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 14850

Article: 14850
Subject: Bus Interface
From: thompson@ren.eecis.udel.edu (Tyrone Thompson)
Date: 20 Feb 1999 02:38:00 GMT
Links: << >>  << T >>  << A >>
Hi,

I am doing a Master's thesis in which I am interfacing to the SBus. I have
what I think is a simple question, but I can't seem to get straight. I am 
using a Xilinx part all I need to do is read and write from the SBus to some
registers in the chip. What I can't seem to figure out is how to set this
up in VHDL. I know the Sbus must be bi-directional and the registers will
need to be tri-states, but I keep chasing my tail as I try to code this. In
addition I am using the 5.2/6.0 Foundation tools as my part (X4008) is no 
longer supported under the latest foundation tools.

If anyone can point me in the right direction in terms of examples or
reference material I would appreciate it.

Thanks,

Tyrone

-- 
--------------
thompson@eecis.udel.edu				University of Delaware
Tyrone Thompson					EE Graduate Student

Article: 14851
Subject: Re: "Altera FreeCore Library" back on the web
From: Hamish Moffatt <hamish@rising.com.au>
Date: 20 Feb 1999 04:43:05 GMT
Links: << >>  << T >>  << A >>
Andy Peters <apeters@noao.edu.NOSPAM> wrote:
> And the new M1.x tools don't support the (obsolete) XC73xx chips (we have

Nor XC4000, which is annoying too.


Hamish
-- 
Hamish Moffatt       Mobile: +61 412 011 176       hamish@rising.com.au

Rising Software Australia Pty. Ltd. 
Developers of music education software including Auralia & Musition.
31 Elmhurst Road, Blackburn, Victoria Australia, 3130
Phone: +61 3 9894 4788  Fax: +61 3 9894 3362  USA Toll Free: 1-888-667-7839
Internet: http://www.rising.com.au/
Article: 14852
Subject: Re: multiple clock domain problem
From: ems@riverside-machines.com.NOSPAM
Date: Sat, 20 Feb 1999 11:24:07 GMT
Links: << >>  << T >>  << A >>
On Fri, 19 Feb 1999 09:36:58 +0200, Utku Ozcan <ozcan@netas.com.tr>
wrote:

<snipped>

you don't really give enough information, but if your post-layout
simulation differs from your other two simulations, and the static
analyser hasn't reported any timing errors, then this should mean that
your timing constraints are incorrect. the post-layout simulation is,
in most cases, simply a test of your constraints. 

if your ncf file has no domain information then this is likely to be
your problem. if you want a fix, you should post a description of how
the clocks are related, the constraints you give to synplify, and the
resulting ncf.

evan

Article: 14853
Subject: Re: multiple clock domain problem
From: ems@riverside-machines.com.NOSPAM
Date: Sat, 20 Feb 1999 11:25:44 GMT
Links: << >>  << T >>  << A >>
On Fri, 19 Feb 1999 18:46:18 GMT, frannhagen@my-dejanews.com wrote:

>I find it tricky sometimes to deal with the unpredictable process execution
>in VHDL. Adding a small phase differance (i.e. 1ns) could help. If your
>flow of data is in one direction (i.e. 12.5 -> 25) only this works.

process execution in VHDL is unpredictable only in the sense that
process execution order, within a single delta cycle, is random.
however, the whole point of the delta cycle model is that this doesn't
matter (it does matter in verilog, but that's a different story).

if you've written a model where execution order does matter, then the
model is incorrect (excluding trivial cases such as generating reports
in different orders) and you should fix it, rather than adding
arbitrary phase delays. do you have any examples?

evan

Article: 14854
Subject: Re: multiple clock domain problem
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Sat, 20 Feb 1999 16:38:59 +0200
Links: << >>  << T >>  << A >>
> you don't really give enough information, but if your post-layout
> simulation differs from your other two simulations, and the static
> analyser hasn't reported any timing errors, then this should mean that
> your timing constraints are incorrect. the post-layout simulation is,
> in most cases, simply a test of your constraints.
> 
> if your ncf file has no domain information then this is likely to be
> your problem. if you want a fix, you should post a description of how
> the clocks are related, the constraints you give to synplify, and the
> resulting ncf.
> 
> evan

  Design entry: Verilog
  Synthesizer: Synplify 5.0.8
  P&R: Design Manager M1.5 Alliance
  Technology: Xilinx XC40150XV/-0.9/BG352

  Ok. Let me explain my problem in a more detail.

              _________              _________
   Address   |         | Address    |         |----> A[x:0]
  ---------->| Logic A |----------->| Logic B |               (Dual Port RAM)
             | 12.5 MHz|            |  25 MHz |----> WR_EN
   Data      |  (FSM)  | Data       |  (FSM)  |
  ---------->|_________|----------->|_________|----> D

  Address and Data at the output of Logic A are valid for just one clock
  period (in this case, at 12.5 MHz). Logic A is running at positive edge.
  Logic B accepts this signals at the falling edge of the 25 MHz clock
  and generates appropriate address, data and write enable signals to
  the Dual Port RAM.

  Logic A and Logic B FSM's are designed and verified in functional level.
  Synplify is given actual frequency values in SDC file and Global Frequency
  Value has been set to 0 (zero MHz). We got the best result under this
  condition. XNF netlist of Synplify is given ngduild and ngd2ver to get
  the post synthesis Verilog netlist (this is recommended by Synplicity).
  SDC file includes actual frequency values to the external clocks and
  divided clocks (12.5 MHz internally derived from 25 MHz external clock).

  Functional simulation and post synthesis simulation are exactly the same.

  Static Timing Analyzer TRCE doesn't produce any timing error. But Design
  Manager warns me that 12.5 MHz clock is using non-dedicated clock resources.
  (Is there any way to dedicate divided clock to a clock network?) Design
  Manager accepts UCF file (NCF file is not given) which includes clock
  constraints and TIMESPEC FROM:TO constraints to define multiple clock
  domain interaction. Without TIMESPEC timing in the UCF file of M1.5,
  the analyzer reported a slower frequency result but with TIMESPEC there
  are new warnings but static timing analyzer gives a better frequency.

  Post layout simulations show that ONLY ADDRESS VECTOR is weird. Address
  Vector of Dual Port RAM has the correct value in a definite clock period
  in two clock cycles later. (This results are with UCF without TIMESPEC):

  Address                   Waveform
                            _____       ____       _
  12.5 MHz Clock        ___/     \_____/    \_____/
                        ____________________________
  Address Input of B    ____________________________ (it doesn't change at all)
                          __    __    __    __    __
  25 MHz Clock          _/  \__/  \__/  \__/  \__/  
                        ____ _____ _________________
  Func & Post synthesis ____X_____X_________________
                        ________________ _____ _____
  Post Layout           ________________X_____X_____

  This result is horrible and we are surprised. Has anyone met such a
  condition before?

  Utku
Article: 14855
Subject: Re: multiple clock domain problem
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Sat, 20 Feb 1999 16:49:02 +0200
Links: << >>  << T >>  << A >>
Forgotten to say:

The problem was

- NCF with clock constraints
- UCF with pin constraints (but no TIMESPEC and no clock constraints)

Following condition was not tested:

- no NCF is given
- UCF with pin constraints, clock contraints and TIMESPEC

regards,

Utku
Article: 14856
Subject: Re: multiple clock domain problem
From: ems@riverside-machines.com.NOSPAM
Date: Sat, 20 Feb 1999 16:35:51 GMT
Links: << >>  << T >>  << A >>
On Sat, 20 Feb 1999 16:38:59 +0200, Utku Ozcan <ozcan@netas.com.tr>
wrote:

<snipped>

1) this won't work if you're only specifying clock constraints in the
NCF, and no further timespecs. the analyser needs you to tell it about
the path from the 25MHz clock, to the 12.5MHz clock, through block A,
to setup for the 25MHz clock at block B (example: how does TRCE know
if you have 1 * 25MHz cycle for this path, or half a cycle?)

2) try the UCF with your timespecs. if this doesn't work, you should
post (a) a schematic showing how the 2 clocks are related to each
other, and what clocks block A and block B, and (b) all your
clock-related constraints from your UCF, and the constraints on all
signals going from block A to block B.

(3) the fact that there's no global buffer on the 12.5MHz clock
obviously doesn't help, but TRCE should pick up any problems if the
constraints are correct.
 
evan

Article: 14857
Subject: Re: multiple clock domain problem
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Sat, 20 Feb 1999 18:55:31 +0200
Links: << >>  << T >>  << A >>
> Yes, instatiate a BUFGP (or whatever the name is for a global buffer)
> at the ouput of the clock-divider. Then you can access the global
> clock routing. I do this sometimes for high fan-out non-clock signals.
> There are 8 global buffers, so if you need to you should use them.

  BUFGP and BUFGS are for XC4000E devices, so there are BUFG, BUFGE
  (Global Early) and BUFGLS (Global Low Skew) buffers in XC4000XV devices.

  But there _is_ buffer... I haven't instantiated a buffer into HDL code.
  Instead I have put a synthesis directive into the HDL code which is
recommended
  by Synplicity folks:

  wire cpu_clk12m /* synthesis syn_keep=1 */;

  This automatically instantiates a buffer at the output of the clock divider.
  It can be seen in RTL view of Synplify FPGA Synthesizer.

> / Jonas Thor

  Utku
Article: 14858
Subject: Re: multiple clock domain problem
From: bob@nospam.thanks (Bob Perlman)
Date: Sat, 20 Feb 1999 17:04:44 GMT
Links: << >>  << T >>  << A >>
Hi - 

On Sat, 20 Feb 1999 16:38:59 +0200, Utku Ozcan <ozcan@netas.com.tr>
wrote:

>  Ok. Let me explain my problem in a more detail.
>
>              _________              _________
>   Address   |         | Address    |         |----> A[x:0]
>  ---------->| Logic A |----------->| Logic B |               (Dual Port RAM)
>             | 12.5 MHz|            |  25 MHz |----> WR_EN
>   Data      |  (FSM)  | Data       |  (FSM)  |
>  ---------->|_________|----------->|_________|----> D
>
>  Address and Data at the output of Logic A are valid for just one clock
>  period (in this case, at 12.5 MHz). Logic A is running at positive edge.
>  Logic B accepts this signals at the falling edge of the 25 MHz clock
>  and generates appropriate address, data and write enable signals to
>  the Dual Port RAM.
>
>  Logic A and Logic B FSM's are designed and verified in functional level.
>  Synplify is given actual frequency values in SDC file and Global Frequency
>  Value has been set to 0 (zero MHz). We got the best result under this
>  condition. XNF netlist of Synplify is given ngduild and ngd2ver to get
>  the post synthesis Verilog netlist (this is recommended by Synplicity).
>  SDC file includes actual frequency values to the external clocks and
>  divided clocks (12.5 MHz internally derived from 25 MHz external clock).
>

I know this doesn't address your question, but I'm curious: why are
you using a 25MHz clock and a 12.5MHz clock?  Why not run everything
off a single 25MHz clock, and create a 12.5 MHz clock enable to feed
to the CE inputs of logic A?  In fact, you could use your existing
internally-divided 12.5MHz clock as the enable (or enable*, actually).
It wouldn't need to be on a global low-skew net, as long as it sets up
to the destination FFs in a 25MHz clock period.

I realize there are reasons to use divide-down clocks, power savings
being one.  But such clocks are often unnecessary. 

Regards,
Bob Perlman

-----------------------------------------------------
Bob Perlman
Cambrian Design Works
Digital Design, Signal Integrity
http://www.best.com/~bobperl/cdw.htm
Send e-mail replies to best<dot>com, username bobperl
-----------------------------------------------------
Article: 14859
Subject: pcb design costs
From: rk <stellare@NOSPAMerols.com>
Date: Sat, 20 Feb 1999 12:53:02 -0500
Links: << >>  << T >>  << A >>
hi guys and gals,

need to do a reality check (check got hit with a whopper bill this
week), what do people pay for pcb design costs, per hour?  or
equivalently, a non-dense, moderately populated 10" x 11" board,
components on one side, 6 layer (from netlist in to artwork out), with
virtually all signals done with autorouter? (it's an easy board).

thanks,

rk

p.s. no solicitations please

Article: 14860
Subject: Re: P&R times for Altera10K200E and Virtex
From: Jamie Lokier <spamfilter.feb1999@tantalophile.demon.co.uk>
Date: 20 Feb 1999 18:59:48 +0000
Links: << >>  << T >>  << A >>
Ray Andraka writes:
> Those are dismal clock rates for a 10K-3.

I know, thanks.

Data points: The average fan-in for the Handel-C version is about 3.25,
while for the AHDL version it is about 2.86.  The AHDL version of the
program runs about twice as fast as the Handel-C version.  (33MHz
vs. 16.5MHz).

A small prototype of the Handel-C program is happy at 40MHz.  It slows
down a lot as more is added to the program.  This is _not_ due to
increased logic depth.  It is due to routing delays.

> Typically arithmetic stuff slows it down, but even there you should
> get about 40 MHz without cliques with 20 bit carry chains in the
> logic.

I disagree.  Time for carry chains, arithmetic etc. are negligable in
this application.  Most of the delays are in I/O pipelines interfacing
to external devices, and internal control/multiplexing logic.

I am getting >40MHz for small programs, (>70MHz for specific things like
async FIFOs), but the clock rate drops drastically as the chip fills up.
There's not much I can see to do about it, except use a bigger or faster
chip.

I see a drastic slowdown as the chip fills, even though there is no
additional logic depth.  I can guarantee that because I connect together
different program modules through Handel-C channels, and I have a very
good idea what form the logic takes through those connections.  Hint: a
row of registers is guaranteed between each module's input and its
output, so the logic depth is bounded as more more modules are
connected together.

There is considerable additional fan-*out* of some control signals; will
this make a difference on a 10K?

BTW, the Handel-C version does not use any hierarchy in the netlist,
while the AHDL version does.  Does this make a difference in your
experience?  It is something I will be experimenting with soon, anyway.

> Sounds like you have many layers of combinatorial stuff between the
> registers.

Not really.  There is plenty of combinational stuff, but the
register-to-register logic depth does not increase much as the chip is
filled, and the clock rate still slows down a lot.

The critical delays are through the RAM elements.  I didn't turn on the
"CUT_OFF_RAM_REGISTERED_WE_PATHS" for this analysis, but I know that
turning it on does increase the max frequency reported by timing
analysis.  (For the Handel-C program.  It makes no difference to the
AHDL program).  It is possible timing analysis is reporting bogus
timings because a lot of the combinational relationships do not have to
execute in a single cycle.

The combinatorial stuff is mostly multiplexers, multiplexers and more
multiplexers.  Control logic accounts for the rest.  There's hardly any
arithmetic -- just a few 16 bit carry chains, and they're lost in the
noise.  The multiplexers and control logic are implicit in Handel-C, and
it is hard to design a large program where you know exactly what is
connected to what.  (It is easy to design a _small_ program where you
know what's what, but this application isn't one).

The AHDL version is about twice as fast as the Handel-C version (33MHz
vs. 16.5MHz).  However, the AHDL version was specifically tuned to
multiplex just two signals in each pipeline stage, hence 1 LC logic
depth.  I am told this tweaking took most of the AHDL development time,
and is unmaintainable -- quite possibly, it would be cheaper to buy a
faster FPGA than do the design tuning.

> Perhaps you should spend more time on designing to the
> architecture than playing with the constraints.

Been doing both, ta.  See above.  I have no intention of writing such
unmaintainable code in Handel-C.

I have gone as far as I can with "designing to the architecture",
several complete rewrites in fact.  And yes I do know the architecture
very well.

You'll notice my article talks about turning _off_ constraints, and that
it's better to tweak the logic than to use timing constraints.  I have
done some basic retiming of the output from the Handel-C compiler to get
better timings, and it does work.  Is messy though.

To be honest, the Flex10K architecture is not really suited to the
application IMO and/or tools, but it is what I have to work with ATM.

The slowdown as more pipelined logic is added is a real problem.
However, adding more non-pipelined logic makes it even worse.  I suspect
an FPGA with more local routing (e.g. a matrix where all adjacent blocks
have some local routing) would scale better with this sort of program,
as the chip fills up.  But I have no data to substantiate it.  Does
anyone else?

-- Jamie
Article: 14861
Subject: Re: multiple clock domain problem
From: Ken McElvain <ken@synplicity.com>
Date: 20 Feb 1999 14:18:18 PST
Links: << >>  << T >>  << A >>
The syn_keep attribute on your wire tells Synplify to isolate the driver and loads

of the net.  It does this by inserting a "virtual" buffer between the driver and
the
loads that is retained while Synplify is processing your design.  This virtual
buffer is what you see in the RTL view.  You should check the net in the
technology
mapped view to see if a clock buffer was inserted.  From your posting it sounds
like no clock buffer was inserted.   For very small clock nets that are internally

generated (<= 16 loads) Synplify will not insert a clock buffer.  The overhead
of getting to a global clock buffer and back to the flip-flops can be worse than
just
routing the signal.  You can find the 12.5 Mhz clock in the <design>.dly file and
see
what the delay was to each clock pin on this clock.  If you see a large delay or
large variation in delay they that could be your problem.

Possible fixes are:

    1) Make the routing of the clock net high priority to shorten the routes.
    2) Floorplan the flip flops in your FSM and the clock divider into a small
region
    of the chip.
    3) The best solution.  Don't divide the clock.  Use an enable instead.  This
was
        Bob Perlman's suggestion and is a good one.  This will eliminate clock
skew
        problems.


Utku Ozcan wrote:

> > Yes, instatiate a BUFGP (or whatever the name is for a global buffer)
> > at the ouput of the clock-divider. Then you can access the global
> > clock routing. I do this sometimes for high fan-out non-clock signals.
> > There are 8 global buffers, so if you need to you should use them.
>
>   BUFGP and BUFGS are for XC4000E devices, so there are BUFG, BUFGE
>   (Global Early) and BUFGLS (Global Low Skew) buffers in XC4000XV devices.
>
>   But there _is_ buffer... I haven't instantiated a buffer into HDL code.
>   Instead I have put a synthesis directive into the HDL code which is
> recommended
>   by Synplicity folks:
>
>   wire cpu_clk12m /* synthesis syn_keep=1 */;
>
>   This automatically instantiates a buffer at the output of the clock divider.
>   It can be seen in RTL view of Synplify FPGA Synthesizer.
>
> > / Jonas Thor
>
>   Utku

Article: 14862
Subject: Re: multiple clock domain problem
From: thor@sm.luth.se (Jonas Thor)
Date: Sun, 21 Feb 1999 00:39:36 GMT
Links: << >>  << T >>  << A >>
On Sat, 20 Feb 1999 16:38:59 +0200, Utku Ozcan <ozcan@netas.com.tr>
wrote:
Hej!

>  Static Timing Analyzer TRCE doesn't produce any timing error. But Design
>  Manager warns me that 12.5 MHz clock is using non-dedicated clock resources.
>  (Is there any way to dedicate divided clock to a clock network?) 

Yes, instatiate a BUFGP (or whatever the name is for a global buffer)
at the ouput of the clock-divider. Then you can access the global
clock routing. I do this sometimes for high fan-out non-clock signals.
There are 8 global buffers, so if you need to you should use them.

/ Jonas Thor
Article: 14863
Subject: Free Tool For FPGA ??
From: wong ying <wongying@hkstar.com>
Date: Sun, 21 Feb 1999 12:25:08 +0800
Links: << >>  << T >>  << A >>
If you are a student, a Chinese student or live in Hong Kong or China
you can not get a license due to the company policy!

The web site is http://www.

Sorry, I don't know because they no allowed me to get the license.

Wong Ying

From: Dennis Reynolds <dennisr@model.com>
To: wongying@hkstar.com <wongying@hkstar.com>
Date: Fai,19 Feb 1999 AM 12:46
Subject: RE: License Request


Hi Wong Ying,

Thank you for your response.

I apologize but the first license sent to you was in error.  Because of
Model Technology policy,  ModelSim is not available for educational or
student use in China.  We regret we are unable to comply with your
request.

Best regards,
Dennis Reynolds


Article: 14864
Subject: Eval Activ-VHDL only for 30 day :(
From: Krzycho <condorek@friko3.onet.pl>
Date: Sun, 21 Feb 1999 18:04:43 +0100
Links: << >>  << T >>  << A >>
Hello
I'm an electronics and telecomunications student. I'm interested in VHDL
and I've got an evaluation version of Activ-VHDL (for Windows) tools
(from Aldec www). Unfortunately this version works only for 30 days.
This period is too small becouse I'm not very good in VHDL and I want to
learn it with Aldec tools.
So I'm looking for a good idea how to extend this period :)

Mail me please: condorek@friko3.onet.pl
Article: 14865
Subject: Under-clocking SDRAM
From: bmathew@hotmail.com
Date: Sun, 21 Feb 1999 17:24:26 GMT
Links: << >>  << T >>  << A >>
Hi,

Does anyone know if it is possible to clock SDRAM (any vendor)
at 1 MHZ or less so as to interface it with an FPGA based
hardware emulator ?

Thanks,
Binu

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    
Article: 14866
Subject: connecting 2 FPGA
From: cpegfa@uxmail.ust.hk (Leprechaun)
Date: 21 Feb 1999 20:27:29 GMT
Links: << >>  << T >>  << A >>
Hi all,

The situation are as follows:
=============================
I am working on a circuit which needs 2 XC4028XL-1 to contain it.
part of the circuit involve the generation of 2 slower clock 
signals(say, B & C) from an input clock (say, A). And then there are 
logics in both chips that depend on these 3 clocks.
In order to reduce the race between the clocks, I use the input clock A
to one chip, past through the input buffer, clock buffer and then
the output buffer to form a clock A' of same rate. Thus, the delay between
B & C will be much less than directly use A, B and C.

Problems are as follows :
=========================
when I test both chip separately, they works fine. However, if I connect
them together using 2 flat cables. There are lots of glitches appear on 
the clocks. I suspect that the glithes are induced in the flat cables but
it is only part of the story.
I use a logic analyzer to look at at A, A', B & C, find out that glitches
appear on A', B & C but not in A. However, B and C can no longer keep 
there shape. It seems that glithes are propagated to the clock generation
circuitry. (but remember, A is clean).
I've struggled with this problem for quite long, originally I don't 
expect it is that difficult. (I think many ppl have connect several FPGAs
together before) But it turns out that underestimate it......

Can anyone suggest any precautions when connecting 2 FPGAs together?

Thank you very much.

Rgds,

Oliver

Article: 14867
Subject: Re: Free circuit design
From: Rainer Dorsch <rainer@rainer.informatik.uni-stuttgart.de>
Date: 21 Feb 1999 21:38:39 +0100
Links: << >>  << T >>  << A >>

There is a low volume mailing-list concernig this subject:

opendesign@circu.its.tudelft.nl




-- 
Rainer Dorsch
Abt. Rechnerarchitektur  e-mail:rainer.dorsch@informatik.uni-stuttgart.de
Uni Stuttgart            Tel.: 0711-7816-215
Article: 14868
Subject: Re: Eval Activ-VHDL only for 30 day :(
From: mw <mawa@nikocity.de>
Date: Mon, 22 Feb 1999 10:13:29 +0100
Links: << >>  << T >>  << A >>
Krzycho schrieb:
> 
> Hello
> I'm an electronics and telecomunications student. I'm interested in VHDL
> and I've got an evaluation version of Activ-VHDL (for Windows) tools
> (from Aldec www). Unfortunately this version works only for 30 days.
> This period is too small becouse I'm not very good in VHDL and I want to
> learn it with Aldec tools.
> So I'm looking for a good idea how to extend this period :)
> 
> Mail me please: condorek@friko3.onet.pl

Check out HARD DISK COPY:

"HARD DISK COPY is a hard disk copying utility which copies from

hard disk or image file to another hard disk or image file. HARD

DISK COPY is ideal for backup, hard disk upgrade, systems

transfer and multiple systems set up."

http://ourworld.compuserve.com/homepages/dcf_hdcp

- genearte HD image
- install AVHDL, after 30 days:
- restore HD image
- install AVHDL
...

Regards
mw

Article: 14869
Subject: Re: PLX9050 Dev. Software
From: Iain Rankin <IRankin@iss-dsp.com>
Date: Mon, 22 Feb 1999 09:52:12 +0000
Links: << >>  << T >>  << A >>
Apparently its a pretty straight forward thing to do...
Your S/W will lookup the registry and determine
what memory locations are available for writing to.

A friend of mine tried it and after some time to got something
to work. OK to write to/from memory is easy, but IRQ
handling etc., needs some expert code.

When i ventured into the world of the PLX9050 i got
the HDK from KrfTech. Its about 500quid for the
95/98 version and 1000quid for NT support. Boy
did it make my life easy!!. It was certainly worth the
money...it even comes with a wizard that can actually
generate code for you. Krftech are apparently bringing
out a version of their HDK that supports a number of
other platforms including Linux...

My application involved using 13 boards with the PLX9050
chip. I had so many problems using the 9050 across a PCI
bridge that as a last resort i decided to use the pin compatible
PLX9052 device - it workedfine.

You might want to look at the following sites:

http://www.bluewatersystems.com/products.html

or

http://www.krftech.com

or

hhtp://www.greymatter.co.uk

Iain


Article: 14870
Subject: Xilinx Virtex
From: Daryl Bradley <dwb105@ohm.york.ac.uk>
Date: Mon, 22 Feb 1999 10:45:20 +0000
Links: << >>  << T >>  << A >>
Anyone had any experience using the new Xilinx Virtex chips along with
any of the development boards as yet?

Cheers

Daryl Bradley

--

Bio-Inspired Architectures
Department of Electronics
University of York, UK
http://www-users.york.ac.uk/~dwb105


Article: 14871
Subject: Re: Eval Activ-VHDL only for 30 day :(
From: Cameron Watt <cwatt@one.net.au>
Date: Mon, 22 Feb 1999 22:37:14 +1100
Links: << >>  << T >>  << A >>
How did you go? I am in the same position, Bach Electronic Eng and needing a
longer evaluation period. I tried contacting them about educational deals
etc. but the best they could come up with was 55% discount, and that makes
it $3600.

Let me know how you do, and I will let you know of anything I discover.

Cheers from Australia,

Cameron Watt.

Krzycho wrote:

> Hello
> I'm an electronics and telecomunications student. I'm interested in VHDL
> and I've got an evaluation version of Activ-VHDL (for Windows) tools
> (from Aldec www). Unfortunately this version works only for 30 days.
> This period is too small becouse I'm not very good in VHDL and I want to
> learn it with Aldec tools.
> So I'm looking for a good idea how to extend this period :)
>
> Mail me please: condorek@friko3.onet.pl

Article: 14872
Subject: Re: Inferring IOFFs with FPGA Express 3.x and Foundation 1.5i
From: "Darrin Nagy" <dnagy@ndsuk.com>
Date: Mon, 22 Feb 1999 12:38:38 -0000
Links: << >>  << T >>  << A >>
BTW...

Asynchronous resets can be used on FFs that you want in IOB's by using the
GSR net.  If you look in the data sheet the GSR net is used in IOBs (fig.41
p4-44).  For VHDL designs I find the best way to do this is to instantiate
the STARTUP block and connect your reset to the GSR net.  Not very portable
I know but at least it lets you know what state your input FFs are going to
be after reset or configuration.   Note the last point is useful if your
using a uP to configure your devices long after a hardware reset has been
deasserted.  Also note the GSR net is active high.


Darrin Nagy


Article: 14873
Subject: Re: Eval Activ-VHDL only for 30 day :(
From: Krzycho <condorek@friko3.onet.pl>
Date: Mon, 22 Feb 1999 15:13:05 +0100
Links: << >>  << T >>  << A >>
Hi

Regards wrote:
 
> Check out HARD DISK COPY: 
> "HARD DISK COPY is a hard disk copying utility which copies from
> hard disk or image file to another hard disk or image file. HARD
> DISK COPY is ideal for backup, hard disk upgrade, systems
> transfer and multiple systems set up."

I haven't got a lot of free space on HD. And I'm looking for an easier
way.

> http://ourworld.compuserve.com/homepages/dcf_hdcp

This URL is bad :(( 

			Krzycho
Article: 14874
Subject: Re: Eval Activ-VHDL only for 30 day :(
From: Krzycho <condorek@friko3.onet.pl>
Date: Mon, 22 Feb 1999 15:24:27 +0100
Links: << >>  << T >>  << A >>
Cameron Watt wrote:
> 
> Let me know how you do, and I will let you know of anything I discover.

I think it is necessary to change some register (Windows '98) but witch
???
Somebody tell me that he knows easier way than changing registers.
This's all what I know to this point.
I tried to change system clock bat I gained nothing by this.


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