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
2017JanFebMarApr2017

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 4600

Article: 4600
Subject: Course/fine grain netlists?
From: P Nibbs <pnibbs@icd.com.au>
Date: Wed, 20 Nov 1996 17:08:05 +1100
Links: << >>  << T >>  << A >>
Hi,

Can anyone tell me what the difference is between a course-grain netlist
and a fine-grain netlist.
eg what is a fine/course grain?

What effect does synthesis, optimization have on this, and how can this
effect the design process?

Thanks,

Phil.

Article: 4601
Subject: Be a Beta Tester!
From: btr@trenet.com
Date: Wed, 20 Nov 1996 06:39:52 GMT
Links: << >>  << T >>  << A >>
Be a Beta Tester!

http://www.trenet.com/btr
Article: 4602
Subject: Re: Configuring Xilinx XC4000 device in Asynch. Peripheral Mode
From: fliptron@netcom.com (Philip Freidin)
Date: Wed, 20 Nov 1996 08:53:19 GMT
Links: << >>  << T >>  << A >>

In article <3291FDAB.611D@pa.msu.edu> Steve Gross <gross@pa.msu.edu> writes:
>I have a couple of questions about configuring a Xilinx XC4000-series

The rest of Steve's article is at the end of this message.

Steve, the problem you are having is that the done pin going high is
not the last thing that needs to happen before the chip is ready to
run. On page 2-30 of the 1994 data book (4-63 of the 1996 one) are the
diagrams that reveal the secrets (well some of them) as to why you
need to give the chip some extra clocks. As you can see, there is a 5
state shift register state machine that sequences the chip after an
internal signal (Full) goes high, together with a signal called
Length Count going high (This is the output of the length count
comparator that is checking the length count that was at the beginning
of the bitstream. on the far right there is an internal signal called 
'finished' that is the real end of things. This state machine is advanced 
by configuration clocks, and in the Asynchrous Peripheral Mode, this is
generated internaly, whenever you give the chip another byte of data.

If the length count is met, then the state machine starts up, and DONE
goes high at time C1 (from Q0), see page 2-29 / 4-61. Here's where things
are going wrong for you: If this happens near the end of shifting a
byte that you have loaded via Asynchrous Peripheral Mode, then the
remaining clocks needed to release I/O and the GSR don't happen because
there are no more data bits to be shifted. Loading the extra byte (I
usuall use an FF byte) gives the logic 8 extra clocks, which will be
enough to cycle the state machine to the finished state.

All the best
Philip Freidin

>device via Asynchrous Peripheral Mode.  I have a design in which 
>I configure an XC4013L in Asynch. Peripheral Mode over VME.  The 
>download appears to work (in the sense that ERR* never goes low, and
>DONE goes high at the appropriate time), but I am having a spot of
>confusion regarding the actual startup.  Details:
>
>I am using NeoCAD (err...Xilinx) FPGA Foundry to create Motorola 
>Exormacs format (S-Record) data which is read by a VME single-board
>computer which in turn writes the bitstream to the XC4013L device
>(which is located on a separate card, reachable from the SBC via VME).
>This data is 30995 bytes (247960 bits) long, which exactly matches
>the "PROM size" required for this device ('96 data book table 22,
>page 4-57).  I write the data to the device using 30995 writes, 
>checking RDY/BUSY* and ERR/INIT* as necessary.  After the 30995th
>write, DONE goes high as expected, but the newly-configured device
>does not actually complete startup (all I/O pins are still 3-stated,
>I am using the "default" startup order, activate DONE first, then
>release internal GTS, then release internal GSR).  I need to write
>one more byte of data to the device (a 30996th byte, it doesn't seem
>to matter what the byte contents are) to get the I/O's to activate.
>
>I wanted to understand why I needed this 30996th write (which didn't
>really make sense to me), so I started digging.  The only funny
>thing I could find was that the "length count" field generated
>by NeoCAD (umm...Xilinx) bitgen, and placed in the bitstream, 
>was NOT 247960 (as I had naively expected) but was actually 247958.
>Does this make sense?  What does Xilinx MakeBits put in as a length
>count for a single 4013 device?  Is this a known bitgen bug?  
>
>One more thing.  This 247958 value DOES work for a different 4013L
>(on the same board) which configures itself in Master Serial Mode,
>reading out of an XC17256 SPROM.  
>
>Thanks in advance for any possible tips.  Peter, do you have any
>insight?  
>
>-Steve Gross	gross@pa.msu.edu


Article: 4603
Subject: Re: Advantage of third party software?
From: david@fpga.demon.co.uk (David Pashley)
Date: Wed, 20 Nov 96 17:13:29 GMT
Links: << >>  << T >>  << A >>
In article <32915B1F.3D2D@icd.com.au> pnibbs@icd.com.au "P Nibbs" writes:

"Hi,
"
"I have just started investigating designing using fpga's, cplds etc and
"was wondering if there are any advantages at all in purchasing third
"party software (ie viewlogic, mentor, cadence, veribest etc).
"
<snip>

You will have to purchase the back-end place-and -route tool from 
the vendor(s) you wish to use; there is no longer any vendor-
independent source of these tools since Xilinx bought NeoCAD.

The disadvantage of using the vendor's own front-end toolset is that 
while they may work well for that vendor, they will encourage you to 
use proprietary libraries, which will mean that you cannot then 
migrate to another vendor, or compare implementation results with 
another vendor.

The third-party tools will be more expensive, but offer the 
potential of vendor-independent design via standards such as LPM 
(Library of Parameterized Macros) and the use of HDLs.

David Pashley


Article: 4604
Subject: Re: Advantage of third party software?
From: stefan_doll@usa.racal.com (Stefan Doll)
Date: Wed, 20 Nov 1996 17:15:01 GMT
Links: << >>  << T >>  << A >>
timolmst@cyberramp.net wrote:


>>As I understand it, I must use their tools (Max+Plus and XACT) to place
>>and route, obtain timing etc). These tools seem to come with schematic
>>capture, vhdl simulators, synthesis engines etc (a complete design
>>environment?).

>Xilinx tools don't COME WITH anything! You pay for all of it,
>cafeteria style. The basic XACT system is just the "back-end" place
>and route stuff. You have to buy your fornt-end design entry tool
>seperate, at additional cost.

With regard to VHDL synthesis: not all synthesizers are equally good
at their job.
In Altera's case I know of some restrictions regarding the
synthesizable subset e.g. an assignment like "A <= (others => 'Z');"
would synthesize with Synopsys but not with Altera. Also there may be
differences with regard to area and speed of the synthesized design,
but I didn't evaluate these tools thoroughly enough to make a
statement about that.


-Stefan






Article: 4605
Subject: POSITION: VHDL ASIC Designer
From: ecla@world.std.com (alain arnaud)
Date: Wed, 20 Nov 1996 17:16:36 GMT
Links: << >>  << T >>  << A >>
We are a small networking company looking for four designers for
the design of an ASIC using VHDL and for design verification.

The following qualifications are required:

	- Previous ASIC experience
	- VHDL
	- Synopsys synthesis
	- Design verification

	There are possibilities for permanent employment or as a 
	consultant.

	 PRINCIPALS ONLY!

	 Fax your resume to (617)527-1325 or email to arnaud@ecla.com




Article: 4606
Subject: ViewLogic PRO series under win95
From: "Gordon McGregor" <g.mcgregor@eee.strath.ac.uk>
Date: 20 Nov 1996 17:38:42 GMT
Links: << >>  << T >>  << A >>
Hi, 
	does anyone happen to know if PRO series is compatable with
	windows 95 ?  I am running Procapture from the Xilinx PROflow
	task launcher and as soon as I select add component the system
	crashes.

	If you have managed to get PRO series working with win95 could
	you get in touch, please ?

-- 
Gordon McGregor
---------------------------------------------------------------------
 Communications Division            Email: g.mcgregor@eee.strath.ac.uk  
 Electronic & Electrical Eng. Dept. http://drl4.eee.strath.ac.uk/~gmcg
 University of Strathclyde          Tel:  +44 (0)141 548 2250
 Glasgow G1 1XW                     Fax:  +44 (0)141 552 4968

Article: 4607
Subject: Re: VHDL code editor for Windows NT.
From: Mike Harrison <mikesh@concentric.net>
Date: Wed, 20 Nov 1996 15:12:04 -0500
Links: << >>  << T >>  << A >>
Vincent Rowley wrote:
> 
> Hi,
> 
> We are looking for a VHDL code editor for Windows NT platform.
> 
> All information is welcome.
> 
> If you reply to the group, send also a copy to the address
> vrowley@hexavision.com .
> 
> Thanks,
> 
> Vincent Rowley
> 
> --
> -----------------------------------------------------------------------
> --  Vincent Rowley                 --  HexaVision Technologies Inc.  --
> --                                 --  2050 Rene-Levesque ouest      --
> --  Email: vrowley@hexavision.com  --  Bureau 101                    --
> --  Tel:   418-686-5000 (233)      --  Sainte-Foy, Quebec, Canada    --
> --  Fax:   418-686-5043            --  G1V 2K8                       --
> -----------------------------------------------------------------------

Hi Vincent,

We use a product called Sledgehammer, which is an HDL addon to CodeWright.
I have been quite satisfied as the editor is very easily customized - I 
feel that emulation with whatever the user is most familiar is an extremely
important attribute.  We only code Verilog here, but it color codes keywords
from both HDL's as well as smart indenting, etc. and emulates Brief, VI, and
Epsilon, as well as "Notebook" type editors.  The company is VHDL Tech.
Group @ 610-882-3130.  They are probably on the web somewhere too!  Hope 
this helps.

Regards,
Mike Harrison
OIS 
Northville, MI
Article: 4608
Subject: Re: Async with FPGA?
From: Michael Baxter <mbaxter@best.com>
Date: Wed, 20 Nov 1996 13:05:13 -0800
Links: << >>  << T >>  << A >>
"Micropipelines", Ivan Sutherland. In Communications of the ACM, 
32(6):720-738, 1989.

While I haven't looked exhaustively for the paper, there are nonetheless 
some interesting things on the web regarding the transition signalling 
and micropipelines. Here's some of what I found:

http://docs.dcs.napier.ac.uk/DOCS/GET/bailey91b/document.html

http://paradise.ucsd.edu/class/ece260c/
Article: 4609
Subject: CFP: FCCM'97 IEEE Symp on Custom Computing Machines
From: jmarnold@radix.net (Jeffrey Arnold)
Date: 20 Nov 1996 21:26:03 GMT
Links: << >>  << T >>  << A >>
                     C A L L    F O R    P A P E R S

			   THE FIFTH ANNUAL
	     IEEE SYMPOSIUM ON CUSTOM COMPUTING MACHINES
			   Napa, California
			  April 16-18, 1997

			 http://www.fccm.org


PURPOSE: To bring together researchers to present recent work in the
use of Programmable Logic Devices as reconfigurable computing
elements.  This symposium will focus primarily on the current
opportunities and problems in this new and evolving technology for
computing.  Contributions are solicited on all aspects of custom
computing, including but not limited to:

Architecture of reconfigurable computing systems, including
coprocessors, attached processors, and hybrids.

Languages, compilation techniques, tools, and environments for
programming;

Application domains;

Prototyping for architecture emulation;

Use of custom computing in education.

SUBMISSIONS: Authors are invited to send submissions (4 copies, 10
pages maximum) by January 10, 1997, to Jeffrey Arnold.  Notification
of acceptance will be sent in early March.  Final papers will be due
on the first day of the symposium.  After the Symposium a proceedings
will be published by the IEEE Computer Society Press.

Authors may also submit PostScript or Microsoft Word manuscripts by
FTP.  For instruction on electronic submission, please see the Web
page or contact Jeffrey Arnold (jma@super.org).

SPONSORSHIP: The IEEE Computer Society and the Technical Committee on
Computer Architecture.

CO-CHAIRS:
Kenneth L. Pocek
Intel
Mail Stop RN6-18
2200 Mission College Boulevard
Santa Clara, California  95052
Voice: 408-765-6705  Fax: 408-765-5165
kenneth_pocek@ccm11.sc.intel.com

Jeffrey M. Arnold
10686 Mira Lago Terrace
San Diego, CA 92131
Voice: 619-547-9257  Fax: 619-547-9010
jma@super.org


PROGRAM COMMITTEE:
Peter Athanas, Virginia Tech.
Donald Bouldin, University of Tennessee, Knoxville
Duncan Buell, Center for Computing Sciences
Michael Butts, Quickturn Design Systems, Inc.
Pak Chan, Univ. California, Santa Cruz			
Apostolos Dollas,  Technical Univ. of Crete		
Scott Hauck, Northwestern Univ.
Brad Hutchings, Brigham Young Univ.
Tom Kean, Xilinx, Inc. (U.K). 
Phil Kuekes, HP Labs.
Wayne Luk, Imperial College
Mark Shand, Digital Equipment (Paris)
Satnam Singh, Univ. of Glasgow
Stephen Smith, Altera Corp.

Article: 4610
Subject: FPGA Gate Counts: No Truth in Advertising
From: david@lowrance.com
Date: Wed, 20 Nov 1996 17:01:09 -0600
Links: << >>  << T >>  << A >>
A while back one of my co-workers brought me an article he'd clipped 
out of a magazine.  Actel had a new part that would hold 20K gates 
with 100% utilization... or so they claimed.  At the time we were just 
getting started on a small ASIC, about 14K gates.  "Great", I said.  
"We'll be able to create a prototype and use that one FPGA to test our 
entire design."  It was not to be.  The problem was that the 20K gate 
part would barely hold 10K 'real' gates.

This wasn't my first run-in with an FPGA company whose parts turned 
out to be smaller in real life than advertised.  Last year an ASIC 
which we wanted to prototype in a Xilinx 4000 series part turned out 
the same way.  The FPGA industry has come a long way by learning from 
the ASIC industry.  Design methodologies are constantly improving and 
support for HDLs gaining momentum.  Can someone tell me why the FPGA 
industry is having such a hard time learning to count?

David Gardner

--
    David N. Gardner
    ASIC Design Engineer           Lowrance Electronics
    david@lowrance.com             12000 E. Skelly Drive.
    918 437-6881 x8583             Tulsa OK, 74128

-----------------------------------------------------------------------
This article was posted to Usenet via the Posting Service at Deja News:
http://www.dejanews.com/           [Search, Post, and Read Usenet News]
Article: 4611
Subject: Free Money
From: "Peter J. Rieck" <prieck@csra.net>
Date: Wed, 20 Nov 1996 17:25:26 -0800
Links: << >>  << T >>  << A >>
MAKE MONEY BY DOING NOTHING!!
Take five minutes to read this and it WILL change your life.

  The Internet has grown tremendously. It doubles in size every 4
months.
  think about it. You see those 'Make.Money.Fast' posts more and more.
  That's ... because it WORKS !  So I thought, all those new users might
  make it work. And I decided to try it out, a few months ago.  Besides,
  whats $5.00, I spend more than that in the morning on my way to work
  on coffee and cigs for the day. So I sent in my money and posted.
  Everyone was calling it a scam, but there are SO many new users from
  AOL, Netcom, etc. they will join in and make it work for you.

  Well, two weeks later, I began recieving bucks in the mail!  I
couldn't
  believe it! Not just a little, I mean big bucks!  At first only a few
  hundred dollars, then a week later, a couple of thousand, then BOOM.
By
  the end of the fourth week, I had recieved nearly $47,000.00. It came
  from all over the world. And every bit of it perfectly legal and on
the
  up and up. I've been able to pay off all my bills and still had enough
  left over for a nice vacation for me and my family.

  Not only does it work for me, it works for other folks as well.
Markus
  Valppu says he made $57,883 in four weeks. Dave Manning claims he made
  $53,664 in the same amount of time. Dan Shepstone says it was only
  $17,000 for him. Do I know these folks? No, but when I read how they
  say they did it, it made sense to me. Enough sense that I'm taking a
  similar chance with $5 of my own bucks. Not a big chance, I admit--but
  one with incredible potential, because $5 is all anyone ever invests
in
  this system. Period. That's all Markus, Dave, or Dan invested, yet
  their $5 netted them tens of thousands of dollars each, in a safe,
  legal, completely legitimate way. Here's how it works in 3 easy steps:

 STEP 1.

 Invest your $5 by writing your name and address on five seperate pieces
 of paper along with the words: "PLEASE ADD ME TO YOUR MAILING LIST."
(In
 this way, you're not just sending a dollar to someone; you're paying
for
 a legitimate service.) Fold a $1 bill, money order, or bank note inside
 each paper, and mail them by standard U. S. Mail to the following five
 addresses:

  1-  Natalie Jansen
      Lancveldlaan 18
      5671 CN Nuenen
      Holland


  2-   Chad Collier
      2785 Cold Springs Rd. #49
      Placerville, CA  95667

 3-  Steve Boltinghouse
      1009 Bird St.
      Hannibal, MO  63401

  4-  Jeremy Bero
      2021 M SW
      Miami, OK 74354

  5-   Andrew Rieck
       119 Wood Creek Dr.
        Martinez, GA 30907
        USA

STEP 2.

     Now remove the top name from the list, and move the
     other names up.This way, #5 becomes #4 and so on.
     Put your name in as the fifth one on the list.

 STEP 3.

     Post the article to at least 250 newsgroups. There are at
     least 19000 newsgroups at any given moment in time.
     Try posting to as many newsgroups as you can. Remember
     the more groups you post to, the more people will see your
     article and send you cash!

 STEP 4.

     You are now in business for yourself, and should start seeing
     returns within 7 to 14 days! Remember, the Internet is new
     and huge. There is no way you can lose.

     Now here is how and why this system works:

     Out of every block of 250 posts I made, I got back 5 responses.
     Yes, thats right,only 5. You make $5.00 in cash, not checks or
     money orders, but real cash with your name at #5.

     Each additional person who sent you $1.00 now also makes 250
     additional postings with your name at #4, 1000 postings. On
     average then, 50 people will send you $1.00 with your name at
     #4,....$50.00 in your pocket!

     Now these 50 new people will make 250 postings each with your
     name at #3 or 10,000 postings. Average return, 500 people= $500.
     They make 250 postings each with your name at #2= 100,000
     postings=5000 returns at $1.00 each=$5,000.00 in cash!

     Finally, 5,000 people make 250 postings each with your name at
     #1 and you get a return of $60,000 before your name drops off
     the list.And that's only if everyone down the line makes only 250
     postings each! Your total income for this one cycle is $55,000.

     From time to time when you see your name is no longer on the list,
     you take the latest posting you can find and start all over again.

                 The end result depends on you. You must follow through
                 and repost this article everywhere you can think of.
                 The more  postings you  make, the more cash ends up in
                 your mailbox. It's too easy and too cheap to pass up!!!

     So thats it. Pretty simple sounding stuff, huh? But believe me, it
     works. There are millions of people surfing the net every day, all
     day, all over the world. And 100,000 new people get on the net
     every day. You know that, you've seen the stories in the paper.
     So, my friend, read and follow the simple instructions and play
     fair. Thats the key, and thats all there is to it. Print this out
     right now so you can refer back to this article easily. Try to keep
     an eye on all the postings you made to make sure everyone is
     playing fairly. You know where your name should be.

     If you're really not sure or still think this can't be     for
real,      then don't do it. But please print this article and pass it
     along to someone you know who really needs the bucks, and see
     what happens.

     REMEMBER....HONESTY IS THE BEST POLICY.YOU DON'T
     NEED TO CHEAT THE BASIC IDEA TO MAKE THE BUCKS!
     GOOD LUCK TO ALL, AND PLEASE PLAY FAIR AND YOU WILL
     WIN AND MAKE SOME REAL INSTANT FREE CASH!

 *** By the way, if you try to deceive people by posting the messages
 with your name in the list and not sending the bucks to the people
 already included, you will not get much. I know someone who did this
 and only got about $150 (and that's after two months). Then he sent
 the 5 bills, people added him to their lists, and in 4-5 weeks he had
 over $10,000!

                 TRY IT AND YOU'LL BE HAPPY!!!  :o) !!!!!!!!!!
Article: 4612
Subject: Re: FPGA Gate Counts: No Truth in Advertising
From: Brad Taylor <blt@emf.net>
Date: Wed, 20 Nov 1996 18:15:12 -0800
Links: << >>  << T >>  << A >>
david@lowrance.com wrote:
> 
> A while back one of my co-workers brought me an article he'd clipped
> out of a magazine.  Actel had a new part that would hold 20K gates
> with 100% utilization... or so they claimed.  At the time we were just
> getting started on a small ASIC, about 14K gates.  "Great", I said.
> "We'll be able to create a prototype and use that one FPGA to test our
> entire design."  It was not to be.  The problem was that the 20K gate
> part would barely hold 10K 'real' gates.
> 
> This wasn't my first run-in with an FPGA company whose parts turned
> out to be smaller in real life than advertised.  Last year an ASIC
> which we wanted to prototype in a Xilinx 4000 series part turned out
> the same way.  The FPGA industry has come a long way by learning from
> the ASIC industry.  Design methodologies are constantly improving and
> support for HDLs gaining momentum.  Can someone tell me why the FPGA
> industry is having such a hard time learning to count?

The short answer is greed.

The longer answer is that there really is no good way to relate FPGA
features to ASIC nand gates.  When the answer is undefined, the
marketing guys will always promote the most optimistic estimate. It
probably got started like this.

Marketing:   How big is that new FPGA thing?
Engineering: Well, it has 400 4LUTS:DFFs and registers in the IO blocks.
Marketing:   No, I mean how big is it?
Engineering: Well we have two design which just fit. One is from a 6K
             ASIC and the other is from a 4K ASIC.
Marketing:   (Lets see 6+4 = 10) OK, gotta go.


But seriously folks:
Can someone tell me what is a real gate? 

FPGA feature metrics
-gates/DFF?
-gates/4LUT?
-gates/4SRAM?
-gates/carry_generator?
-gates/tristate_bus_line?
-gates/interconnect_switch?

Software metrics
-gates/16 bit operator?
-gates/line_of_code?

System metrics
-gates/unrouted_fpga
-gates/routed_fpga
-gates/fpga_at_10MHz
-gates/fpga_at_50MHz
-ASIC_gate/FPGA_gate

Financial metrics 
- VHDL_gates/hr
- Schematic_gates/hr
- Debugged_gates/hr
- delivered_to_a_customer_and_used_gates/hr
- Gates/mm_silicon
- Gates/pin
- Gates/$


Just messing around while I wait for a 10K_gate_FPGA to route:

routing metric = 10K_gates/29min * 50%full= 10K gates/hr. 
design metric  =  2 days =  5k/16hrs = 300 gates/hr
debug metric   = I wish I knew!

-
Brad

Article: 4613
Subject: Re: VHDL code editor for Windows NT.
From: Rynier van der Watt <arvdwatt@csir.co.za>
Date: Thu, 21 Nov 1996 06:59:51 +0200
Links: << >>  << T >>  << A >>
Vincent Rowley wrote:
> 
> Hi,
> 
> We are looking for a VHDL code editor for Windows NT platform.
> 
> All information is welcome.
> 
> If you reply to the group, send also a copy to the address
> vrowley@hexavision.com .
> 
> Thanks,
> 
> Vincent Rowley
> 
> --
> -----------------------------------------------------------------------> --  Vincent Rowley                 --  HexaVision Technologies Inc.  --
> --                                 --  2050 Rene-Levesque ouest      --
> --  Email: vrowley@hexavision.com  --  Bureau 101                    --
> --  Tel:   418-686-5000 (233)      --  Sainte-Foy, Quebec, Canada    --
> --  Fax:   418-686-5043            --  G1V 2K8                       --
> -----------------------------------------------------------------------I use Winedit, shareware available from :

http://www.windowware.com

Very usefull editor also includes VHDL code highlighting.

I have written some macros to do compilation and error tracing for 
Modeltec and Viewsynthesis from within the editor, also VHDL templates 
and Automatic test bench generation. If you want these E-mail me at the 
following address:

arvdwatt@csir.co.za

Regards
Rynier van der Watt

Article: 4614
Subject: FPGA TEST BOARDS
From: jjfakas@erols.com
Date: Wed, 20 Nov 1996 21:07:58 -0800
Links: << >>  << T >>  << A >>
I am looking for a low cost FPGA test board which will run on an IBM PC 
and will take a X4010 or X5010 FPGA. I am on a low budget and can't 
afford thousands of dollars. Does anyone know of such a beast? I already 
have the XILINX routing software and synthesis tools.  

Thanks in advance,

Jim
Article: 4615
Subject: Lattice ISP Question
From: Richard Staley <rjs@hep.ph.bham.ac.uk>
Date: 21 Nov 1996 10:24:31 GMT
Links: << >>  << T >>  << A >>
Has anyone made their own download cable for programming devices from an IBM PC
parallel board?
If so , (and it was successful) could you describe the circuit?

Thanks in advance , Richard.

Article: 4616
Subject: Which Mentor Graphics synthesis tool?
From: jorge <jorge@radar-sci.jpl.nasa.gov>
Date: Thu, 21 Nov 1996 10:50:16 GMT
Links: << >>  << T >>  << A >>
Hello and good day,

I'm new to VHDL and even more so to Mentor Graphics tools.

My question is:
Will MG continue to support autologic II?  Or, will they(MG) drop it all
together since I here MG is now going to use EXEMPLAR's synthesis tools?

As a side note, does anyone know when Xilinx xc4ke will be compatible
with Mentor Graphics 8.5?   (I'm not sure if b1 or b2?)

TIA,
--jfg
Article: 4617
Subject: Re: VHDL code editor for Windows NT.
From: John Maher <jmaher@vhdl.com.au>
Date: 21 Nov 1996 11:15:20 GMT
Links: << >>  << T >>  << A >>
Check out the VHDL Editor at http://www.vhdl.com.au

Setanta ED is a VHDL aware editor offering:

- syntax colouring
- template expansion
- Automatic TestBench Generation
- Automatic model insertions 
(hundreds of free models available)
- Automatic Error tracking


A 45 day evaluation licence is available, at 
http://www.vhdl.com.au

-John Maher
Technical Director - VHDL System Solutions P/L 

Article: 4618
Subject: Re: Lattice ISP Question
From: 96kwong@uxmail.ust.hk (Wong Man Kit)
Date: 21 Nov 1996 14:07:50 GMT
Links: << >>  << T >>  << A >>
Richard Staley (rjs@hep.ph.bham.ac.uk) wrote:
: Has anyone made their own download cable for programming devices from an IBM PC
: parallel board?
: If so , (and it was successful) could you describe the circuit?

: Thanks in advance , Richard.

What's your problem exactly? Why not buy download cable from Lattice?
If yuo are interested in ISP design, you are welcome to contact our
UK's Lattice office for help.

--
******************************************************************************
* Wong Man Kit                                                        黃文傑 *
* Applications Engineer                                               工程師 *
* Lattice Semiconductor Asia Ltd                        萊迪思半導體有限公司 *
* Rm 201, 72 Tat Chee Avenue,                                       香港九龍 *
* HKITC, Kowloon,                                             達之路七十二號 *
* Hong Kong                                         香港工業科技中心二零一室 *
* (852)-2319-2929                                            (852)-2319-2929 *
******************************************************************************
* Email address     *      mankit_wong@latticesemi.com      *       電郵地址 *
*                   *        96kwong@alumni.ee.ust.hk       *                *
* Home page         *   http://eelindc1.ee.ust.hk/~eekwong  *           網頁 *
******************************************************************************
Article: 4619
Subject: Re: ViewLogic PRO series under win95
From: Rene Bakker <rbakker@amolf.nl>
Date: Thu, 21 Nov 1996 17:11:36 +0100
Links: << >>  << T >>  << A >>
Gordon McGregor wrote:
> 
> Hi,
>         does anyone happen to know if PRO series is compatable with
>         windows 95 ?  I am running Procapture from the Xilinx PROflow
>         task launcher and as soon as I select add component the system
>         crashes.

It's not compatible with win95. It's almost not compatible with win32s
either. It sort of a dos program which tends to look like windows. I
would
call it PRObug. But that has changed with the new version of ViewLogic's
products: WorkView office.
If you are using Xilinx XACT 6.0.0 or 6.0.1 just stick to win3.11 and
wait
for the update of Xilinx XACT 1Q1997. This update should work with win95
and
NT.
If you already have ViewLogic Workview Office 7.2 you can go to win95
but
the Xilinx Hardware Debugger probably won't work. (I heard of someone
who
told me that it didn't work with a screen saver and did without the
screen
saver.)

> 
>         If you have managed to get PRO series working with win95 could
>         you get in touch, please ?
> 
You would have to use magic to do that.

-- Rene Bakker


P.S. Downgrading from Win95 back to win3.11 isn't fun. 
I have some experience

-- 
------------------------------------------------------------------------
| Rene Bakker         | FOM Institute for Atomic and Molecular Physics |
| Digital engineer    | Electronica & Informatica                      |
|                     | Kruislaan 407                                  |
|                     | 1098 SJ  Amsterdam                             |
|                     | The Netherlands                                |
|                     |                                                |
|                     | Tel:    +31-(0)20-6081234                      |
|                     | Fax:    +31-(0)20-6684106                      |
|                     | E-mail: rbakker@amolf.nl                       |
------------------------------------------------------------------------
Article: 4620
Subject: Re: FPGA Gate Counts: No Truth in Advertising
From: david@lowrance.com
Date: Thu, 21 Nov 1996 11:11:10 -0600
Links: << >>  << T >>  << A >>
In article <3293BB30.15B5@emf.net>,
    Brad Taylor <blt@emf.net> wrote:
> 
> The short answer is greed.
> The longer answer is that there really is no good way to relate FPGA
> features to ASIC nand gates.  When the answer is undefined, the
> marketing guys will always promote the most optimistic estimate. It
> probably got started like this.
> Marketing:   How big is that new FPGA thing?
> Engineering: Well, it has 400 4LUTS:DFFs and registers in the IO blocks.
> Marketing:   No, I mean how big is it?
> Engineering: Well we have two design which just fit. One is from a 6K
>              ASIC and the other is from a 4K ASIC.
> Marketing:   (Lets see 6+4 = 10) OK, gotta go.
> But seriously folks:
> Can someone tell me what is a real gate? 
> FPGA feature metrics
> -gates/DFF?
> -gates/4LUT?
> -gates/4SRAM?
> -gates/carry_generator?
> -gates/tristate_bus_line?
> -gates/interconnect_switch?
> Software metrics
> -gates/16 bit operator?
> -gates/line_of_code?
> System metrics
> -gates/unrouted_fpga
> -gates/routed_fpga
> -gates/fpga_at_10MHz
> -gates/fpga_at_50MHz
> -ASIC_gate/FPGA_gate
> Financial metrics 
> - VHDL_gates/hr
> - Schematic_gates/hr
> - Debugged_gates/hr
> - delivered_to_a_customer_and_used_gates/hr
> - Gates/mm_silicon
> - Gates/pin
> - Gates/$
> Just messing around while I wait for a 10K_gate_FPGA to route:
> routing metric = 10K_gates/29min * 50%full= 10K gates/hr. 
> design metric  =  2 days =  5k/16hrs = 300 gates/hr
> debug metric   = I wish I knew!
> -
> Brad
> 

Your right.  This can be a confusing problem and anybody who wants to 
could come up with new ways of measuring their parts, new metrics that
give whatever appearance they wish.  I do however think it is possible 
to take a logical approach to this issue.

Paul Hardy, a Xilinx FAE, emailed me with his thoughts on the subject.
He included a link to a Xilinx app note which discusses how Xilinx 
counts gates ( http://www.xilinx.com/xapp/xapp059.pdf ).  

The first approach Xilinx took to calculating the Maximum Logic Gates
was to analyse the fundamental building block of their architecture,
the CLB, and determine how it relates to the logic gates of a typical
ASIC vendor.  Here's what they came up with for a 4000 series part:

   CLB Resource               Gates
   -----------------------------------
   4-input LUT (2 per CLB)     1 to  9
   3-input LUT                 1 to  6
   flip-flop (2 per CLB)       7 to 12 
   TOTAL                      17 to 48
   
We can also look at other vendors this way too.  For example, the
Actel 3200 family.  These parts are based on a finer grained 
architecture than the Xilinx 4000 series.  The fundamental block
of the 3200 series part is a Logic Module:

   LM Resource                Gates
   -----------------------------------
   AND                         2
   OR                          2
   4 input mux                 7
   flip flop (in half of LMs)  7 to 12 
   TOTAL                       14.5 to 17

These numbers give us an idea of what is possible (with the perfect 
design). The obvious problem though is that there is no such thing as a 
perfect design.  In the real world we'll never have a design that 
uses every element of every FPGA logic block.  These numbers can still
be very useful however for comparing different FPGA structures.

But the original question still stands.  If I have a design that takes
X number of ASIC NAND gates then how do I know what FPGAs it will fit
into.  I think the only way to determine this is through empirical 
data.  If we were to take a variety of designs of known size (in terms
of ASIC NAND gates) and route them into an FPGA then we could determine
a real-life number representing the average number of gates per logic
module.  This would truely be useful.  And by the way Xilinx claims to
have done this.  According to that app note I mentioned previously 
Xilinx claims they can get an average of 26 gates per CLB.  In my 
experience the actual number is _much_ lower, on the order of 10.  And
yes this means that my design which utilized a 4010 maxed out at about
4K gates... significantly lower than Xilnix's minimum number of 7K.

I should note that buried in the Actel documentation is a table that
gives "Average Gates Per Module".  According to this table a 3200 series
part will hold about 4.8 gates per module.  This is very close to what
I've seen which was about 4.5.  But for the 3200 series part they claim
holds 20K gates if you calculate its size using the 4.8 number from 
their own documentation you only get about 12K. 

I want to encourage everyone to post a reply to this message, including
vender FAEs or reps.  I want to hear about the experiances of others and 
to know what kinds of utilization other designers have seen with Xilinx, 
Actel, or any other FPGA.

David Gardner




-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet
Article: 4621
Subject: Re: FPGA Gate Counts: No Truth in Advertising
From: tomstan@worldnet.att.net (Tom Standley)
Date: Thu, 21 Nov 1996 18:45:49 GMT
Links: << >>  << T >>  << A >>
david@lowrance.com wrote:

>A while back one of my co-workers brought me an article he'd clipped 
>out of a magazine.  Actel had a new part that would hold 20K gates 
>with 100% utilization... or so they claimed.  At the time we were just 
>getting started on a small ASIC, about 14K gates.  "Great", I said.  
>"We'll be able to create a prototype and use that one FPGA to test our 
>entire design."  It was not to be.  The problem was that the 20K gate 
>part would barely hold 10K 'real' gates.
>
>

Couldn't something like the PREP benchmarks (http://www.prep.org) be
used to compare FPGA's?  I'm just starting to look into this but it
appears that they have several standard designs with synthesis results
for several parts from a number of vendors.

Wouldn't a comparison of the synthesis results of several parts using
the same tool set give an indication of the effective size of the part
for the type of design indicated by the benchmark?
Article: 4622
Subject: Re: Async with FPGA?
From: hauck@eecs.nwu.edu (Scott A. Hauck)
Date: Thu, 21 Nov 1996 13:37:14 -0600
Links: << >>  << T >>  << A >>
In article <32937289.4E25@best.com>, Michael Baxter <mbaxter@best.com> wrote:

> "Micropipelines", Ivan Sutherland. In Communications of the ACM, 
> 32(6):720-738, 1989.
> 
> While I haven't looked exhaustively for the paper, there are nonetheless 
> some interesting things on the web regarding the transition signalling 
> and micropipelines. Here's some of what I found:
> 
> http://docs.dcs.napier.ac.uk/DOCS/GET/bailey91b/document.html
> 
> http://paradise.ucsd.edu/class/ece260c/

For those interested in asynchronous circuits in general, as well as some
of the issues involved in mapping asynchronous logic to FPGAs, take a look
at
http://www.ece.nwu.edu/~hauck/asynch.html .

This page includes copies of my 1995 Proceedings of the IEEE Survey of
Asynchronous Design Methodologies (including micropipelines), as well as a
paper on Montage, the first FPGA specifically designed for asynchronous
circuits.

Also, Eric Brunvand has done a lot of work on mapping asynchronous
circuits onto Actel FPGAs.  He doesn't seem to have his papers online, but
the proper references are off his homepage
(http://www.cs.utah.edu/~elb/home.html).

Scott
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
|               Scott A. Hauck, Assistant Professor                         |
|  Dept. of ECE                        Voice: (847) 467-1849                |
|  Northwestern University             FAX: (847) 467-4144                  |
|  2145 Sheridan Road                  Email: hauck@ece.nwu.edu             |
|  Evanston, IL  60208                 WWW: http://www.ece.nwu.edu/~hauck   |
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
Article: 4623
Subject: Electronics question
From: John Rinck <rinck@actel.com>
Date: Thu, 21 Nov 1996 12:52:46 -0800
Links: << >>  << T >>  << A >>
I know this is an FPGA newsgroup, so I post here because a) I don't
know where else to, and b) I figure there would be a great
concentration of EEs in this group.

Anyway, here is my question:

Are there any formulas in electronics that incorporate the
ratio of threshold voltages (Vto) between p-channel and n-channel
devices?  Specifically, (Vtp/Vtn), or (Vtn/Vtp).  

The reason I ask is that I am doing a sensitivity study in
which I increase Vtp by 0.1V while at the same time decrease
Vtn by 0.1V, and I want some way to correlate these two
parameters.  I just blindly thought of Vtp/Vtn but couldn't
think of why that is a good formula.

For that matter, are there any useful equations at all that
use both Vtn AND Vtp in them in any way?

I know that if I only vary either Vtn OR Vtp that I could
use the percent difference, but what about varying two relatively
independent variables such as these?

Thanks,

John

-- 
John Rinck         rinck@actel.com
Web:  http://www.wp.com/RINCK/ (for the Silicon Valley Apartment Guide)
---------------------------------------------------------------
Article: 4624
Subject: Re: Lattice ISP Question
From: Ed Barrett <ed.barrett@worldnet.att.net>
Date: Thu, 21 Nov 1996 13:54:25 -0800
Links: << >>  << T >>  << A >>
Richard Staley wrote:
> 
> Has anyone made their own download cable for programming devices from an IBM PC
> parallel board?
> If so , (and it was successful) could you describe the circuit?
> 
> Thanks in advance , Richard.

Actually, LAttice documents this cable in their, databook and ISP Manual. 
It is really quite simple and should be no problem to build.


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
2017JanFebMarApr2017

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