the higher the radiation levels. they run experiments by putting dosimetry on airplanes, recently saw a paper where they loaded it onto a concorde (presented at ieee nsrec '98). of course, when you do some work in a radiation facility, they assure you that the dose you got on the cross-country flight was higher than the dose you'll get there. yeah, right. :-) anyways, contact michele for rate info. i think they're having a session in another conference this fall on the topic, perhaps at sematech. if anyone wants info, drop me an email, i got the info somewhere. also, at ieee nsrec, saab and xilinx had a paper on a topic related to this: Neutron Single Event Upsets In SRAM-Based FPGAs Abstract - SRAM-based FPGAs have been studied for their sensitivity to atmospheric high energy neutrons. FPGAs with the supply voltage 5V and 3.3V were irradiated by 0-11, 14 and 100 MeV neutrons and showed a very low SEU susceptibility. this is available on-line at: http://rk.gsfc.nasa.gov/papers.htm ____________________________________________________________________ > My understanding of this is that you still get radiation hits at gnd level, just not as > many of them. > > What is statistically significant depends on your point of view. > > I also recall seeing something about Radiation triggering LATCHUP - how do the redundant > engineering approaches handle that problem :-) not very well. gotta be able to isolate power, etc., etc., etc. typically a latchup detection circuit is used that can control power. system design must be able to tolerate the light's blinking by saving state, restoring, etc. this type of stuff is done. > - jg rk p.s. someone spelled 'reliability' wrong in the subject :-)Article: 11403
Hi, This is ABEL code for grey code up counter (CGU14) in Lattice Macro Library. Hope this can help you. Regards, Man Kit MODULE CGU14 CLK,CD,EN,PS PIN; D3..D0,LD PIN; Q3..Q0 PIN ISTYPE 'REG_D'; N_Q3..N_Q0 NODE ISTYPE 'COM'; count = [Q3..Q0]; data = [D3..D0]; n_count = [N_Q3..N_Q0]; EQUATIONS count.CLK = CLK; count.AR = CD; count.D = n_count; N_Q0 = (Q0 & !LD & !EN) # (D0 & LD) # (!Q3 & !Q2 & !Q1 & !LD & EN) # (!Q3 & Q2 & Q1 & !LD & EN) # (Q3 & Q2 & !Q1 & !LD & EN) # (Q3 & !Q2 & Q1 & !LD & EN) # PS; N_Q1 = (Q1 & !LD & !EN) # (D1 & LD) # (Q1 & !Q0 & !LD & EN) # (!Q3 & !Q2 & Q0 & !LD & EN) # (Q3 & Q2 & Q0 & !LD & EN) # PS; N_Q2 = (Q2 & !LD & !EN) # (D2 & LD) # (!Q3 & Q1 & !Q0 & !LD & EN) # (Q2 & !Q1 & !LD & EN) # (Q2 & Q0 & !LD & EN) # PS; N_Q3 = (Q3 & !LD & !EN) # (D3 & LD) # (Q2 & !Q1 & !Q0 & !LD & EN) # (Q3 & Q0 & !LD & EN) # (Q3 & Q1 & !LD & EN) # PS; END Marc Heuler <marc@aargh.franken.de> wrote in article <+-pMC*BFo@aargh.franken.de>... > I want to implement a Gray code counter (16 bit) in ABEL HDL for the > Lattice 1016 device. > > Are there gray-code-counter examples on the net? >Article: 11404
>In the past, I have always assumed that the 'hide it away' approach to >security in fpgas was the only one available. However, I recently came >accross a much more interesting approach. It is possible to use >the inherent volatility of sram-based fpgas as an asset to security. >If you programme the design into the device in the factory, battery >back it up to keep it in-situ for normal use, then fit tamper sensors >on the casework of your product (microswitches, tilt switches >whatever). These tamper sensors need to 'evaporate' the design >by resetting the device. In this way an attack on the product will >find nothing of a proprietory nature. This approach can be taken >further as regards any microprocessor program storage or whatever. This self-destruct feature has been used for many years in line encryptors used in banking etc comms. One has switches on the lid, the PCB is potted in epoxy with sensing wire wrapped round it, light sensitive devices inside the box (to detect someone opening the box and looking inside), etc. There is a battery+capacitor and the EPROM/EEPROM/SRAM/whatever gets destroyed if tampering is detected. -- 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: 11405
This is a multi-part message in MIME format. --------------19E5BB001702AC4EF3F31801 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello reader, i would announce two development boards of Alcatel Telecom. The two boards are useful for ASIC prototyping and simulation. We already used the boards to verify DSP algorithms written in VHDL, to test the behaviour of PLL circuits, to built test equippment (complex signal generators written in VHDL) for our fab and for other applications. Short description of the boards: HW_SIM * 200.000 gates logical resources (4*Altera EPF10K50) (300.000 gates if equpipped with 1EPF10K70) * 2 Slots for SIMM memory modules (30 pin types) * 2 on board oscillators (48,640 MHz, 16,384MHz) * PBA size 233mm*210mm * breadboard area * reset circuit * ... DEV_KIT * 20.000 gates logical resource (2*Altera EPF81188) * on board oscillator (16,384 MHz) * PBA size 233mm*160mm (possible splitting 2 times 100mm*160mm) * breadboard area * reset circuit * ... For more information about the development boards contact one of the persons listed below. Financial Questions: R.Prestin@alcatel.de Technical Questions: brod@lts.sel.alcatel.de Best regards Lothar Brodbeck p.s. An user wrote: ... We successfully used the HW_SIM development board to verify our complex signal processing algorithm. ... The programming and the handling of the evaluation board stands out. ... You can find a description and pictures of the boards using the following address: http://www.dev.alcatel.de/telecom/asd/test_hw/topdevboard.htm --------------19E5BB001702AC4EF3F31801 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Lothar Brodbeck Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Lothar Brodbeck n: Brodbeck;Lothar org: Alcatel Telecom adr: Lothar Brodbeck;;Department AS/HEC;Lorenzstrasse 10;70435 Stuttgart;;Germany email;internet: brod@lts.sel.alcatel.de title: Development Engineer tel;work: +49 711 821 47334 tel;fax: +49 711 821 45069 note;quoted-printable:Telephone +49 711 821 47334=0D=0A= Fax +49 711 821 45068 x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------19E5BB001702AC4EF3F31801--Article: 11406
Hello to everbody !!! I want to insert C functions in VHDL code .I have heard that this can be done using foreign architectures,but how do I have to write my C code in order to get the parameters passed by the VHDL code and how can I return values ? Thank you very much in advanced . Benjamin -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member ForumArticle: 11407
>>>>>>>>> SPAM deleted by Archive OwnerArticle: 11408
gd@nospamheliontech.com (Graeme Durant) writes: >On 8 Aug 1998 00:32:13 GMT, "Robert Semenoff" ><robert_semenoffNOSPAM@phoenix.com> wrote: >>I am trying to determine the most secure way of implementing >>proprietary algorithms. >In the past, I have always assumed that the 'hide it away' approach to >security in fpgas was the only one available. However, I recently came >accross a much more interesting approach. It is possible to use >the inherent volatility of sram-based fpgas as an asset to security. >If you programme the design into the device in the factory, battery >back it up to keep it in-situ for normal use, then fit tamper sensors >on the casework of your product (microswitches, tilt switches >whatever). These tamper sensors need to 'evaporate' the design >by resetting the device. In this way an attack on the product will >find nothing of a proprietory nature. This approach can be taken >further as regards any microprocessor program storage or whatever. It depends on what level of sophistication you're expecting from your attackers. DRAM contents can be recovered hours after power is removed, even longer if you store it at low temperatures. SRAM is even worse, in extreme cases it's possible to recover the state months later. I touch on this briefly towards the end of http://www.cs.auckland.ac.nz/~pgut001/pubs/secure_del.html, one day I'll finally get around to extending this section a bit. Peter.Article: 11409
I need to implement in an fpga an algorithm that will divide an integer by 3. The dividend length is still to be determined but will be somewhere between 20 and 30 bits, and the divisor is always the number 3. Does anyone know an efficient combinatoric algorithm that can accomplish this? Thanks in advance, KenArticle: 11410
Hi, We have just launched a new Software Metrics site at http://www.totalmetrics.com We have an interest in software measurement for realtime and embedded systems. There is some information on training courses in this area and we will be posting papers on this subject soon. Look at our html newsletter for more information. http://www.totalmetrics.com/newsletter/news9807.htm If you have any experience, links or news you would like to share in this area, please forward it to editor@totalmetrics.com Thank you, Jin Ng editor@totalmetrics.comArticle: 11411
> It depends on what level of sophistication you're expecting from your > attackers. DRAM contents can be recovered hours after power is removed, even > longer if you store it at low temperatures. SRAM is even worse, in extreme > cases it's possible to recover the state months later. I touch on this > briefly towards the end of > http://www.cs.auckland.ac.nz/~pgut001/pubs/secure_del.html, one day I'll > finally get around to extending this section a bit. > > Peter. The military cryptos I dealt with some 12-13 years ago re-wrote the memory several times with a random pattern as part of the destruct sequence. The SRAM FPGA configuration memory cell is more akin to a shift register made of flip-flops than to SRAM memory. I'm not sure how that affects retention of residual levels (I suspect it makes the residual even stronger). I don't see any reason you couldn't write in a couple of dummy configurations during the destruct sequence to obliterate the real configuration though. -- -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: 11412
>I need to implement in an fpga an algorithm that will divide an integer >by 3. // PRE: a>=0 q = 0; r = a+1; while (r>=4) { x = r/4; // can be implemented by shifting y = r%4; // can be implemented by ANDing q = q+x; r = x+y; } // POST: a==3*q+r && 0<=r && r<3 // that is: q==a/3 -- Jeroen Fokker | jeroen@cs.uu.nl dept.of Computer Science, Utrecht University | tel.+31-30-2534118 PObox 80089, 3508TB Utrecht, the Netherlands | fax.+31-30-2513791Article: 11413
See the latest on FPGAs, CPLDs, and related software on The Programmable Logic Jump Station at http://www.optimagic.com/ The Programmable Logic Jump Station is a comprehensive set of links to nearly all matters related to programmable logic. Featuring: --------- --- Frequently-Asked Questions (FAQ) --- Programmable Logic FAQ - http://www.optimagic.com/faq.html A great resource for designers new to programmable logic. --- FPGAs, CPLDs, FPICs, etc. --- Recent Developments - http://www.optimagic.com Find out the latest news about programmable logic. Device Vendors - http://www.optimagic.com/companies.html FPGA, CPLD, SPLD, and FPIC manufacturers. Device Summary - http://www.optimagic.com/summary.html Who makes what and where to find out more. Market Statistics - http://www.optimagic.com/market.html Total high-density programmable logic sales and market share. --- Development Software --- Free and Low-Cost Software - http://www.optimagic.com/lowcost.html Free, downloadable demos and evaluation versions from all the major suppliers. Design Software - http://www.optimagic.com/software.html Find the right tool for building your programmable logic design. Synthesis Tutorials - http://www.optimagic.com/tutorials.html How to use VHDL or Verilog. --- Related Topics --- FPGA Boards - http://www.optimagic.com/boards.html See the latest FPGA boards and reconfigurable computers. Design Consultants - http://www.optimagic.com/consultants.html Find a programmable logic expert in your area of the world. Research Groups - http://www.optimagic.com/research.html The latest developments from universities, industry, and government R&D facilities covering FPGA and CPLD devices, applications, and reconfigurable computing. News Groups - http://www.optimagic.com/newsgroups.html Information on useful newsgroups. Related Conferences - http://www.optimagic.com/conferences.html Conferences and seminars on programmable logic. Information Search - http://www.optimagic.com/search.html Pre-built queries for popular search engines plus other information resources. The Programmable Logic Bookstore - http://www.optimagic.com/books.html Books on programmable logic, VHDL, and Verilog. Most can be ordered on-line, in association with Amazon.com . . . and much, much more. Bookmark it today!Article: 11414
Kenneth W. Wagner wrote: > I need to implement in an fpga an algorithm that will divide an integer > by 3. The dividend length is still to be determined but will be > somewhere between 20 and 30 bits, and the divisor is always the number > 3. > > Does anyone know an efficient combinatoric algorithm that can accomplish > this? > > Thanks in advance, > Ken Approximation to N/3: [Image] Hope you can read the above. The quantities in parentheses only needs to be evaluated once, all others are shifts and adds. Extend the ellipses to desired output accuracy. Do summation from right to left to minimize size of adders. Alternatively, use KCM (constant coefficient multiplier), which breaks dividend into 4-bit groups, feeds each group k( = 0..m) into 16 x (4k+n) LUT, then sums all LUT outputs. Read further about KCMs at Xilinx web site. The KCM is more efficient in a Xilinx part than an unrolled version of the adder only circuit. Both adder and KCM may be 'rolled-up'. >>>>>>>>> File binary deleted by Archive OwnerArticle: 11415
Kenneth W. Wagner wrote: > I need to implement in an fpga an algorithm that will divide an integer > by 3. The dividend length is still to be determined but will be > somewhere between 20 and 30 bits, and the divisor is always the number > 3. > > Does anyone know an efficient combinatoric algorithm that can accomplish > this? > > Thanks in advance, > Ken You can treat this as a constant multiply by 1/3. The multiplication can then be done as a look up of partial products. Each 4 bits of the dividend goes into a look up that is as wide as necessary to get the required output precision. Each look up outputs 0/3, 1/3, 2/3, ... ,15/3. The outputs of each lookup are then shifted and summed in an adder tree. The tree depth is proportional to log(input bits). This can be entirely combinatorial, can be pipelined, or can be broken down into a sequential algorithm depending on your performance needs. The xilinx core generator can generate pre-placed optimized constant multipliers for you. -- -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: 11416
Keneth, 1/4+1/16+1/64+1/256+1/1024+...=1/3. So shift your number by 2, 4, 6, 8, etc. and add everithing. Catalin Baetoniu Kenneth W. Wagner wrote: > I need to implement in an fpga an algorithm that will divide an integer > by 3. The dividend length is still to be determined but will be > somewhere between 20 and 30 bits, and the divisor is always the number > 3. > > Does anyone know an efficient combinatoric algorithm that can accomplish > this? > > Thanks in advance, > KenArticle: 11417
There are multiple positions open for the following direct hire position. Interested individuals should responde to RSizingSol@aol.com Description: You will design chips and/or boards and assist with sytem integrations. Products will include the design of high bandwidth electro-optic interfaces, sensors, laser modulators/controllers, photodectors, micro-positioning control circuits, microprocessor subsystems and power/ground systems. Work will be performed using state of the art design and analysis tools. Experience with most of the following is required: Design/simulation/test using TTL, ECL, PECL and CMOS technologies Design/simulation/test using FPGA (alters), and PLD devices (HDL languge experience is desired) Design, simulation and test of microprocessor or DSP subsystems (some programming experience is desired) EDA tools (e.g. viewlogic, mentor, candence) This position requires relocation to western NY. Excellent benefits and salary with goal sharing option and relocation.Article: 11418
John L. Smith wrote: > Kenneth W. Wagner wrote: > >> I need to implement in an fpga an algorithm that will divide an >> integer >> by 3. The dividend length is still to be determined but will be >> somewhere between 20 and 30 bits, and the divisor is always the >> number >> 3. >> >> Does anyone know an efficient combinatoric algorithm that can >> accomplish >> this? >> >> Thanks in advance, >> Ken > [Image] Extending this to 20 bits would require 3 (LUT) + 11.5(Adder) more CLBs for total of 49. " 24 3 + 13.5 65.5 28 3 + 15.5 84 32 3 + 17.5 104.5 Unfortunately, the fully parallel implementation of the KCM is O(N^2) due to the increasing size of the adders required. A rolled up (serialized) version would be O(N), but would also take O(N) time. I hope no one minds the JPG, it's how the cut and paste worked. >>>>>>>>> Stupid file binary deleted by Archive OwnerArticle: 11419
Hi, I was wondering if anyone knew how to simulate the power consumption from a Xilinx FPGA. I have completed the design(first draft) and would like to estimate if I can get further power consumption with modifications to the design. TIA, Phil.Article: 11420
You can further reduce additions. With only 4 additions you can divide 32 bits by 3, the divisions are free. 1/4 + [1/4] / 4 + [1/4 + [1/4] / 4 ] / 16 + [1/4 + 1[1/4] / 4 + [1/4 + [1/4] / 4 ] / 16] / 256 + ... -- Erwin Oertli mailto:oertli@inf.ethz.ch http://www.cs.inf.ethz.ch/~oertli Catalin wrote in message <35D098E9.764ABCC0@nt.com>... >Keneth, > >1/4+1/16+1/64+1/256+1/1024+...=1/3. So shift your number by 2, 4, 6, 8, etc. >and add everithing. > >Catalin Baetoniu > >Kenneth W. Wagner wrote: > >> I need to implement in an fpga an algorithm that will divide an integer >> by 3. The dividend length is still to be determined but will be >> somewhere between 20 and 30 bits, and the divisor is always the number >> 3. >> >> Does anyone know an efficient combinatoric algorithm that can accomplish >> this? >> >> Thanks in advance, >> Ken > > >Article: 11421
In article <6qpe8a$uqp$1@nnrp1.dejanews.com>, <benjamin.carrion-schaefer@hl.siemens.de> wrote: >Hello to everbody !!! > >I want to insert C functions in VHDL code .I have heard that this can be done >using foreign architectures,but how do I have to write my C code in order to >get the parameters passed by the VHDL code and how can I return values ? > >Thank you very much in advanced . Depends entirely on which simulator you are using. -- /* _ */main(int k,char**n){char*i=k&1?"+L*;99,RU[,RUo+BeKAA+BECACJ+CAACA" /* / ` */"CD+LBCACJ*":1[n],j,l=!k,m;do for(m=*i-48,j=l?m/k:m%k;m>>7?k=1<<m+ /* | */8,!l&&puts(&l)**&l:j--;printf(" \0_/"+l));while((l^=3)||l[++i]);} /* \_,hris Brown -- All opinions expressed are probably wrong. */Article: 11422
Nice, Can one systematically find similar fast converging recursive algorithms for scaling by other fixed but freely chosen numbers? For example: What if one would like to scale a CORDIC generated result back to same reference level as the inputs had? (Rect. to Polar conversion) Output should then be a fixed multiply by 0.607253 (for 16 bit precision) I tried this once and I think I came close to 16 bit precision with 4 (19 bit) adders, but I found the shifts and add/subtracts by trial and error, not by formal analysis. I always wondered if it could be reduced to 3 adders, but didn't find time to take it any further. How did you find the series expansion you used? (maybe I should dig out my old math books to find this, Sorry if the answer is obvious) Harald Vefling Vingmed Sound AS Erwin Oertli wrote: > > You can further reduce additions. With only 4 additions you can divide 32 > bits by 3, the divisions are free. > > 1/4 > + [1/4] / 4 > + [1/4 + [1/4] / 4 ] / 16 > + [1/4 + 1[1/4] / 4 + [1/4 + [1/4] / 4 ] / 16] / 256 > + ... > > -- > Erwin Oertli > mailto:oertli@inf.ethz.ch > http://www.cs.inf.ethz.ch/~oertli > > Catalin wrote in message <35D098E9.764ABCC0@nt.com>... > >Keneth, > > > >1/4+1/16+1/64+1/256+1/1024+...=1/3. So shift your number by 2, 4, 6, 8, > etc. > >and add everithing. > > > >Catalin Baetoniu -- snipArticle: 11423
i did a Z80 board about 15 years ago, with test code in a windowed eprom. there was one occasion when i erased the eprom, and wrote in a completely different program. the board wasn't working correctly, and I eventually froze the eprom. amazingly, for a couple of minutes, the processor executed the old, 'erased', code sequence, before reverting to the new code. it only happened once - i never managed to duplicate it. evanArticle: 11424
I'm looking for a comparison regarding achievable FFT-speed between FPGA vs. DSP-solutions. For instance: DSP TMS C6x can manage a 1024 point-FFT in 104 µs. Which FPGA can achieve which speed? Is there a comprehensive table in the web? Who can help me? Thomas
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