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 67600

Article: 67600
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
Date: Mon, 15 Mar 2004 21:59:07 +0100
Links: << >>  << T >>  << A >>
If I follow f x C x Vsquared = 100 MHz x 20 pF x 3V x 3V = 20 mW
the current would be about I = P / U ~= 7 mA
But 7mA is much better as 20mA regular Crystal !

I don't find any Crystal with this spec.
Maybe you know where to find that?

Thanks for any advice.
Larry

Peter Alfke wrote:

> Don't get your hopes too high:
> f x C x Vsquared =
> 100 MHz x 20 pF x 3V x 3V = 20 mW
> Do you need crystal precision and stability?
> Peter Alfke
> 
> 
>>From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
>>Newsgroups: comp.arch.fpga
>>Date: Mon, 15 Mar 2004 19:49:13 +0100
>>Subject: low power Oscillator for Xilinx CoolrunnerII
>>
>>Hi all,
>>
>>We are doing a low power product and we need to have an 100Mhz onboard.
>>Regular 20mA crystal oscillator will be too high current.
>>
>>Do you have any idea about the how to generate 100MHz, low current, for
>>an collrunnerII based product?
>>
>>Thanks
>>Larry
>>www.amontec.com
>>
> 
> 

------------ And now a word from our sponsor ----------------------
For a quality mail server, try SurgeMail, easy to install,
fast, efficient and reliable.  Run a million users on a standard
PC running NT or Unix without running out of power, use the best!
----  See http://netwinsite.com/sponsor/sponsor_surgemail.htm  ----

Article: 67601
Subject: Re: Programmed ground pins v physical grounding (Xilinx CPLD)
From: chris <>
Date: Mon, 15 Mar 2004 13:10:49 -0800
Links: << >>  << T >>  << A >>
You won't be any better off programming unused pins to ground, and I think you 
might burn more power that way since TTL type logic burns less power pulled up 
than down. Let them float (don't program them at all). 


Article: 67602
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: Peter Alfke <peter@xilinx.com>
Date: Mon, 15 Mar 2004 13:12:29 -0800
Links: << >>  << T >>  << A >>
50 ppm means garden-variety cheap xtal. Nothing else gets you that
quality...
Peter Alfke

> From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
> Newsgroups: comp.arch.fpga
> Date: Mon, 15 Mar 2004 21:51:58 +0100
> To: Peter Alfke <peter@xilinx.com>
> Subject: Re: low power Oscillator for Xilinx CoolrunnerII
> 
> Peter Alfke wrote:
>> Don't get your hopes too high:
>> f x C x Vsquared =
>> 100 MHz x 20 pF x 3V x 3V = 20 mW
>> Do you need crystal precision and stability?
>> Peter Alfke
>> 
>> 
>>> From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
>>> Newsgroups: comp.arch.fpga
>>> Date: Mon, 15 Mar 2004 19:49:13 +0100
>>> Subject: low power Oscillator for Xilinx CoolrunnerII
>>> 
>>> Hi all,
>>> 
>>> We are doing a low power product and we need to have an 100Mhz onboard.
>>> Regular 20mA crystal oscillator will be too high current.
>>> 
>>> Do you have any idea about the how to generate 100MHz, low current, for
>>> an collrunnerII based product?
>>> 
>>> Thanks
>>> Larry
>>> www.amontec.com
>>> 
>> 
>> 
> Hi and thanks Peter,
> 
> No, we don't need any special precision or stability, +-50ppm schould be
> enough.
> Do you have any idea or advice ?
> 
> Laurent
> 


Article: 67603
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: Peter Alfke <peter@xilinx.com>
Date: Mon, 15 Mar 2004 13:14:14 -0800
Links: << >>  << T >>  << A >>
I was just pointing out that ultra-low power and 100 MHz do not go together,
unless you run very small voltage swing.
Peter

> From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
> Newsgroups: comp.arch.fpga
> Date: Mon, 15 Mar 2004 21:59:07 +0100
> To: Peter Alfke <peter@xilinx.com>
> Subject: Re: low power Oscillator for Xilinx CoolrunnerII
> 
> If I follow f x C x Vsquared = 100 MHz x 20 pF x 3V x 3V = 20 mW
> the current would be about I = P / U ~= 7 mA
> But 7mA is much better as 20mA regular Crystal !
> 
> I don't find any Crystal with this spec.
> Maybe you know where to find that?
> 
> Thanks for any advice.
> Larry
> 
> Peter Alfke wrote:
> 
>> Don't get your hopes too high:
>> f x C x Vsquared =
>> 100 MHz x 20 pF x 3V x 3V = 20 mW
>> Do you need crystal precision and stability?
>> Peter Alfke
>> 
>> 
>>> From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
>>> Newsgroups: comp.arch.fpga
>>> Date: Mon, 15 Mar 2004 19:49:13 +0100
>>> Subject: low power Oscillator for Xilinx CoolrunnerII
>>> 
>>> Hi all,
>>> 
>>> We are doing a low power product and we need to have an 100Mhz onboard.
>>> Regular 20mA crystal oscillator will be too high current.
>>> 
>>> Do you have any idea about the how to generate 100MHz, low current, for
>>> an collrunnerII based product?
>>> 
>>> Thanks
>>> Larry
>>> www.amontec.com
>>> 
>> 
>> 
> 
> ------------ And now a word from our sponsor ----------------------
> For a quality mail server, try SurgeMail, easy to install,
> fast, efficient and reliable.  Run a million users on a standard
> PC running NT or Unix without running out of power, use the best!
> ----  See http://netwinsite.com/sponsor/sponsor_surgemail.htm  ----


Article: 67604
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: Uwe Bonnes <bon@elektron.ikp.physik.tu-darmstadt.de>
Date: Mon, 15 Mar 2004 21:19:41 +0000 (UTC)
Links: << >>  << T >>  << A >>
Amontec Team, Laurent Gauch <laurent.gauch@amontecdeleteallcaps.com> wrote:
: If I follow f x C x Vsquared = 100 MHz x 20 pF x 3V x 3V = 20 mW
: the current would be about I = P / U ~= 7 mA
: But 7mA is much better as 20mA regular Crystal !

: I don't find any Crystal with this spec.
: Maybe you know where to find that?

Use 1.8 Volt VCCO for the oscillator and use the clock doubling meachnisme
to run with 50 MHz. With the LVCMOS18_ANY attribute, the 1.8 Volt VCCO
should be achievable on any bank...

Bye
-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------



Article: 67605
Subject: Re: Programmed ground pins v physical grounding (Xilinx CPLD)
From: Peter Alfke <peter@xilinx.com>
Date: Mon, 15 Mar 2004 13:42:29 -0800
Links: << >>  << T >>  << A >>

Vitual ground by means of a strong (the strongest possible) output,
externally and internally tied to ground, does not consume any power at all.
Zilch.
Remember, these are CMOS driversn even when they are called "LVTTL"!
Peter Alfke


> From: chris <>
> Organization: (none)
> Newsgroups: comp.arch.fpga
> Date: Mon, 15 Mar 2004 13:10:49 -0800
> Subject: Re: Programmed ground pins v physical grounding (Xilinx CPLD)
> 
> 
> You won't be any better off programming unused pins to ground, and I think
> you might burn more power that way since TTL type logic burns less power
> pulled up than down.  Let them float (don't program them at all).





Article: 67606
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: "Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com>
Date: Mon, 15 Mar 2004 22:52:28 +0100
Links: << >>  << T >>  << A >>
We are coupling Philips LPC22114 ARM7TDMI-s with a coolrunnerII. All the 
IOs bus (3.3V) of the LPC2114 will be connected to the Coolrunner, so we 
cannot reserve 1.8V for any bank on the CoolRunnerII (we will lose too 
many IOs on the coolrunnerII).

LPC2114 is used for algorithm part and nakia 3310 LCD interface part 
(via spi), and EEPROM access (I2C).
CoolrunnerII will be use for high frequency custom features, and for 
Compact Flash connection for data logger purpose in fat 12 or 16 format.
We are thinking to have a 80-100MHz onboard, to control custom process 
at 160-200MHz. For that, we will use the dual-edge feature of the 
CoolrunnerII.

Any advice about 80-100Mhz low power Crystal ?

Larry

Uwe Bonnes wrote:

> Amontec Team, Laurent Gauch <laurent.gauch@amontecdeleteallcaps.com> wrote:
> : If I follow f x C x Vsquared = 100 MHz x 20 pF x 3V x 3V = 20 mW
> : the current would be about I = P / U ~= 7 mA
> : But 7mA is much better as 20mA regular Crystal !
> 
> : I don't find any Crystal with this spec.
> : Maybe you know where to find that?
> 
> Use 1.8 Volt VCCO for the oscillator and use the clock doubling meachnisme
> to run with 50 MHz. With the LVCMOS18_ANY attribute, the 1.8 Volt VCCO
> should be achievable on any bank...
> 
> Bye


Article: 67607
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: Jim Granville <no.spam@designtools.co.nz>
Date: Tue, 16 Mar 2004 10:56:15 +1300
Links: << >>  << T >>  << A >>
Amontec Team, Laurent Gauch wrote:
> Hi all,
> 
> We are doing a low power product and we need to have an 100Mhz onboard.
> Regular 20mA crystal oscillator will be too high current.
> 
> Do you have any idea about the how to generate 100MHz, low current, for 
> an collrunnerII based product?

Look at something like the DS1087 clock generator ?
Specs appx 5mA, and <= 66Mhz, so the clock double in CR2 would allow 100MHz.

Osc Power is very Vcc dependant, so if you are mW-paranoid, I'd look
at a separate low voltage logic device, run on minimal Vcc.

This philips device is also interesting
http://www.philipslogic.com/products/lvc/pdf/74lvc1gx04.pdf

.. but the data does not show hysteresis in the buffer,
nor linear-mode (Xtal usage) Icc/Vcc/Freq info, so you'd
need to trial one.
  If Freq is not important, a tiny logic ring oscillator with
trimmed min vcc would get close to minimum power.
-jg


Article: 67608
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: Uwe Bonnes <bon@elektron.ikp.physik.tu-darmstadt.de>
Date: Mon, 15 Mar 2004 21:57:01 +0000 (UTC)
Links: << >>  << T >>  << A >>
Amontec Team, Laurent Gauch <laurent.gauch@amontecdeleteallcaps.com> wrote:
: We are coupling Philips LPC22114 ARM7TDMI-s with a coolrunnerII. All the 
: IOs bus (3.3V) of the LPC2114 will be connected to the Coolrunner, so we 
: cannot reserve 1.8V for any bank on the CoolRunnerII (we will lose too 
: many IOs on the coolrunnerII).

I gave you the hint with LVCMOS18_ANY... 

Here from the XILINX database:
> All CoolRunner-II devices have the ability to share any I/O bank with 1.8V
> LVCMOS inputs. The only limitation is that this pin may not utilize a
> Schmitt Trigger.  

-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Article: 67609
Subject: Re: Which should I use, Floorplanner or PACE
From: "Tim" <tim@rockylogic.com.nooospam.com>
Date: Mon, 15 Mar 2004 22:14:10 -0000
Links: << >>  << T >>  << A >>
Philip Freidin wrote:
> There seem to be at least 6 ways of specifying area constraints
> for a XC2V6000 design I am working on:
>
> 1) The constraints editor
>
> 2) My favorite editor UltraEdit, on the UCF file
>
> 3) The Floorplanner
>
> 4) PACE
>
> 5) FPGA editor
>
> 6) Embedded in the Verilog source

My personal preference is usually to embed the constraints in the
HDL source, but I use VHDL which allows for this sort of stuff to
be derived by algorithm.  You don't have that capability in Verilog,
so I would write a small script to generate the UCF - I would use
Perl but Python or whatever would do just as well.

Of course the UCF syntax should have been defined as a TCL script
in the first instance.  Maybe now would be a good time for the X
guys to add an optional UCF2 syntax?



Article: 67610
Subject: Re: low power Oscillator for Xilinx CoolrunnerII
From: "Leon Heller" <leon_heller@hotmail.com>
Date: Mon, 15 Mar 2004 22:28:46 -0000
Links: << >>  << T >>  << A >>

"Amontec Team, Laurent Gauch" <laurent.gauch@amontecDELETEALLCAPS.com> wrote
in message news:4055fa0a$1@news.vsnet.ch...
> Hi all,
>
> We are doing a low power product and we need to have an 100Mhz onboard.
> Regular 20mA crystal oscillator will be too high current.
>
> Do you have any idea about the how to generate 100MHz, low current, for
> an collrunnerII based product?

How about a Butler oscillator using a 100 MHz overtone crystal? It should be
possible to design this to take a lot less current.

Leon



Article: 67611
Subject: Re: Device/Board Selection (CPU Design)
From: John <nospam@nospam.com>
Date: Mon, 15 Mar 2004 22:38:42 GMT
Links: << >>  << T >>  << A >>
> The easiest might be PCI104.  The PC104 bus is based on the ISA
> standard.  They added a connector for PCI and called it PC104+.

Although the specification is interesting it doesn't seem practical. I'm 
basically looking for something cheap which is compatible with off the 
shelf components. Standard PCI seems to be the easiest in that respect.

> Otherwise you can do a search on passive PCI backplane.  I don't have
> info myself, but a google search turned up 18000 hits.  

Yea, I looked through google but most of the results were junk or 
companies I've never heard of. I was just wondering if anyone had any 
reccomendation.

CPCI and PMC don't make sense for this (in my opinion) because they give 
up the whole benefit (in terms of this project) of using PCI. The idea 
here is low cost with compatibility with standard x86-PC-style components.


> I almost forgot, don't let the mention of a PICMG slot fool you.  That
> is the designation of where your CPU card will plug in.  This format is
> set up to allow for both PCI and ISA coming off the CPU card, hence the
> different name.  

Yes. I noticed that when looking for a backplane. It seems that most CPU 
cards use that format. Is this simply to interface with ISA slots on the 
backplane or does using both busses give some advantage?

> I still think you will be better off making a CPU board to plug into a
> socket 370 or socket A.  That gives you a lot of very low cost options
> and is still supported widely.  It will be a lot less work making a
> board with fewer chips too.  

Could you explain your reasoning here? It seems that integrating into a 
socket like those above would be more difficult than simply using PCI. I 
browsed the site you linked to and it seems that the only standard 
socket that they offer (without doing custom work) is Socket 7. What 
would the board  gain by plugging into a CPU socket as opposed to PCI? 
Are the components which are integrated into a motherboard (eg: Socket 
7) well documented to the point where I could interface with them from 
an FPGA plugged directly into the CPU Socket? If so, then it might be 
more cost effective to manufacture a board which included just an FPGA 
and a package conversion to a popular socket format. However, I don't 
see any such products avaliable for purchase (meaning that it would have 
to be custom fabricated). It seems to me that FPGAs on PCI cards 
(complete with things such as memories and programming interfaces) are 
quite avaliable and can be had at a lower cost. The other thing is that 
it is much easier to find products for PCI than it is for Socket 7. 
Although Socket 7 has lasted for quite some while I doubt that it will 
be around as long as PCI. This would mean a new board when Socket 7 
becomes hard to find (as it already is in the PC world).

Thanks, your input is greatly appreciated!

Article: 67612
Subject: Re: copy protection on FPGA using embedded serial number
From: Lasse Langwadt Christensen <langwadt@ieee.org>
Date: Tue, 16 Mar 2004 00:42:35 +0100
Links: << >>  << T >>  << A >>
john jakson wrote:
 >
snip
> 
> I have to wonder whether the circuit guys went to the trouble of
> anticipating these sorts of attacks. This really would only concern a
> certain agency  in possesion of enemy crypto system using these parts,
> but good JB movie material.
> 
> back to reading

Or those who intend to use it for creditcards or worse "rechargeble"
plastic cards to replace cash ..

-Lasse


Article: 67613
Subject: Re: copy protection on FPGA using embedded serial number
From: nweaver@ribbit.CS.Berkeley.EDU (Nicholas C. Weaver)
Date: Tue, 16 Mar 2004 00:24:28 +0000 (UTC)
Links: << >>  << T >>  << A >>
In article <adb3971c.0403150030.306de652@posting.google.com>,
john jakson <johnjakson@yahoo.com> wrote:
>I have to wonder whether the circuit guys went to the trouble of
>anticipating these sorts of attacks. This really would only concern a
>certain agency  in possesion of enemy crypto system using these parts,
>but good JB movie material.

There are a LOT of secrets billable at anywhere between $1k and $100k
and $1M which might very well end up in hostile hands.

In addition to teh other ones mentioned:

Sattelite receivers SHOULD all have a unique secret which can be
revoked.  But an attacker who gets this could go clone-happy and make
a dozen (or hundreds) of illegitimate receivers.

If they have more common secrets (eg, shared key across a lot of
units), they can't be revoked and an attacker who gets one can go
really clone-happy.

When dealing with security, one can't use absolutes, but $-costs and
related issues.
-- 
Nicholas C. Weaver                                 nweaver@cs.berkeley.edu

Article: 67614
Subject: UCF or XCF - which one to use ?
From: Sudhir.Singh@email.com (Sudhir Singh)
Date: 15 Mar 2004 17:15:36 -0800
Links: << >>  << T >>  << A >>
Hello,
Would anyone be able to provide me with some info on which constraints
file format to use for Xilinx designs - UCF or XCF? Do I need to use
both or just one will do for a design?

Thanks
Sudhir Singh

Article: 67615
Subject: Re: Altera, Cyclone: pin not connected warning
From: gregs@altera.com (Greg Steinke)
Date: 15 Mar 2004 17:55:36 -0800
Links: << >>  << T >>  << A >>
erojr <janos.nojunk.nospam.ero@cern.nojunk.nospam.ch> wrote in message > 
> I think, if you declare in the Unused pin tab  As inputs, tri-stated 
>  
> (This is the exact declaration! The other options are  As outputs, 
> driving ground  and  As outputs, driving an unspecified signal  - I
>  
> would really like to know what is this later exactly!), you end up in 
> your hardware a large number of unconnected input pins. Just in the last 
> 
> thread about unconnected  TRST  pin several people pointed out, that 
> 
> input pins MUST NOT remain unconnected, as they might start to oscillate 
> 
> in high frequency, possibly destroying your FPGA. Or am I mistaken?
> 
> I did not find any indication in Altera Datasheets if the input pins 
> have internal weak pullup resistors in general (the TRST pin should have 
> 
> one, by definition). This also means if the unused pins aren t connecte
> d 
> at all they must not be declared  As inputs, tri-stated .
> 
> Could somebody explain here what is the correct procedure?
> 
> Thanks,
> 
> Janos Ero
> CERN Div. EP

Newer Altera FPGAs (APEX II, Cyclone, Stratix) have a programmable
pull up on the IOs. This is turned on through the Quartus software.
The typical value is 25 kohms.

There's also an option for a bus hold. This weakly holds the IO in the
last state, either high or low. It acts as a ~7 kohm resistor to VCCIO
or GND, depending on what the pin was last driven to. The
pullup/pulldown is weak enough that it can be overcome by an external
driver.

Sincerely,
Greg Steinke
gregs@altera.com
Altera Corporation

Article: 67616
Subject: what technology is the mcnc.genlib in the SIS package
From: paraagv@hotmail.com (paraag)
Date: 15 Mar 2004 19:31:31 -0800
Links: << >>  << T >>  << A >>
what technology is the mcnc.genlib in the SIS package and how do I
come to know of this

Is it 0.18, 0.35, 0.5 etc.
pLEASE CAN ANYONE HELP


PARAAG VAISHAMPAYAN
UNC CHARLOTTE

Article: 67617
Subject: Re: Device/Board Selection (CPU Design)
From: rickman <spamgoeshere4@yahoo.com>
Date: Mon, 15 Mar 2004 23:25:03 -0500
Links: << >>  << T >>  << A >>
John wrote:
> 
> Yea, I looked through google but most of the results were junk or
> companies I've never heard of. I was just wondering if anyone had any
> reccomendation.

No, but the first page I found had several valid hits and three ads. 
Passive backplanes are very popular with industrial computers and a
there lot of Asian makers which drive the prices down. 


> Yes. I noticed that when looking for a backplane. It seems that most CPU
> cards use that format. Is this simply to interface with ISA slots on the
> backplane or does using both busses give some advantage?

No, in fact, I had not noticed that there are also some pure PCI
backplanes, so that should be better for you.  


> > I still think you will be better off making a CPU board to plug into a
> > socket 370 or socket A.  That gives you a lot of very low cost options
> > and is still supported widely.  It will be a lot less work making a
> > board with fewer chips too.
> 
> Could you explain your reasoning here? It seems that integrating into a
> socket like those above would be more difficult than simply using PCI. I
> browsed the site you linked to and it seems that the only standard
> socket that they offer (without doing custom work) is Socket 7. What
> would the board  gain by plugging into a CPU socket as opposed to PCI?
> Are the components which are integrated into a motherboard (eg: Socket
> 7) well documented to the point where I could interface with them from
> an FPGA plugged directly into the CPU Socket? If so, then it might be
> more cost effective to manufacture a board which included just an FPGA
> and a package conversion to a popular socket format. However, I don't
> see any such products avaliable for purchase (meaning that it would have
> to be custom fabricated). It seems to me that FPGAs on PCI cards
> (complete with things such as memories and programming interfaces) are
> quite avaliable and can be had at a lower cost. The other thing is that
> it is much easier to find products for PCI than it is for Socket 7.
> Although Socket 7 has lasted for quite some while I doubt that it will
> be around as long as PCI. This would mean a new board when Socket 7
> becomes hard to find (as it already is in the PC world).

First, you need to do more than just look at web pages.  ISI is a
company that you need to talk with to get info on just what they do and
don't have.  They have a full line of PGA adapters.  It is unlikely that
they can't fit another socket.  
If you plug into the CPU socket, you don't need to build any of the
other stuff at all.  I am sure you can get adequate info on socket 7. 
Intel had to document it for the chip makers.  This would be part of the
chip spec.  If not Intel, then AMD likely has a spec.  They made parts
for it too, didn't they?  Why do you need a socket that will be long
lived?  Changing your FPGA design for a different socket should be no
big deal if you need to do that in a couple of years.  Are you going to
mass produce this thing? 

I have not seen any PCI cards with FPGAs setup to be bus masters, but if
they are there that would be fine.  Keep in mind that it will need
memory on the card since PCI transfers are much slower than SDRAM or
DDR.  

The other difference is that a PC motherboard will have all the IO
features you need built in.  It is not easy to find interfaces for
keyboards, etc on PCI cards.  But I guess you can add anything you want
via USB.  

I have not spent much time thinking about this.  I just know that a
mother board gives you everything a high speed CPU needs (including low
voltage power in large doses).  A PCI card will have to do a lot of work
to replace that.  

I may have said this once before, or maybe I said it to John, but you
should do your design work before you even both to consider how or what
hardware to put it in.  You likely have months of work ahead of you
designing the HDL.  Finding hardware should be the easy part.  

Heck, if push comes to shove, you can use one of my boards if you will
be happy with an XC3S400.  It will have SDRAM and SBSRAM, (you can
ignore the DSP chip) but only a PC104 interface, no PCI... at least not
this version.  

Good Luck.  :)

-- 

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

Article: 67618
Subject: Re: UCF or XCF - which one to use ?
From: "Kelvin @ SG" <kelvin8157@hotmail.com>
Date: Tue, 16 Mar 2004 13:43:33 +0800
Links: << >>  << T >>  << A >>
UCF is used with the command NGDBuild, XCF is used with XST. I managed to do
some projects
without using any XCF constraints, so I guess XCF is not compulsory...

Kelvin



Sudhir Singh <Sudhir.Singh@email.com> wrote in message
news:e3db8a89.0403151715.4960252b@posting.google.com...
> Hello,
> Would anyone be able to provide me with some info on which constraints
> file format to use for Xilinx designs - UCF or XCF? Do I need to use
> both or just one will do for a design?
>
> Thanks
> Sudhir Singh



Article: 67619
Subject: Re: Device/Board Selection (CPU Design)
From: hmurray@suespammers.org (Hal Murray)
Date: Tue, 16 Mar 2004 06:17:37 -0000
Links: << >>  << T >>  << A >>
> No, in fact, I had not noticed that there are also some pure PCI
> backplanes, so that should be better for you.

What's a "pure" PCI backplane?

Where do the clocks come from and/or who does the bus arbitration?

Got a sample URL?

-- 
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.


Article: 67620
Subject: Re: EABīs in ACEX 1K devices
From: "Simon Peacock" <nowhere@to.be.found>
Date: Tue, 16 Mar 2004 19:28:21 +1300
Links: << >>  << T >>  << A >>
Is there anything to do ???  you could try re-dreaming up a different
algorithm... or use the RAM better... Can you fit a bigger device on the
same footprint.?

Can you get a piggy back board built to tide you over to the next PCB
release...

P.S. its important to check your resources before designing something... or
you'll design a PC with 640 k of RAM and expect it to run windows

Simon


"roland voraberger" <madcap@sbox.tugraz.at> wrote in message
news:c348n4$23a9qs$1@ID-209114.news.uni-berlin.de...
> hello!
>
> iīve got big problems with RAM in a ACEX 1k design. i canīt get a fit
> because i use too much (16/ 12) EABīs but in total number of bits iīm
> far beyond the maximum number of the bits available.
> what can i do (software is max+plus II 10.1)
>
> thanks in advance, roland
>
>



Article: 67621
Subject: Re: xilinx jtag problems
From: Pawan <ps@bits-pilani.ac.in>
Date: Mon, 15 Mar 2004 22:40:09 -0800
Links: << >>  << T >>  << A >>
The "trigger" in chipsope pro refers to external trigger(manually applied to 
board) or is an internal trigger(generated by the software)? Whichever is true, 
what are the steps to see the waveforms on the waveform window in chipscope. 
When I set the trigger condition, the software starts "uploading data" from fpga 
but doesnot collect any samples and it keeps on showing"upload data" message and 
my fpga gets heated up.I have been going nuts on this, somebody please help. 


Article: 67622
Subject: Re: Release asynchrounous resets synchronously
From: msm30@yahoo.com (William Wallace)
Date: 15 Mar 2004 23:00:18 -0800
Links: << >>  << T >>  << A >>
process(clk,iReset)
begin
 if(iReset='1') then
   ResetSR="00";
 elsif(clk'event and clk = '1') then
   ResetSR(0) <= '1';
   ResetSR(1) <= ResetSR(0);
 end if;
end process;

RESET <= ResetSR(1);

simplesm : process(clk,RESET)
begin
  if(Reset = '1') then -- NOTE, this is not iReset
    state <= S001;
  elsif(clk'event and clk='1') then
    case state is =>
      when S001 =>
        state <= S010;
      when S010 =>
        state <= S100;
      when S100 =>
        state <= S001;
    end case
  end if
end process simplesm;

Ray Andraka <ray@andraka.com> wrote in message news:<404DEFB5.6806938E@andraka.com>...
> The GSR net is asynchronous.  No matter how slow your clock is, unless you synchronize the reset to the
> clock OUTSIDE OF THE FPGA, you can't guarantee all flip-flops in the design will see the release of
> reset on the same clock edge.  This is because without syncronization to the clock, there is no
> guarantee the release of reset won't happen close enough to a clock edge to make some flip-flops see it
> before the clock edge, and others after the clock edge.  You can't synchronize it inside the FPGA
> because GSR affects all the flip-flops.  A side issue is the fact that the GSR net is very slow compared
> to the speed the rest of the FPGA logic can operate at.
> 
> William Wallace wrote:
> 
> > Jeff Cunningham <jcc@sover.net> wrote in message news:<m8I2c.274$_43.199987@newshog.newsread.com>...
> > > William Wallace wrote:
> > > > THE IMPORTANCE OF PROVIDING A RESET THAT LEAVES THE RESET
> > > > SYNCHRONOUSLY EVEN IF IT DRIVES THE ASYNCHRONOUS RESET...
> > >
> > > You cannot rely on async reset at startup for initializing one hot state
> > > machines and the like...
> >
> > Please explain why.  Meanwhile, releasing the asynchronous reset
> > synchronously, and ensuring that the reset path delay and setup is
> > less than a clock period, would work.
> >
> > Please re-read my original post, and tell me exactly where I am wrong.
> 
> --
> --Ray Andraka, P.E.
> President, the Andraka Consulting Group, Inc.
> 401/884-7930     Fax 401/884-7950
> email ray@andraka.com
> http://www.andraka.com
> 
>  "They that give up essential liberty to obtain a little
>   temporary safety deserve neither liberty nor safety."
>                                           -Benjamin Franklin, 1759

Article: 67623
Subject: Chipscope
From: Pawan <ps@bits-pilani.ac.in>
Date: Mon, 15 Mar 2004 23:06:22 -0800
Links: << >>  << T >>  << A >>
have added the ILA and the ICON core to my design... but when i run the analyser 
it simply hang and says "waiting to be uploaded". Is this the problem with the 
setting of the triggers. do i need to give the chip (FPGA kit) any external 
input.... how does it take these triggers.... 


Article: 67624
Subject: Re: Release asynchrounous resets synchronously
From: msm30@yahoo.com (William Wallace)
Date: 15 Mar 2004 23:07:02 -0800
Links: << >>  << T >>  << A >>
"It is the control logic, and only control logic that will take off
running that matters."

I agree that matters.  E.g., a state machine that has some INIT states
before it transitions to an IDLE state.  State Machines that are
waiting for some external input to change may be less prone to
problems.


rickman <spamgoeshere4@yahoo.com> wrote in message news:<404D54D8.7AC8882E@yahoo.com>...
> William Wallace wrote:
> > 
> > rickman <spamgoeshere4@yahoo.com> wrote in message
> > 
> > > ...The async global reset is not designed to be fast.  So
> > > you have to make sure that your design will work if the FFs come out of
> > > reset on different clock cycles.
> > 
> > That is what I was getting at with the static timing analysis checking
> > the reset line to ensure that all flip flops with an asynchronous
> > reset (driven from the output of a FF) see the same first clock.  If
> > the static timing analysis tools can do that, you don't need to waste
> > combinatorial real-estate on synchronous resets.
> 
> The analysis tool can't make the GSR network faster than it is.  It is
> not designed to be fast, it is designed to minmize its use of resources
> (area).  I don't have a number for the speed of the path, but even when
> FPGA logic topped out at 50 MHz, we were warned not to rely on the GSR
> net to release all the FFs on the same clock cycle.  
> 
> Normally you don't need to worry with the reset release on most of your
> logic.  Things like data path registers don't care about the reset.  It
> is the control logic, and only control logic that will take off running
> that matters.  In most of my designs, the control logic waits for some
> input signal anyway.  I don't build too many FSMs that start off by
> counting.  The ones that do can be set to go through a set of gray codes
> initially so that they are reset release insensitive.  
> 
> If you can get the reset to release the entire design in the same clock
> cycle, then you are ok.  But you can't use one of the standard clocked
> FFs to do that.  It will have to be an external FF that controls the
> timing of the release.  
> 
> -- 
> 
> Rick "rickman" Collins
> 
> rick.collins@XYarius.com
> Ignore the reply address. To email me use the above address with the XY
> removed.
> 
> Arius - A Signal Processing Solutions Company
> Specializing in DSP and FPGA design      URL http://www.arius.com
> 4 King Ave                               301-682-7772 Voice
> Frederick, MD 21701-3110                 301-682-7666 FAX



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