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
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

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 13100

Article: 13100
Subject: Re: Xilinx COREgen and Leonardo troubles...
From: Le mer Michel <michel.lemer@ago.fr>
Date: Mon, 16 Nov 1998 10:18:09 +0100
Links: << >>  << T >>  << A >>
Jacob W Janovetz wrote:

> Hello,
>
>    I've "built" a multiplier (10-bit by 8-bit) component
> from COREgen (for Solaris -- Linux would be preferred, by the way).
> I get the component output as mult10x8.xnf.
>
>    Now, I try to instantiate that in VHDL under Leonardo.
> I get the following error from 'ngdbuild':
>
> WARNING:basnb:79 - Pin mismatch between block "themult", TYPE="mult10x8", and
>    file "<MY_DIRECTORY>/mult10x8.ngo" at pin
>    "prod(0)".  Please make sure that all pins on the instantiated component
>    match pins in the lower-level design block.  (Pin-name matching is
>    case-insensitive.)
>
> prod(0) is the LSB of the product pin.  Now, I've read a few
> things about this error (namely solution 2234) in the help.  It
> says to change the bus config under logiblox.ini.  However, I look
> at the coregen.ini file under $COREGEN/wkg and there is nothing to
> address this.  In addition, there are no setup menus in COREgen to
> help me out.
>
>    What needs to be done here?  If someone knows what's going on,
> I can send my VHDL, XNF, EDIF files or what not to show how I'm
> doing things.  Thanks for any assistance.
>
>     Cheers,
>     Jake
>
> --
>    janovetz@uiuc.edu    | Once you have flown, you will walk the earth with
>  University of Illinois | your eyes turned skyward, for there you have been,
>                         | there you long to return.     -- da Vinci
>         PP-ASEL         | http://www.ews.uiuc.edu/~janovetz/index.html

Hello.

I use Coregen and Galileo (now same vendor as Leonardo bur less powerfull).
I use the file.vhx whith Galileo, file.vhd for functionnal simulation and the
file.xnf with M1.4. Galileo sees only an empty entity (without architecture) and
the M1.4 links the components. No problems.
Anyway, do you have problem with bus only or with all signals?
May be the way for writing bus is different between coregen and leonardo (see
option). Do you use <> or ()  or thomesing else for bus? (data<0>, data(0),
...).

Good
luck.
Michel.

Article: 13101
Subject: Re: Xilinx COREgen and Leonardo troubles...
From: Mark Snook <mark.snook@arm.com>
Date: Mon, 16 Nov 1998 09:38:44 +0000
Links: << >>  << T >>  << A >>
Jacob W Janovetz wrote:
>    I've "built" a multiplier (10-bit by 8-bit) component
> from COREgen (for Solaris -- Linux would be preferred, by the way).
> I get the component output as mult10x8.xnf.
> 
>    Now, I try to instantiate that in VHDL under Leonardo.
> I get the following error from 'ngdbuild':
> 
> WARNING:basnb:79 - Pin mismatch between block "themult", TYPE="mult10x8", and
>    file "<MY_DIRECTORY>/mult10x8.ngo" at pin
>    "prod(0)".  Please make sure that all pins on the instantiated component
>    match pins in the lower-level design block.  (Pin-name matching is
>    case-insensitive.)

I believe the problem here is due to a mismatch between the CoreGen
netlist (XNF) and the output of your synthesis tool. Although you think
your pin names are the same, one tool writes the pins as a bus and the
other as individual ports. There are a number of ways to approach this,
but I expect your synthesis tool has an option to write the netlist with
ports expanded rather than as busses. However then you need to watch out
for the way that the tool expands busses. For example one tool may be
writing busa(0) and the other busa<0> or busa_0.

A solution that I have used with Exemplar Leonardo is to load the
appropriate library, read in the CoreGen XNF file and write it out again
as a VHDL netlist. The VHDL can then be read in when you analyze the
rest of the VHDL for your project. Make sure you set a don't touch on
the CoreGen block so that it is not optimized with the rest of the
design. Finally the design that is written out as everything you need
and you don't rely on the Xilinx merger to get the port mapping right.

This may sound like a pain, but you usually only need to do the XNF to
VHDL netlist conversion once for each CoreGen model.

I am interested to here if anyone has a better solution.

Mark
Article: 13102
Subject: Re: Big-Endian vs Little-Endian
From: yves@px.uk.com (Yves Tchapda)
Date: Mon, 16 Nov 1998 09:45:35 GMT
Links: << >>  << T >>  << A >>

Hi Shail,
Technically speaking, there is no  advantage of Big-Endianness over
Little-Endianness. Intel processors are predominantly Little-Endian,
while Motorola processors are Big-Endian. Choosing one over the other
depends primarily on the overall system. If you are designing a
subsystem which will be connected to an Intel platform, it makes sense
to choose  Little-Endianness. It is important to note that although
many bridges (e.g., PCI-to-local bus) have on the fly byte-swapping,
this only maintains address invariance. If a program running on one
32-bit platform tries to access 32-bit data produced by another
program running on  another 32-bit platform with different endianness,
you still have to perform explicit conversion. Address invariance
means that, when accessing an aggregate data structure (e.g., record),
the relative address of the fields are maintained. If latency isn't a
problem, it doesn't matter which ever Endianness you choose.

To reply to your second point, not all RISC processors are Big-Endian.
The i960 family of processors are Little-Endian although some of them
have the capability of Endian conversion.

Dr Yves Tchapda
Design Engineer
Power X
England
Article: 13103
Subject: Re: Example of clock circuit needed !
From: msimon@tefbbs.com
Date: Mon, 16 Nov 1998 10:24:48 GMT
Links: << >>  << T >>  << A >>
You are thinking of a phase adder in a technique called DDS or Direct
Digital Synthesis.

Simon
=============================
Le mer Michel <michel.lemer@ago.fr> wrote:

>ovilup wrote:
>
>> Hello !
>>
>> I am working on an I2C controller. Now, I am designing the
>> internal clock generator. I have an 1.5 MHz internal clock,
>> from which I have to generate the 100 KHz, 90 KHz, 44 KHz
>> 1.5 KHz SCL clocks.
>>
>> Any examples of such an clock generator would be appreciated !
>>
>> Thank you in advance.
>> OL
>
>Hello
>
>I do not know exactly what it is but I heard about the direct numeric
>synthesis. It is use in the signal generators to privide a wave of a
>specific frequency.
>
>Bye.
>Michel.
>

Design Your Own MicroProcessor(tm) http://www.tefbbs.com/spacetime/index.htm
Article: 13104
Subject: Re: Example of clock circuit needed !
From: Brian Boorman <XZY.bboorman@harris.com>
Date: Mon, 16 Nov 1998 09:07:23 -0500
Links: << >>  << T >>  << A >>

--------------2E0E5DD694A1EA8757417BEC
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

You don't mention whether you need all these clock rates at same time, or one at
a time. Also what is the acceptable % error on those clock rates? 100 Khz is 1.5
Mhz divided by 15, you can get 88.23 Khz dividing by 17, 44.11 Khz by dividing
by 34. 1.5 Khz by dividing by 1000. If you only need one clock at a time, use a
programmable counter where a bit-vector selects what counter preload value to
use. For getting 50% duty cycle from odd divisor, check out my article in the
August 15, 1997 issue of EDN magazine, page 73.

Brian C. Boorman
Harris RF Communications
Rochester, NY 14610
XYZ.bboorman@harris.com
<Remove the XYZ. for valid address>

Le mer Michel <michel.lemer@ago.fr> wrote:

> >ovilup wrote:
> >
> >> Hello !
> >>
> >> I am working on an I2C controller. Now, I am designing the
> >> internal clock generator. I have an 1.5 MHz internal clock,
> >> from which I have to generate the 100 KHz, 90 KHz, 44 KHz
> >> 1.5 KHz SCL clocks.
> >>
> >> Any examples of such an clock generator would be appreciated !
> >>
> >> Thank you in advance.
> >> OL
>

--------------2E0E5DD694A1EA8757417BEC
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<HTML>
You don't mention whether you need all these clock rates at same time,
or one at a time. Also what is the acceptable % error on those clock rates?
100 Khz is 1.5 Mhz divided by 15, you can get 88.23 Khz dividing by 17,
44.11 Khz by dividing by 34. 1.5 Khz by dividing by 1000. If you only need
one clock at a time, use a programmable counter where a bit-vector selects
what counter preload value to use. For getting 50% duty cycle from odd
divisor, check out my article in the August 15, 1997 issue of EDN magazine,
page 73.

<P>Brian C. Boorman
<BR>Harris RF Communications
<BR>Rochester, NY 14610
<BR>XYZ.bboorman@harris.com
<BR>&lt;Remove the XYZ. for valid address>

<P>Le mer Michel &lt;michel.lemer@ago.fr> wrote:
<BLOCKQUOTE TYPE=CITE>>ovilup wrote:
<BR>>
<BR>>> Hello !
<BR>>>
<BR>>> I am working on an I2C controller. Now, I am designing the
<BR>>> internal clock generator. I have an 1.5 MHz internal clock,
<BR>>> from which I have to generate the 100 KHz, 90 KHz, 44 KHz
<BR>>> 1.5 KHz SCL clocks.
<BR>>>
<BR>>> Any examples of such an clock generator would be appreciated !
<BR>>>
<BR>>> Thank you in advance.
<BR>>> OL
<BR><A HREF="http://www.tefbbs.com/spacetime/index.htm"></A>&nbsp;</BLOCKQUOTE>
</HTML>

--------------2E0E5DD694A1EA8757417BEC--

Article: 13105
Subject: Re: Xilinx COREgen and Leonardo troubles...
From: Brian Boorman <XZY.bboorman@harris.com>
Date: Mon, 16 Nov 1998 09:18:44 -0500
Links: << >>  << T >>  << A >>
Leonardo has a property called EDIF_WRITE_ARRAYS, which can be set to TRUE or
FALSE. You can also specify how Leonardo writes out bus identifiers (i.e. <> ()
[]). Xilinx tools do accept EDIF netlists. Try looking at this method as a solution
to your problem.

Mark Snook wrote:


> I believe the problem here is due to a mismatch between the CoreGen
> netlist (XNF) and the output of your synthesis tool. Although you think
> your pin names are the same, one tool writes the pins as a bus and the
> other as individual ports. There are a number of ways to approach this,
> but I expect your synthesis tool has an option to write the netlist with
> ports expanded rather than as busses. However then you need to watch out
> for the way that the tool expands busses. For example one tool may be
> writing busa(0) and the other busa<0> or busa_0.
>
>
> Mark



--
Brian C. Boorman
Harris RF Communications
Rochester, NY 14610
XYZ.bboorman@harris.com
<Remove the XYZ. for valid address>


Article: 13106
Subject: Re: Xilinx COREgen and Leonardo troubles...
From: gd@nospam.heliontech.com (Graeme Durant)
Date: Mon, 16 Nov 1998 14:33:00 GMT
Links: << >>  << T >>  << A >>
On 14 Nov 1998 04:04:50 GMT, janovetz@ews.uiuc.edu (Jacob W Janovetz)
wrote:

>Hello,
>
>   I've "built" a multiplier (10-bit by 8-bit) component
>from COREgen (for Solaris -- Linux would be preferred, by the way).
>I get the component output as mult10x8.xnf.
>
>   Now, I try to instantiate that in VHDL under Leonardo.  
>I get the following error from 'ngdbuild':
>
>WARNING:basnb:79 - Pin mismatch between block "themult", TYPE="mult10x8", and
>   file "<MY_DIRECTORY>/mult10x8.ngo" at pin
>   "prod(0)".  Please make sure that all pins on the instantiated component
>   match pins in the lower-level design block.  (Pin-name matching is
>   case-insensitive.)
>


I think your problem here is to do with the differing bus naming
conventions used by Xilinx tools and Exemplar.  I have not tried using
CoreGen macros with Leonardo, but I have used LogiBlox cells and
experienced similar problems.  In LogiBlox, first you have to select
'Mentor' from the vendor type.  This ensures that the right sort of
bus numbering brackets are used (eg. <> or [] or () ).  CoreGen has
only one output bus option which seems to be set correctly for
Exemplar with <> brackets.  

Second, you must use the Exemplar '-nobus' switch during synthesis.
This expands all busses into bunches of 'single wires' in the netlist,
and is the format in which Xilinx tools work.  If you don't do this,
the netlist will have traditional busses, and the CoreGen generated
block will not; thus causing the mismatch.

I guess the latter requirement is the cause of your problems.

Hope that helps.

Graeme Durant
HELION Technology Limited - Programmable Logic Consultants
The Granary, Home End, Fulbourn, Cambridge CB1 5BS.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
mailto:gd@heliontech.com
http://www.heliontech.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Article: 13107
Subject: Re: Xilinx COREgen and Leonardo troubles...
From: janovetz@ews.uiuc.edu (Jacob W Janovetz)
Date: 16 Nov 1998 14:52:44 GMT
Links: << >>  << T >>  << A >>

Mark,

  Over the weekend, I was able to get the tools to cooperate.  I
just wrote a script to revise the .xnf files from COREgen into the
right format.  And I can still do all this without using a GUI.
Why?  BECAUSE I'M NOT USING WINDOWS!!!!!  Yyeeeee Haaa!
  (Sorry, I love UNIX...)

   Cheers,
   Jake



Mark Snook <mark.snook@arm.com> writes:

>Jacob W Janovetz wrote:
>>    I've "built" a multiplier (10-bit by 8-bit) component
>> from COREgen (for Solaris -- Linux would be preferred, by the way).
>> I get the component output as mult10x8.xnf.
>> 
>>    Now, I try to instantiate that in VHDL under Leonardo.
>> I get the following error from 'ngdbuild':
>> 
>> WARNING:basnb:79 - Pin mismatch between block "themult", TYPE="mult10x8", and
>>    file "<MY_DIRECTORY>/mult10x8.ngo" at pin
>>    "prod(0)".  Please make sure that all pins on the instantiated component
>>    match pins in the lower-level design block.  (Pin-name matching is
>>    case-insensitive.)

>I believe the problem here is due to a mismatch between the CoreGen
>netlist (XNF) and the output of your synthesis tool. Although you think
>your pin names are the same, one tool writes the pins as a bus and the
>other as individual ports. There are a number of ways to approach this,
>but I expect your synthesis tool has an option to write the netlist with
>ports expanded rather than as busses. However then you need to watch out
>for the way that the tool expands busses. For example one tool may be
>writing busa(0) and the other busa<0> or busa_0.

>A solution that I have used with Exemplar Leonardo is to load the
>appropriate library, read in the CoreGen XNF file and write it out again
>as a VHDL netlist. The VHDL can then be read in when you analyze the
>rest of the VHDL for your project. Make sure you set a don't touch on
>the CoreGen block so that it is not optimized with the rest of the
>design. Finally the design that is written out as everything you need
>and you don't rely on the Xilinx merger to get the port mapping right.

>This may sound like a pain, but you usually only need to do the XNF to
>VHDL netlist conversion once for each CoreGen model.

>I am interested to here if anyone has a better solution.

>Mark
--
   janovetz@uiuc.edu    | Once you have flown, you will walk the earth with
 University of Illinois | your eyes turned skyward, for there you have been,
                        | there you long to return.     -- da Vinci
        PP-ASEL         | http://www.ews.uiuc.edu/~janovetz/index.html
Article: 13108
Subject: Re: placement&routing problems
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Mon, 16 Nov 1998 17:14:17 +0200
Links: << >>  << T >>  << A >>
> A few of points:
> 
> 1. One of the improvements to Par/v1.5 is that ram blocks are recognised
> as such, and are supposedly placed in a sensible manner. This should
> remove the need for RLOC'ing.

  This is right. We have installed Service Pack and we haven't met
  any such problems now.
 
> 2. It sounds like Xilinx have suggested the -ir map option to get round
> some mapping bug's in version 1.5. However, if the -ir option is used
> relative placement information is lost, so RLOC'ing wouldn't get you
> anywhere.  I suggest getting the M1.5 service pack anyway.

  So, if I don't meed those errors, then I don't need to use RLOC'ing,
  do I?
 
> 3. If the clb's with rams are intended to also have flip-flops, it might
> be worth while checking in EPIC that they are both going in the same
> clb.

  The FF's in CLB's are not used in RAM implementations, as far as I
  have understood Xilinx Data Book 1998. In EPIC Editor, when I have
  looked inside CLB's, it seems that edge-triggering has been implemented
  either in F-G Generators or somewhere on chip.

> 4. Contrary to popular belief, structured hierarchical relative
> placement is possible with VHDL too. In fact it's easier and quicker.

  Frankly speaking, I am not interested in VHDL-Verilog difference.

  The problem is, if my big amount of RAM can be a drawback for routing
  (we have core, which consists of a logic of 600-650 CLB's and internal
  RAM's of 1000-1200 CLB's; so the amount of RAM is ca. %55 of core but
  not any device resources).

  Now my question is, is such an amount of RAM is really a drawback for
  placement&routing? One engineer from Canada told me that such a big
  amount of internal RAM is beyond the limits of Xilinx internal resource
  capabilities, since it needs astronomical resources. But we have never
  met some information on RAM concept for Xilinx technology. If you know
  any good information occasionally, that we seem to have missed it,
  would you please inform me?

  The 1000-1200 CLB's is not a single block RAM but is made up of
  several 32x8, 128x1, 128x3, 128x7 and 128x8 dual-port RAM's and 64x8
  synchronous single-port RAM's. It was not necessary to implement internal
  RAM's in this way. One of the internal RAM structure had to be implemented
  as 128x40 storage structure but we thought it would be a good way to
  partition it by instantiating 5 times 128x8 dual-port RAM modules.
  It is not so easy to redesign RAM organization. If possible, I want
  to have your experiences about RAM organizations and those effects on
  placement&routing. Since a placement&routing efforts take at least 3-4
  days on Ultra-10/300 MHz/128 Mb, your experiences will be invaluable.
  Would you please share your RAM organization concepts and their influences
  on placement&routing?

  What we have learned are: Design Manager is smart enough to place internal
  RAM's according to minimal routing requirements. We have observed that
  128x1 dual-port RAM structures have been placed as much "square-based"
  as possible (we have seen it at EPIC Editor).

  We have mapped our chip (600-650 CLB's for logic, 1000-1200 CLB's for internal
  RAM's) into several devices and here are the results (it took several
  weeks to get these results):

Device          Speed   Packet  Total CLB's     Mapped  Amount  Unrouted 
		Grade		of device	CLB's		connections
========================================================================
XC4062XL        -3      HQ304   2304 (48x48)    2570    %111    2-2000 
XC4085XLA       -09     HQ304   3136 (56x56)    2580    %85     2-1000
XC40110XV       -09     BG352   4096 (64x64)    2580    %63     2
XC40150XV       -09     BG352   5184 (72x72)    2580    %50     2
XC40200XV       -09     BG432   7056 (84x84)    2590    %37     2
XC40250XV       -09     BG432   8464 (92x92)    2600    %30     2

  Some people also recommended to use Floorplanner. How can we place 128x1?
  Is the tool smart enough to "recognize" our placement? Does it really improve
  the performance significantly? We must expert to use Floorplaner, I think,
  and even for such a complicated chip.

  Utku
Article: 13109
Subject: Re: Looking for a good documentation on FPGA
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Mon, 16 Nov 1998 17:26:59 +0200
Links: << >>  << T >>  << A >>
Steven K. Knapp wrote:
> 
> You can find some tutorial information via The Programmable Logic Jump
> Station at http://www.optimagic.com/tutorial.html.

  this link doesn't work

  utku

> 
> Also, you may want to check out The Programmable Logic Bookstore at
> http://www.optimagic.com/books.html#VHDL.
> 
> -----------------------------------------------------------
> Steven K. Knapp
> OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally"
> E-mail:  sknapp@optimagic.com
>    Web:  http://www.optimagic.com
> -----------------------------------------------------------
> 
> ovilup wrote in message <01be0d93$78946fa0$4162e2c1@timteh.dnttm.ro>...
> >Dear all,
> >
> >I am looking for a good documentation on programming FPGA's and
> >ASIC's in VHDL. I have good aknowledges of VHDL, and I intend to learn
> >more about FPGA's and ASIC's.
> >
> >Thank you.
> >
Article: 13110
Subject: Xilinx 4k programming
From: serard@lslsun.epfl.ch (Sebastien Erard)
Date: 16 Nov 1998 16:28:07 +0100
Links: << >>  << T >>  << A >>
Hi,

I have to programme some xc40125xv from a 68360. The FPGAs are in
synchronous paripheral mode. D[7..0] from the FPGAs are connected
to D[31..24] from the 68k. CCLK is generated from a MAX7128, as the
others control signals (/prog and /init). /init has an external
pull up tied to 3v3.

To programm, I do:

- set /prog and /init down,
- wait a few us (about 50 us),
- release /init and set /prog high,
- wait for /init to go high,
- send a first CCLK pulse (320ns high),
- put an 8 bits data on the bus and send 8 CCLK pulses
  (data are remove before the seconde pulse, pulses are generated by
   a control register in the MAX7128),
- look if /init has gone down (no wait),
- send the next byte (last datas are already in, due to 68k timmings)
- and so on (send byte and look /init) for complete data (349630 bytes)

- if /init goes down during the process (error!), stop it and show how
  many bytes were sent

- at the end, look if done pin is high (fed up)


and I have a probleme: after 100-150 bytes (it allways changes), the 
process stop, due to a low /init. and I just don't understand why.
If someone could help me, I would be very gratefull, because it's 
for my diploma project and I'm stuck there and I don't know how to
go on.

Thanks every body, Sebastien
Article: 13111
Subject: Re: Xilinx Foundation vs. Altera Max Plus
From: "Steve" <reply.through.newsgroup@paranoid.com>
Date: Mon, 16 Nov 1998 15:35:26 GMT
Links: << >>  << T >>  << A >>
glenn kubota wrote in message <364E1429.91883AE8@earthlink.net>...
>i've got a project where i'm looking at using CPLDs from either Altera
>or Xilinx.  it looks like they both have chips that will fit the bill.
>my question, however, is what're the relative pros and cons of Xilinx
>Foundation vs. Altera Max Plus? what's easier to learn?  what will serve
>me better in the future?
>
>thanks,
>glenn k.
>
>

I've been using both vendors for mainly CPLD designs.  They both have
their strengths in both tools and silicon, and they don't remain static.
That
said here are some of my thoughts.

Tools:
Maxplus II is easier to use, but Xilinx has made big improvements since
introducing foundation, and Altera is not without their weaknesses.  If you
are doing any serious VHDL design you will find the Altera VHDL somewhat
limited.  Since adopting FPGA Express, Xilinx has pulled ahead here.  Of
course if you have the budget, they both support 3rd party synthesis tools.
Also you will find the Altera simulator pretty basic.  I chose to buy my own
VHDL simulator anyway, so that's not an issue for me.

Chips:
The 7K parts vs the 9500's is a bit of a saw off.  The 7K has some blazingly
fast clock rates, but the 9500's product term sharing system can make for
some very nice wide logic functions.  The 7K also has a neat feature where
you can bypass the logic and get sub 3ns setup times!  Have to watch the
newer parts though.  It looks like the 9500XL's may have addressed the
low clock frequency problem.


Good Luck
Steve



Article: 13112
Subject: Re: Looking for a good documentation on FPGA
From: Jer-Sheng Chen <chenjs@ece.nwu.edu>
Date: Mon, 16 Nov 1998 10:06:08 -0600
Links: << >>  << T >>  << A >>
Try http://www.optimagic.com/tutorials.html


On Mon, 16 Nov 1998, Utku Ozcan wrote:

> Date: Mon, 16 Nov 1998 17:26:59 +0200
> From: Utku Ozcan <ozcan@netas.com.tr>
> Newsgroups: comp.arch.fpga, comp.cad.synthesis, comp.lang.vhdl
> Subject: Re: Looking for a good documentation on FPGA
> 
> Steven K. Knapp wrote:
> > 
> > You can find some tutorial information via The Programmable Logic Jump
> > Station at http://www.optimagic.com/tutorial.html.
> 
>   this link doesn't work
> 
>   utku
> 
> > 
> > Also, you may want to check out The Programmable Logic Bookstore at
> > http://www.optimagic.com/books.html#VHDL.
> > 
> > -----------------------------------------------------------
> > Steven K. Knapp
> > OptiMagic, Inc. -- "Great Designs Happen 'OptiMagic'-ally"
> > E-mail:  sknapp@optimagic.com
> >    Web:  http://www.optimagic.com
> > -----------------------------------------------------------
> > 
> > ovilup wrote in message <01be0d93$78946fa0$4162e2c1@timteh.dnttm.ro>...
> > >Dear all,
> > >
> > >I am looking for a good documentation on programming FPGA's and
> > >ASIC's in VHDL. I have good aknowledges of VHDL, and I intend to learn
> > >more about FPGA's and ASIC's.
> > >
> > >Thank you.
> > >
> 
> 

Article: 13113
Subject: Is there an alternative to Altera EPM5128 OTP?
From: "Arnold Beland" <acbel@worldnet.att.net>
Date: 16 Nov 1998 16:41:01 GMT
Links: << >>  << T >>  << A >>
I have been using Altera's EPM5128 devices since they came out.  I am using
increasing quantities of them at $55 a pop.  Is there a cheaper alternative?

Arnold Beland
acbel@worldnet.att.net



Article: 13114
Subject: Re: Big-Endian vs Little-Endian
From: shailbains007@my-dejanews.com
Date: Mon, 16 Nov 1998 17:18:00 GMT
Links: << >>  << T >>  << A >>
In article <364ff418.2246316251@firewall>,
  yves@px.uk.com (Yves Tchapda) wrote:
>
> Hi Shail,
> Technically speaking, there is no  advantage of Big-Endianness over
> Little-Endianness. Intel processors are predominantly Little-Endian,
> while Motorola processors are Big-Endian. Choosing one over the other
> depends primarily on the overall system.

I'm not totally convinced. Why do Intel designers choose an Endianness that
is, well, counter-intutive? Take a look at an x86 memory dump to know what I
mean...  Are you sure there's no difference in the gate counts required to
make a system Small or Big Endian? Maybe one of them requires much fewer
gates to implement...

--Shail Bains
Silicon Automation Systems Ltd.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    
Article: 13115
Subject: Re: Xilinx 4k programming
From: fliptron@netcom.com (Philip Freidin)
Date: Mon, 16 Nov 1998 17:42:55 GMT
Links: << >>  << T >>  << A >>

After the leading string of eight '1' bits, is the pattern 0,0,1,0 (see
page 4-49 of the 1/98 data book). then 24 length count bits, then 4 more
'1's. Total of 40 bits of header. If you are clocking everything
correctly, you should see the exact same stuff with a 1 clock delay on the
DOUT pin (right next to the DIN pin). Check that the header is coming out
and it looks correct. The out going header is EXACTLY what the chip
thought it got as an incomming header, so if it is incorrect, then the
chip internally has not got a valid header.

The random failure position suggests a clocking problem, such as a 
non-clean rising and falling edge of the clock. You need to check this 
with a 300MHz (or faster) scope. Remember to check the falling edge, as a 
glitch on the falling edge can look like a rising edge. Given your 
problem description, this seems quite likely.

Philip Freidin.


In article <36504487.0@dinews.epfl.ch> serard@lslsun.epfl.ch (Sebastien Erard) writes:
>Hi,
>
>I have to programme some xc40125xv from a 68360. The FPGAs are in
>synchronous paripheral mode. D[7..0] from the FPGAs are connected
>to D[31..24] from the 68k. CCLK is generated from a MAX7128, as the
>others control signals (/prog and /init). /init has an external
>pull up tied to 3v3.
>
>To programm, I do:
>
>- set /prog and /init down,
>- wait a few us (about 50 us),
>- release /init and set /prog high,
>- wait for /init to go high,
>- send a first CCLK pulse (320ns high),
>- put an 8 bits data on the bus and send 8 CCLK pulses
>  (data are remove before the seconde pulse, pulses are generated by
>   a control register in the MAX7128),
>- look if /init has gone down (no wait),
>- send the next byte (last datas are already in, due to 68k timmings)
>- and so on (send byte and look /init) for complete data (349630 bytes)
>
>- if /init goes down during the process (error!), stop it and show how
>  many bytes were sent
>
>- at the end, look if done pin is high (fed up)
>
>
>and I have a probleme: after 100-150 bytes (it allways changes), the 
>process stop, due to a low /init. and I just don't understand why.
>If someone could help me, I would be very gratefull, because it's 
>for my diploma project and I'm stuck there and I don't know how to
>go on.
>
>Thanks every body, Sebastien


Article: 13116
Subject: Re: Big-Endian vs Little-Endian
From: muzok@nospam.pacbell.net (muzo)
Date: 16 Nov 1998 10:22:32 PST
Links: << >>  << T >>  << A >>
actually intuitiveness depends on how you define it :-). It can be argued that
little endian is more intuitive. The Least Significant byte is at the LS address
is what you get with little endian which frankly makes sense. You are thinking
from the point of view of a human and that POV is not suitable for computers. You
should think about increasing byte addresses giving you more of the bytes. Also
when you sign extend a number it is nice if its address doesn't change.

shailbains007@my-dejanews.com wrote:

>In article <364ff418.2246316251@firewall>,
>  yves@px.uk.com (Yves Tchapda) wrote:
>>
>> Hi Shail,
>> Technically speaking, there is no  advantage of Big-Endianness over
>> Little-Endianness. Intel processors are predominantly Little-Endian,
>> while Motorola processors are Big-Endian. Choosing one over the other
>> depends primarily on the overall system.
>
>I'm not totally convinced. Why do Intel designers choose an Endianness that
>is, well, counter-intutive? Take a look at an x86 memory dump to know what I
>mean...  Are you sure there's no difference in the gate counts required to
>make a system Small or Big Endian? Maybe one of them requires much fewer
>gates to implement...
>
>--Shail Bains
>Silicon Automation Systems Ltd.
>
>-----------== Posted via Deja News, The Discussion Network ==----------
>http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

muzo

WDM & NT Kernel Driver Development Consulting <muzok@pacbell.net>
Article: 13117
Subject: Re: Big-Endian vs Little-Endian
From: russmay@ditmco.com (Russell May)
Date: Mon, 16 Nov 1998 19:02:39 GMT
Links: << >>  << T >>  << A >>
I have worked with both Big-Endian and Little-Endian types.
I see no particular advantage to either, except one or the
other is usually required to match up with the software used
to compile or assemble programs, and ICE equipment for debugging.
There might be an advantage to one or the other when matching
data structures to hardware in specific designs (especially with
address/data buses instead of separate address and data buses),
but this has not been important in my experience.

I do agree that byte memory dumps for structures larger than
bytes (especially for bit fields which span between bytes)
are generally less intelligible with Little-Endian types, but
that appears to be the only advantage of Little-Endian types.

Russell May

On Mon, 16 Nov 1998 17:18:00 GMT, shailbains007@my-dejanews.com wrote:

>In article <364ff418.2246316251@firewall>,
>  yves@px.uk.com (Yves Tchapda) wrote:
>>
>> Hi Shail,
>> Technically speaking, there is no  advantage of Big-Endianness over
>> Little-Endianness. Intel processors are predominantly Little-Endian,
>> while Motorola processors are Big-Endian. Choosing one over the other
>> depends primarily on the overall system.
>
>I'm not totally convinced. Why do Intel designers choose an Endianness that
>is, well, counter-intutive? Take a look at an x86 memory dump to know what I
>mean...  Are you sure there's no difference in the gate counts required to
>make a system Small or Big Endian? Maybe one of them requires much fewer
>gates to implement...
>
>--Shail Bains
>Silicon Automation Systems Ltd.
>
>-----------== Posted via Deja News, The Discussion Network ==----------
>http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

Article: 13118
Subject: Re: Big-Endian vs Little-Endian
From: russmay@ditmco.com (Russell May)
Date: Mon, 16 Nov 1998 19:18:21 GMT
Links: << >>  << T >>  << A >>
I have worked with both Big-Endian and Little-Endian types.
I see no particular advantage to either, except one or the
other is usually required to match up with the software used
to compile or assemble programs, and ICE equipment for debugging.
There might be an advantage to one or the other when matching
data structures to hardware in specific designs (especially with
address/data buses instead of separate address and data buses),
but this has not been important in my experience.

I do agree that byte memory dumps for structures larger than
bytes (especially for bit fields which span between bytes)
are generally less intelligible with Little-Endian types, but
that appears to be the only disadvantage of Little-Endian types.

Russell May

On Mon, 16 Nov 1998 17:18:00 GMT, shailbains007@my-dejanews.com wrote:

>In article <364ff418.2246316251@firewall>,
>  yves@px.uk.com (Yves Tchapda) wrote:
>>
>> Hi Shail,
>> Technically speaking, there is no  advantage of Big-Endianness over
>> Little-Endianness. Intel processors are predominantly Little-Endian,
>> while Motorola processors are Big-Endian. Choosing one over the other
>> depends primarily on the overall system.
>
>I'm not totally convinced. Why do Intel designers choose an Endianness that
>is, well, counter-intutive? Take a look at an x86 memory dump to know what I
>mean...  Are you sure there's no difference in the gate counts required to
>make a system Small or Big Endian? Maybe one of them requires much fewer
>gates to implement...
>
>--Shail Bains
>Silicon Automation Systems Ltd.
>
>-----------== Posted via Deja News, The Discussion Network ==----------
>http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    


Article: 13119
Subject: Re: Modifying Disk serial number in boot sector....anyone have any problems with it?
From: "Austin Franklin" <dark4room@ix.netcom.com>
Date: 16 Nov 1998 21:07:25 GMT
Links: << >>  << T >>  << A >>
> I can't verify the details of offset and commands. But I can verify that
> the process works.

Does that work for NT?????

Thanks,

Austin

Article: 13120
Subject: newbie question about timing
From: jai <jk840@hotmail.com>
Date: Mon, 16 Nov 1998 16:13:14 -0500
Links: << >>  << T >>  << A >>
Hi,

I intend to implement some of my circuit functionality on a Xilinx
Spartan FPGA. The signals are time critical but I know it can be
implemented in this FPGA.
My question is that does the actual pinout make a difference on the
timing? (I want to do board layout first and then work on the FPGA
configuration so I was wondering if I could arbitarily assign I/O pins).

TIA,
Jai.
Article: 13121
Subject: Re: Software tool
From: "Bob Deasy" <bobd%nospam%@model.com>
Date: Mon, 16 Nov 1998 15:14:30 -0800
Links: << >>  << T >>  << A >>
Not that ModelSim is the most expensive but if you are interested in a good
ratio of price-feature-performance functionality then check out
www.model.com.

APS wrote in message <364DE427.2C8C962D@associatedpro.com>...
>Check out the PeakVHDL suite at http://www.associatedpro.com It can handle
all
>the major FPGA vendors, has synthesis and simulation, and is a bit less
money
>than the ALDEC suite.
>
>Ido Kleinman wrote:
>
>> Dear all,
>>
>> I am looking for a good VHDL development (compiling, simulation,
gate-level
>> sim/FSM support, Testbench generator) software tool. I need it to have a
>> comfortable interface to maxplus2 for exporting compiled designs for
>> synthesis in Altera's devices.
>> I've been looking around lately and I've evaluated Aldec's "Active-VHDL"
and
>> quite happy with it, but I have a slow internet connection, therefore,
>> before I start downloading tens of megabytes again, I would like to know
if
>> there are any other compact software tools worth looking at...?
>>
>> Anyone got any experience with Aldec's tool?
>> Are the big ones (Synopsys, Examplar..) worth the investment?
>>
>> --
>>
>>  Ido Kleinman.
>>  kleinn@mail.biu.ac.il
>
>--
>__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/
>
>Richard Schwarz, President              EDA & Engineering Tools
>Associated Professional Systems (APS)   http://www.associatedpro.com
>3003 Latrobe Court                      richard@associatedpro.com
>Abingdon, Maryland 21009
>Phone: 410.569.5897                     Fax:410.661.2760
>
>__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/
>
>


Article: 13122
Subject: CE Student Seeks Research Pointers
From: douglasut@aol.com (DouglasUT)
Date: 16 Nov 1998 23:54:05 GMT
Links: << >>  << T >>  << A >>
My name is Douglas.  I've been a student of Computer Science and Electronics
Engineering Technology for several years now.  Recently I've moved to Ogden,
UT.

In the near future, I plan to attend Weber State University to finish my BSCS. 
As a
special Independant-Study project, I'm going to attempt a port of the Linux
operating
system to the Zilog Z380, a 32-bit successor to the Zilog Z80.  

I'm looking for references to PROTECTED MEMORY, how it is implemented,
and perhaps some design examples.  I will attempt to also design an external
memory management unit to implement memory protection for this processor,
using an FPGA or several MSI/LSI and CPLD components.  At the present time,
I'm trying to gather information, but have found almost nothing -- application
notes
or theory, which would better-explain an actual implementation of protected
memory
for a new processor.

Any references, contacts, or pointers you might provide would be
most-appreciated.

Cordially,

Douglas Beattie Jr.                http://www2.whidbey.net/~beattidp/

Article: 13123
Subject: Re: newbie question about timing
From: "Ken Coffman" <kcoffman@intermec.com>
Date: Mon, 16 Nov 1998 16:05:06 -0800
Links: << >>  << T >>  << A >>
Group the signals in a way that make sense for the circuit board and I don't
think you'll have problems with speed (you won't split a data bus all around
the package, would you? I don't think so, you'll group them together). Do at
least one place and route to make sure you don't have a bad pin assignment
(like a signal pin connected to power or ground, or something like that).
jai wrote in message <3650956A.7DF7@hotmail.com>...
>Hi,
>
>I intend to implement some of my circuit functionality on a Xilinx
>Spartan FPGA. The signals are time critical but I know it can be
>implemented in this FPGA.
>My question is that does the actual pinout make a difference on the
>timing? (I want to do board layout first and then work on the FPGA
>configuration so I was wondering if I could arbitarily assign I/O pins).
>
>TIA,
>Jai.


Article: 13124
Subject: Re: Low Cost FPGA Development Tools
From: Jeff Vallier <jeffrey.vallier@smi.siemens.com>
Date: Mon, 16 Nov 1998 16:12:09 -0800
Links: << >>  << T >>  << A >>
I just saw an $89 starter kit for students from Xilinx in a San Jose
bookstore last weekend. I didn't get the deytails. Go to www.clbooks.com. If
you work for a company, your local Xilinx rep should have free to low-cost CD
rom starter packages.

Peff

Hlebasko@t-online.de wrote:

> I am looking for information for some low cost FPGA development tools for
> learning.  I am trying to improve my "skill set" on my own and I can't
> afford multiple thousand dollar tool sets.
>
> Thanks.
>
> Joseph Hlebasko





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
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

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