Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMar2019

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

Custom Search

Messages from 52350

Article: 52350
Subject: Re: Interfacing to a PC using EPP parallel port
From: "Blake" <nobody@nowhere.com>
Date: Fri, 07 Feb 2003 06:56:44 GMT
Links: << >>  << T >>  << A >>
I'm going to try again...

Check out our QuickUSB module at www.quickusb.com.  Its designed for
applications just like you're describing.  It gives you a high-speed
parallel port input into your PC without having to learn USB.

Regards,
Blake

"Ernest Jamro" <jamro@agh.edu.pl> wrote in message
news:3E39483A.3090702@agh.edu.pl...
> I have done a similar project and it does not work on
> every computer.
>
> 1. EPP signals are often not TTL compatible on some computers in EPP mode,
> 2. PCs motherboard chip set (or PP cable) does not work correctly on
> some PCs (e.g. when the transferred data change from 0x00 to 0xFF
> the EPP data_stobeN goes high even when waitN signal is still low)
> 3. some PC does not support EPP mode at all.
> 4. Transfer depends strongly on a PC you've got.
>
> You need not bother about the DMA - use a standard memory transfer
> unless your PC must do some other critical calculation.
>
> Good luck anyway
>
> Ernest Jamro
>
> Bob Fischer wrote:
> > I will be testing an FPGA design that is intended to drive a PC for
> > initial checkout and later to an embedded computer using parallel
> > port.  I selected the EPP protocol as it looks like it can support
> > what I need to do.
> >
> > The FPGA will output 10 bytes of data to the PC each cycle of
> > operation.  The data consists of five 14 bit values output in two
> > bytes each.  The FPGA will be performing about 40,000 cycles per
> > second.  Think of each cycle as a 25 us frame.  Data collection (about
> > 4 us), processing (about 7-8 us) occurs for the first 11-12 us of each
> > frame.  When the data is ready the parallel port Interrupt line is
> > asserted.
> >
> > The burst rate during the available 13 us data output portion is
> > around 770 Khz.  The times have already been verified in the
> > simulations.  For the simulation I used an 800 ns byte cycle.  The
> > testbench emulates the PC by responding to the Interrupt, invoking the
> > byte cycle timing as expected from the PC by cycling the Data Strobe
> > line (400 ns low then 400 ns high for each byte).  The FPGA responds
> > with Waits and presentation of data bytes at the time defined for the
> > EPP port.  I used the timing found in web site
> > www.beyondlogic.org/epp/epp.htm
> >
> > The output of the FPGA is configured for TTL levels, slow transitions.
> >  I intend to pipe the FPGA directly to the DB connector and through a
> > 3 ft parallel cable to the PC parallel port.
> >
> > In the PC we will DMA the data to memory and accumulate it for several
> > seconds.  A display program will access that memory and generate
> > graphs, etc for visual analysis of the performance and results.
> >
> > Does this approach to PC interfaceing sound feasible?  Has anyone out
> > there any prior experience they would like to share?  Some Do's and
> > Don'ts?
> >
> > Bob Fischer
> > FPGA independent designer
>



Article: 52351
Subject: Re: USB2 or firewire or 100Mb ethernet link to FPGA design
From: "Blake" <nobody@nowhere.com>
Date: Fri, 07 Feb 2003 06:57:13 GMT
Links: << >>  << T >>  << A >>
The web site is at www.quickusb.com.

Blake

"Nobody" <nobody@nowhere.com> wrote in message
news:CGI0a.3289$ta2.1025@nwrddc01.gnilink.net...
> Terry,
>
> Check out our QuickUSB module and kits.  It is designed to do exactly what
> you're looking for.
>
> Thanks,
> Blake
>
> "Theron Hicks (Terry)" <hicksthe@egr.msu.edu> wrote in message
> news:3E431E87.905ECF6F@egr.msu.edu...
> > Hi,
> >     I am looking at design upgrade of an existing instrumentation
> > project.  Currently the design talks over a high speed bus to a rather
> > expensive (~$1600 US) parrallel digital input board.  The data rate
> > would be on the order of 50M bits per second not including any
> > overhead.  I do not think I want to go to an all FPGA based solution.
> > Currently the FPGA in the system is the smallest Spartan2e series
> > device.  By the way, quantities are very small, on the order of less
> > than 50 pieces per year.
> >
> >     So here are my questions...
> >
> > 1.    Does anyone happen to know what the USB2 or firewire is rated for
> > in terms of the longest cable length?  I had thought USB2 was limited to
> > about 2 meters but I have seen USB cables about 5 meters long recently.
> > A longer length interconnect would appear to be desireable.  I know that
> > ethernet is good for several hundred meters.  That would be far beyond
> > my needs.
> >
> > 2.    Has anyone had any experience with either USB2 or 100mB or
> > firewire as an interconnect to an FPGA based design?
> >
> > 3.    Do you happen to have any recommendations as to a possible
> > off-the-shelf solution (either a small board or a 1 or two chip
> > solution, ideally something with a demo board available)?
> >
> > Thanks,
> > Theron Hicks
> >
>
>



Article: 52352
Subject: Byteblaster II
From: "Thorsten Bunte" <t.bunte@beckhoff.de>
Date: Fri, 7 Feb 2003 08:15:11 +0100
Links: << >>  << T >>  << A >>
Hello,

is there a schematic available for Alteras Byteblaster II.
In contrast to Byteblaster and Byteblaster MV the datasheet of Byteblaster
II does not show any schematic of the circuit.

Thanks,
Thorsten



Article: 52353
Subject: Multicontext FPGA
From: "RC" <rado3@poczta.onet.pl>
Date: Fri, 7 Feb 2003 09:41:51 +0100
Links: << >>  << T >>  << A >>
Hello,

Can someone tell me which company sells reconfigurable multicontext FPGA
devices?

Thanks in advance,
Regards,
Rado



Article: 52354
Subject: blockram initialization
From: "Alphaboran" <alphaboran@yahoo-no-spam.com>
Date: Fri, 7 Feb 2003 13:08:15 +0200
Links: << >>  << T >>  << A >>
Hello all,

I generate a blockram by using the Xilinx core generator. I edited a coe
file for the memory initalization values and works fine, if I want to change
the  values of the blockram can I change the values directly in the edf file
or I must generate the core again but with other coe file?

Thanks in advance.

Best Regards,
Harris



Article: 52355
Subject: Re: USB2 or firewire or 100Mb ethernet link to FPGA design
From: Aurash Lazarut <aurash@xilinx.com>
Date: Fri, 07 Feb 2003 11:23:06 +0000
Links: << >>  << T >>  << A >>
Terry,

You can use EDK ( embedded dev. kit) and spartan2e (xc2s400e or
xc2s600e) and build a small system around MicroBlaze
with Ethernet 10/100 MAC and use BRAM for the SW (tcp/ip stack) I've
used this "receipe" and it works fine.

Aurash

"Theron Hicks (Terry)" wrote:
> 
> Hi,
>     I am looking at design upgrade of an existing instrumentation
> project.  Currently the design talks over a high speed bus to a rather
> expensive (~$1600 US) parrallel digital input board.  The data rate
> would be on the order of 50M bits per second not including any
> overhead.  I do not think I want to go to an all FPGA based solution.
> Currently the FPGA in the system is the smallest Spartan2e series
> device.  By the way, quantities are very small, on the order of less
> than 50 pieces per year.
> 
>     So here are my questions...
> 
> 1.    Does anyone happen to know what the USB2 or firewire is rated for
> in terms of the longest cable length?  I had thought USB2 was limited to
> about 2 meters but I have seen USB cables about 5 meters long recently.
> A longer length interconnect would appear to be desireable.  I know that
> ethernet is good for several hundred meters.  That would be far beyond
> my needs.
> 
> 2.    Has anyone had any experience with either USB2 or 100mB or
> firewire as an interconnect to an FPGA based design?
> 
> 3.    Do you happen to have any recommendations as to a possible
> off-the-shelf solution (either a small board or a 1 or two chip
> solution, ideally something with a demo board available)?
> 
> Thanks,
> Theron Hicks

-- 
 __
/ /\/\ Aurelian Lazarut
\ \  / System Verification Engineer
/ /  \ Xilinx Ireland
\_\/\/
 
phone:	353 01 4032639
fax:	353 01 4640324

Article: 52356
Subject: Re: blockram initialization
From: Aurash Lazarut <aurash@xilinx.com>
Date: Fri, 07 Feb 2003 11:41:44 +0000
Links: << >>  << T >>  << A >>
Harris,

Yes it's possible to change in the edif file but depending on your bram
"geometry" it's very likely to be not "very obvious" what (and how) you
should change. By the other hand your edif it will not match the XCO +
COE file and if you regenerate the core you'll lose the changes. My
recommendation is to edit the *.coe and regenerate.

PS. if you want to change the content of the BRAM on the fly inside your
design (for testing purpose) you can take advantage of the second port
(if is available) and put together a small design with PicoBlaze + uart
and "refresh" the blockram content.  

Aurash

Alphaboran wrote:
> 
> Hello all,
> 
> I generate a blockram by using the Xilinx core generator. I edited a coe
> file for the memory initalization values and works fine, if I want to change
> the  values of the blockram can I change the values directly in the edf file
> or I must generate the core again but with other coe file?
> 
> Thanks in advance.
> 
> Best Regards,
> Harris

-- 
 __
/ /\/\ Aurelian Lazarut
\ \  / System Verification Engineer
/ /  \ Xilinx Ireland
\_\/\/
 
phone:	353 01 4032639
fax:	353 01 4640324

Article: 52357
Subject: Virtex-II Pro PowerPC cache memory as main program/data storage?
From: serebr@mailandnews.com (Valeri Serebrianski)
Date: 7 Feb 2003 03:42:28 -0800
Links: << >>  << T >>  << A >>
In Virtex-II Pro datasheet stated that PowerPC processor has 16 KByte
of instruction and 16 KByte of data cache onboard. For my application
that amount of memory is more than enough (processor speed is far more
valuable).
Is this possible to store both the program and data tables solely in
those caches without any use of Virtex-II Pro SelectRAM or external
RAM? It assumes that initial content for both caches will be stored in
main Virtex-II Pro configuration RAM and will be loaded at startup.

Valeri Serebrianski.

Article: 52358
Subject: Re: NIOS and ACEX1K
From: "Nial Stewart" <nial@spamno.nialstewart.co.uk>
Date: Fri, 7 Feb 2003 12:14:32 -0000
Links: << >>  << T >>  << A >>
> Hi
> I'm trying to implement a NIOS CPU on a ACEX1K100-1 from Altera.
> The only problem that I have is the fmax that I can use; I need 33MHz
while
> Quartus II is able to use a fmax of 25Mhz.
> ACEX1K family should be able to support  PCI interface ( 33Mhz and 64Mhz)
so
> I don't understand while the fmax is 25Mhz.
> Can somebody help me?.
> Thanks

You're comparing two completely different things.

The PCI interfaces are probably highly optimised for speed.

The NIOS is a prarameterisable module (I presume) that the tools will
do the best they can with.

What clock constraints are you using?

I've a poster "Nios 2.0 New Features" which lists the 32
bit reference design as running at 37MHz and the 16 bit
reference at 43 MHz. Have you tried the reference designs
for comparison?

Have you added much user logic/any custom instructions to
your design?

Nial.

------------------------------------------------
Nial Stewart Developments Ltd
FPGA and High Speed Digital Design
www.nialstewartdevelopments.co.uk






Article: 52359
Subject: Quartus II problems
From: "Roger" <rogerwilson@hotmail.com>
Date: Fri, 7 Feb 2003 12:33:25 -0000
Links: << >>  << T >>  << A >>
I've been having some problems with Quartus II 2.2 and I'd be interested to
hear if anyone else has had similar experiences:

1) For a while I have been successfully using the Quartus II simulator but
recently a strange phenomenon has stopped me. I define the input waveforms
in the .vwf file and run the simulator. All appears to run OK but the output
waveforms in the Simulator messages window do not appear correctly. For
example, the input clock signal just appears to be low all the time.
Scrolling in time is only possible by dragging the marker along - using the
left or right arrows doesn't work. Changes in the input waveforms do occur
when these random positions are visited. It's as if I'm zoomed in to a tiny
resolution and can only see a small area of the waveforms. Trying to zoom
out has no effect.

2) Quartus II hangs up now and again (Win XP professional, P4 1.8GHz, 384MB)
sometimes when the compiler is running, sometimes during simulation.
Re-installing from the CD as changed nothing. It would be neat to blame all
this on file corruption but I don't believe it is.

Any suggestions anyone?

Thanks.

Rog.



Article: 52360
Subject: Partial Reconfiguration - Virtex-E
From: Rainer Schmidt <rainersc@hni.upb.de>
Date: Fri, 07 Feb 2003 14:17:08 +0100
Links: << >>  << T >>  << A >>
Hi,

I would like to do Partial Reconfiguration on a Virtex-E. I need a bus 
macro for the communication between the reconfigurable modules. Xilinx 
Application note XAPP290 says, that Xilinx provides these macros. 
Unfortunately the example of this application note only contains bus 
macros for Virtex and Virtex-II FPGAs. These don't work for me.

Can anybody tell me, where I can find the bus macro for Virtex-E?

Thanks,
Rainer


Article: 52361
Subject: Carry Save Adder
From: Lars Unger <larsu@ida.ing.tu-bs.de>
Date: Fri, 7 Feb 2003 14:17:14 +0100
Links: << >>  << T >>  << A >>
Hi there,

I am working on a realization of a Carry-Save-Adder for a Xilinx Virtex device.
Up to now i was able to implement it using 2 LCs per Bit. However I read in the
data book that arithmetic logic allows a 1-bit full adder per LC. This works
fine for a Ripple-Carry-Adder. Is there any way to utilize the carry logic for
the Carry-Save-Adder, too ?

Best wishes,
Lars.
-- 
GnuPG public key:
http://www.ida.ing.tu-bs.de/~larsu/larsu_ida_ing_tu-bs_de.key

Article: 52362
Subject: Re: low pass FIR filter in FPGA
From: news@rtrussell.co.uk
Date: Fri, 7 Feb 2003 13:23:56 +0000 (UTC)
Links: << >>  << T >>  << A >>
In comp.arch.fpga Garry Allen <garrya@ihug.com.au> wrote:

: input data is 9 bit samples at 33.325714 MHz (or possibly 31.104 MHz)
: 1 dB ripple allowed in pass band (0 - 7MHz), minimum of 65 dB down in
: the stop band. (8 MHz)

I agree with the estimate of about 70 taps.  My favorite filter
synthesis tool (proprietory) gave the following figures:

  Sampling frequency = 33.325714 MHz
  Number of taps = 64
  Pass-band ripple (0 - 7 MHz) = 0.986 dB
  Stop-band attenuation (>8 MHz) = -65.2 dB

This is with a non-quantised filter.  As another post correctly
remarked, 9-bits is nothing like enough for a 65 dB dynamic
range (that would need about 11 bits).  You also need to check
whether the ripple spec. is peak or peak-peak.

Richard.
http://www.rtrussell.co.uk/

Article: 52363
Subject: Re: NIOS and ACEX1K
From: "Mahesh M. Bandi" <mab77+@pitt.edu>
Date: Fri, 7 Feb 2003 08:26:14 -0500
Links: << >>  << T >>  << A >>

Hi,

Just a suggestion, have you tried using FPGA Advantage instead of Quartus
II?

Thank You,
Best Regards,
Mahesh

On Thu, 6 Feb 2003, Giaccaglini Giorgio wrote:

> Date: Thu, 6 Feb 2003 12:49:38 +0100
> From: Giaccaglini Giorgio <g.giaccaglini@libero.it>
> Newsgroups: comp.arch.fpga
> Subject: NIOS and ACEX1K
> 
> Hi
> 
> I'm trying to implement a NIOS CPU on a ACEX1K100-1 from Altera.
> 
> The only problem that I have is the fmax that I can use; I need 33MHz while
> Quartus II is able to use a fmax of 25Mhz.
> 
> ACEX1K family should be able to support  PCI interface ( 33Mhz and 64Mhz) so
> I don't understand while the fmax is 25Mhz.
> 
> Can somebody help me?.
> 
> Thanks
> 
> 
> 
> 


Article: 52364
Subject: Divide clock frequency by 1.5: output duty cycle is not 50%
From: llaa57@yahoo.com (llaa57)
Date: 7 Feb 2003 05:32:37 -0800
Links: << >>  << T >>  << A >>
I implemented the circuit described in the application note "Unusual
clock dividers" written by Peter Alfke:
http://www.xilinx.com/xcell/xl33/xl33_30.pdf.
I used a Xilinx XC9536 and the input clock is generated by an
oscillator (SCO-061S 48MHz) by Sunny.
My problem is that the output clock's duty cycle is 33% and not 50% as
expected. Why? Is the CPLD unsuitable for this circuit?

Many thanks in advance.

Article: 52365
Subject: Re: USB2 or firewire or 100Mb ethernet link to FPGA design
From: "Theron Hicks" <hicksthe@egr.msu.edu>
Date: Fri, 7 Feb 2003 09:34:04 -0500
Links: << >>  << T >>  << A >>
Blake,
    Looks like a potential...
1.    Can I get a hold of a parts list?  I am wondering what it would cost
to implement the hardware on my circuit board if I buy your license pack.
2.    I notice the information references an Altera FPGA.  Is there anything
inherently requiring an Altera part for the application?  (For example,
altera specific intelectual property.)
3.    As you folks seem to be expert, can you tell me what the maximum
length for USB2 is?
4.    If I buy your daughter card, is the card USB2 certified?  In other
words can I get away with using the USB2 logo?  This is not critical, just a
question.

Thanks,
Theron Hicks

"Blake" <nobody@nowhere.com> wrote in message
news:dNI0a.3299$ta2.1475@nwrddc01.gnilink.net...
> The web site is at www.quickusb.com.
>
> Blake
>
> "Nobody" <nobody@nowhere.com> wrote in message
> news:CGI0a.3289$ta2.1025@nwrddc01.gnilink.net...
> > Terry,
> >
> > Check out our QuickUSB module and kits.  It is designed to do exactly
what
> > you're looking for.
> >
> > Thanks,
> > Blake
> >
> > "Theron Hicks (Terry)" <hicksthe@egr.msu.edu> wrote in message
> > news:3E431E87.905ECF6F@egr.msu.edu...
> > > Hi,
> > >     I am looking at design upgrade of an existing instrumentation
> > > project.  Currently the design talks over a high speed bus to a rather
> > > expensive (~$1600 US) parrallel digital input board.  The data rate
> > > would be on the order of 50M bits per second not including any
> > > overhead.  I do not think I want to go to an all FPGA based solution.
> > > Currently the FPGA in the system is the smallest Spartan2e series
> > > device.  By the way, quantities are very small, on the order of less
> > > than 50 pieces per year.
> > >
> > >     So here are my questions...
> > >
> > > 1.    Does anyone happen to know what the USB2 or firewire is rated
for
> > > in terms of the longest cable length?  I had thought USB2 was limited
to
> > > about 2 meters but I have seen USB cables about 5 meters long
recently.
> > > A longer length interconnect would appear to be desireable.  I know
that
> > > ethernet is good for several hundred meters.  That would be far beyond
> > > my needs.
> > >
> > > 2.    Has anyone had any experience with either USB2 or 100mB or
> > > firewire as an interconnect to an FPGA based design?
> > >
> > > 3.    Do you happen to have any recommendations as to a possible
> > > off-the-shelf solution (either a small board or a 1 or two chip
> > > solution, ideally something with a demo board available)?
> > >
> > > Thanks,
> > > Theron Hicks
> > >
> >
> >
>
>



Article: 52366
Subject: Re: USB2 or firewire or 100Mb ethernet link to FPGA design
From: "Theron Hicks" <hicksthe@egr.msu.edu>
Date: Fri, 7 Feb 2003 09:54:48 -0500
Links: << >>  << T >>  << A >>
Aurash,
   Good idea but 2 problems.

1.    The cost for the chip is about $110 compared to about $15 for the
xc2s50e that I am using.  This may be a minor cost differential when I add
the cost of a fully integrated ethernet solution.  However, I suspect that
an integrated solution can be had for substantially less than $95 US.

2.    The real show stopper for me is the package for this part.  It is a
BGA package.  I know that you folks say that the BGA is no big deal but for
me it is.  I am just now getting used to fine pitch leaded packages.

However, thanks for the response.  I will keep it in mind if the rest of the
responses are less ideal.

Thanks,
Theron Hicks


"Aurash Lazarut" <aurash@xilinx.com> wrote in message
news:3E43971A.72669198@xilinx.com...
> Terry,
>
> You can use EDK ( embedded dev. kit) and spartan2e (xc2s400e or
> xc2s600e) and build a small system around MicroBlaze
> with Ethernet 10/100 MAC and use BRAM for the SW (tcp/ip stack) I've
> used this "receipe" and it works fine.
>
> Aurash
>
> "Theron Hicks (Terry)" wrote:
> >
> > Hi,
> >     I am looking at design upgrade of an existing instrumentation
> > project.  Currently the design talks over a high speed bus to a rather
> > expensive (~$1600 US) parrallel digital input board.  The data rate
> > would be on the order of 50M bits per second not including any
> > overhead.  I do not think I want to go to an all FPGA based solution.
> > Currently the FPGA in the system is the smallest Spartan2e series
> > device.  By the way, quantities are very small, on the order of less
> > than 50 pieces per year.
> >
> >     So here are my questions...
> >
> > 1.    Does anyone happen to know what the USB2 or firewire is rated for
> > in terms of the longest cable length?  I had thought USB2 was limited to
> > about 2 meters but I have seen USB cables about 5 meters long recently.
> > A longer length interconnect would appear to be desireable.  I know that
> > ethernet is good for several hundred meters.  That would be far beyond
> > my needs.
> >
> > 2.    Has anyone had any experience with either USB2 or 100mB or
> > firewire as an interconnect to an FPGA based design?
> >
> > 3.    Do you happen to have any recommendations as to a possible
> > off-the-shelf solution (either a small board or a 1 or two chip
> > solution, ideally something with a demo board available)?
> >
> > Thanks,
> > Theron Hicks
>
> --
>  __
> / /\/\ Aurelian Lazarut
> \ \  / System Verification Engineer
> / /  \ Xilinx Ireland
> \_\/\/
>
> phone: 353 01 4032639
> fax: 353 01 4640324



Article: 52367
Subject: Re: Partial Reconfiguration - Virtex-E
From: Ray Andraka <ray@andraka.com>
Date: Fri, 07 Feb 2003 15:10:27 GMT
Links: << >>  << T >>  << A >>
The Virtex macro should work in virtexE

Rainer Schmidt wrote:

> Hi,
>
> I would like to do Partial Reconfiguration on a Virtex-E. I need a bus
> macro for the communication between the reconfigurable modules. Xilinx
> Application note XAPP290 says, that Xilinx provides these macros.
> Unfortunately the example of this application note only contains bus
> macros for Virtex and Virtex-II FPGAs. These don't work for me.
>
> Can anybody tell me, where I can find the bus macro for Virtex-E?
>
> Thanks,
> Rainer

--
--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
401/884-7930     Fax 401/884-7950
email ray@andraka.com
http://www.andraka.com

 "They that give up essential liberty to obtain a little
  temporary safety deserve neither liberty nor safety."
                                          -Benjamin Franklin, 1759



Article: 52368
Subject: Re: Carry Save Adder
From: Ray Andraka <ray@andraka.com>
Date: Fri, 07 Feb 2003 15:14:09 GMT
Links: << >>  << T >>  << A >>
Not without using or blocking the second half of the slice.   What is the
motivation for doing a carry save adder here?  The ripple carry arithmetic uses
half the area (because of the carry chain logic) and a tree made of FPGA ripple
carry adds is faster than one made of carry save adds because of the hit you take
on the routing.  See my multipliers page on my website for a bit more detail (under
wallace tree multipliers).

Lars Unger wrote:

> Hi there,
>
> I am working on a realization of a Carry-Save-Adder for a Xilinx Virtex device.
> Up to now i was able to implement it using 2 LCs per Bit. However I read in the
> data book that arithmetic logic allows a 1-bit full adder per LC. This works
> fine for a Ripple-Carry-Adder. Is there any way to utilize the carry logic for
> the Carry-Save-Adder, too ?
>
> Best wishes,
> Lars.
> --
> GnuPG public key:
> http://www.ida.ing.tu-bs.de/~larsu/larsu_ida_ing_tu-bs_de.key

--
--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
401/884-7930     Fax 401/884-7950
email ray@andraka.com
http://www.andraka.com

 "They that give up essential liberty to obtain a little
  temporary safety deserve neither liberty nor safety."
                                          -Benjamin Franklin, 1759



Article: 52369
Subject: FFT Size and speed
From: "gallenm" <gallenm@ic24.net>
Date: Fri, 7 Feb 2003 16:02:02 -0000
Links: << >>  << T >>  << A >>
Hi

Can anyone give me an estimate of the size of
a 1024 pt FFT and the maximum frequency that
it can run ?

Will it fit on a 200K logic element fpga which does not
have any onboard multipliers ?

Is their any good rules of thumb for estimating the size of
these FFTs ?

Thanks
Alan




Article: 52370
Subject: Re: Partial Reconfiguration - Virtex-E
From: Rainer Schmidt <rainersc@gmx.de>
Date: Fri, 07 Feb 2003 17:26:34 +0100
Links: << >>  << T >>  << A >>
I tried the Virtex macro and it worked after I patched the family and 
device field in the .nmc file.

Thanks for your response!
Rainer

Ray Andraka wrote:

> The Virtex macro should work in virtexE
> 
> Rainer Schmidt wrote:
> 
> 
>>Hi,
>>
>>I would like to do Partial Reconfiguration on a Virtex-E. I need a bus
>>macro for the communication between the reconfigurable modules. Xilinx
>>Application note XAPP290 says, that Xilinx provides these macros.
>>Unfortunately the example of this application note only contains bus
>>macros for Virtex and Virtex-II FPGAs. These don't work for me.
>>
>>Can anybody tell me, where I can find the bus macro for Virtex-E?
>>
>>Thanks,
>>Rainer
>>
> 
> --
> --Ray Andraka, P.E.
> President, the Andraka Consulting Group, Inc.
> 401/884-7930     Fax 401/884-7950
> email ray@andraka.com
> http://www.andraka.com
> 
>  "They that give up essential liberty to obtain a little
>   temporary safety deserve neither liberty nor safety."
>                                           -Benjamin Franklin, 1759
> 
> 
> 


Article: 52371
Subject: Re: Group Multiple tables
From: vbetz@altera.com (Vaughn Betz)
Date: 7 Feb 2003 08:28:16 -0800
Links: << >>  << T >>  << A >>
Tullio Grassi <tullio@umd.edu> wrote in message news:<3E3EE388.9080604@umd.edu>...
> Roberto Gallo wrote:
> >     Hello there,
> > 
> >     I am desiging a device that uses table lookups on processing of
> > information (Writing an AES core). However  as many as sixteen 256x8 bits
> > tables should be used. These tables are equal in content, however they have
> > independent indexing.
> >     I was wondering about how to group them, so that I could use FPGAs
> > memory intead of registers. Is it possible to use a multi-output memory? How
> > to?
> >     I am using a APEX20K200 with Quartus II 2.0 and Leonardo and writing in
> > VHDL.
> > 
> >     Thank you,
> >         Roberto Gallo.
> > 
> > 
> 
> I think that LUTs fit particularly well the xilinx
> distributed RAM. This feature is not available
> on Altera devices (I've heard xiinx has a patent on it).
> 
> -- 
> 
> Tullio Grassi
> 
> ======================================
> Univ. of Maryland - Dept. of Physics
> College Park, MD 20742 - US
> Tel +1 301 405 5970
> Fax +1 301 699 9195
> ======================================


In the device you're in (20K family), you should use one ESB per
table, so you'll need 16 ESBs.  The 20K family has only one read port
per ESB, so you have to use one ESB per table.

In Stratix, the M4K RAMs have two read/write ports, so for a ROM like
this you can use both ports to do reads.  This means you can fit two
of your tables in each M4K RAM, and you would only need 8 M4Ks.  APEX
II also has rams with two read/write ports, so you can use this trick
to use only 8 RAMs in it too.

Distributed RAM would not be a good solution for this application. 
Each LUT in distributed RAM has one read port, and only implements a
16x1 RAM.  To build a 256 x 8 memory would take 16 * 8 = 128 LUTs just
to get enough memory bits, and you would also need an extra LUT for
each bit of width to do the final bit of address muxing.  So 128 + 8 =
136 LUTs per table, and you would again need 16 tables, since you only
have one read port per ROM.  That's 2176 LUTs.  You're much better off
using the dedicated RAM structures.

The other posters had good summaries of how to instantiate the RAMs
you need.

Vaughn Betz

Article: 52372
Subject: Re: Divide clock frequency by 1.5: output duty cycle is not 50%
From: "John_H" <johnhandwork@mail.com>
Date: Fri, 07 Feb 2003 16:45:01 GMT
Links: << >>  << T >>  << A >>
Consider that your 48MHz clock has a period just over 20ns.  If your
oscillator has a 50% duty cycle then you have edges (rising or falling) that
can produce activity in synchronous logic at just over 10 ns intervals.
This 10 ns value is the resolution that you can apply to your generated
edges.

To have a divide by 1.5, your desired frequency of 32MHz will have a period
of just over 30 ns.  Since your edge placement can only be on the 10 ns
boundaries, you cannot get a 50% duty cycle.

The only way to get a divide by 1.5 with 50% duty cycle is to use a PLL or
DLL which isn't part of your CPLD or have another means - such as a delay
line - to shift an edge by 5 ns or so.  It isn't recommended that you try to
build up 5 ns of delay inside the CPLD because delay times will change with
temperature, voltage, manufacturing batch, phase of the moon and what not.



"llaa57" <llaa57@yahoo.com> wrote in message
news:5fde4855.0302070532.7af08737@posting.google.com...
> I implemented the circuit described in the application note "Unusual
> clock dividers" written by Peter Alfke:
> http://www.xilinx.com/xcell/xl33/xl33_30.pdf.
> I used a Xilinx XC9536 and the input clock is generated by an
> oscillator (SCO-061S 48MHz) by Sunny.
> My problem is that the output clock's duty cycle is 33% and not 50% as
> expected. Why? Is the CPLD unsuitable for this circuit?
>
> Many thanks in advance.



Article: 52373
Subject: LFSR: Galois and Fibonacci
From: chris.p.ward@ntlworld.com (Chris Ward)
Date: 7 Feb 2003 08:46:28 -0800
Links: << >>  << T >>  << A >>
I am correct in thinking Galois and Fibonacci LFSRs are mathematically
equivalent?

If I have a generator polynomial and I implement the two architectures
do I need to do anything different to obtain the same sequences from
the two?

Thanks
Chris

Article: 52374
Subject: Re: USB2 or firewire or 100Mb ethernet link to FPGA design
From: "John_H" <johnhandwork@mail.com>
Date: Fri, 07 Feb 2003 16:51:24 GMT
Links: << >>  << T >>  << A >>
Are you looking at just a direct point-to-point connection between your
proprietary hardware?  If you use USB2 or Ethernet, there's overhead
involved that you don't need if all you're doing is piping data.

"Theron Hicks (Terry)" <hicksthe@egr.msu.edu> wrote in message
news:3E431E87.905ECF6F@egr.msu.edu...
> Hi,
>     I am looking at design upgrade of an existing instrumentation
> project.

[snip]





Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMar2019

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

Custom Search