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 22425

Article: 22425
Subject: Timing Analyzer drains System recources
From: Eddy <jdo@somewere.else>
Date: Tue, 09 May 2000 09:28:48 +0200
Links: << >>  << T >>  << A >>
Hello,

Does anybody know what going on? Using the foundation timing analyzer
just a few runs drains my system recources on my Win98 system. Exiting
the program is the only option left to get back about 70%(!!!) of the
resources. Especially scrolling the report eats up resources. Does
anybody else experience this problem?

greetings,
Edwin Baaij

Article: 22426
Subject: Altera Megafunction in Exemplar Leonardo
From: "Thorsten Bunte" <t.bunte@beckhoff.de>
Date: Tue, 9 May 2000 09:37:49 +0200
Links: << >>  << T >>  << A >>
Hi,

I want to use the Megafunction SCFIFO created with Maxplus 2.
The properties are:
- width: 1Bit
- deep: 8
- both sides are synchronized by one clock
- outputs: empty
- asynchronous clear
- show-ahead synchronous FIFO mode

I am targeting a FLEX6k.

After synthesis in Leonardo I get the message:
Warning, component scfifo has no visible entity binding.

During place&route the following message appears:
Warning: Illegal memory cell name "scfifo_1_8_on_on"
Error: can't find design file "scfifo_1_8_on_on"

If I use only Maxplus 2 everything works fine.

Is it possible to use megafunctions created with MaxPlus 2 any how?
How can I use these functions with Leonardo?

I hope someone can help me.

Thanks,
Thorsten


Article: 22427
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: "Peter J. Kootsookos" <p.kootsookos@remove_this.ieee.org>
Date: Tue, 9 May 2000 08:56:06 +0100
Links: << >>  << T >>  << A >>
Joel Kolstad <Joel.Kolstad@USA.Net> wrote

> But is he active in multi-threaded programming under Windows NT or perhaps
> VMS?  We've got plenty of people with their CS degrees at work, and only a
> small fraction of them could tell you what a mutex (or semaphore) is.

I've never heard it before now, but the MSDN library does, indeed, reference
it:

"An object of class CMutex represents a "mutex" - a synchronization object
that allows one thread mutually exclusive access to a resource. Mutexes are
useful when only one thread at a time can be allowed to modify data or some
other controlled resource. For example, adding nodes to a linked list is a
process that should only be allowed by one thread at a time. By using a
CMutex object to control the linked list, only one thread at a time can gain
access to the list."

Ciao,

Peter K.


--
Peter J. Kootsookos
Wb: www.clubi.ie/PeterK





Article: 22428
Subject: pipeline shiftreg in virtex
From: William LenihanIii <lenihan3we@earthlink.net>
Date: Tue, 09 May 2000 08:19:40 GMT
Links: << >>  << T >>  << A >>
I have some shift registers in a Xilinx Virtex design and between the
synthesis tool (FPGA Compiler II) and the Xilinx Alliance P&R (2.1i), it
is placing these shift registers inside the Look-up tables, not in the
'regular' registers in the slices/CLBs -- which is where I need them
since they are acting as pipeline registers to help break up the long
travel time from one side of the chip to the other -- and forcing them
inside a "SRL16" of one CLB isn't going to do that.

Is there a way of coaxing the synthesis and/or P&R tool to put shift
registers in a resource of the designers' choosing (without manual
instantiation of SRL16's vs. FDCE's)?


--
========================
William Lenihan
lenihan3we@earthlink.net
========================


Article: 22429
Subject: Re: Xilinx Block Select Ram+ and LeonardoSpectrum
From: Phil Endecott <phil_endecott@spamcop.net>
Date: Tue, 09 May 2000 09:36:00 +0000
Links: << >>  << T >>  << A >>
Hi Andy,

I also struggled to understand this cryptic statement.  I believe that
the case that the Xilinx hardware supports but that Leonardo can't infer
is when one or other of the two ports is a read/write port.  So you are
allowed:

 - a single read/write port
 - one read port and one write port

but not

 - two read/write ports
 - one read port and one read/write port
 - one write port and one read/write port

HOWEVER I'd strongly advise against trying to infer ANY Xilinx block
RAMs
using Leonardo at the moment.  I uncovered a serious bug in this back in
January and have yet to hear of a fix.  I suggest that you just
instantiate the Xilinx component directly.  (Even when ram inferencing
does work in Leonardo it is horribly slow.)

Regards,

--Phil.


Andy Krumel wrote:
> 
> The Xilinx Spartan II and Vertex datasheets claims (and Mentor documentation
> backs it up) that:
> "Currently, LeonardoSpectrum does not infer dual port RAMs that read both
> read and write address."
> 
> A dual port RAM has independent clocks and addresses which support
> simultaneous read from the same/different locations and simultaneous
> read/write to different locations.
> 
> The possible interpretations seem to be:
> 1) LS does not infer dual port RAMs.
> 2) LS infers dual port RAMs but only one read is allowed at a time, ie
> either port A or port B.
> 3) #2 except restriction in place only when writing data.
> 
> My best guess is #2.
> 
> Thanks,
> Andy
Article: 22430
Subject: Good books on FEC
From: shahzad2512@my-deja.com
Date: Tue, 09 May 2000 09:47:45 GMT
Links: << >>  << T >>  << A >>
I want to implement FEC techniques in Xilinx FPGAs. Could someone
please list some good books on FEC like Viterbi, Reed-Solomon and Turbo
codes. What are the names of the Research magazines that publish the
new research in these fields.
I will be thankful
Regards,
SHAH


Sent via Deja.com http://www.deja.com/
Before you buy.
Article: 22431
Subject: Re: How to Prevent theft of FPGA design
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Tue, 09 May 2000 11:48:06 +0100
Links: << >>  << T >>  << A >>


Jim Granville wrote:

> CThis is actually very similar to what I had in mind - a combination of
> shifter and PROM/Tables
> ( a lot of our IP has this type of code, and I have often thought it
> woud be a dog to
> reverse engineer )
>
> The only difference is whether you allow external, or self 'seed' of the
> shifter.
> I guess a public source version of this could allow users either option,
> or maybe
> more a seed select, than a seed load ( too open ).
>
> So, if we accept that this raises the bar sufficently on CPLD and
> handshake attack, that
> leaves just bitstream attack on the FPGA.
>
> There are potential weaknesses here :-
>
> a) If just one copy exists in FPGA, correlation of PROM to bitstream
> location starts to
> look simpler. A Hacker can get 100's of your bitstreams quite easily,
> plus generate their
> own reference ones.
> b) I am sure a std HDL compile ( common source code ) will tend to place
> cells in
> a similar bitstream order, again aiding hacking.
>
>  For a) the design could 'fill' the spare space in the FPGA with
> multiple copies of
> this engine - now, many bits are changing.
>  For b) the FPGA vendors would need to add a scramble place/route
> option, so that within a
> block, the CLBs change order, from compile to compile.
>
> >
> > If anybody wants to try to break this I can set up a challenge test.
>
> Send me an email, and we can discuss this further.
>
> - jg

As several people have pointed out this is still suseptible to JBits attack.

(1) Use JBits to get a map of the routed design. [Can JBits do this ? If not then readback the
configuration via the JTAG i/f.].

(2) Find the pin that takes the input from the CPLD.

(3) Trace the logic forward from this pin.

>>From then in the simplest case the hacker could,

(4) Find the output from the scrambling logic that sends a go/no-go signal to the rest of the
FPGA.

(5) Use JBits to modify the the bitstream so that the output from (4) is permanently ``o.k.''.

This could probably be made very difficult by using multiple output bits through look-up tables.
So the remaining vulnerability is (3) which, in principle, will enable our determined - & by now
v. pissed off - hacker to re-create the CPLD.  First step in protecting this would be to include
a lot of ``dummy'' logic in the FPGA so that to re-create it in the CPLD would require a bigger
one than on the original board.

So we now have 2 new conditions required for our protection scheme:

o The output from the FPGA logic should affect the design in a sufficiently complex way that
will put off all but the most absolutely determined hacker.

o The CPLD & FPGA mangling logic should be asymmetric to the extent that knowledge of the FPGA
will not allow the CPLD to be recreated - or only with great difficulty.  The current proposed
scheme of running the same state machine in both CPLD & FPGA and comparing results will fall
down here.

What we need is a pair of functions E (in CPLD) and D (in FPGA) such that E followed by D
produces a known result. This result could be a fixed output value checked every so often but it
would have to be fairly long to resist probing. A better idea might be to generate an MFM
encoded stream where a legal E/D pair will produce a known pattern of edge drop-outs.

Article: 22432
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: j.kwisthout@observator.com (Johan Kwisthout)
Date: Tue, 09 May 2000 12:08:21 GMT
Links: << >>  << T >>  << A >>
On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote:

>Result of check: I asked a CS friend (he has a masters degree and is
>active in the profession) if he knew or could guess what a mutex is. He
>had a vague memory that it is a species of mosquito.
>
>Jerry

Funny, I study CS in the Netherlands and the theory of operation
systems (like synchronisation algorithms, multitasking theory) was
part of the first half of the bachalors program, dealing with
semaphores, mutexes, critical regions, producers/consumers, deadlocks
etcetera. Guess it depends on the university where you're studying...

Johan.

<snip>

Article: 22433
Subject: Nets or regs
From: "Tomasz Brychcy" <tbrychcy@sensor.ime.pz.zgora.pl>
Date: Tue, 9 May 2000 14:54:54 +0200
Links: << >>  << T >>  << A >>
Hello,

What's synthesisable model (verilog) is better:

1) model which contains only nets.
2) model which contains only registers.

Both models are exactly the same.

Thanks for each reply

tbrychcy@sensor.ime.pz.zgora.pl


Article: 22434
Subject: Re: Programming FPGA
From: Theron Hicks <hicksthe@egr.msu.edu>
Date: Tue, 09 May 2000 09:02:32 -0400
Links: << >>  << T >>  << A >>


Jon Elson wrote:

> "J. Boss" wrote:
>
> > Hi,
> >
> > I've just made a very easy design, which I want to implement in a Xilinx
> > SpartanXL FPGA using a processor-bus. I'm using the Foundation Series F2.1i.
> > The design is schematic-based. The implementation-file has been succesfully
> > created (bit-file). When I upload the program to the FPGA, everything works
> > fine (INIT remains low during programming and DONE becomes high after all
> > bits have been send and after this at least four CCLK-cycles follow!). Also
> > all timing-specs of programming the FPGA are applied. But when I measuring
> > some output pins, I just don't see what I would expect.
>
> Well, I have a bit of experience with the XC9500 parts, but I'm just moving
> into the FPGA world, too.  I did see some knowledge base articles on the
> correct settings for the clock source selection for programming to function
> correctly.  But, DONE shouldn't go high unless the download blocks have
> passed the checksum, so it sounds like the programming was correctly
> loaded into the device.  Is there some simply combinatorial signal you can
> check?  My guess is that the internal clock is set up wrong, and is not
> taking the clock from the source you intend it to.
>
> Jon

In my case I had to switch the done to clock pulse 4 instead of clock pulse 1.
This was using a serial EEPROM to load code.  It doesn't sound like that is your
problem but you might check it out.

Article: 22435
Subject: Re: Non-BGA High Pin Count FPGA/CPLD
From: Rickman <spamgoeshere4@yahoo.com>
Date: Tue, 09 May 2000 09:04:35 -0400
Links: << >>  << T >>  << A >>
Greg Neff wrote:
> 
> In article <8f76nd$rp4$1@sun27.hrz.tu-darmstadt.de>,
>   Uwe Bonnes <bon@elektron.ikp.physik.tu-darmstadt.de> wrote:
> > Hallo,
> >
> > are there any high pin count ( around 220 usable IO pins) FPGA/CPLD
> > available in Non-BGA package?
> >
> > Thanks for pointers.
> >
> > Bye
> >
> > --
> > Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de
> >
> > Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
> > --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------
> >
> 
> It looks like some Xilinx XC4000E and XC4000XL devices are available in
> 304pin QFPs, with 256 I/Os.  Check with Xilinx for real availability.


Because of the high power and ground pin counts on most packages, you
will need to go larger than 256 package pins to get 220 IOs. The
smallest package I can find that will give you this is a 304 pin QFP. It
looks like the Xilinx families have up to 256 IOs in this package. 

Another option is a PGA, but they are quite expensive. They also don't
seem to be as commonly used. 

If you want high IO count in smaller devices, the Lucent OR2 family
gives you the best bang for the CLB. The OR2C12A through OR2C40A provide
252 IOs in the 304 pin QFP. Several of these devices also come in 3 volt
versions. 


-- 

Rick Collins

rick.collins@XYarius.com

remove the XY to email me.



Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design

Arius
4 King Ave
Frederick, MD 21701-3110
301-682-7772 Voice
301-682-7666 FAX

Internet URL http://www.arius.com
Article: 22436
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: "PeterS" <peters AT aston desIGNS comMERCIAL>
Date: Tue, 9 May 2000 14:06:53 +0100
Links: << >>  << T >>  << A >>

Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>...
>On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote:
>
>>Result of check: I asked a CS friend (he has a masters degree and is
>>active in the profession) if he knew or could guess what a mutex is. He
>>had a vague memory that it is a species of mosquito.
>>
>>Jerry
>
>Funny, I study CS in the Netherlands and the theory of operation
>systems (like synchronisation algorithms, multitasking theory) was
>part of the first half of the bachalors program, dealing with
>semaphores, mutexes, critical regions, producers/consumers, deadlocks
>etcetera. Guess it depends on the university where you're studying...
>
>Johan.
>
><snip>
>
I studied CS some 20 years ago and also covered semaphores, but don't ever
recall the term mutex, until I came across it a couple of years ago doing
some NT stuff. Is "mutex" purely Microsoft-speak for what everyone else
calls a semaphore?

I don't like the use of "flag" in such specific terms - to me, a flag is
simply an
indication of state, with no protection mechanism for setting or resetting.

BTW Didn't Dijkstra first describe semaphores, and wasn't he from the
Netherlands?
 (not that its got anything to do with anything, much ;-)



Article: 22437
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: john.stewart@removethis.mindspring.com (John Stewart)
Date: Tue, 09 May 2000 13:13:04 GMT
Links: << >>  << T >>  << A >>
On Tue, 9 May 2000 08:56:06 +0100, "Peter J. Kootsookos"
<p.kootsookos@remove_this.ieee.org> wrote:

>I've never heard it before now, but the MSDN library does, indeed, reference
>it:

	I learned about mutex semaphores while reading the Design of
OS/2 by H.M. Deitel and M.S. Kogan back in 1992.  So the term is
definately older than win32.  Whether it is older than Microsoft(one
time partner in OS/2's development) I do not know.

		John Stewart


Article: 22438
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: "David Brown" <david.nospam@westcontrol.com>
Date: Tue, 9 May 2000 15:20:01 +0200
Links: << >>  << T >>  << A >>
Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>...
>On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote:
>
>>Result of check: I asked a CS friend (he has a masters degree and is
>>active in the profession) if he knew or could guess what a mutex is. He
>>had a vague memory that it is a species of mosquito.
>>
>>Jerry
>
>Funny, I study CS in the Netherlands and the theory of operation
>systems (like synchronisation algorithms, multitasking theory) was
>part of the first half of the bachalors program, dealing with
>semaphores, mutexes, critical regions, producers/consumers, deadlocks
>etcetera. Guess it depends on the university where you're studying...
>

It certainly does.  Some universities and courses will cover things like
Windows programming, which does not really have anything to do with CS, but
which looks good to PHBs for later employment.  Others will teach you the
theory of computer systems, and will cover semaphores, etc., in a manner
which can then be applied to any system.  Part-time courses and newer
universities tend towards the former, while older, more accademic
universities tend towards the later (I know that's a terrible,
unsubstantiated generalisation, but it is basically true).

Dutch universities will of course tend towards the theoretical - after all,
a high proportion of the work in synchronisation was done by Dutch
mathematicians such as Djikstra (spelling?).


>Johan.
>
><snip>
>


Article: 22439
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: "David Brown" <david.nospam@westcontrol.com>
Date: Tue, 9 May 2000 16:02:45 +0200
Links: << >>  << T >>  << A >>
Prof. Andrew Tanenbaum refers to a binary semaphore named "mutex" in
Operating Systems Design and Implementation (the book with the entire source
code to Minix, precursor and insperation for Linux, as an Appendix) in 1987.
The reference is to solve a buffer situation with three semaphores - "full"
and "empty" initialised to 0, and a binary semaphore "mutex" initialised to
1 to ensure mutually exclusive access to the buffer.

So the term "mutex" has probably been generalised from particular uses such
as this one.


John Stewart wrote in message <39180b20.30634157@news.mindspring.com>...
>On Tue, 9 May 2000 08:56:06 +0100, "Peter J. Kootsookos"
><p.kootsookos@remove_this.ieee.org> wrote:
>
>>I've never heard it before now, but the MSDN library does, indeed,
reference
>>it:
>
> I learned about mutex semaphores while reading the Design of
>OS/2 by H.M. Deitel and M.S. Kogan back in 1992.  So the term is
>definately older than win32.  Whether it is older than Microsoft(one
>time partner in OS/2's development) I do not know.
>
> John Stewart
>
>


Article: 22440
Subject: Xilinx fpga board schematics?
From: myself@magma.ca (myself)
Date: Tue, 09 May 2000 14:24:25 GMT
Links: << >>  << T >>  << A >>
Hi does anyone have a simple schematic for a Xilinx fpga board.

perhaps the 4000 series fpga and a power supply, clock and some i/o &
or connectors?

I will be buying the student edition book and sw  when the new version
comes out in July. I think there are schematics for there demo board
in the book, but i would like to design a pcb to test out some designs
now.

p.s. I use "pads power pcb" but can import most ascii files schmatic
and or PCB. or with a word or pdf file I could create my own schematic
and pcb.

Thanks any help is appreciated.

My email address if it is easyer to send there is "martinb@magma.ca"

Article: 22441
Subject: Re: Xilinx Block Select Ram+ and LeonardoSpectrum
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Tue, 09 May 2000 15:31:33 +0100
Links: << >>  << T >>  << A >>


Phil Endecott wrote:

> Hi Andy,
>
> I also struggled to understand this cryptic statement.  I believe that
> the case that the Xilinx hardware supports but that Leonardo can't infer
> is when one or other of the two ports is a read/write port.  So you are
> allowed:
>
>  - a single read/write port
>  - one read port and one write port
>
> but not
>
>  - two read/write ports
>  - one read port and one read/write port
>  - one write port and one read/write port
>
> HOWEVER I'd strongly advise against trying to infer ANY Xilinx block
> RAMs
> using Leonardo at the moment.  I uncovered a serious bug in this back in
> January and have yet to hear of a fix.  I suggest that you just
> instantiate the Xilinx component directly.  (Even when ram inferencing
> does work in Leonardo it is horribly slow.)
>
> Regards,
>
> --Phil.
>
> Andy Krumel wrote:
> >
> > The Xilinx Spartan II and Vertex datasheets claims (and Mentor documentation
> > backs it up) that:
> > "Currently, LeonardoSpectrum does not infer dual port RAMs that read both
> > read and write address."
> >
> > A dual port RAM has independent clocks and addresses which support
> > simultaneous read from the same/different locations and simultaneous
> > read/write to different locations.
> >
> > The possible interpretations seem to be:
> > 1) LS does not infer dual port RAMs.
> > 2) LS infers dual port RAMs but only one read is allowed at a time, ie
> > either port A or port B.
> > 3) #2 except restriction in place only when writing data.
> >
> > My best guess is #2.
> >
> > Thanks,
> > Andy

I asked the same question of Synplify some time ago.  They said that Block RAM
inference will be allowed via some sort of template in a future release. Basically
AFAIK the only way to define a true dual port RAM at the Verilog  level is to have
the reg array that defines the RAM written in 2 independent always blocks.
This is o.k. for simulation but is not allowed in the synthesis tools. In other
words the obvious definition:

reg [15:0] mem [256:0]

always @(clk_a) begin
    dout_a <= mem[addr_a];
    if (we_a)
       mem[addr_a] <= din_a
end
always @(clk_b) begin
    dout_b <= mem[addr_b];
    if (we_b)
       mem[addr_b] <= din_b
end


will fail synthesis, try it yourself.

Article: 22442
Subject: HELP - what to choose?
From: Andy Holt <andyh@city.ac.uk>
Date: Tue, 09 May 2000 15:37:41 +0100
Links: << >>  << T >>  << A >>
This is the sort of thing I would expect to be an FAQ, but there doesn't
seem to be one for this group.

I have been thinking about "playing" with an FPGA both from the view of
learning about an interesting-looking technology and with the hope of
constructing an emulation of a '60s mainframe (more about this later).

I am looking for advice on low-cost ways of doing this (Let's say price
ceiling of about £200 [$300]). It seems that I am going to need two main
things:
* A package of software.
* A prototyping/evaluation board.

Taking the second of these first there seem to be few choices available
(without paying lots of $$$) -

The most obvious seems to be the XS40 from Xess using the Xilinx 4000
series devices. (http://www.xess.com)

Less expensive with more features on board is the Atmel FPGA Starter Kit
from Kanda (http://www.kanda.com)

If I qualify for the special deal (I am a member of staff at a
University), the Altera Design Laboratory Package looks very
attractively priced - if I can't get the deal it is likely to be too
expensive. (http://www.altera.com)

The only other low-priced boards available seem to be the Australian
ones from Burch Electronic Designs (http://www.burched.com.au) that were
recently advertised on this group. However they seem to be very "bare"
and more designed for building real-world prototypes than for learning.
All versions also come with FPGAs of minimal capacity (almost certainly
too small for my "big" project).


As for software, the Kanda and Atmel packages come with some, for the
Xess one I would also have to spend another $100 for the Foundation
student edition.

** so, first question: any known "gotcha's" with the above alternatives? 
  [ISTR a recent hint that the Atmel software was weak in one respect -
it is noticeable that their web site seems to say almost nothing about
its functionality - and the low cost version of Foundation doesn't
include VHDL?]
Are there other reasonable options?


The other main question I have concerns estimating how big an FPGA I
would need for the mainframe emulation. I assume that the "usable gate"
counts for all devices tend to be as much marketing as technical
statements. I have detailed (but only "almost complete") descriptions of
the logic design for the mainframe that I am interested in (ICT 1905 -
aka FP6000) and I can be reasonably confident that it has less than 6000
gates including FPU ... probably less than 4000 without. Is this likely
to fit in a "10000 gate" FPGA?, a "20000 gate" one, or whatever?



More questions later :-)


Andy Holt
Systems Consultant
City University
London, England
andyh@city.ac.uk
Article: 22443
Subject: Re: HELP - what to choose?
From: Andy Holt <andyh@city.ac.uk>
Date: Tue, 09 May 2000 15:43:15 +0100
Links: << >>  << T >>  << A >>


Andy Holt wrote:
> ...

ooops, so busy trying to avoid any stupid misteaks that I forgot to say
"please" ... I'm not usually so rude - I hope

Andy
Article: 22444
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: OneStone <onestone@chariot.net.au>
Date: Wed, 10 May 2000 00:31:58 +0930
Links: << >>  << T >>  << A >>
PeterS wrote:
> 
> Johan Kwisthout wrote in message <3917fdf7.20732581@obsserver>...
> >On Mon, 08 May 2000 22:30:04 -0400, Jerry Avins <jya@ieee.org> wrote:
> >
> >>Result of check: I asked a CS friend (he has a masters degree and is
> >>active in the profession) if he knew or could guess what a mutex is. He
> >>had a vague memory that it is a species of mosquito.
> >>
> >>Jerry
> >
> >Funny, I study CS in the Netherlands and the theory of operation
> >systems (like synchronisation algorithms, multitasking theory) was
> >part of the first half of the bachalors program, dealing with
> >semaphores, mutexes, critical regions, producers/consumers, deadlocks
> >etcetera. Guess it depends on the university where you're studying...
> >
> >Johan.
> >
> ><snip>
> >
> I studied CS some 20 years ago and also covered semaphores, but don't ever
> recall the term mutex, until I came across it a couple of years ago doing
> some NT stuff. Is "mutex" purely Microsoft-speak for what everyone else
> calls a semaphore?
> 
> I don't like the use of "flag" in such specific terms - to me, a flag is
> simply an
> indication of state, with no protection mechanism for setting or resetting.

The original system of semaphores, for signalling, used flags. A
semaphore is simply a signal, as is a flag, flags are not necessarily
binary, originally semaphore, the method of singalling, each with
different context sensitive meaning. Semaphore pre-dates Dijkstra et al.
I believe it originated in the 18th century.

Semaphore  noun [U] 
             a system of communication using two mechanical arms or
             hand-held flags which are moved into different positions to
             represent different letters, numbers or symbols 
             Semaphore was widely used at sea, before the advent of
             electricity. 

             (figurative) When I lived opposite her we would send
             semaphore signals (=messages without speaking) to each
             other from our bedroom windows. 

Al
Article: 22445
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: Bill Williams <Innkeeper@iabv.com>
Date: Tue, 09 May 2000 10:07:39 -0500
Links: << >>  << T >>  << A >>
Clearly your friend has a bright future at Microsoft.
Article: 22446
Subject: FPGA Development - EDA Administration
From: Jo <JCrump@alphatec.uk.com>
Date: Tue, 09 May 2000 08:18:38 -0700
Links: << >>  << T >>  << A >>
I am working with a global comms organisation who have
several sites across the Uk and who are growing by
acquisition all the time. I am trying to find the following:
This can be based in Cambridge or the Midlands.
Please email me at JCrump@alphatec.uk.com
EDA ADMINISTRATION

This is a newly created role which is central to the
creation of an EDA group. The successful candidate will
assist in building a new EDA team. This group will sit
above the project teams and support large numbers of design
projects totalling around 400+ engineers. This support will
involve training, assisting complex design work,
introducing and developing new tools.

This challenging role provides technical support in all
areas of ASIC and FPGA development. The job encompasses all
aspects of administering and managing a complex electronic
design automation infrastructure based on verilog in a
distributed Sun-Solaris environment headquartered in
Birmingham. The role is hands-on and proactive. In addition
to day-to-day support, the successful candidate will be
capable of taking the initiative to find and promote
solutions related to EDA issues and development policies as
they arise.
Experience:
&#61623; In depth Cadence/ Synopsis administration skills
(1 year+)
&#61623; Sound UNIX skills (preferably Solaris)
&#61623; NT experience advantageous
&#61623; Shell scripting (Skill, Perl, Tcl)
&#61623; General Configuration Management skills and an
understanding of complete HW development lifecycle.




* Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping.  Smart is Beautiful
Article: 22447
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: "PeterS" <peters AT aston desIGNS comMERCIAL>
Date: Tue, 9 May 2000 16:36:35 +0100
Links: << >>  << T >>  << A >>
>> I don't like the use of "flag" in such specific terms - to me, a flag is
>> simply an
>> indication of state, with no protection mechanism for setting or
resetting.
>
>The original system of semaphores, for signalling, used flags. A
>semaphore is simply a signal, as is a flag, flags are not necessarily
>binary, originally semaphore, the method of singalling, each with
>different context sensitive meaning.

Hmmm, interesting sentence. Wonder what it means ;-)

Chains of simple signal flags were also used to "post" entire messages from
the rigging
of sailing ships (like Nelson's "England expects this day..." signal at
Trafalgar),
but there was nothing, I guess to stop anyone else (ie no interlock) putting
up a
message from their own rigging, which is why I dislike the interchangeable
use of
"flag" and "semaphore".

>I believe it originated in the 18th century.
>
>Semaphore  noun [U]
>             a system of communication using two mechanical arms or
>             hand-held flags which are moved into different positions to
>             represent different letters, numbers or symbols
>             Semaphore was widely used at sea, before the advent of
>             electricity.
>
>             (figurative) When I lived opposite her we would send
>             semaphore signals (=messages without speaking) to each
>             other from our bedroom windows.
>
>Al

I live near to a restored semaphore tower dating from the time of the
Napoleonic wars
which signalled from the Admiralty in central London down a chain of towers
to
Portsmouth and Plymouth.( Just off the A3 outside the M25.)
It has working arms, and an interesting display model of an earlier
system which used six flat, square shutters which could be individually
displayed
upright, face-on, or rotated horizontal. Somewhat OT, I realise, but a
fascinating
monument. Oh yes, a short message took less than 15 minutes from London to
Portsmouth, weather permitting (50 odd miles, 80+ km).




Article: 22448
Subject: virtex e lvds clock recovery
From: Carsten =?iso-8859-1?Q?N=F6ding?= <carsten.noeding@uni-mainz.de>
Date: Tue, 09 May 2000 17:43:42 +0200
Links: << >>  << T >>  << A >>
Does anyone know how to recover the clock from the lvds data stream
within a Xilinx Virtex E device? Any ideas how to do this without
external devices?

Thanks in advance.
Article: 22449
Subject: Re: ANNOUNCE: Embedded Systems Glossary and Bibliography
From: Rennie Allen <rallen@computermotion.com>
Date: Tue, 09 May 2000 08:59:00 -0700
Links: << >>  << T >>  << A >>
OneStone wrote:
> 
> I probably just move in much older circles than you,

Hmmm, I'm feeling pretty old right now, thanks for making me feel 
younger ;-)

> before the computer
> world went insane with jargon.

What year, exactly, was that ?  Was that the year Grace Hopper coined 
the term "bug" ?  Seems to me, that any substantial technology will 
inevitably require its own terminology.  I have no problem with this, I
simply want whatever terminology that is developed to be clear, and 
concise (btw: I don't consider "bug" to be in the category of clear and 
concise, although it is quaint).

> Mutual exclusion mechanism smacks of the
> same sort of mentality that turned housewifes into domestic engineers,
> and directory enquiries operators into information consultants. Since

I disagree.  MUTual EXclusion is the function of the device. 
Personally,
I find functional naming clearer than say, the term "bit flag" (which is 
incredibly overloaded, and gives no clue as to the intended function of
the
device - and in no way conveys the requirement for atomic state change).

Contrast the rationale behind this, to that behind the term "domestic 
engineer".  The term "domestic engineer" is more ambiguous than
housewife. 
I contend that mutex is *less* ambiguous than "bit flag", since a mutex
can only be one thing, while a "bit flag" could be many things (much as 
housewife can mean only one thing, while "domestic engineer" could mean 
many things). 

> pthreads are a POSIX multithreading standard I have never had cause to
> read about them extensively or use them in any embedded systems. And a
> review in Amazon showed that only the Butenhof book seemed to be really
> popular, the numbers sold on this subject hardly makes the book
> 'popular'.

Hmmm, Amazon.com, as dearly as they may want us to believe so, is not
the 
only source of technical reference material on the planet, hence your
data
(with regards to popularity) may be incomplete. 

Certainly, I didn't mean to imply that these books were popular relative
to,
say Stephen King novels :-)

> Perhaps reasonably popular amongst POSIX programmers, but my

A reasonable guess would be that every real-time Posix programmer, has a
copy
of at least one pthreads book.
 
> HC12, HC16, and even my MCORE systems aren't running POSIX, definitely
> none of my 8 bitters do.

Understood.  My original point was simply that I don't believe the
Embedded 
Systems Glossary was being obscure, simply by including the term mutex;
in 
fact, it appears to be a necessary inclusion (since the purpose of a
glossary 
is to explain terms).  Your point was that the term "bit flag" should be 
included, and while this would be nice, I can understand why someone
attempting
to produce a concise glossary would avoid this term (due to the fact it
is 
extensively overloaded).

Rennie


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