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 6175

Article: 6175
Subject: Verilog vs VHDL once again.
From: no_spam@Glue.umd.edu (David T. Wang)
Date: 22 Apr 1997 23:57:17 GMT
Links: << >>  << T >>  << A >>
Hello, I'd like to solicit some opinions once again for the age old
Verilog vs VHDL question.  Definitely NOT trying to start another holy war,
but like to hear opinions about these languages.  I'm simply trying to
solicit these opinions on behalf of my advisor.

  Here at the University of Maryland, we've been using an HDL from Japan,
known as Parthenon/SFL (and octools) to go from a high level hardware
description to Silicon.  Those designs are then sent to MOSIS for fab.

  However, since SFL isn't well known, or widely used here in the US, 
We're probably going to emphasize either VHDL or verilog more.  Currently,
we're using VHDL under Mentor Graphics for design work, though we haven't
submitted anything for MOSIS fabrication.  Soon, we will be getting
Cadence.  

  Our target device will be various types of FPGA's that we are just
starting to get, Motorola, Xilinx, and perhaps a few others, We would
also like to be able to compile down to a VLSI layout, and submit it
to MOSIS for fabrication.  The general feeling by just about everyone
I have spoken with is that Verilog with Cadence, with synopsis Libraries
is the direction that we should be going.  

  Any suggestions?  opinions?

  please post or email davewang@wam.umd.edu, or kazuo@eng.umd.edu
 


Article: 6176
Subject: ISP CPLD from AMD or Cypress???
From: Timothy Del Sol <tdelsol@primenet.com>
Date: 22 Apr 1997 21:37:00 -0700
Links: << >>  << T >>  << A >>


Hello all,

I have code for a couple of 22V10 PALs written in PALASM.  I want to
consolidate the PALs into a CPLD and add some other logic.  To do this I
have to convert the PALASM code to VHDL.  My problem is I'm looking into
using the AMD's ISP Mach family or the Cypress ISP Flash370 family.  I
need to know what are the pros and cons for both parts.  Which one has
better the better design tools? The Mach software is free.  The Flash370
is $99.  Both have ISP.  What's the price and performance difference
between the two?

Thanks,

Del Sol

Article: 6177
Subject: Re: ISP CPLD from AMD or Cypress???
From: Kayvon Irani <kirani@cinenet.net>
Date: Tue, 22 Apr 1997 22:40:59 -0700
Links: << >>  << T >>  << A >>
Timothy Del Sol wrote:
> 
> Hello all,
> 
> I have code for a couple of 22V10 PALs written in PALASM.  I want to
> consolidate the PALs into a CPLD and add some other logic.  To do this I
> have to convert the PALASM code to VHDL.  My problem is I'm looking into
> using the AMD's ISP Mach family or the Cypress ISP Flash370 family.  I
> need to know what are the pros and cons for both parts.  Which one has
> better the better design tools? The Mach software is free.  

	Tim, which MACH software are you talking about? Last time I checked into 
	buying the complete VHDL toolset + fitter + timing analyzer for MACH5 the
	cost was over 5 grand! I guess the problem is AMD does not produce any of
	these tools not even the fitters for new devices. As far as Cypress tools,
	they produce the entire toolset and therefore can sell it at a lower price
	hoping they'll make that up in silicon sales. I have seen both toolsets and
	they both have their high and low points but overall they both do a descent
	job.

	Regards,
	Kayvon Irani
	Los Angeles, Ca
Article: 6178
Subject: Re: ISP CPLD from AMD or Cypress???
From: "David Wang" <lionking@cheerful.com>
Date: 23 Apr 1997 05:56:59 GMT
Links: << >>  << T >>  << A >>
Timothy Del Sol <tdelsol@primenet.com> wrote in article
<335D90EB.31E7@primenet.com>...
>  
> better the better design tools? The Mach software is free.  The Flash370
> is $99.  Both have ISP.  What's the price and performance difference
> between the two?
> 

Hi, Timothy:
I think Lattice ispLSI/pLSI is your best choice!
All lattice CPLDs have ISP function, and Lattice ISP is free now! (the unit
price of ispLSI device is the same as non-ISP pLSI device).
Lattice ispLSI device's DATA retension is more than 20 years, and more than
10,000 Erase/Reprogram cycles, (AMD only 100 cycles).
The price is also lower than other vendor's. (in my country:-)
For sw, Lattice now has free CD of ISP Synario Starter sw, you can request
it from your local Lattice sales office or distis/reps.

You can find more information from (http://www.latticesemi.com)

good luck!
David
Article: 6179
Subject: Reconfigurable Computing
From: pirger@astrosun.tn.cornell.edu (Bruce Pirger)
Date: 23 Apr 1997 06:00:17 GMT
Links: << >>  << T >>  << A >>
Greetings all:

	To those reading this group with a large interest in RC, is there
a different forum dedicated to RC?  There appears to the little discussion
about RC, existing platforms, developments, plans given DARPA ACS funding,
etc.  
        Please fill me (us) in.


Thanks,

Bruce

Article: 6180
Subject: Re: FPGA gate counting: No truth in advertising
From: Kayvon Irani <kirani@cinenet.net>
Date: Tue, 22 Apr 1997 23:03:41 -0700
Links: << >>  << T >>  << A >>
> Therefore, our pASIC1 lineup looks like this:
> 
> pASIC1 device    # logic cells    2-input NANDs    usable gates
> QL8x12B              96              2,976             1,056 (1k)
> QL12x16B            192              5,952             2,112 (2k)
> QL16x24B            384             11,904             4,224 (4k)
> QL24x32B            768             23,808             8,448 (8k)
>  
	Any comments on how these compare to CPLD gates/logic cell counts? Recently,
	I had a design done in VHDL with a bunch of up/down counters (count up by two 	
	down by one) and some simple I/O logic. The design fit in QL12x16B with 84% 	
	utilization; the same design also fit in a Xilinx xc95108(108 logic cells) with
	98% utilization. Both designs were run through the same synthesis tool with
	area optimization optin selected. Any comments?

	Kayvon Irani,
	Los Angeles
Article: 6181
Subject: CFP Design and Test (FPGA issue)
From: fmeyer@cs.tamu.edu (Jackie Meyer)
Date: 23 Apr 1997 12:26:35 GMT
Links: << >>  << T >>  << A >>
DEADLINE:  MAY 1

                        Call For Papers:  FPGA
                  IEEE Design and Test of Computers
                            Special Issue
                   Spring 1998, Volume 15, Number 1
     Web Page: http://www.cs.tamu.edu/faculty/lombardi/datsi.htm

IEEE Design and Test of Computers seeks original manuscripts for a
theme issue on Field Programmable Gate Arrays (FPGAs) scheduled to
appear in the first issue of 1998.  Articles concerning applied
research and practical experience reports are solicited.  The topics
of interest include, but are not limited to:

   FPGA Fabrication and Technology.  Manufacturing, process control,
   yield enhancement, and novel architectures for and device
   technology to support field programmable chips.

   Exploitation of Field Programmability.  Configurable computer
   architectures, rapid prototyping, programmable interconnect
   architectures, field configurable memories, programmable I/O
   systems, and novel implementations.

   Reliable Online Implementations.  Online testing, built-in
   self-test, concurrent testing, design for rapid testability,
   latency reduction, fault containment, verification of
   reprogramming, online reconfiguration approaches, and design for
   reconfigurability.

   Synthesis Approaches and Tools.  Partitioning, logic minimization
   and technology mapping, placement and routing, test generation,
   verification of synthesis, and design for synthesizability.

Submitted articles must not have been previously published or
currently submitted for publication elsewhere.  Authors should submit
their original work to the guest editor by May 1, 1997, formatted
according to the instructions below.  Notification of acceptance will
be sent September 1, 1997.  Camera-ready copy for accepted papers will
be due November 1, 1997.

Submit articles by May 1, 1997 to:

Prof. Fabrizio Lombardi, Guest Editor
Department of Computer Science                  Phone:  (409) 845--5464
Texas A & M University                            Fax:  (409) 847--8578
College Station TX  77843-3112            E-Mail:  lombardi@cs.tamu.edu

Important dates:

        May 1, 1997:  Submission deadline
  September 1, 1997:  Authors notified of acceptance
                      with requested revisions
   November 1, 1997:  Final copy due to Design & Test Managing Editor
        Spring 1998:  Publication in IEEE Design and Test of Computers

Submission requirements:

Send six (6) copies of the manuscript, in English, to the guest
editor.  Manuscripts are not to exceed 35 double-spaced pages,
inclusive of figures and tables, in A4 or 8.5 by 11 inches.  Type size
must be at least 12 point.  Each copy of the manuscript must contain a
cover page with author contact information (name, postal address,
telephone number, and e-mail address) and a 100-word abstract.
Manuscripts must be cleared for publication.  Accepted manuscripts
will be edited for technical content, structure, style, clarity, and
grammar.  Detailed information for authors can be found at the Computer
Society D&T website at http://www.computer.org/pubs/d&t/d&t.htm or in
the Spring 1996 issue of Design & Test.
Article: 6182
Subject: Download Xilinx Fpga (XC4***)
From: jmaus@elis.rug.ac.be (Jurgen Maus)
Date: 23 Apr 1997 13:56:07 GMT
Links: << >>  << T >>  << A >>

Can somebody tell me if it`s possible to download a bitstream to a XC4000
component with a Xchecker cable that`s supposed to be used for a XC3000
or a XC2000 component?

If it`s possible, how can I do it?

Article: 6183
Subject: Re: ISP CPLD from AMD or Cypress???
From: "Dan Kuechle" <dan_kuechle@i-tech.com>
Date: 23 Apr 1997 14:07:49 GMT
Links: << >>  << T >>  << A >>


Kayvon Irani <kirani@cinenet.net> wrote in article
<335DA0EB.7240@cinenet.net>...
> Timothy Del Sol wrote:
> > 
> > Hello all,
> > 
> > I have code for a couple of 22V10 PALs written in PALASM.  I want to
> > consolidate the PALs into a CPLD and add some other logic.  To do this
I
> > have to convert the PALASM code to VHDL.  My problem is I'm looking
into
> > using the AMD's ISP Mach family or the Cypress ISP Flash370 family.  I
> > need to know what are the pros and cons for both parts.  Which one has
> > better the better design tools? The Mach software is free.  
> 

Last time I checked the AMD mach ISP parts were many times more expensive
than the Cypress equilevents.  I've used the Cypress CY7C374i parts with
the
$99 software (plus a $99 programming pod) and have been very happy with
them.
Lots of routing resources.
Article: 6184
Subject: Re: Reconfigurable Computing
From: jmarnold@potomac.znet.com (Jeffrey M. Arnold)
Date: 23 Apr 1997 07:24:01 -0700
Links: << >>  << T >>  << A >>
In article <5jk8hh$jh2@newsstand.cit.cornell.edu>,
Bruce Pirger <pirger@astrosun.tn.cornell.edu> wrote:
>Greetings all:
>
>	To those reading this group with a large interest in RC, is there
>a different forum dedicated to RC?  There appears to the little discussion
>about RC, existing platforms, developments, plans given DARPA ACS funding,
>etc.  
>        Please fill me (us) in.

This (comp.arch.fpga) is the usenet forum for discussion of
reconfigurable computing, although it's difficult to tell from the
content these days.  The original charter of c.a.f states:

  The unmoderated newsgroup comp.arch.fpga will be open to discussions
  on all topics related to the use of reconfigurable Field
  Programmable Gate Arrays (FPGAs) as computational engines.
  Appropriate topics include, but are not limited to:
      system architecture
      FPGA device architecture
      languages and compilation techniques
      tools
      software environments
      applications

At the time this newsgroup was formed (during the "request for
discussion" period) there was a great deal of discussion about the
name of the group.  Some felt that "FPGA" was too technology specific,
while others maintained that since there was no concensus name for the
field, FPGA would suffice if the group were put under the "comp.arch"
hierarchy.  So here it is.

With the increasing interest in custom computing as evidenced by the
record attendance at FCCM and the DARPA ACS program, perhaps we will
see an increase in releated postings in comp.arch.fpga.

-jeff
-- 
Jeffrey M. Arnold		jma@super.org or jmarnold@znet.com
10686 Mira Lago Terrace		Tel: 619-547-9257
San Diego, CA 92131		Fax: 619-547-9010
USA
Article: 6185
Subject: IWLS 97: Advance Program and Registration
From: sharad@new-delhi.Princeton.EDU (Sharad Malik)
Date: 23 Apr 1997 17:43:54 GMT
Links: << >>  << T >>  << A >>
The registration information and the advance program for the International
Workshop on Logic Synthesis (IWLS) '97 is available off the web at:

	http://www.ee.princeton.edu/iwls97.html

Sharad Malik.
-- 
Sharad Malik                            sharad@ee.princeton.edu
Associate Professor                     609-258-4625
Dept. of Electrical Engineering         609-258-3745 Fax
Princeton University                    http://www.ee.princeton.edu/~sharad
Article: 6186
Subject: Re: ISP CPLD from AMD or Cypress???
From: Andy Whitehouse <a.whitehouse@dmv.co.uk>
Date: Wed, 23 Apr 1997 21:21:35 GMT
Links: << >>  << T >>  << A >>
Kayvon Irani wrote:
> 
> Timothy Del Sol wrote:
> >
>         they both have their high and low points but overall they both do a descent
>         job.
> 
>         Regards,
>         Kayvon Irani
>         Los Angeles, Ca

Descent job ?

You mean they fall over often ?
Article: 6187
Subject: XC52xx and Hardware Debugger
From: stuart.summerville@practel.com.au (Stuart Summerville)
Date: Thu, 24 Apr 1997 02:39:12 GMT
Links: << >>  << T >>  << A >>

Hi all,
Below is a report I sent to xilinx support about a problem I am having
with the hardware debugger. 
While Xilinx tech support have been most cooperative, I still have not
had the problem resolved. If anyone can help me on this, I would be
most appreciative - time is a-waistin'.....

--------------------------------------------------------------------------------------------------------------------
Version of Software: V7.12 XactStep V6.0.1
OS: WFW311/Win32S13C
HW Config: P5-150, 48MB RAM, 2G Disks
Installation: Local HD
Notes: The whole translation/compilation process in XDM works fine, 
and I have included the Readback symbol in the schematic. Note that 
the RDCLK symbol is removed during the compilation process. I assume 
that the compilation automatically uses CCLK as stated in the 
documentation.

Problem:
======
 I have a question re. the hadrware debugger that comes with Xact
V6.0.1. When I load the bitstream file (root.bit) into the debugger, I
get the message:

 "The logic allocation file (.ll file) for the specified design was 
not found. Debugging options will b disabled"

The file (root.ll) is in the directory 
(d:\test\test\v1_8\rev1\root.ll) and has standard file attributes.
Even if I copy the file to other directories higher up the tree, or
into Xact/Data as suggested in the Hardware & Peripherals Guide, I
have no success.  This Guide says to ensure that the file is in either
the xact/data directory (why, when each different version of the
design would probably have different logic layouts????) or the current
dir (?? presumably the v1_8\rev1 directory quoted above.). 

Am I looking at the wrong warning in the Guide? I am looking at 
Warning 005 "LL file filename.ll is not found. Design verification
will not work correctly.",  P3-69.  The warning has the same apparent
meaning, but differs in detail: The message that comes up in the
debugger does not quote a warning number, and  is worder slightly
differently.

I have looked through Xilinx Answers on Xilinx's web site, & it 
merely points me to this info in the Hardware & Peripherals guide.

--------------------------------------------------------------------------------------------------------------------

Regards, Stu.
---------------------------------------------
Stuart Summerville      
Project Engineer         
Practel International
442 Torrens Road, Kilkenny, SA 5009
Tel: (61.8) 8268 2196  Fax: (61.8) 8268 2882
Email: stuart.summerville@practel.com.au  
---------------------------------------------
Article: 6188
Subject: Primes / long integer multiplication
From: Mike Butts <mbutts@realizer.com>
Date: Wed, 23 Apr 1997 20:03:39 -0700
Links: << >>  << T >>  << A >>
At FCCM, co-chair Ken Pocek of Intel challenged us to go after
the search for Mersenne prime numbers, which are the biggest 
of all known primes.  A Mersenne prime is of the form 2^n-1.
The current biggest is 2^1398269-1, found by the Great
Internet Mersenne Prime Search project, which has lots of people
using their PCs' spare cycles to run prime-searching free
shareware. http://www.mersenne.org

The hard compute problem is to multiply really big integers
very fast.  Really, really big.  Millions of bits long each.
Ordinary multiplication is an order-n^2 algorithm, which
blows up on such numbers.  There are many better algorithms,
including some based on using FFTs.  (See Knuth Vol. 2)

Most of the algorithm development lately has naturally been 
directed towards algorithms that run well on conventional 
computers, which have lots of memory, fast pipelined floating 
point, but narrow, fixed-length words and the dreaded 
memory-processor bottleneck <choke>.

With FCCMs, that is FPGA-based machines, those constraints
are reversed.  We can program thousand-bit multipliers,
parallel FFT accelerators, cellular automata, systolic
arrays, whatever.  What are good algorithms to use for
very long word multiplication on FPGAs?  Ones which get
faster when more FPGAs are added?  Ones which get really
fast with tons of FPGAs?  FPGAs are always growing, you know.

Who will be the first to discover the largest prime so
far using FPGAs?

   --Mike
Article: 6189
Subject: Re: Pentium Pro Worth it for Altera Max Plus?
From: charon@pacifier.com (Charon)
Date: 24 Apr 1997 06:56:37 GMT
Links: << >>  << T >>  << A >>
George (gzs@explorer2.clark.net) wrote:
: 
: >Both available memory and processor speed will enhance your performance,
: 
: how about these dual or quad pentium or pentium-pro boards
: 
: is it realistic to expect a speedup for the Altera or Xilinx
:   software compared with the uni-processor machines if running
:   NT - (im a unix person I do not know what kind of multi-tasking
:   NT really has for running other jobs simultaneously)

I think it was discussed some time ago -- Max+Plus II doesn't 
take advantage of more than one CPU.  But if you multitask, 
one CPU can be used exclusively for compiling while the other's 
free to do other things.

--- 
Reply-To: and From: fields have been altered to avoid spam.

Article: 6190
Subject: Re: Download Xilinx Fpga (XC4***)
From: fliptron@netcom.com (Philip Freidin)
Date: Thu, 24 Apr 1997 09:37:50 GMT
Links: << >>  << T >>  << A >>

The Xchecker cable, DLC4, can be used with XC2000, XC3000, XC4000, and 
XC5000, products. The FPGA is setup for Serial Slave mode configuration, and
in the case of the XC4000, you will need to connect the following signals 
from the DLC4 to your target system:

Vcc, Gnd, CCLK, D/P (connect to the done pin, with a pullup resistor as 
well), Din, Prog, and Init. The rest of the DLC4 signals can be left 
unconnected. This is shown in my "Xact Hardware and Peripherals Guide".
The hard copy has the info on page 5-18. The online copy (with V6.0.1 SW)
has it on page 3-20


All the best,
Philip Freidin.

In article <5jl4dn$lp9@infoserv.rug.ac.be> jmaus@elis.rug.ac.be (Jurgen Maus) writes:
>
>Can somebody tell me if it`s possible to download a bitstream to a XC4000
>component with a Xchecker cable that`s supposed to be used for a XC3000
>or a XC2000 component?
>
>If it`s possible, how can I do it?
>


Article: 6191
Subject: EDIF200
From: Thomas Finke <finke>
Date: 24 Apr 1997 14:33:53 GMT
Links: << >>  << T >>  << A >>
Hi all,

I am looking for a full discription of the EDIF200 netlist-standard.
Does anybody know, where to get it ?

Regards, Thomas                 email: finke@iti.mu-luebeck.de

Article: 6192
Subject: Re: EDIF200
From: Christian Cartland - Glover <cartland@apg.philips.co.uk>
Date: Thu, 24 Apr 1997 14:58:39 GMT
Links: << >>  << T >>  << A >>
Thomas Finke wrote:
> 
> Hi all,
> 
> I am looking for a full discription of the EDIF200 netlist-standard.
> Does anybody know, where to get it ?
> 
> Regards, Thomas                 email: finke@iti.mu-luebeck.de

You could try the edif web site at http://www.edif.org/

Christian Glover.
email: cartland@apg.philips.co.uk
Article: 6193
Subject: Re: prep benchmarks for FPGAs
From: waynet@goodnet.com (Wayne Turner)
Date: Thu, 24 Apr 97 15:33:33 GMT
Links: << >>  << T >>  << A >>
In article <peter-1404971535550001@appsmac-1.xilinx.com>, peter@xilinx.com 
(Peter Alfke) wrote:
>In article <5iocgn$thg$1@news.goodnet.com>, waynet@goodnet.com (Wayne
>Turner) wrote:

--- begin included text ---------------------------------------------------
[...]
>If you promised to measure, you measure honestly and don't use any magic
>tricks with disappearing logic.
>
>Peter Alfke, speaking for himself
--- end included text ---------------------------------------------------

I've included the above text so everyone knows what my followup below was 
directly referring to, namely Mr. Alfke's desire for honesty in measurement.  
He left that part out of his post for some reason.  The following was my post 
about disappearring logic and honesty in measurement:

>> Then do you also disagree with how your company seems to report macrocell 
>> usage in your 9500 family, where even if all products terms for a macrocell 
>> are borrowed

He also left out the rest of this statement, which continues:

"the report shows the macrocell as not used, though there are no 
products term left in it?  I found that to be an interesting measurement 
itself.   Correct me if I'm wrong, but if I'm not then what is your opinion   
on that?"

>That was a cheap shot.

Really?  Mr. Alfke's first response to my question was the private unsolicited 
email he sent me after my post, which is included here in its entirety:

--- begin included email --------------------
"Wayne, I agree with you, if it's the way you say.
Borrowing from a macrocell is, of course a "soft" thing. If I borrow         
one or two terms, I have not really killed it. But borrowing all five         
makes the macrocell useless. 
I will look into this. I am all for honesty.
Thanks for pointing this out.
Peter Alfke"
--- end included email ----------------------

At the time, I appreciated his response, and he obviously didn't feel my 
question was a "cheap shot" and even thanked me for pointing it out.  I later 
emailed him and merely asked if he had any new news on the situation, and he 
responded with this post which he also emailed to me.

>I had complained about Altera using deceptive methods in PREP, where
>potential users are supposed to find a fair and honest comparison between
>technologies and manufacturers.
>
>Wayne then compares that to the XC9500 fitter reporting only the number of
>macrocells used, without consideration of how many product-terms were
>³stolen² from the neighbor(s).

Actually, my post was in reference to "disappearring logic", which is what you 
occassionally get if you try to use the remaining macrocells that seem to be 
unused in a XC9500 design report file.  You may think you have half the 
macrocells left, but when you try to use them, they disappear because the 
logic in the front of them was gone already.

>But that is a detailed report that you get AFTER you have done the design,
>and the same report also specifies product terms and I/Os used. The number
>of macrocells used is just one ingredient, and there really is no fair way
>to report it any differently.
>
>1. I/O pins limit how many signals may be brought into and generated by
>the device. 
>
>2. Macrocells limit the size of the design with regard to how many logic
>functions may be implemented. 
>
>3. Product terms determine the complexity of the design with regard to
>the complexity of each implemented logic function.
>
>Any CPLD design, regardless of vendor, may be I/O limited, macrocell
>limited, or product-term limited. Therefore, we report I/O, macrocell,
>and product term usage in the header of the fitter report.

I still fail to see how a device can have any "macrocells" left if all of the 
product terms are used.  And I also wonder if anyone has seen an apps guy use 
the macrocell usage statistic to show less device usage over a competitor for 
the same design.  Let's face it, when parts are numbered by number of 
macrocells, macrocell usage is what people are going to look at first.  The 
problem is, most people assume a macrocell has LOGIC associated with it; it's 
generally not just an unused register.

>To take this a step further, exporting all 5 product terms from a macrocell
>does not necesssarily make that macrocell unusable. That same macrocell
>that exported all its product terms, can import product terms from other
>macrocells that are implementing less complex logic functions. ( I have
>heard that the French import some red wine from California ).

A quote from Mr. Alfke's email to me, also shown above:

"Borrowing from a macrocell is, of course a "soft" thing. If I borrow         
one or two terms, I have not really killed it. But borrowing all five         
makes the macrocell useless. "

So is it useless or not?  You've said both...

And one cannot indefinitely borrow product terms; eventually you run out of 
macrocells, right?  Otherwise all of those chain-letters to ask 10 friends to 
send you a dollar would really work and we would all be millionaires ;)

>This subject has no relationship to Alteras ³manipulation of the truth² in
> PREP.

I would think all manipulations of the truth are related...

The point is, I personally feel you are probably an honest person yourself, 
but your perception is colored to the point where you think YOUR company 
doesn't do the same marketing types of things as you beat on your competitors 
for supposedly doing.

As someone else said so well in an earlier post, "let he who is without a 
marketing department cast the first stone"...

Wayne
Article: 6194
Subject: Re: Instatiation of Xilinx Primitives in VHDL?
From: Kate Meilicke <kate.meilicke@xilinx.com>
Date: Thu, 24 Apr 1997 11:42:09 -0400
Links: << >>  << T >>  << A >>

Yves Blaquiere wrote:

> How can these designs, generated with xbloxgen and memgen, be
> simulated in Synopsys. I think that instantiation is useless if their
> behavior cannot be simulated before place and route.

The XACT 5.2.1 software doesn't provide simulation models for
xbloxgen and memgen.  The only solution with this version of
software is to do a quick place and route - router/placer effort
of 1, back annotate and do simulation.  It is not a good solution.

The real fix is the next version of software - M1.  xbloxgen and
memgen are replaced with LogiBLOX.  LogiBLOX provides Verilog
or VHDL for functional simulation.

Talk to your local FAE or Sales Rep about getting the new
version of software.

Kate Meilicke
Xilinx FAE
Article: 6195
Subject: Re: System Level Integration on Deep submicron FPGAs
From: Kate Meilicke <kate.meilicke@xilinx.com>
Date: Thu, 24 Apr 1997 12:19:44 -0400
Links: << >>  << T >>  << A >>
Kayvon Irani wrote:
> 
> Hello every one:
>Are the FPGAs in a position to offer a cost-effective hassle-free
>         solution to system level integration compared to ASICs and gate arrays for
>         densities below 100K and volumes of less than 5K? Can the tools easily handle
>         different IP cores from different providers? Has any one implemented multiple
>         IP cores in one FPGA? Any comments are appreciated.


Check out the Xilinx web page:

http://www.xilinx.com

Click on Systems Level Integration and then Alliance Core Program. 
There are several
Third party vendors that have cores for Xilinx FPGAs.

Kate Meilicke
Article: 6196
Subject: Escalade anyone?
From: Dirk Brandis <"brandis<SPAM>"@dlcc.com>
Date: Thu, 24 Apr 1997 09:32:46 -0700
Links: << >>  << T >>  << A >>
Anyone have experience with the Escalade Design Book (v 2.1) VHDL front
end tools?  I'm aware the earlier releases had problems, but how about
the new one?

NOTE: remove <SPAM> from reply email address if responding via email

THANKS!!
Article: 6197
Subject: Re: prep benchmarks for FPGAs
From: gah@u.washington.edu (G. Herrmannsfeldt)
Date: 24 Apr 1997 17:25:16 GMT
Links: << >>  << T >>  << A >>
Some time ago, I was working on an FPGA implementation of a systolic
array, which was a serial array of adders and maximum functions.
I implemented a saturating adder and max function as xilinx macros.
The design also had a lookup table for each unit cell.  Well, 
a unit cell was five 16 bit saturating adders, six 16 max functions,
and an n by 8 lookup table.  I could get four such unit cells into a XC4013.

It was interesting to see what optimizations PPR could do.  

Earlier in the design, I was using adders and maximum functions designed
by a verilog compiler, which required, and allowed, more optimization.

The main optimization that is allowed is that the comparison required by
a maximum function is all generated by the carry logic, and the LUT are
free for the multiplexer.  This is how I designed it as a macro.  But between
the verilog compiler and ppr, it did manage to reuse some, but not all, of 
the LUT in the subtractor for doing the compare.


Anyway, I think that any optimizations a compiler can find in real problems
should be legal in benchmarks.  If they have to be hand optimized, then
that probably isn't fair.  But then some real desings need hand optimizing,
also.

Maybe, like the SPEC processor benchmarks, some real designs need to be
used.  How about an implementation of the Intel 8086?  For smaller ones,
the i4004?  Maybe other new or old processors could be used, too.

Maybe there are better real-world designs, but I think, like SPEC, they
should reflect what people really use them for.

-- glen
Article: 6198
Subject: Call for participation, Advanced PLD & FPGA Day UK and Sweden
From: Peter Clarke <pclarke@lfields.demon.co.uk>
Date: Thu, 24 Apr 1997 21:46:46 +0100
Links: << >>  << T >>  << A >>

             /\
            /  \        /\
           /\   \     /    \
          /  \   \  /        \
         /    \   /            \
        /      \  \            /
       /        \   \        /
      /__________\   \\    / 
     /------------\   \ \/
    /              \   \ 
   /                \   \
------             ----------
 
THE ADVANCED PLD & FPGA DAYS, 97 


                     Call for participation at the 

		     7th Annual Advanced PLD & FPGA 

 		       Conference and Exhibition, 

		 May 13th, May 14th, The Ascot Pavilion
			Ascot Racecourse, UK., 

                  May 20th, Stockholmsmassan, Stockholm



The event is organized by Miller Freeman plc.
30 Calderwood St., London SE18 6QH


 
Details and on-line registration are available via the World Wide Web at 
URL = http://www.pld97.co.uk


May 13th

2.00 - 5.00pm 	TRAINING MODULE 
		"Introduction to PLDs and FPGAs"
		Workshop run by University of Kent

14th May

9.00am		Registration & Exhibition opens

9.20am		Introduction and best paper award for APFD '96

9.30 - 10.00am	KEYNOTE PRESENTATION
		"Challenges for the CPLD and FPGA market"
		Prof. Gabriele Saucier
		Institut National Polytechnique Grenoble

The remainder of the day is divided into two streams denoted A and B

10.20am	A 	"Efficient implementation of telecoms IP cores in system 
		programmable gate arrays", Stef Niewiadomski, Actel

	B	"Advanced high-level HDL design techniques for 		
		programmable logic", Darron May, ALT Technologies

10.50am	A	"Three FPGA systems and subsystems clocked at 200MHz",
		Peter Alfke, Xilinx

	B	"How much VHDL simulation is needed for a complex FPGA", 
 		M. S. Abrahams, TransEDA

11.20am		C O F F E E   B R E A K 

11.50am	A	"Flek 10K: extending the boundares of programmable 	
		logic", Tim Symons, Altera

	B	"Design methodologies for dynamically reconfigurable 	
		FPGAs", Joel Rosenberg, Atmel

12.20pm	A	"The 100Kgate plus FPGA design challenge", 
		Steve Collis, Mentor Graphics

	B	"Video motion tracking using Harp and Handel: a case 	
		study", Jonathon Saul, Oxford University Computing 	
		Laboratory

12.50pm		L U N C H   B R E A K	

2.15pm	A	"Implementation of a PCI-bus target interface in a high 
		density FPGA", Adam Elbirt, Viewlogic Systems 

	B	"FPGAs: the value of reprogrammability", 
		Thomas Oelsner, Quicklogic

2.45pm	A	"Complex RTL-to-silicon time scales using 		
		laser-programmable gate arrays" 
		Nick Heaton, Excel Consultants

	B	"A designers guide to selecting in-system programming"
		Mark Aaldering, Philips Semiconductors

3.15pm		B R E A K

3.45pm	A	"Techniques and tools for FPGA-based DSP design"
		Bradly Fawcett, Xilinx

	B	"SCUBA: a tool for efficient ORCA FPGA data-path/memory 
		synthesis", Tom Smart, Lucent Technologies

4.15pm	A	"CPLDs and FPGAs: is there really a difference",
		Peter Trott, Vantis

	B	"Timing-driven optimization of FPGAs using top-down 	
		design", Doug Perry, Exemplar Logic

4.45pm		C O N F E R E N C E   E N D S

-- 
Peter Clarke
Programme Co-ordinator

Article: 6199
Subject: Re: prep benchmarks for FPGAs
From: Ray Andraka <randraka@ids.net>
Date: Thu, 24 Apr 1997 18:18:05 -0400
Links: << >>  << T >>  << A >>
> [...]
> >If you promised to measure, you measure honestly and don't use any magic
> >tricks with disappearing logic.
> >
> >Peter Alfke, speaking for himself
> --- end included text 

"Magic tricks with disappearing logic" is what reconfigurable computing
is all about!  So, does that make me dishonest??? (deliberately taken
completely out of context)

-Ray Andraka, P.E.
Chairman, the Andraka Consulting Group
401/884-7930     Fax 401/884-7950
email randraka@ids.net
http://www.ids.net/~randraka


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