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
Austin Franklin wrote: > Ray, thanks. > > So now I have to go setup every guy in the lab and the production floor, > who used to be able to just use a little 486 notebook w/ a 20M disk and > DOS, with some version of Windoze, a mouse, and a project > directory...instead of just XChecker and a .bit file on a floppy... To me, > this seems silly to have made it 'so' complicated when it used to be so > easy. > > Also, is there a way of doing it from the command line, as opposed to > having to bring up the GIU, select 'stuff' etc.? > > Austin > I'm not sure about the h/w debugger but you can run the JTAG programmer tool from the command line. However I had problems gettingthe 1.5i version to run under Win95. Works o.k from the NT machine even though this needs 2 extender cables to reach the lab bench.Article: 18701
Hello, I am hoping that someone has a really good model for a pullup in a VHDL test bench that can convert a bidirectional signal between an 'H' and a '1'. I realize that I can just assign and 'H' to a signal in the test bench to pull up the signal when not driven by my device under test (DUT). The problem is that the DUT has a state machine that looks for the bidirectional signal (that is its input and output pin) to be a '1' during certain states. As you could guess, the state machine never progresses in simulation due to the 'H' on the signal. The device performs well in-circuit because the resistor on the circuit board pulls the signal to a logic 1 on the board. The issue is pulling up the bidirectional signal with something like an 'H' and converting that to a logic '1' in simulation so that the internal DUT "sees" a '1' when just pulled up by the test bench pullup. Any help is greatly appreciated. Please post the response on the newsgroup so that it may help others, rather than replying by email.Article: 18702
Altera Corporation (San Jose, CA) has begun to ship the first in its 20KE line of programmable logic devices (PLDs): the EP20K400E. The device implements 400,000 gates (one million maximum system gates), and is based on the 0.18 micron, six-layer-metal, 1.8-V process. It features on-chip content-addressable memory (CAM), low-voltage differential signaling (LVDS), and phase locked loops (PLLs), which allows designers to create System-on-a-Programmable-Chip silicon (SOPC) silicon for applications like Layer 3 routers and switches, wideband CDMA baseband signal processing, and ATM cell processing and traffic management. http://www2.embeddedtechnology.com/welcome/ngckArticle: 18703
hi ray, Ray Andraka wrote: > Rich, > > I'm pretty sure the current crop of xilinx and altera parts could handle this data rate > with minimal pipelining. I've done an FIR filter in xilinx with 16 bit wide parallel > arithmetic that runs at around 150 MHz in a Virtex -4 (slowest speed grade). I've also > done 16 bit arithmetic designs in 4000XLA-07's that run in the high 140 MHz range. His > input is 155Mb/s and it is paralleled down to a maximum data rate of 78 MHz. The > encoder output is 10 bits, so a reasonable guess is any arithmetic inside is likely to > be less than 10 bits. The maximum width into the encoder is 4 bits, so if it is > combinatorial, the logic is one LUT deep in both Altera and Xilinx. Obviously, not > knowing the details of the encoder blocks, we can't really tell how complicated they > are, but based on the observations of the input and output widths a first guess > indicates that it is possible. the original poster seems to want a non-volatile solution for the lab ... i'm not sure exactly why ... but i believe that most modern fpgas could probably handle this without too much trouble. i haven't run timing analyses yet on the new sx-a stuff, which would be the latest actel part, and a bit fresher than the q-logic devices which is a few years old, at least with respect to technology (almost two years ago, with non-speed graded devices, i was able to get 500 MHz+ ripple counter performance, so they're not exactly slow pokes). they seem to be spending more time on hard-wired cores, adding value that way, a bit different approach then the actelians. i do know from some lab experiments on early sx-a prototypes [day job] that they are quite fast. i think that, if i get some time, i will download the software for sx-a (a 3 hour download here so not at the top of my list of things to do today) and perhaps run some analyses to see how fast a circuit like this could run, making some assumptions for the encoder. architecturally, they don't have the carry chain like the xilinx 4k and virtex devices, which is fast, since it's hardwired. for those not familiar with the devices, starting in sx, and continuing in sx-a, they have some higher speed routing resources, direct connect (no antifuses) and fast connect (one antifuse) in addition to the regular routing - note that starting in sx they changed antifuse structure, cutting the antifuse resistance by approximately 1 order of magnitude while moving away from the channelled architecture. however, without the carry chain, a ripple adder will be relatively slow and a look ahead technique would be needed. of course, i'm not sure exactly what's in his encoder, perhaps not even math. now there were some talks talking about speed/power vs. different devices at mapld '99, i'll have to get those proceedings (a few papers still in need of finalization) out so i can look at what they came up with :-) ----------------------------------------------------- > I haven't looked at the Actel selection lately, so I don't know what kinds of speed you > can expect there. I do know there is no carry chain, so you have to be careful in the > selection of the logic used in artihmetic. your're right, in sx and sx-a there is no carry chain. i don't believe there is one in their gatefield-designed devices either (someone yell if i got that wrong). for their sram devices, (not yet released), there's a description in the following architecture paper ( http://rk.gsfc.nasa.gov/richcontent/papers/actel_sram_architecture99.pdf ). they moved to lut's, although at a different size then the xilinx guys, using a mixture of 2 and 3-input luts. they also have different types of routing, but it's been a while since i've read that paper closely. if only i had more time ... :( ------------------------------------------------------------------------------- > A possibility might be to use a CPLD for this. Most of the CPLDs are non-volatile, and > many support in system reprogrammability. If the system is as simple as represented > here, I see no reason it wouldn't fit into one of the many CPLDs out there. Again, the > data rates are within the realm of possibility for the currently available devices with > careful design. again, this shouldn't be hard to fit into a number of different devices. for a lab instrument, not sure why one would have a preference for a "rom" or "ram" types, unless they want to keep the design secure with minimum hassle <rk carefully avoids the long design security thread> well, back to work, rkArticle: 18704
You can modify your vhdl to check if the signal in question is equal to an 'H' or a '1'. if ( ( sig1 = '1' ) or ( sig1 = 'H' ) ) then Not too elegant, but that's how I do it. It will synthesize the same but it will simulate the way you want. Pete Dudley Chris <csbeasley@nospworldnet.att.net> wrote in message news:01bf2a26$35b8f1a0$0a24450c@c1... > Hello, I am hoping that someone has a really good model for a pullup in a > VHDL test bench that can convert a bidirectional signal between an 'H' and > a '1'. > > I realize that I can just assign and 'H' to a signal in the test bench to > pull up the signal when not driven by my device under test (DUT). The > problem is that the DUT has a state machine that looks for the > bidirectional signal (that is its input and output pin) to be a '1' during > certain states. As you could guess, the state machine never progresses in > simulation due to the 'H' on the signal. The device performs well > in-circuit because the resistor on the circuit board pulls the signal to a > logic 1 on the board. > > The issue is pulling up the bidirectional signal with something like an 'H' > and converting that to a logic '1' in simulation so that the internal DUT > "sees" a '1' when just pulled up by the test bench pullup. > > Any help is greatly appreciated. Please post the response on the newsgroup > so that it may help others, rather than replying by email. > >Article: 18705
On 8 Nov 1999 20:17:06 GMT, "Chris" <csbeasley@nospworldnet.att.net> wrote: >Hello, I am hoping that someone has a really good model for a pullup in a >VHDL test bench that can convert a bidirectional signal between an 'H' and >a '1'. > >I realize that I can just assign and 'H' to a signal in the test bench to >pull up the signal when not driven by my device under test (DUT). The >problem is that the DUT has a state machine that looks for the >bidirectional signal (that is its input and output pin) to be a '1' during >certain states. As you could guess, the state machine never progresses in >simulation due to the 'H' on the signal. The device performs well >in-circuit because the resistor on the circuit board pulls the signal to a >logic 1 on the board. > >The issue is pulling up the bidirectional signal with something like an 'H' >and converting that to a logic '1' in simulation so that the internal DUT >"sees" a '1' when just pulled up by the test bench pullup. > >Any help is greatly appreciated. Please post the response on the newsgroup >so that it may help others, rather than replying by email. Look at the to_X01() function in ieee.std_logic_1164. It converts '1' and 'H' to '1'; '0' and 'L' to '0' and the rest get converted to 'X'. Regards, Allan.Article: 18706
Viewlogic provides a jed2ahdl utility though I don't know if it's reliable or if it's still available given the developments of AMD -> Vantis -> Lattice. Adam Michael Ayton wrote: > Is it possible to read back the contents of an older 16L8 pld and make a > schematic from the jedec ??? -- "Sometimes I think the surest sign that there's intelligent life on other planets is that none of it has tried to contact us." - Calvin, "Calvin and Hobbes"Article: 18707
Is it possible to read back the contents of an older 16L8 pld and make a schematic from the jedec ???Article: 18708
Anyone know if it is possible to reverse engineer a 16L8??? and get some sort of code from it other than the JEDEC? ThanksArticle: 18709
Hello all Can anyone solve the two problems found in ViewLogic's Workview office 7.5: (1) When I use FPGA Express (included in WVOffice 7.5) to synthesize my design, it generates the EDIF file for Altera's MAXplus to fit into the device. The generated EDIF file flattens my bus (defined in VHDL) , so when I do backannotation (post simulation), I have to re-define my test bench in Speedwave (a VHDL simulator) for doing backannotation. For example, I defined in VHDL: FREQ:in std_logic_vector(35 downto 0); Then in the backannotation file, it becomes FREQ35: in std_logic; FREQ34: in std_logic; FREQ33: in std_logic; FREQ32: in std_logic; FREQ31: in std_logic; .. FREQ1: in std_logic; FREQ0: in std_logic; It will make every designer very upset to do much typing. I have checked the files, it is the problem in EDIF file generated by FPGA Express. Can you solve the bug? (2) Besides, FPGA Express can't synthesize the VHDL LPM generated files, which are generated from Altera's MaxPlus. The error message is: (access violation) FEXP caused an invalid page fault in module DPMW.DLL at 0137:1005eec7. Registers: EAX=02d0c074 CS=0137 EIP=1005eec7 EFLGS=00010206 EBX=00000000 SS=013f ESP=0269fad0 EBP=00000000 ECX=02eb8d78 DS=013f ESI=02dbfe08 FS=3977 EDX=02f0c770 ES=013f EDI=02eb8d8c GS=0000 Bytes at CS:EIP: 8b 3b 52 e8 81 6b fb ff 83 c4 04 89 30 8b 44 24 Stack dump: 02ec6158 02ca1c08 02feaa08 00780d08 02d0c4ac 02eb8d78 02eb8d50 02f0c770 02f0c790 02eb8d78 00000000 02f0c750 02f0c820 02ca1c08 0000001c 00780d08 Below is the VHDL file (multip.vhd) -- megafunction wizard: %LPM_MULT% -- GENERATION: STANDARD -- VERSION: WM1.0 -- MODULE: lpm_mult -- ============================================================ -- File Name: multip.vhd -- Megafunction Name(s): -- lpm_mult -- ============================================================ -- ************************************************************ -- THIS IS A WIZARD GENERATED FILE. DO NOT EDIT THIS FILE! -- ************************************************************ -- Copyright (C) 1988-1999 Altera Corporation -- Any megafunction design, and related net list (encrypted or decrypted), -- support information, device programming or simulation file, and any other -- associated documentation or information provided by Altera or a partner -- under Altera's Megafunction Partnership Program may be used only to -- program PLD devices (but not masked PLD devices) from Altera. Any other -- use of such megafunction design, net list, support information, device -- programming or simulation file, or any other related documentation or -- information is prohibited for any other purpose, including, but not -- limited to modification, reverse engineering, de-compiling, or use with -- any other silicon devices, unless such use is explicitly licensed under -- a separate agreement with Altera or a megafunction partner. Title to -- the intellectual property, including patents, copyrights, trademarks, -- trade secrets, or maskworks, embodied in any such megafunction design, -- net list, support information, device programming or simulation file, or -- any other related documentation or information provided by Altera or a -- megafunction partner, remains with Altera, the megafunction partner, or -- their respective licensors. No other licenses, including any licenses -- needed under any third party's intellectual property, are provided herein. LIBRARY ieee; USE ieee.std_logic_1164.all; LIBRARY lpm; USE lpm.lpm_components.all; ENTITY multip IS PORT ( dataa : IN STD_LOGIC_VECTOR (15 DOWNTO 0); datab : IN STD_LOGIC_VECTOR (15 DOWNTO 0); clock : IN STD_LOGIC ; result : OUT STD_LOGIC_VECTOR (31 DOWNTO 0) ); END multip; ARCHITECTURE SYN OF multip IS SIGNAL sub_wire0 : STD_LOGIC_VECTOR (31 DOWNTO 0); COMPONENT lpm_mult GENERIC ( LPM_WIDTHA : NATURAL; LPM_WIDTHB : NATURAL; LPM_WIDTHP : NATURAL; LPM_WIDTHS : NATURAL; INPUT_B_IS_CONSTANT : STRING; LPM_REPRESENTATION : STRING; USE_EAB : STRING; LPM_PIPELINE : NATURAL ); PORT ( dataa : IN STD_LOGIC_VECTOR (15 DOWNTO 0); datab : IN STD_LOGIC_VECTOR (15 DOWNTO 0); clock : IN STD_LOGIC ; result : OUT STD_LOGIC_VECTOR (31 DOWNTO 0) ); END COMPONENT; BEGIN result <= sub_wire0(31 DOWNTO 0); lpm_mult_component : lpm_mult GENERIC MAP ( LPM_WIDTHA => 16, LPM_WIDTHB => 16, LPM_WIDTHP => 32, LPM_WIDTHS => 32, INPUT_B_IS_CONSTANT => "NO", LPM_REPRESENTATION => "SIGNED", USE_EAB => "OFF", LPM_PIPELINE => 1 ) PORT MAP ( dataa => dataa, datab => datab, clock => clock, result => sub_wire0 ); END SYN; -- ============================================================ -- CNX file retrieval info -- ============================================================ -- Retrieval info: PRIVATE: WidthA NUMERIC "16" -- Retrieval info: PRIVATE: WidthB NUMERIC "16" -- Retrieval info: PRIVATE: WidthS NUMERIC "32" -- Retrieval info: PRIVATE: WidthP NUMERIC "32" -- Retrieval info: PRIVATE: OptionalSum NUMERIC "0" -- Retrieval info: PRIVATE: AutoSizeResult NUMERIC "1" -- Retrieval info: PRIVATE: B_isConstant NUMERIC "0" -- Retrieval info: PRIVATE: SignedMult NUMERIC "1" -- Retrieval info: PRIVATE: USE_EAB NUMERIC "0" -- Retrieval info: PRIVATE: ConstantB NUMERIC "0" -- Retrieval info: PRIVATE: ValidConstant NUMERIC "1" -- Retrieval info: PRIVATE: Latency NUMERIC "1" -- Retrieval info: PRIVATE: aclr NUMERIC "0" -- Retrieval info: PRIVATE: LPM_PIPELINE NUMERIC "1" -- Retrieval info: PRIVATE: optimize NUMERIC "0" -- Retrieval info: CONSTANT: LPM_WIDTHA NUMERIC "16" -- Retrieval info: CONSTANT: LPM_WIDTHB NUMERIC "16" -- Retrieval info: CONSTANT: LPM_WIDTHP NUMERIC "32" -- Retrieval info: CONSTANT: LPM_WIDTHS NUMERIC "32" -- Retrieval info: CONSTANT: INPUT_B_IS_CONSTANT STRING "NO" -- Retrieval info: CONSTANT: LPM_REPRESENTATION STRING "SIGNED" -- Retrieval info: CONSTANT: USE_EAB STRING "OFF" -- Retrieval info: CONSTANT: LPM_PIPELINE NUMERIC "1" -- Retrieval info: USED_PORT: dataa 0 0 16 0 INPUT NODEFVAL dataa[15..0] -- Retrieval info: USED_PORT: result 0 0 32 0 OUTPUT NODEFVAL result[31..0] -- Retrieval info: USED_PORT: datab 0 0 16 0 INPUT NODEFVAL datab[15..0] -- Retrieval info: USED_PORT: clock 0 0 0 0 INPUT NODEFVAL clock -- Retrieval info: CONNECT: @dataa 0 0 16 0 dataa 0 0 16 0 -- Retrieval info: CONNECT: result 0 0 32 0 @result 0 0 32 0 -- Retrieval info: CONNECT: @datab 0 0 16 0 datab 0 0 16 0 -- Retrieval info: CONNECT: @clock 0 0 0 0 clock 0 0 0 0 -- Hong Kong Sent via Deja.com http://www.deja.com/ Before you buy.Article: 18710
In article <80871c$58t$1@nnrp1.deja.com>, Leslie Yip <leslie.yip@asmpt.com> writes > Hello all > >Can anyone solve the two problems found in ViewLogic's Workview office >7.5: > >(1) When I use FPGA Express (included in WVOffice 7.5) > to synthesize my design, it generates the EDIF file >for Altera's MAXplus to fit into the device. >The generated EDIF file flattens my bus (defined in VHDL) >, so when I do backannotation (post simulation), I have >to re-define my test bench in Speedwave (a VHDL simulator) for doing >backannotation. > >For example, I defined in VHDL: >FREQ:in std_logic_vector(35 downto 0); > >Then in the backannotation file, it becomes >FREQ35: in std_logic; >FREQ34: in std_logic; >FREQ33: in std_logic; >FREQ32: in std_logic; >FREQ31: in std_logic; >.. >FREQ1: in std_logic; >FREQ0: in std_logic; > >It will make every designer very upset to do much typing. >I have checked the files, it is the problem in EDIF file generated by >FPGA Express. > >Can you solve the bug? > This isn't a bug, it's a feature! (Sorry). You can try changing the bus style in the VHDL output settings of FPGA Express, but my experience was that it did not help. Also, I found that the latest version of FPGA Express (3.3) is better at getting busses right. To fix it in VHDL, you can write a configuration. Here is an example from our Comprehensive VHDL course design flow exercise. -- -- exflow/src/cfgxilinx_synopsysfe3.vhd -- -- Configuration for gate level simulation with output from Synopsys -- FPGA Express Version 3 (Viewlogic). -- -- Rev Author Date Comment -- ==================================================================== -- 001 APF 990219 Created ISSUE1.43 -- use work.all; configuration CfgGate of chiptb is for Bench -- test bench architecture name for all : chip use entity work.chip(Structure) -- (gate level architecture name) port map (nReset => nReset, Enable => Enable, Load => Load, UpDn => UpDn, Data0 => Data(0), Data1 => Data(1), Data2 => Data(2), Data3 => Data(3), Data4 => Data(4), Clk25M => Clk25M, TestClk => TestClk, nTestMode => nTestMode, nDispA0 => nDispA(0), nDispA1 => nDispA(1), nDispA2 => nDispA(2), nDispA3 => nDispA(3), nDispA4 => nDispA(4), nDispA5 => nDispA(5), nDispA6 => nDispA(6), nDispB0 => nDispB(0), nDispB1 => nDispB(1), nDispB2 => nDispB(2), nDispB3 => nDispB(3), nDispB4 => nDispB(4), nDispB5 => nDispB(5), nDispB6 => nDispB(6), nLed0 => nLed(0), nLed1 => nLed(1), nLed2 => nLed(2), nLed3 => nLed(3), nLed4 => nLed(4), nLed5 => nLed(5), nLed6 => nLed(6), nLed7 => nLed(7) ); end for; end for; end; As you can see, this one was actually written for Viewlogic! As I say, with the later versions, I didn't need it any more. Also, if your circuit changes too much, you may have to write a new configuration. >(2) Besides, FPGA Express can't synthesize the VHDL LPM generated >files, which are generated from Altera's MaxPlus. > >The error message is: (access violation) >FEXP caused an invalid page fault in >module DPMW.DLL at 0137:1005eec7. >Registers: This is one for Viewlogic - I have found them very helpful fixing bugs. kind regards Alan -- Alan Fitch DOULOS Ltd. Church Hatch, 22 Market Place, Ringwood, BH24 1AW, Hampshire, UK Tel: +44 (0)1425 471 223 Email: alan.fitch@doulos.com Fax: +44 (0)1425 471 573 ** Visit THE WINNING EDGE www.doulos.com **Article: 18711
On Thu, 14 Oct 1999 01:31:17 -0400, "abdulqadir alaqeeli" <alaqeeli@bobcat.ent.ohiou.edu> wrote: >Hi, >Does anyone know where I find a PCI Board with a Virtex Chip? >I found the following two companies: >Avnet: http://www.em.avnet.com/semi/marketing/xlx-19990526semw.html >Embedded Solutions Ltd ( RC1000-PP) > >I need any helpful information so I can choose a good board. > >Abdul > > > Try the company call NallaTech whose web site is http:/www.nallatech.com -- Replace NOJUNK with graham.millar to reply.Article: 18712
In article <7vum2f$lji$1@lure.pipex.net>, "david garnett" <dave.garnett@metapurple.co.uk> wrote: > I'm trying to make a bus master PCI interface with an AMCC5933 and a > Coolrunner PLD, and I'm having real problems getting dma transfers to work > for more than a few words - specifically, transfers of greater than about 16 > words usually hang the pci bus after a few (or many) transfers. Any help, > information or suggestions of a suitable ng would be very welcome ! > Personally, I wouldn't bother. I worked on a 5933 design 3 years ago. Several bugs in the 5933 resulted in the entire project being aborted. The most serious of these is that when operating in master mode, if the PCI target happens to request a retry on the last item of your burst then the following events happen. 1) Your FIFO is marked empty (which it is) 2) You load the next target address (because your're ready to do the next DMA and you think your FIFO is empty) 3) The 5933 uses THIS NEW ADDRESS to write the retried item which is sitting in an undocumented 'holding' register off the end of the FIFO - i.e. writes the last item of data for your last DMA burst to the first address of your new DMA. 4) The 5933 then goes and increments the target address register and delivers the rest of your DMA data to addr+4 etc. It is not possible to work around this bug without snooping the PCI bus (and consequently breaking the loading rules) to determine that retry has happened at the end of your DMA. I have not seen this bug either acknowledged by AMCC or reported fixed. I'd also be happy to see AMCC say that this bug has been found and fixed, but unless it has, your design will be doomed so far as DMA mastering is concerned. Sent via Deja.com http://www.deja.com/ Before you buy.Article: 18713
"Leslie Yip (/ Loui)" wrote: > > Hello all > > Can anyone solve the two problems found in ViewLogic's Workview office > 7.5: > > (1) When I use FPGA Express (included in WVOffice 7.5) > to synthesize my design, it generates the EDIF file > for Altera's MAXplus to fit into the device. > The generated EDIF file flattens my bus (defined in VHDL) > , so when I do backannotation (post simulation), I have > to re-define my test bench in Speedwave (a VHDL simulator) for doing > backannotation. > > For example, I defined in VHDL: > FREQ:in std_logic_vector(35 downto 0); > > Then in the backannotation file, it becomes > FREQ35: in std_logic; > FREQ34: in std_logic; > FREQ33: in std_logic; > FREQ32: in std_logic; > FREQ31: in std_logic; > .. > FREQ1: in std_logic; > FREQ0: in std_logic; > > It will make every designer very upset to do much typing. > I have checked the files, it is the problem in EDIF file generated by > FPGA Express. > > Can you solve the bug? > > (2) Besides, FPGA Express can't synthesize the VHDL LPM generated > files, which are generated from Altera's MaxPlus. > > The error message is: (access violation) > FEXP caused an invalid page fault in > module DPMW.DLL at 0137:1005eec7. > Registers: > EAX=02d0c074 CS=0137 EIP=1005eec7 EFLGS=00010206 > EBX=00000000 SS=013f ESP=0269fad0 EBP=00000000 > ECX=02eb8d78 DS=013f ESI=02dbfe08 FS=3977 > EDX=02f0c770 ES=013f EDI=02eb8d8c GS=0000 > Bytes at CS:EIP: > 8b 3b 52 e8 81 6b fb ff 83 c4 04 89 30 8b 44 24 > Stack dump: > 02ec6158 02ca1c08 02feaa08 00780d08 02d0c4ac 02eb8d78 02eb8d50 02f0c770 > 02f0c790 02eb8d78 00000000 02f0c750 02f0c820 02ca1c08 0000001c 00780d08 > > Below is the VHDL file (multip.vhd) > > -- megafunction wizard: %LPM_MULT% > -- GENERATION: STANDARD > -- VERSION: WM1.0 > -- MODULE: lpm_mult > > -- ============================================================ > -- File Name: multip.vhd > -- Megafunction Name(s): > -- lpm_mult > -- ============================================================ > -- ************************************************************ > -- THIS IS A WIZARD GENERATED FILE. DO NOT EDIT THIS FILE! > -- ************************************************************ > > -- Copyright (C) 1988-1999 Altera Corporation > > -- Any megafunction design, and related net list (encrypted or > decrypted), > -- support information, device programming or simulation file, and > any other > -- associated documentation or information provided by Altera or a > partner > -- under Altera's Megafunction Partnership Program may be used only > to > -- program PLD devices (but not masked PLD devices) from Altera. > Any other > -- use of such megafunction design, net list, support information, > device > -- programming or simulation file, or any other related > documentation or > -- information is prohibited for any other purpose, including, but > not > -- limited to modification, reverse engineering, de-compiling, or > use with > -- any other silicon devices, unless such use is explicitly > licensed under > -- a separate agreement with Altera or a megafunction partner. > Title to > -- the intellectual property, including patents, copyrights, > trademarks, > -- trade secrets, or maskworks, embodied in any such megafunction > design, > -- net list, support information, device programming or simulation > file, or > -- any other related documentation or information provided by > Altera or a > -- megafunction partner, remains with Altera, the megafunction > partner, or > -- their respective licensors. No other licenses, including any > licenses > -- needed under any third party's intellectual property, are > provided herein. > > LIBRARY ieee; > USE ieee.std_logic_1164.all; > LIBRARY lpm; > USE lpm.lpm_components.all; > > ENTITY multip IS > PORT > ( > dataa : IN STD_LOGIC_VECTOR (15 DOWNTO 0); > datab : IN STD_LOGIC_VECTOR (15 DOWNTO 0); > clock : IN STD_LOGIC ; > result : OUT STD_LOGIC_VECTOR (31 DOWNTO 0) > ); > END multip; > > ARCHITECTURE SYN OF multip IS > > SIGNAL sub_wire0 : STD_LOGIC_VECTOR (31 DOWNTO 0); > > COMPONENT lpm_mult > GENERIC ( > LPM_WIDTHA : NATURAL; > LPM_WIDTHB : NATURAL; > LPM_WIDTHP : NATURAL; > LPM_WIDTHS : NATURAL; > INPUT_B_IS_CONSTANT : STRING; > LPM_REPRESENTATION : STRING; > USE_EAB : STRING; > LPM_PIPELINE : NATURAL > ); > PORT ( > dataa : IN STD_LOGIC_VECTOR (15 DOWNTO 0); > datab : IN STD_LOGIC_VECTOR (15 DOWNTO 0); > clock : IN STD_LOGIC ; > result : OUT STD_LOGIC_VECTOR (31 DOWNTO 0) > ); > END COMPONENT; > > BEGIN > result <= sub_wire0(31 DOWNTO 0); > > lpm_mult_component : lpm_mult > GENERIC MAP ( > LPM_WIDTHA => 16, > LPM_WIDTHB => 16, > LPM_WIDTHP => 32, > LPM_WIDTHS => 32, > INPUT_B_IS_CONSTANT => "NO", > LPM_REPRESENTATION => "SIGNED", > USE_EAB => "OFF", > LPM_PIPELINE => 1 > ) > PORT MAP ( > dataa => dataa, > datab => datab, > clock => clock, > result => sub_wire0 > ); > > END SYN; > > -- ============================================================ > -- CNX file retrieval info > -- ============================================================ > -- Retrieval info: PRIVATE: WidthA NUMERIC "16" > -- Retrieval info: PRIVATE: WidthB NUMERIC "16" > -- Retrieval info: PRIVATE: WidthS NUMERIC "32" > -- Retrieval info: PRIVATE: WidthP NUMERIC "32" > -- Retrieval info: PRIVATE: OptionalSum NUMERIC "0" > -- Retrieval info: PRIVATE: AutoSizeResult NUMERIC "1" > -- Retrieval info: PRIVATE: B_isConstant NUMERIC "0" > -- Retrieval info: PRIVATE: SignedMult NUMERIC "1" > -- Retrieval info: PRIVATE: USE_EAB NUMERIC "0" > -- Retrieval info: PRIVATE: ConstantB NUMERIC "0" > -- Retrieval info: PRIVATE: ValidConstant NUMERIC "1" > -- Retrieval info: PRIVATE: Latency NUMERIC "1" > -- Retrieval info: PRIVATE: aclr NUMERIC "0" > -- Retrieval info: PRIVATE: LPM_PIPELINE NUMERIC "1" > -- Retrieval info: PRIVATE: optimize NUMERIC "0" > -- Retrieval info: CONSTANT: LPM_WIDTHA NUMERIC "16" > -- Retrieval info: CONSTANT: LPM_WIDTHB NUMERIC "16" > -- Retrieval info: CONSTANT: LPM_WIDTHP NUMERIC "32" > -- Retrieval info: CONSTANT: LPM_WIDTHS NUMERIC "32" > -- Retrieval info: CONSTANT: INPUT_B_IS_CONSTANT STRING "NO" > -- Retrieval info: CONSTANT: LPM_REPRESENTATION STRING "SIGNED" > -- Retrieval info: CONSTANT: USE_EAB STRING "OFF" > -- Retrieval info: CONSTANT: LPM_PIPELINE NUMERIC "1" > -- Retrieval info: USED_PORT: dataa 0 0 16 0 INPUT NODEFVAL dataa[15..0] > -- Retrieval info: USED_PORT: result 0 0 32 0 OUTPUT NODEFVAL > result[31..0] > -- Retrieval info: USED_PORT: datab 0 0 16 0 INPUT NODEFVAL datab[15..0] > -- Retrieval info: USED_PORT: clock 0 0 0 0 INPUT NODEFVAL clock > -- Retrieval info: CONNECT: @dataa 0 0 16 0 dataa 0 0 16 0 > -- Retrieval info: CONNECT: result 0 0 32 0 @result 0 0 32 0 > -- Retrieval info: CONNECT: @datab 0 0 16 0 datab 0 0 16 0 > -- Retrieval info: CONNECT: @clock 0 0 0 0 clock 0 0 0 0 > > -- > Hong Kong > > Sent via Deja.com http://www.deja.com/ > Before you buy. Hi, I think I can help you with your Netlist. After Chip-Optimization with FPGA-Express you can mark the optimized chip and push the right mouse button. There you find 'export netlist'. A window pops up where you can define the bus style of the exported netlist. Actually You have 'EXPAND'. Try another one which is appropriate to Max-Plus. Now press OK and the netlist is written once again but this time with the selected bus notation. After every synthesis you will have to do this step by yourself. I think there was a way to change the default settings but i don't know it anymore. Maybe another one knows about this. Bye, Michael mlschmid@iis.fhg.deArticle: 18714
<!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> Hello Anton, <br>there are a few solutions in my opinion. You can solve this task using <br>both gdf&tdf and tdf only based design entry. You wrote about AHDL <br>implementation only, but this is not the easiest way. <br>However, in fact, what you look for is a frequency divider with 2/3 <br>ratio. <br>My proposition is: build a frequency divider with 1/3 ratio and then <br>multiply it by 2 (or vice versa). <br>Another way is to create a special synchronous machine based on counter <br>with combinatorial feed-back circuitry forcing the counter to work with <br>an appropriated cycle. <br>Contact me if you are still interested in the solution. <p>PS. You send a question with no more details (e.g. there is no <br>information about duty cycle), so real solution will be the best when <br>design's constraints are fully defined. What is it to be, your design I <br>mean? Which and what version of Altera's software do you use? <p>Regards, <br>Jessie <p>Anton Erasmus wrote: <blockquote TYPE=CITE>Hi, <p>I am trying to get a 16MHz Frequency from a 24MHz Clock on a Altera <br>EPM7128. The 24MHz Clock is connected to one of the global clock <br>inputs. Can anyone help me in doing this in AHDL ? <p>Regards <br> Anton Erasmus</blockquote> </html>Article: 18715
Hello all, I'm looking for a VHDL/verilog model for a sample rate converter, probably for FPGA implementations. For e.g., I want to design a module for upconverting 44.1kHz/22.05kHz/11.025kHz audio samples to 48kHz. Could somebody help me in getting some pointers for this. Thanks in advance, -- Regards, -Srikanth. Texas Instruments, Inc. gurrapu@ti.com 972-480-2318Article: 18716
You can find a fairly comprehensive list of Virtex boards on The Programmable Logic Jump Station at http://www.optimagic.com/boards.html. -- ----------------------------------------------------------- Steven K. Knapp OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally" E-mail: sknapp@optimagic.com Web: http://www.optimagic.com ----------------------------------------------------------- Graham Millar <NOJUNK@gecm.com> wrote in message news:3827e9f4.437697816@news.geccs.gecm.com... > On Thu, 14 Oct 1999 01:31:17 -0400, "abdulqadir alaqeeli" > <alaqeeli@bobcat.ent.ohiou.edu> wrote: > > >Hi, > >Does anyone know where I find a PCI Board with a Virtex Chip? > >I found the following two companies: > >Avnet: http://www.em.avnet.com/semi/marketing/xlx-19990526semw.html > >Embedded Solutions Ltd ( RC1000-PP) > > > >I need any helpful information so I can choose a good board. > > > >Abdul > > > > > > > Try the company call NallaTech whose web site is > http:/www.nallatech.com > > -- > Replace NOJUNK with graham.millar to reply.Article: 18717
Thanks to all who have replied, or may still reply to my post.Article: 18718
I'm new to using ORCAD (I have tried mailing them/using their notice board with no response so far) - does anyone have any experience of using the Express synthesis for simpleplds (actually GAL22V10)? I have a little behavioural VHDL FSM which analyzes ok; when I try to build it Express generates an error message telling me I have too many product terms for a couple of rows. On inspecting the vhdl netlist its generated, there are actually not many PTs in the expressions its complaining about - well within the powers of a 22V10! Is there a common user error which could generate this? Or is this a known problem for which a patch is available? (I have v. 7.1) Thanks for any advice, Graham SeamanArticle: 18719
This is a multi-part message in MIME format. --------------DED6A3B2420B0CAB7336F407 Content-Type: multipart/alternative; boundary="------------7FB0D021144509B516D3919E" --------------7FB0D021144509B516D3919E Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Rick Filipkiewicz wrote: > Brian Philofsky wrote: > > > > > > > > > If you wish to simulate the design with logic level delays but not > > wire delays, this is possible if you run the synthesized netlist > > through the Xilinx M1 tools to the Map stage. Run the ncd file that > > Map outputs through ngdanno and then run the nga file from ngdanno > > through ngd2vhdl. The command sequence will look like: > > > > ngdbuild <design_netlist> > > map -o logic_sim.ncd <design>.ngd > > ngdanno logic_sim.ncd > > ngd2vhdl logic_sim.nga > > > > > > The output of ngd2vhdl will be a structual VHDL file and an SDF file > > containing only logic delays. > > > > > > > >> Hi, > >> > >> we would like to simulate an FPGA design after logic synthesis (for > >> Xilinx XC4036) but before doing any place and route. I.e. we want to > >> > >> simulate the delays of the logic, but not of the wires. > >> > >> Is there a way to do this (with design compiler and vss)? > >> > >> I understand that we can write a vhdl netlist from design compiler > >> after > >> synthesis. But this netlist is not simulatable, e.g. it contains > >> iob_4000 components for which we do not have simulation models. > >> > >> Any help on this would be appreciated, > >> > >> Joe > >> > > > > If you want to go via the MAP route you will have to compile & use the > Xilinx simulation primitive library ``simprims''. I don't know the state > of the VHDL one but I've found a fair number of problems with the > Verilog ones. At least in 2.1i the Verilog LUT models work but there are > still errors & issues in some of the others. > > For the other alternative of going via an FPGA Express netlist might > mean using the ``unisims'' library which, as far as I can see from my > Verilog experiance, is much cruder than the simprims one. The reason I suggested the Map route is because the original request was to simulate the logic delays but not routing delays. Most netlists produced by synthesis tools do not have any delay information therefore only a unit delay simulation would be possible. By processing the design through map, you will get true logic level delays but not routing delays. Performing a post-synthesis simulation is fine if you are interested in only a unit delay simulation and will save you a few steps through our tools although in general the netlist translation and map stage do not take too much CPU time. True you will have to use the SIMPRIM libraries with this approach rather than the UNISIM libraries however I don't understand the problems you mention. I successfully use both types of simulation models all the time for both VHDL and Verilog simulations. The only major things that come to mind that you may be referring to are in the 1.5 release of the LUT models (for Verilog), you must drive all design inputs or sourced LUTs will go to a simulation value of X. All that was needed to be done is to tie undriven design inputs to a known value. With the 2.1 release, that requirement has been lifted and you should only see X's on the outputs of undriven LUTs when that input has a direct correlation to the result. The only other problems I generally see when a global reset is not performed on the design before beginning simulation. The methodology for global reset has been changed for Verilog in the 2.1 release so that it will work better with pre-compiled libraries which seems to be the trend of most Verilog simulators today. This methodology is explained in our documentation. Without this global reset, it is possible the design might come up in a unknown state. I am not trying to say the Xilinx libraries are without fault however they should work with most simulation/design scenarios. There are always going to be corner cases which can not be easily accounted for and as those are encountered, we will try to fix them. If you would like to go into more detail as to the problems you mention, I would like to hear them. Although I am not directly involved with the simulation group, I can pass on any comments/experiences you or anyone else might have about the simulation libraries. I would like to change this perception that they are error-prone. Thanks, -- Brian Philofsky -- Xilinx Design Engineer -- ------------------------------------------------------------------- / 7\'7 Brian Philofsky (brian.philofsky@xilinx.com) \ \ ` Xilinx Design Engineer hotline@xilinx.com / / 2100 Logic Drive 1-800-255-7778 \_\/.\ San Jose, California 95124-3450 1-408-879-5199 ------------------------------------------------------------------- --------------7FB0D021144509B516D3919E Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> Rick Filipkiewicz wrote: <blockquote TYPE=CITE>Brian Philofsky wrote: <p>> <br>> <br>> <br>> If you wish to simulate the design with logic level delays but not <br>> wire delays, this is possible if you run the synthesized netlist <br>> through the Xilinx M1 tools to the Map stage. Run the ncd file that <br>> Map outputs through ngdanno and then run the nga file from ngdanno <br>> through ngd2vhdl. The command sequence will look like: <br>> <br>> ngdbuild <design_netlist> <br>> map -o logic_sim.ncd <design>.ngd <br>> ngdanno logic_sim.ncd <br>> ngd2vhdl logic_sim.nga <br>> <br>> <br>> The output of ngd2vhdl will be a structual VHDL file and an SDF file <br>> containing only logic delays. <br>> <br>> <br>> <br>>> Hi, <br>>> <br>>> we would like to simulate an FPGA design after logic synthesis (for <br>>> Xilinx XC4036) but before doing any place and route. I.e. we want to <br>>> <br>>> simulate the delays of the logic, but not of the wires. <br>>> <br>>> Is there a way to do this (with design compiler and vss)? <br>>> <br>>> I understand that we can write a vhdl netlist from design compiler <br>>> after <br>>> synthesis. But this netlist is not simulatable, e.g. it contains <br>>> iob_4000 components for which we do not have simulation models. <br>>> <br>>> Any help on this would be appreciated, <br>>> <br>>> Joe <br>>> <br>> <p>If you want to go via the MAP route you will have to compile & use the <br>Xilinx simulation primitive library ``simprims''. I don't know the state <br>of the VHDL one but I've found a fair number of problems with the <br>Verilog ones. At least in 2.1i the Verilog LUT models work but there are <br>still errors & issues in some of the others. <p>For the other alternative of going via an FPGA Express netlist might <br>mean using the ``unisims'' library which, as far as I can see from my <br>Verilog experiance, is much cruder than the simprims one.</blockquote> <p>The reason I suggested the Map route is because the original request was to simulate the logic delays but not routing delays. Most netlists produced by synthesis tools do not have any delay information therefore only a unit delay simulation would be possible. By processing the design through map, you will get true logic level delays but not routing delays. Performing a post-synthesis simulation is fine if you are interested in only a unit delay simulation and will save you a few steps through our tools although in general the netlist translation and map stage do not take too much CPU time. <p>True you will have to use the SIMPRIM libraries with this approach rather than the UNISIM libraries however I don't understand the problems you mention. I successfully use both types of simulation models all the time for both VHDL and Verilog simulations. The only major things that come to mind that you may be referring to are in the 1.5 release of the LUT models (for Verilog), you must drive all design inputs or sourced LUTs will go to a simulation value of X. All that was needed to be done is to tie undriven design inputs to a known value. With the 2.1 release, that requirement has been lifted and you should only see X's on the outputs of undriven LUTs when that input has a direct correlation to the result. The only other problems I generally see when a global reset is not performed on the design before beginning simulation. The methodology for global reset has been changed for Verilog in the 2.1 release so that it will work better with pre-compiled libraries which seems to be the trend of most Verilog simulators today. This methodology is explained in our documentation. Without this global reset, it is possible the design might come up in a unknown state. <p>I am not trying to say the Xilinx libraries are without fault however they should work with most simulation/design scenarios. There are always going to be corner cases which can not be easily accounted for and as those are encountered, we will try to fix them. <p>If you would like to go into more detail as to the problems you mention, I would like to hear them. Although I am not directly involved with the simulation group, I can pass on any comments/experiences you or anyone else might have about the simulation libraries. I would like to change this perception that they are error-prone. <p>Thanks, <br> <p>-- Brian Philofsky <br>-- Xilinx Design Engineer <br> <br> <pre>-- ------------------------------------------------------------------- / 7\'7 Brian Philofsky (brian.philofsky@xilinx.com) \ \ ` Xilinx Design Engineer hotline@xilinx.com / / 2100 Logic Drive 1-800-255-7778 \_\/.\ San Jose, California 95124-3450 1-408-879-5199 -------------------------------------------------------------------</pre> </html> --------------7FB0D021144509B516D3919E-- --------------DED6A3B2420B0CAB7336F407 Content-Type: text/x-vcard; charset=us-ascii; name="brianp.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Brian Philofsky Content-Disposition: attachment; filename="brianp.vcf" begin:vcard n:Philofsky;Brian tel;fax:(408) 879-4442 tel;work:1-800-255-7778 x-mozilla-html:TRUE org:<BR><H1 ALIGN="CENTER"><img src="http://www.xilinx.com/images/xlogoc.gif" alt="Xilinx" ALIGN="CENTER"> Design Center version:2.1 email;internet:brianp@xilinx.com title:<H3 ALIGN="CENTER"><img src="http://bennyhills.fortunecity.com/deadparrot/108/homer.gif" alt="Homer" align="center"> Design Engineer adr;quoted-printable:;;2100 Logic Drive=0D=0ADept. 2510;San Jose;CA;95124-3450;USA x-mozilla-cpt:;25776 fn:<H3 ALIGN="CENTER">Brian Philofsky end:vcard --------------DED6A3B2420B0CAB7336F407--Article: 18720
On Tue, 09 Nov 1999 10:21:32 -0600, Srikanth Gurrapu <gurrapu@ti.com> scribbled: >Hello all, > >I'm looking for a VHDL/verilog model for a sample rate converter, >probably for FPGA implementations. For e.g., I want to design >a module for upconverting 44.1kHz/22.05kHz/11.025kHz audio samples to >48kHz. > >Could somebody help me in getting some pointers for this. >Thanks in advance, > >-- >Regards, >-Srikanth. >Texas Instruments, Inc. >gurrapu@ti.com >972-480-2318 > > > i suppose you could start with AD1892 data sheet MartinArticle: 18721
Srikanth , Sounds big - big fifo etc... Once you have the 44.1K to 48K upconversion the others are easy since they are covered by the least-common denominator for the 2 larger numbers (divide by 2 or 4). Try: http://www.analog.com/publications/whitepapers/products/AD1890.html for theory of operation & background on polyphase filtering & related sample conversion. Gd. luck, Bill Bill Groves Sachem Acoustics R&D Srikanth Gurrapu <gurrapu@ti.com> wrote in message news:38284A06.404A4C5A@ti.com... > Hello all, > > I'm looking for a VHDL/verilog model for a sample rate converter, > probably for FPGA implementations. For e.g., I want to design > a module for upconverting 44.1kHz/22.05kHz/11.025kHz audio samples to > 48kHz. > > Could somebody help me in getting some pointers for this. > Thanks in advance, > > -- > Regards, > -Srikanth. > Texas Instruments, Inc. > gurrapu@ti.com > 972-480-2318 > > >Article: 18722
This is a multi-part message in MIME format. --------------14E1AB51380722305725271D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit test --------------14E1AB51380722305725271D Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Brian Boorman Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Brian Boorman n: Boorman;Brian org: Harris RF Communications adr: 1680 University Ave;;;Rochester;NY;14610;USA email;internet: XYZ.bboorman@harris.com title: Electrical Design Engineer tel;work: (716) 242-3744 tel;fax: (716) 242-3150 x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard --------------14E1AB51380722305725271D--Article: 18723
I have decided to use the CAN bus for a variety of control projects. I would appreciate any suggestions. Thanks in advance, Arnold BelandArticle: 18724
This is a multi-part message in MIME format. --------------9983C74A8453C89DFACA2AC0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello, Can U please mention where I would be able to find what kind of Fitter algorithms are being used to map the Netlists into the FPGA architecture for Xilinx family.. I am interested in finding out how the fitter algorithm actually transfers the netlist into the device.. If there are any relevant web pages regarding this, can you please let me know. Thanks and regards Anup --------------9983C74A8453C89DFACA2AC0 Content-Type: text/x-vcard; charset=us-ascii; name="anup.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for anup kumar raghavan Content-Disposition: attachment; filename="anup.vcf" begin:vcard n:Anup Kumar;Raghavan tel;home:0061-7-38761962 tel;work:0061-7-33658849 x-mozilla-html:FALSE url:www.csee.uq.edu.au org:University of Queensland;Computer Science and Electrical Engineering adr:;;;;;;Australia version:2.1 email;internet:anup@elec.uq.edu.au fn:Anup Kumar end:vcard --------------9983C74A8453C89DFACA2AC0--
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