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 17300

Article: 17300
Subject: Re: License sharing for synopsys/cadence/modeltech
From: Tim Davis <timdavis@tdcon.com>
Date: Mon, 19 Jul 1999 17:26:26 -0600
Links: << >>  << T >>  << A >>
1. I guess the real answer is what will happen when you are caught?
2. If the company already has the software then there wouldn't be a copyright
violation. Otherwise for them to have a copy of the software would be a
copyright violation -- That is the criminal part of the equation. For them that
is.
3. If you are charging for the use of the license then you would probably have
to move to Borneo and put on a sweaty ape suit to hide from the S Police.

chipfactory@hotmail.com wrote:

> Hi Folks:
> When we read the documentation for Synopsys we found out that
> a license server can also be somewhere in the internet. Now we
> came up with the idea to connect our Synopsys/Cadence/Modeltech
> license server to the internet and share our licenses with others.
> During the night here in Europe our licenses are not used.
> Would there someone be interested and is this legal?
>
> Best regards
> Jack
>
> email: chipfactory@hotmail.com

--

Tim Davis

Article: 17301
Subject: Re: Frequency multiplier in XC4000
From: Peter Alfke <peter@xilinx.com>
Date: Mon, 19 Jul 1999 16:50:41 -0700
Links: << >>  << T >>  << A >>


bill morris wrote:

> Has anybody tried to implement a frequency multiplier in XC4k? I need to
> implement a frequency multiplier by 2^n (n=1,2,3,4). I know VIRTEX contain
> DLL's which can implement that but unfortunately I have to use XC4kE.
>
> Please help!

Bill, I drew a circuit that might do the trick:

There is an internal oscillator consisting of a 2-input NAND gate driving a
number of non-inverters feeding back to one input.
The other input is driven from the Q of a Control  flip-flop clocked by your
incoming clock, its D is permanently High.

You use the internal oscillator to clock a counter that is one bit too long (
e.g. 4 bits for divide by 8 )
The MSB of this counter drives the asynchronous Reset input of the  Control
flip-flop. The Qbar output of the Control flip-flop is an active High
asynchronous Reset input to all counter bits, except its an asynchronous PRESET
for the LSB.

That should do it.

Incoming clock edge sets Control flip-flop, thus starting oscillator and raking
away the asynchronous preset to a count value of 1.
When counter reaches 2-to-the-n, the MSB clears the CONTROL flip-flop, which in
turn presets the counter to a 1 and holds it there until the next rising edge of
your main ( slow) clock.

Let me know when you got any further with this.

Regards

Peter

Article: 17302
Subject: Re: License sharing for synopsys/cadence/modeltech
From: "Andy Peters" <apeters@noao.edu.NOSPAM>
Date: Mon, 19 Jul 1999 18:40:36 -0700
Links: << >>  << T >>  << A >>
My two cents:

Ignoring possible export restrictions (due to encryption, "national
security," etc.), you've paid for x number of license tokens.  Why should
the vendor care where they're run?  As long as the number of simultaneous
users doesn't exceed the number of licenses you've paid for, what's the
difference?

I have the (floating) FlexLM license daemon running on my NT box at work.  I
can dial into our network from home.  My home machine has a static IP
address, and I can running the Xilinx tools and FPGA Express without any
problems.  All of the design files are on my home machine; the only network
traffic is what's needed to deal with the license.

While on the topic of licensing, what about the following:  for one seat,
node-locked licenses are generally about half the price of a floating
license.  Say you have a Unix network, and you install the license server
and the tools on a machine called gromit.  Give any potential users a login
on gromit.  To use the tools, all one has to do is rlogin to gromit and do
your work.  (Cross-mount drives and make it even easier to handle files
across the network.)  So, if you have a handful of people who need access to
the software and they don't require full-time access, purchasing one
node-locked license and doing the rlogin seems like one way of saving some
cash.

'course, ya can't do that trick on NT.  And, of course, the Unix licenses
are generally about double the price of the NT licenses, so...

--
-----------------------------------------
Andy Peters
Sr Electrical Engineer
National Optical Astronomy Observatories
950 N Cherry Ave
Tucson, AZ 85719
apeters (at) noao.edu

"You want partial credit?  You build bridge, bridge falls down - no partial
credit."
-- Dr A. Chang, professor of Mechanical Engineering at Stevens Institute of
Technology

chipfactory@hotmail.com wrote in message
<37964ef8.605363216@news.maltanet.net>...
>Hi Folks:
>When we read the documentation for Synopsys we found out that
>a license server can also be somewhere in the internet. Now we
>came up with the idea to connect our Synopsys/Cadence/Modeltech
>license server to the internet and share our licenses with others.
>During the night here in Europe our licenses are not used.
>Would there someone be interested and is this legal?
>
>Best regards
>Jack
>
>email: chipfactory@hotmail.com
>


Article: 17303
Subject: Re: License sharing for synopsys/cadence/modeltech
From: "Wade D. Peterson" <peter299@maroon.tc.umn.edu>
Date: Mon, 19 Jul 1999 21:28:27 -0500
Links: << >>  << T >>  << A >>
R. Mark Gogolewski <gogo@netcom.com> wrote in message
news:7n00fp$g5u@dfw-ixnews11.ix.netcom.com...
[snip]
> Let's say that the license agreements didn't specifically
> have wording to make sure that this usage is restricted.
> IMO, I would think that everyone in this industry - designers
> and vendors - would agree that software licenses were
> never priced with this type of usage in mind.
>
> Does it _have_ to strictly be in violation of a license
> agreement before it feels like stealing?  Let's be
> realistic, if everyone did this, vendors would have to
> either technically do something to stop it, or drastically
> change the price model.
>
> Thoughts?
[snip]

I don't think that 'feels like stealing' has anything to do with it.  A license
is a binding agreement between a buyer and seller.  If the license says that
they can do this, then by all means they should do it.  If the license says they
can't do it, then shey should not do it.  If the license does not anticipate
this possibility, then they should probably spend a couple of hundred bucks to
get an attorney's opinion on the agreement and go from there.

As far as being a good idea or not, I think that some of the EDA tool companies
have decided that this is a good idea.  I was down at the DAC conference in New
Orleans last month, and there was a company preparing to sell  'timeshare
licenses' over the internet.  Apparently, they were working with Synopsys (and
others) on the idea.  If you need synthesize or simulate on Synopsys, and don't
want to spend the $30K for a license, then you could use theirs on a 'per use'
basis.  I think it's a great idea.

--
Wade D. Peterson
Silicore Corporation
3525 E. 27th St. No. 301, Minneapolis, MN USA 55406
TEL: (612) 722-3815, FAX: (612) 722-5841
URL: http://www.silicore.net/  E-MAIL: peter299@maroon.tc.umn.edu



Article: 17304
Subject: Xilinx/Synopsys License Problem
From: Nicholas Brown <nbrownNOT@gulfaccess.com>
Date: Mon, 19 Jul 1999 23:07:05 -0400
Links: << >>  << T >>  << A >>
Hello,
    I have just purchased the Xilinx Foundation Series 1.5 Student
Edition, a XS40 and a XS95 board  from Xess Corp. Unfortunately, the
license software dosnít seem to work and Xilinx now claims that they
donít support the student edition. I was hoping that someone here knows
a solution to the problem.
    Everything seems to work except the VHDL synthesis. When attempting
to synthesize a VHDL entry I get:

Pcm :Synopsys server initialization
Dpm :Invalid Host(-9,57)
Pcm :Cannot find a valid license for Synopsys synthesis

    I have spent several hours attempting to get this to work. Initially
I
used Drive number 0B87-7DBF with the Xilinx license generator which is
the id for the D drive where the Foundation Series software resides (as
reported by Vol C: in DOS.)
This did not work, so I manually changed the license.dat file so the id
was 1878-08CD which is the id the master drive, C:. This did not work.
Then I downloaded Hostid.exe from Synopsys which gave the host id as
f8d709c10000. This also did not work. The Autoexec.bat dose contain:
C:\flexlm\license.dat. Any insight to the problem would be appreciated.
Thanks in advance
-Nick


Article: 17305
Subject: Re: Xilinx/Synopsys License Problem
From: Ray Andraka <randraka@ids.net>
Date: Mon, 19 Jul 1999 23:33:12 -0400
Links: << >>  << T >>  << A >>
Last time I checked, VHDL wasn't included in the student edition, of course
that was a while ago now.

Nicholas Brown wrote:

> Hello,
>     I have just purchased the Xilinx Foundation Series 1.5 Student
> Edition, a XS40 and a XS95 board  from Xess Corp. Unfortunately, the
> license software dosnít seem to work and Xilinx now claims that they
> donít support the student edition. I was hoping that someone here knows
> a solution to the problem.
>     Everything seems to work except the VHDL synthesis. When attempting
> to synthesize a VHDL entry I get:
>



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


Article: 17306
Subject: Re: License sharing for synopsys/cadence/modeltech
From: gogo@netcom.com (R. Mark Gogolewski)
Date: 20 Jul 1999 05:10:14 GMT
Links: << >>  << T >>  << A >>
In article <7n0mr4$l76$1@news1.tc.umn.edu>,
Wade D. Peterson <peter299@maroon.tc.umn.edu> wrote:
>R. Mark Gogolewski <gogo@netcom.com> wrote in message
>news:7n00fp$g5u@dfw-ixnews11.ix.netcom.com...
>[snip]
>> Let's say that the license agreements didn't specifically
>> have wording to make sure that this usage is restricted.
>> IMO, I would think that everyone in this industry - designers
>> and vendors - would agree that software licenses were
>> never priced with this type of usage in mind.
>>
>> Does it _have_ to strictly be in violation of a license
>> agreement before it feels like stealing?  Let's be
>> realistic, if everyone did this, vendors would have to
>> either technically do something to stop it, or drastically
>> change the price model.
>>
>> Thoughts?
>[snip]
>
>I don't think that 'feels like stealing' has anything to do with it.  A 
>license is a binding agreement between a buyer and seller.  If the license 
>says that they can do this, then by all means they should do it.  If the 
>license says they can't do it, then shey should not do it.  If the license 
>does not anticipate this possibility, then they should probably spend a 
>couple of hundred bucks to get an attorney's opinion on the agreement and 
>go from there.
>
>As far as being a good idea or not, I think that some of the EDA tool 
>companies have decided that this is a good idea.  I was down at the DAC 
>conference in New Orleans last month, and there was a company preparing 
>to sell  'timeshare licenses' over the internet.  Apparently, they were 
>working with Synopsys (and others) on the idea.  If you need synthesize 
>or simulate on Synopsys, and don't want to spend the $30K for a license, 
>then you could use theirs on a 'per use' basis.  I think it's a great idea.


Good comments - and I agree.  I would add, however, that in this last
paragraph, the scenario is that this type of licensing is being
_discussed and negotiated_ - rather than simply being implemented just
because the license agreement didn't anticipate it.

There has been a smattering of discussion both on the newsgroups
and on ESNUG about "getting around" the specifics of licensing.

I believe that there should be some consideration of the intent
of the agreement even if the specific wording doesn't cover it.
While both the vendors and the designers can hire lawyers until
the cost of doing business raises prices through the roof, when
unforseen contingencies arise, the success or failure of the
resolution boils down to the relationship between the vendor and
the client.

Just my $0.02,

Mark

P.S.  I think the idea of specifically licensing software this
way on a pure usage model has a heck of a lot of merit.  My
comments began from my perception that there is a portion of the
audience at large that looks for ways to slip around the rules
as much as they can...
Article: 17307
Subject: What happened to Esprimo?
From: "Anthony Ellis - LogicWorks" <aelogic@iafrica.com>
Date: Tue, 20 Jul 1999 07:10:39 +0200
Links: << >>  << T >>  << A >>
Does anyone know what happened to Esprimo:  www.vhdl.com.
I have their VHDL editor but trying to find out about upgrades etc. is
fruitless as they don't reply to any email.



Article: 17308
Subject: Re: License sharing for synopsys/cadence/modeltech
From: Steve Bird <steveb@vizef.demon.co.uk.island>
Date: Tue, 20 Jul 1999 11:21:25 +0100
Links: << >>  << T >>  << A >>
You and everyone else who knows the port number! This is a known
(potential) security breach for flexLM. Whilst you can control which
external IP addresses can access the internal license server the
increase in site maintenance may not make it worth the effort.


In article <37964ef8.605363216@news.maltanet.net>,
chipfactory@hotmail.com writes
>Hi Folks:
>When we read the documentation for Synopsys we found out that 
>a license server can also be somewhere in the internet. Now we 
>came up with the idea to connect our Synopsys/Cadence/Modeltech 
>license server to the internet and share our licenses with others. 
>During the night here in Europe our licenses are not used.
>Would there someone be interested and is this legal?
>
>Best regards
>Jack
>
>email: chipfactory@hotmail.com
>

Article: 17309
Subject: Re: Xilinx/Synopsys License Problem
From: Nicholas Brown <nbrownNOT@gulfaccess.com>
Date: Tue, 20 Jul 1999 08:42:48 -0400
Links: << >>  << T >>  << A >>
It currently claims to be capable of VHDL. The first instance of VHDL starts on
page 61 of the manual, "Entering the Design Using the VHDL Hardware Description
Language.".

Ray Andraka wrote:

> Last time I checked, VHDL wasn't included in the student edition, of course
> that was a while ago now.
>
> Nicholas Brown wrote:
>
> > Hello,
> >     I have just purchased the Xilinx Foundation Series 1.5 Student
> > Edition, a XS40 and a XS95 board  from Xess Corp. Unfortunately, the
> > license software dosnít seem to work and Xilinx now claims that they
> > donít support the student edition. I was hoping that someone here knows
> > a solution to the problem.
> >     Everything seems to work except the VHDL synthesis. When attempting
> > to synthesize a VHDL entry I get:
> >
>
> --
> -Ray Andraka, P.E.
> President, the Andraka Consulting Group, Inc.
> 401/884-7930     Fax 401/884-7950
> email randraka@ids.net
> http://users.ids.net/~randraka



Article: 17310
Subject: Re: Frequency multiplier in XC4000
From: "Earthlink News" <evansamuel@earthlink.net>
Date: Tue, 20 Jul 1999 08:15:28 -0500
Links: << >>  << T >>  << A >>
Correction to the schematic.  There is an invertor from the counter to the
first stage nand gate.

Evan

evansamuel@earthlink.net





Article: 17311
Subject: "Contract Outsourcing?!"
From: "Blake Nelson" <nelson@cstn.com>
Date: Tue, 20 Jul 1999 09:26:28 -0600
Links: << >>  << T >>  << A >>
http://www.cstn.com

A staffing option that does not involve recruiters or headhunters!!!!!!!

Regards,
Blake



Article: 17312
Subject: Re: Xilinx/Synopsys License Problem
From: "Craig Slorach" <craigs@elec.gla.ac.uk>
Date: Tue, 20 Jul 1999 16:40:44 +0100
Links: << >>  << T >>  << A >>
You should have Synopsys FPGA Express as your synthesis tool for VHDL which
is provided in the 1.5 Student Edition. Do not edit the license.dat file as
it's a special file and isn't that easy to do (there will be a checksum at
the botttom of the file to stop people just changing the configuration file
whenever they wanted to). I haven't any experience with 2 drive
configurations, but would assume that the volume-id would have to be on the
primary boot drive (C)- check on the Xilinx support site for this as they
should have info on the Flex-LM licensing issue- do you have a network card
in your PC?- if you have then host to this as it is a better option since
you don't hit a problem when your hard drive crashes. Other thing to check
is the the 'set' in your autoexec has worked properly, got to a command
prompt and type:

set

which will echo your configuration settings.

Hope this helps and feel free to e-mail if you can't get it running.

Craig




Article: 17313
Subject: Re: Frequency multiplier in XC4000
From: Peter Alfke <peter@xilinx.com>
Date: Tue, 20 Jul 1999 09:41:38 -0700
Links: << >>  << T >>  << A >>
Thanks for the drawing. Close but no cigar!

Eliminate all the nets driven by the two Q outputs.

Then make the Control flip-flop's Q drive the NAND gate and, through an
inverter, the active High counter Reset.
So when the Control flip-flop gets set, it starts the oscillator and allows the
counter to count.

The MSB of the counter, through an inverter, must drive the active Low Reset of
the control flip-flop.
So when the counter reaches two-to-the-n, it immediately resets the control
flip-flop, which stops the oscillator and also resets the counter. There is a
benevolent race condition between the MSB of the counter and the Control
flip-flop.

If there are more questions, we can continue off-line. The whole newsgroup may
not be interested...

Peter Alfke
=================================================
Earthlink News wrote:

> Correction to the schematic.  There is an invertor from the counter to the
> first stage nand gate.
>
> Evan
>
> evansamuel@earthlink.net
>
>  [Image]

Article: 17314
Subject: Solaris vs. NT
From: Mark Kinsley <mkinsley@xs4all.nl>
Date: Tue, 20 Jul 1999 19:56:30 +0200
Links: << >>  << T >>  << A >>
We are currently investigating upgrading of our platform hardware.  We 
have a team of 3 designers each with an out-dated PC, these are
networked with two somewhat obsolete Spark Workstations.  We are
running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to
Quartus planned).

With the good price performance ratio of PC platforms and the ease of
use of NT, we are considdering migrating our toolset to high end NT
machines.  What the heck, we need new PC's anyway.  What does concern me
however is the lack of multi-user support and performance levels.

I would hate to overlook any important issues which should be
considdered in making this decision, so if you have any input please
mail it to me / post it to the group.

...how beneficial is a Multi-CPU system for these tools ?


Regards,
Mark Kinsley
mkinsley @ xs4all.nl
Article: 17315
Subject: Re: Xilinx/Synopsys License Problem
From: "Mark Condit" <mcondit@NOSPAMlittonlaser.com>
Date: Tue, 20 Jul 1999 14:02:38 -0400
Links: << >>  << T >>  << A >>
I think I may know what your problem is.  The Foundation series software
wants to be node locked, using either your hard disk serial number or your
ethernet address.  The problem is that the software doesn't recognize hard
disk serial numbers that start with a zero "0".  I have 1.5i and have been
on temp licenses for the last 8 months.  I'm told Xilinx is working the
problem, but no date for a fix yet. Good luck.





Article: 17316
Subject: Re: License sharing for synopsys/cadence/modeltech
From: jdl@user2.teleport.com (Jay Lessert)
Date: 20 Jul 1999 12:29:54 -0700
Links: << >>  << T >>  << A >>
In article <379f7c9e.221323198@news.rdc2.occa.home.com>,
David Rogoff <drogoff@home.com> wrote:
>Cadence and Synopsys (and probably others) have terms in their license
>agreements and prohibit license servers from being more than a specified
>distance from the machine requesting the license.  This makes it against the
>rules to share licenses between offices of the same company if they are in
>different cities!  I don't think that there is anything that can technically
>stop you from doing it, but you are breaking a legal contract.

Not to belabor the point, but the question of whether, in general, the
rather amusingly restrictive license agreements issued by most
commercial SW vendors constitute "legal contracts" is something
that has yet to be definitively decided in US courts (don't know about
the rest of the world).

I'm not picking on Cadence/Synopsys/Mentor, most of these license
agreements are pretty bad, and many of the PC-oriented licenses are
absurdly bad.  I particularly enjoy the ones that restrict me to a
"single backup copy" of the software-- I take a backup of that SW every
bleeping week, every month I set one of those aside for two years, then
keep one of those per year for three more years (five years total),
something like 35 backup copies total.  :-)

But most customers behave ethically, and *nobody* wants to take one
of these agreements to court, least of all the vendors.
-- 
Jay Lessert              Portland, Oregon USA             jdl@teleport.com
Article: 17317
Subject: Re: Solaris vs. NT
From: janovetz@uiuc.edu (Jake Janovetz)
Date: 20 Jul 1999 20:24:52 GMT
Links: << >>  << T >>  << A >>

First and foremost, you should demand that your tools be
supported by Linux.  Forcing you to go to NT is a huge step
back in usability (as you've noticed).  Without this valuable
feedback, those companies will just keep shoveling support at
NT which is a clearly inferior platform for development of this
sort.

I suppose it all depends on how your designers use the CAD tools.
Do they log in from remote and run large scripts on them (like many
do)?  Do they run scripts to do format conversions (like many do)?
If so, then your designers may be happier with the newer Solaris
boxes.  Solaris machines have come down in price considerably in
the past year or two.  Without support for Linux, I would not even
consider a PC for serious CAD work.


   Cheers,
   Jake


In <3794B84E.49BF@xs4all.nl> Mark Kinsley <mkinsley@xs4all.nl> writes:

>We are currently investigating upgrading of our platform hardware.  We 
>have a team of 3 designers each with an out-dated PC, these are
>networked with two somewhat obsolete Spark Workstations.  We are
>running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to
>Quartus planned).
>
>With the good price performance ratio of PC platforms and the ease of
>use of NT, we are considdering migrating our toolset to high end NT
>machines.  What the heck, we need new PC's anyway.  What does concern me
>however is the lack of multi-user support and performance levels.
>
>I would hate to overlook any important issues which should be
>considdered in making this decision, so if you have any input please
>mail it to me / post it to the group.
>
>...how beneficial is a Multi-CPU system for these tools ?
>
>
>Regards,
>Mark Kinsley
>mkinsley @ xs4all.nl
--
   janovetz@uiuc.edu    | Once you have flown, you will walk the earth with
 University of Illinois | your eyes turned skyward, for there you have been,
                        | there you long to return.     -- da Vinci
        PP-ASEL         | http://www.ews.uiuc.edu/~janovetz/index.html
Article: 17318
Subject: Re: Solaris vs. NT, use Linux
From: "B. Joshua Rosen" <bjrosen@polybus.com>
Date: Tue, 20 Jul 1999 18:36:14 -0400
Links: << >>  << T >>  << A >>
<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
You can do most of your development on Linux. ModelSim is not available
for Linux as far as I know, although you should ask them, but several other
simulators are. I'm using Finsim + Undertow which has a bundle price that
is similar to ModelSim's, and I've been happy with it so far.VCS and Avant's
simulator are also available on Linux. As far as I know there are no synthesis
tools on Linux yet so you might what to get one Solaris box to put on your
network. Linux and Solaris work together naturally. What you don't want
to do is use NT, if you're coming from a Unix environment you will find
that NT is hopelessly inadequate.
<p>Josh
<br>&nbsp;
<br>&nbsp;
<br>&nbsp;
<br>&nbsp;
<p>Mark Kinsley wrote:
<blockquote TYPE=CITE>We are currently investigating upgrading of our platform
hardware.&nbsp; We
<br>have a team of 3 designers each with an out-dated PC, these are
<br>networked with two somewhat obsolete Spark Workstations.&nbsp; We are
<br>running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade
to
<br>Quartus planned).
<p>With the good price performance ratio of PC platforms and the ease of
<br>use of NT, we are considdering migrating our toolset to high end NT
<br>machines.&nbsp; What the heck, we need new PC's anyway.&nbsp; What
does concern me
<br>however is the lack of multi-user support and performance levels.
<p>I would hate to overlook any important issues which should be
<br>considdered in making this decision, so if you have any input please
<br>mail it to me / post it to the group.
<p>...how beneficial is a Multi-CPU system for these tools ?
<p>Regards,
<br>Mark Kinsley
<br>mkinsley @ xs4all.nl</blockquote>
</html>

Article: 17319
Subject: Re: Solaris vs. NT
From: Mark Kinsley <mkinsley@xs4all.nl>
Date: Wed, 21 Jul 1999 02:48:22 +0200
Links: << >>  << T >>  << A >>
Thanks for the input,

I really like Linux and the concept of an open OS, but as far as our
design team goes, we're quite happy with whatever makes doing our design
work the easiest, and let's face it, Linux is unfortunately a few steps
behind MS in this area.  For uneducated hardware designers, system admin
is a touch easier (and more confortable) under NT.  I do agree that
pressure should be placed on the software vendors to support Linux so
that as Linux gets better, all the tools will be available.  ? What
exactly do you define as "a huge step back in useability". ?  What
format conversions are you referring to ?

Currently our designers have Win 95 PC's on which the design capture is
performed (using Codewright & Ease VHDL).  The VHDL output files are
simulated on either the PC or the Spark (we have licenses for both). 
For synthesis, the VHDL files are FTP'd accross to the Spark and
synthesized by running a script file through Leonardo.  The EDIF netlist
output is then placed and routed on either a PC or Spark (we have one
license for both).  I have seen that we can run similar scripts on a PC
platform.

What are the genuine merits of a Solaris solution over an NT solution
?....

1. Useability: ? Here i say NT wins ? Explain why i should change me
view if you disagree.
2. Cost/Performance Ratio:  ?
3. Multi-User: Solaris wins hands-down
4. Tool support:  Seems about the same for NT & Solaris (rumour has it
Quartus initially only supports PCs)
5. Software Licensing: Solaris sometimes more expensive
6. Multi-CPU support: Slightly better under Solaris, but this is more a
software dependent issue, and is only relevant with multiple users (thus
Solaris)
7. Anything else i should be looking at..


Just thinking aloud, and ready to listen...

Mark


Jake Janovetz wrote:
> 
> First and foremost, you should demand that your tools be
> supported by Linux.  Forcing you to go to NT is a huge step
> back in usability (as you've noticed).  Without this valuable
> feedback, those companies will just keep shoveling support at
> NT which is a clearly inferior platform for development of this
> sort.
> 
> I suppose it all depends on how your designers use the CAD tools.
> Do they log in from remote and run large scripts on them (like many
> do)?  Do they run scripts to do format conversions (like many do)?
> If so, then your designers may be happier with the newer Solaris
> boxes.  Solaris machines have come down in price considerably in
> the past year or two.  Without support for Linux, I would not even
> consider a PC for serious CAD work.
> 
>    Cheers,
>    Jake
>
Article: 17320
Subject: Re: Solaris vs. NT
From: rk <stellare@NOSPAM.erols.com>
Date: Tue, 20 Jul 1999 21:19:16 -0400
Links: << >>  << T >>  << A >>
asic & eda today [i believe i got the name right] a while back did a
detailed study of this question, mostly for the asic flow.  they did a head
to head comparison of nt vs. unix and addressed linux in depth.  an
independent lab did it.  also considered was the software packages that you
had to get and their cost.

of course, the #1 thing is the software, really, and what you want to run.
me?  i have linux, '95, NT, and unix boxes, although i do almost all my work
on nt or '95.  just not enough s/w for linux to make it very useful for
design and compatibility for office work isn't adequate on linux [we've run
our own independent tests].  it seems that there have been some new releases
of sim s/w so linux is something to watch.  i don't want to have to maintain
the linux box, don't want to invest the time, and '95 and NT are easy to
maintain.  for asic, you can get all the tools on unix, so if that's your
target probably the way to go, as they are getting cheaper.  for fpga's,
either unix or windoze would work, although i prefer the pc for that,
limiting the amount of different computers i have to deal with [office type
stuff has to be on pc for compatibility with the rest of the world].

if you want fully remote operation, then unix/linux is nice with good
telnet.  i've heard running X over the network from home doesn't hack it
very well.  but if you're running command line and scripts and no graphical
output that'll work.  there are remote s/w add-ons for PC which i've heard
good things about but i haven't tried them.  if you run remote, you might
also be able to run local on home pc and check out license over the net.
that might work fine, too.

for performance, the study showed that there was no big difference between
pc/nt and unix, with NT perhaps a bit ahead.  with the new cpu releases, NT
will probably be a bunch ahead.  the big thing that affects performance,
from the study and my own observations is the size of memory.  get lots of
it.  even if you don't get it all now, make sure your motherboard supports
at least 1 gigabyte.  sometimes the synthesis algorithms just get into a
mode where they want lots of ram, and then they die if they have to hit the
hard disk.

good luck,

rk

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

Mark Kinsley wrote:

> We are currently investigating upgrading of our platform hardware.  We
> have a team of 3 designers each with an out-dated PC, these are
> networked with two somewhat obsolete Spark Workstations.  We are
> running(crawling) ModelSim, Leonardo Spectrum and Maxplus II (Upgrade to
> Quartus planned).
>
> With the good price performance ratio of PC platforms and the ease of
> use of NT, we are considdering migrating our toolset to high end NT
> machines.  What the heck, we need new PC's anyway.  What does concern me
> however is the lack of multi-user support and performance levels.
>
> I would hate to overlook any important issues which should be
> considdered in making this decision, so if you have any input please
> mail it to me / post it to the group.
>
> ...how beneficial is a Multi-CPU system for these tools ?
>
> Regards,
> Mark Kinsley
> mkinsley @ xs4all.nl



Article: 17321
Subject: Re: Solaris vs. NT
From: elh@vu-vlsi.ee.vill.edu (Edward L. Hepler)
Date: 20 Jul 1999 21:45:53 -0400
Links: << >>  << T >>  << A >>
In article <379518D6.27E1@xs4all.nl>, Mark Kinsley  <mkinsley@xs4all.nl> wrote:
>Thanks for the input,
>
>I really like Linux and the concept of an open OS, but as far as our
>design team goes, we're quite happy with whatever makes doing our design
>work the easiest, 

...

>What are the genuine merits of a Solaris solution over an NT solution
>?....
>
>1. Useability: ? Here i say NT wins ? Explain why i should change me
>view if you disagree.
>2. Cost/Performance Ratio:  ?
>3. Multi-User: Solaris wins hands-down
>4. Tool support:  Seems about the same for NT & Solaris (rumour has it
>Quartus initially only supports PCs)
>5. Software Licensing: Solaris sometimes more expensive
>6. Multi-CPU support: Slightly better under Solaris, but this is more a
>software dependent issue, and is only relevant with multiple users (thus
>Solaris)
>7. Anything else i should be looking at..

7   It is much easier to deal with remote access to "server" machines
    (using X windows, etc.) using Solaris/Linux...
8.  Stability/Reliability... Solaris/Linux

Ed Hepler

Dr. Edward L. Hepler
  President,                     Adjunct Professor, 
  VLSI Concepts, Inc.            Villanova University Graduate Courses:
    VLSI and System                  ECE-8440 System Design and Modeling 
      Architecture, Design,          ECE-8445 Advanced Computer Architecture
      and CAD                        ECE-8460 VLSI Design
email: hepler@vlsi-concepts.com  or  elh@ece.vill.edu
voice: (610) 408-9121
fax:   (610) 408-9121
www:   http://www.vlsi-concepts.com     Read: I John 5:10-13

Article: 17322
Subject: Re: Solaris vs. NT
From: Zoltan Kocsi <root@127.0.0.1>
Date: 21 Jul 1999 12:42:28 +1000
Links: << >>  << T >>  << A >>
Mark Kinsley <mkinsley@xs4all.nl> writes:

> With the good price performance ratio of PC platforms and the ease of
> use of NT, we are considdering migrating our toolset to high end NT
> machines.  What the heck, we need new PC's anyway.  What does concern me
> however is the lack of multi-user support and performance levels.

First, demand Linux tools from the vendors (they'll ignore you but do 
it anyway). 
Then, if you choose unix over NT and you are willing to switch tools, 
you can find simulators for Linux but no synthesis or P&R. If you 
simulate a lot relative to synthesis and P&R time, you might be better 
off with a Solaris box running the latter and Linux boxes running the 
former. This way you stay in a productive environment while still keep
the cost low (and you can keep demanding Linux versions :-).

If you go NT, you may get back to that "ease of use" clause later ...

Regards,

Zoltan

-- 
+------------------------------------------------------------------+
| ** To reach me write to zoltan in the domain of bendor com au ** |
+--------------------------------+---------------------------------+
| Zoltan Kocsi                   |   I don't believe in miracles   |  
| Bendor Research Pty. Ltd.      |   but I rely on them.           |
+--------------------------------+---------------------------------+
Article: 17323
Subject: Re: Solaris vs. NT
From: Jerry Zdenek <zdenekjs@interaccess.com>
Date: Wed, 21 Jul 1999 01:22:54 -0500
Links: << >>  << T >>  << A >>
Mark Kinsley wrote:
> 
> Thanks for the input,
> 
> I really like Linux and the concept of an open OS, but as far as our
> design team goes, we're quite happy with whatever makes doing our design
> work the easiest, and let's face it, Linux is unfortunately a few steps
> behind MS in this area.  For uneducated hardware designers, system admin
> is a touch easier (and more confortable) under NT.  I do agree that
> pressure should be placed on the software vendors to support Linux so
> that as Linux gets better, all the tools will be available.  ? What
> exactly do you define as "a huge step back in useability". ?  What
> format conversions are you referring to ?

Depends all on how much automation you do.  Attempting to automate
things under NT *sucks*. 

I really believe in automating as much of the process as possible;  That
way, some user (usually me) can't screw it up or forget a step.  We do
post-layout simulation with an EDIF netlist, and to jump back and forth
between the unix ws (for everything but the Maxplus2 fitter) and the PC
just to fit the FPGA is very annoying. 

What I like to be able to do is make some change to the VHDL, and kick
off the entire process again *without* having to be there.  It's so much
more efficient, esp. if the design is large.  What I define is the
easiest is to make a change to the source and then walk away until I
have simulation results available.

System admin is easier under NT? May be just me, but I don't think so. 
Each NT box has to have the software installed under it, vs 1 for the
unix box.  I  can set up another linux box to add to our group a heck of
a lot faster than I could ever set up another NT box.  All the
software's installed on one NFS server box.  Makes upgrades so much
easier, too.

I spend as much time maintaining one PC as I do the entire unix cluster,
if not more on the one PC.

At least in our setup, Backup of the NT boxes is a pain.  We don't have
any good way of doing it.

And, unfortunately since we run Veribest for the board layout, we're
dependent on an NT box to serve out ODBC data for it. And, as expected,
NT
ends up messing up every week or so and the machine has to be rebooted
for one reason or another.  Then every user in the dept has to shut down
to avoid problems while the machine is rebooted.  We also have to keep
the PCB files on those machines to make performance acceptable.  Makes a
real problem when the box acts up.

> Currently our designers have Win 95 PC's on which the design capture is
> performed (using Codewright & Ease VHDL).  The VHDL output files are
> simulated on either the PC or the Spark (we have licenses for both).
> For synthesis, the VHDL files are FTP'd accross to the Spark and
> synthesized by running a script file through Leonardo.  The EDIF netlist
> output is then placed and routed on either a PC or Spark (we have one
> license for both).  I have seen that we can run similar scripts on a PC
> platform.

Yes, but can you script all of the stages together on NT?  I think not.

We also have written a lot of wrapper tools to make test benches for us,
create stimulus macros, edit incompatibilites between tools, extract
info from report files... etc.  This works so much better under Unix.

BTW, why aren't you putting the VHDL files on a network drive?  Then you
wouldn't have to ftp them.
 
> What are the genuine merits of a Solaris solution over an NT solution
> ?....
> 
> 1. Useability: ? Here i say NT wins ? Explain why i should change me
> view if you disagree.
Depends on your user :).  If you want batch capability or would rather
type than click a lot,  NT doesn't cut it.  The newer GUIs on Linux are
a wash, if not better than NT's.

> 2. Cost/Performance Ratio:  ?
I think on the same hardware, Solaris will beat out NT.  I know linux
does over 95.  Exemplar was slower by 2x under 95 than under linux.

> 3. Multi-User: Solaris wins hands-down
Quite.

> 4. Tool support:  Seems about the same for NT & Solaris (rumour has it
> Quartus initially only supports PCs)
Pretty much a wash in my experience.  

> 5. Software Licensing: Solaris sometimes more expensive
How many dongles can you put off the back of the PC?  Sometimes they
interact with each other, too.  Floating licences are wonderful, esp if
you end up having to go to the factory floor to debug something and need
to make a change.

> 6. Multi-CPU support: Slightly better under Solaris, but this is more a
> software dependent issue, and is only relevant with multiple users (thus
> Solaris)
No, some programs (albeit very few) can run with multiple CPUS...
Don't remember which ones, though.  I think some simulators....

> 7. Anything else i should be looking at..

8. Reliability/Uptime.  NT Loses by a huge margin.
 
> Just thinking aloud, and ready to listen...
> 
> Mark

For me, automating the design process as much as practical is what
makes it the easiest.  That leaves me time (and brain cells) to
concentrate on designing.  Repetitive steps should be done by the
computer, not by me
telling it to do the same exact steps each time.  

I'd rather just type make and go get a soda than sit, click on one pgm,
click several times, click again, start another program, click some
more.....  It's so much more efficient.  Gives me more time to think
about the design.


Jerry
Article: 17324
Subject: Xilinx CPLD ChipViewer
From: Klaus Falser <kfalser@durst.it>
Date: Wed, 21 Jul 1999 06:26:40 GMT
Links: << >>  << T >>  << A >>
Hello,

I installed the Xilinx WebPack, there free CPLD Fitter environment.
The documentations says it should include a chip viewer for the XC9500
family, but I did not see it.
Does anybody know more about?

Best regards
    Klaus
--
Klaus Falser
Durst Phototechnik AG
I-39042 Brixen


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.


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