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 26050

Article: 26050
Subject: Re: FPGA Express strikes again! Xilinx response
From: krw@attglobal.net (Keith R. Williams)
Date: Mon, 02 Oct 2000 03:10:25 GMT
Links: << >>  << T >>  << A >>
On Sun, 1 Oct 2000 20:45:18, Rick Filipkiewicz <rick@algor.co.uk>
wrote:

> 
> 
> Ray Andraka wrote:
> 
> > Like I said, lots of bells and whistles, but nothing that I can see to justify
> > the extra $$$.
> >
> 
> It's the special IPO release. The investors just got a bit freaked by the number
> of amateurs using the tool.

Yikes.  I can be considered a newbie here, but at $20K for a 
license, plus $3K a year to keep the tool working, I don't think 
I qualify as an "amateur".  I agree though. I don't see the bux 
for Pro.  

..Amplify is another issue altogether. I'd like some information
from anyone with experience with it.

----
  Keith


Article: 26051
Subject: Programming Cypress Graphics Clock Generator
From: "Sherdyn" <sherdyn@yahoo.com>
Date: Mon, 2 Oct 2000 17:30:28 +0800
Links: << >>  << T >>  << A >>
Hi,

I have encountered a strange problem with Cypress PLL ICD2061A. The problem
that I have is I couldn't get the Cypress part to program up to my desired
frequency when I divide down a clock before it is output to the PLL.
However, when I replace the divider circuit with a multiplexer, then the
whole thing works. (Of course the parameters to program the PLL in these two
cases are different). I am using Altera part EPF10K30E.

So what could FPGA causes the PLL not to work well? Jitter or something
else. Any clue?

Sherdyn



Article: 26052
Subject: Re: multi-input adders in virtex ?
From: erika_uk@my-deja.com
Date: Mon, 02 Oct 2000 09:57:46 GMT
Links: << >>  << T >>  << A >>
hi,

for my designs, i have never introduced a placement information( apart
from the ones related to the mapping)
when my synchronous deisgn works as expected, i push the timing
constraint up, to know what speed i can achieve

from your experience( and hopefully from other experts ones), apart
from the time saved from implementation process, have you beat timing
driven design with a pre-placement one ?

--Erika


In article <39D77C9B.2CFEF178@andraka.com>,
  Ray Andraka <ray@andraka.com> wrote:
> Absolutely!  If you look at it carefully, the architecture is
actually quite
> regular.  Put the first level adders on every other column, then
insert the
> second layer on every other interleaved column, the third on every
other between
> them and so on.  This is quite fast for small trees.  When the tree
gets large
> enough that the interconnect between levels is too long, you can add
an extra
> register at the inputs of the next layer to break up the long
critical path from
> the previous level through the carry chain.  In the case of an FIR
filter, using
> the partially transposed architecture to absorb the delays, you use
an adder
> chain rather than a tree.  That structure is good for an SDA filter
with bit
> rates of more than 150 MHz in a Virtex-4 (slow speed grade).
>
> erika_uk@my-deja.com wrote:
> >
> > hi ray,
> >
> > for the case of an adder tree do you still believe that building the
> > adders using carry logic is wise ?. The architecture is irregular
and
> > the adders are all vertical
> >
> > --Erika
> >
> > In article <39D6B380.3383A152@andraka.com>,
> >   Ray Andraka <ray@andraka.com> wrote:
> > > Ahh, but in the FIR filter you can go to a partially transposed
> > architecture and
> > > thereby absorb the delays as part of the sample delays you need
> > anyway.  There
> > > is **no_latency_penalty** if you do it right!  It is not the
> > intuitively obvious
> > > approach, though.
> > >
> > > Muzaffer Kal wrote:
> > > >
> > > > I am implementing a programmable FIR and latency is a very big
> > > > problem. So my definition of best is the fastest design with at
most
> > > > one pipeline in the adder tree. Size of no importance.
> > > >
> > > >
> > >
> > > --
> > > -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  or http://www.fpga-guru.com
> > >
> >
> > Sent via Deja.com http://www.deja.com/
> > Before you buy.
>
> --
> -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  or http://www.fpga-guru.com
>


Sent via Deja.com http://www.deja.com/
Before you buy.

Article: 26053
Subject: Multiplication
From: Vipan Kakkar <vipank@cobalt.et.tudelft.nl>
Date: Mon, 02 Oct 2000 15:12:29 +0200
Links: << >>  << T >>  << A >>
Dear Friends:

I would like to know that how to multiply e.g. 1100 with 0.0011. I
already have an array multplier to multiply e.g. 1100 with 0110. But,
now I want to multiply fractions (i.e. .0011) using the same multiplier,
so what modifications should I do in the existing multiplier.

Regards




Article: 26054
Subject: Re: Xilinx XC2018 Design tools
From: Dave Vanden Bout <devb@xess.com>
Date: Mon, 02 Oct 2000 10:32:55 -0400
Links: << >>  << T >>  << A >>
> These parts are long over the hill (They were announced as end of life parts
> some 7 or 8 years ago).  None of the current generation tools support it.  The
> last xilinx tool set to support them was XACT6, which was replaced about 3 years
> ago (and in my case 2 machines ago) by the M1 tools.  If you can find someone
> with a copy, and who is willing to give up the accompanying dongle you may still
> not be able to use the tools unless you resurrect an old machine too.  XACT6 was
> a dos/windows3.1 toolset.  It didn't work very well under windows95.  At this
> point, if your goal is learning, I'd recommend spending the $200 or so for the
> XESS board plus the student edition of the current tools.  The kit comes with
> labs and a text.  In the end, I think you'll wind up spending less and you'll
> have current device experience.

It's my job to keep up with these things so I'll make a slight correction here:  the
XESS XSK-40 kit has been reduced from $199 to $154.  Prentice Hall no longer
includes the labs and hardcopy text in the Xilinx Student Edition so XESS will
provide its own version of these materials at www.xess.com in November.

>
>
> "news.gate.net" wrote:
> >
> > I'm a rather "advanced" hobbyist, looking to get into some FPGA design (for
> > starters, as some glue logic for an ISA PC Card).  I recently stripped an
> > old videoconferencing system and it had about 50 XC2018-P84C chips on it
> > (also some XC3042's and a pair of XC3090's).  It also has a lot of GALs,
> > 57C291's, 57C45's, and some various Altera stuff.  In short, it is a
> > reprogrammable logic goldmine.
> >
> > But, many of the parts are out of production and finding development support
> > for them is difficult.  I have Protel 99 SE and have a simple design input
> > into it (schematic).  But, when I compile it, I get a message about missing
> > PLA2XNF.EXE.  I'm assuming this is part of the Xilinx XACT tools?
> >
> > The problem I have is that nothing currently available from Xilinx seems to
> > support my XC2018 chips.  I prefer to use the Xilinx stuff at this point,
> > because I can store the programming in an EPROM (I have a burner) rather
> > than having to program the actual chip (I don't have a burner capable of
> > doing any programmable logic burning).
> >
> > Does anyone have some pointers as to where I can find support for these
> > devices?  (It appears that the student [and even professional] editions of
> > Xilinx' stuff are geared toward their newer parts).
> >
> > Thanx in advance!
> > Robert Garito
> > rgarito@gate.net
>
> --
> -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  or http://www.fpga-guru.com

--
|| Dr. Dave Van den Bout   XESS Corp.               (919) 387-0076 ||
|| devb@xess.com           2608 Sweetgum Dr.        (800) 549-9377 ||
|| http://www.xess.com     Apex, NC 27502 USA   FAX:(919) 387-1302 ||



Article: 26055
Subject: Re: FPGA Express strikes again! Xilinx response
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Mon, 02 Oct 2000 15:51:38 +0100
Links: << >>  << T >>  << A >>


"Keith R. Williams" wrote:

> On Sun, 1 Oct 2000 20:45:18, Rick Filipkiewicz <rick@algor.co.uk>
> wrote:
>
> >
> >
> > Ray Andraka wrote:
> >
> > > Like I said, lots of bells and whistles, but nothing that I can see to justify
> > > the extra $$$.
> > >
> >
> > It's the special IPO release. The investors just got a bit freaked by the number
> > of amateurs using the tool.
>
> Yikes.  I can be considered a newbie here, but at $20K for a
> license, plus $3K a year to keep the tool working, I don't think
> I qualify as an "amateur".  I agree though. I don't see the bux
> for Pro.
>
> ..Amplify is another issue altogether. I'd like some information
> from anyone with experience with it.
>
> ----
>   Keith

I'm sorry the ``amateur'' comment was meant as a joke about the meaningless addition
of things like ``PRO'' to tools just to make them sound more serious & heavyweight.

Article: 26056
Subject: Re: GPIO on AVNET Xilinx FPGA board? any cables?!?
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Mon, 02 Oct 2000 16:07:14 +0100
Links: << >>  << T >>  << A >>


David Forbes wrote

> That AMP connector is a board-to-board 0.8mm pitch job, so you will not
> find a cable that fits it.
>
> This sounds like a good time to vent some steam in the direction of
> whoever designed a prototypng board with a 0.8mm pitch connector on it.

>

You might try looking in the Samtec catalogue. They have some 0.8 mm stuff that
might be suitable for a bodge. And you can generally get their stuff within a
few days or max 5-6 if its a custom special. Contrast AMP and or their distis:
MOQ = 10-20 GUnits, lead time after your design has qualified for museum
status.

Company connector rules after much bitter experience:

(1) If its not a PC world connector and its not in the Samtec book it doesn't
get used.

(2) If a client insists on some AMP special they source it.



Article: 26057
Subject: "Xilinx Adds FPGA Support to Free Web Design Tools"
From: "Jan Gray" <jsgray@acm.org>
Date: Mon, 02 Oct 2000 15:10:13 GMT
Links: << >>  << T >>  << A >>
"Xilinx, Inc. today announced full support of the entire Spartan-II FPGA
family as well as the 300,000 system gate Virtex XCV300E FPGA in the WebPACK
ISE tool suite. The free downloadable software, previously available only
for Xilinx CPLDs, now offers a zero-cost-of-entry point for designing with
Xilinx FPGAs."

See http://www.xilinx.com/prs_rls/webfpga.html.

(Yippee!)

Jan Gray
Gray Research LLC
www.fpgacpu.org




Article: 26058
Subject: Re: multi-input adders in virtex ?
From: Ray Andraka <ray@andraka.com>
Date: Mon, 02 Oct 2000 16:38:52 GMT
Links: << >>  << T >>  << A >>
Yes, in some cases by a factor of 2.  One case for which I have become somewhat
famous for is the one I presented at the FPGA'00 panel where an FIR filter (same
one we've been discussing here) placed and routed automagically for a 95 MHz bit
clock (missed the time spec which was for 120 MHz).   The exact same design,
when hand placed achieves 147 MHz.  Don't let anyone convince you the tools can
do as good a job with placement as the human brain.

Placement can be done entirely from within the floorplanner, but if you are
going to be reusing pieces, then it is often worth the effort to include the
placement in the design itself so that you don't waste time repeating the layout
on the same macro.

erika_uk@my-deja.com wrote:
> 
> hi,
> 
> for my designs, i have never introduced a placement information( apart
> from the ones related to the mapping)
> when my synchronous deisgn works as expected, i push the timing
> constraint up, to know what speed i can achieve
> 
> from your experience( and hopefully from other experts ones), apart
> from the time saved from implementation process, have you beat timing
> driven design with a pre-placement one ?
> 

-- 
-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  or http://www.fpga-guru.com

Article: 26059
Subject: Re: Multiplication
From: Ray Andraka <ray@andraka.com>
Date: Mon, 02 Oct 2000 16:40:32 GMT
Links: << >>  << T >>  << A >>
No hardware changes needed in the multiplier itself.  All you have to do is
change the interpretation of the position of the radix point.  It is equivalent
to scaling your input and output by 32 in your case.

Vipan Kakkar wrote:
> 
> Dear Friends:
> 
> I would like to know that how to multiply e.g. 1100 with 0.0011. I
> already have an array multplier to multiply e.g. 1100 with 0110. But,
> now I want to multiply fractions (i.e. .0011) using the same multiplier,
> so what modifications should I do in the existing multiplier.
> 
> Regards

-- 
-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  or http://www.fpga-guru.com

Article: 26060
Subject: Re: "Xilinx Adds FPGA Support to Free Web Design Tools"
From: rickman <spamgoeshere4@yahoo.com>
Date: Mon, 02 Oct 2000 13:22:13 -0400
Links: << >>  << T >>  << A >>
Someone should tell the Xilinx product people about this. Looking at the website shows no indication
of any of this on the web pages that describe the WebPack ISE software. 

I also noticed that there are big differences other than just the chips targeted. The existing
WebPack does not include FPGA Express synthesis, (not sure about XST, whatever that is) but the
press release indicates that it now includes a full HDL capability. 

"WebPACK ISE is an Integrated Synthesis Environment?, which includes VHDL, Verilog and ABEL
synthesis, HDL simulation and test vector generation..."

Opps, reading the full press release says, "is scheduled for release in mid-October 2000". So I
guess we will have to wait a few more weeks...

Now if we can just get support for numeric_std!  :)


Jan Gray wrote:
> 
> "Xilinx, Inc. today announced full support of the entire Spartan-II FPGA
> family as well as the 300,000 system gate Virtex XCV300E FPGA in the WebPACK
> ISE tool suite. The free downloadable software, previously available only
> for Xilinx CPLDs, now offers a zero-cost-of-entry point for designing with
> Xilinx FPGAs."
> 
> See http://www.xilinx.com/prs_rls/webfpga.html.
> 
> (Yippee!)
> 
> Jan Gray
> Gray Research LLC
> www.fpgacpu.org

-- 

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

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

Internet URL http://www.arius.com

Article: 26061
Subject: Synthesis failures
From: Andy Peters <"apeters <"@> n o a o [.] e d u>
Date: Mon, 02 Oct 2000 10:45:29 -0700
Links: << >>  << T >>  << A >>
Hey, believe it or not, FPGA Express v3.4 is not the ONLY target for
this bitch-moan.  Synplify Pro v6 fails, too.

Target: Spartan XCS20XL.

I instantiate (and also, with Synplify, infer) a bunch of 16x1 dual-port
RAMs.  The gotcha is that these RAMs need to be clocked on the falling
edge.  Now, after reading the Xilinx Libraries docs, I think that
there's no such library part (falling-edge clocked RAM) for SpartanXL. 
But, looking at the CLB with the FPGA Editor, there's a
clock-polarity-select mux right in front of the RAM clock input,
soooooooo....

Anyways, according to Xilinx tech support, the clock inverter should be
pushed into the CLB.

But BOTH synthesis tools invert the clock in an LUT, then BUFG the
inverted clock.  Therefore, there's no way that the inverter can be
absorbed into the CLB.

This is a bug, plain and simple.

-- a
----------------------------
Andy Peters
Sr. Electrical Engineer
National Optical Astronomy Observatory
950 N Cherry Ave
Tucson, AZ 85719
apeters (at) n o a o [dot] e d u

Article: 26062
Subject: Re: FPGA Express strikes again! Xilinx response
From: Andy Peters <"apeters <"@> n o a o [.] e d u>
Date: Mon, 02 Oct 2000 10:46:50 -0700
Links: << >>  << T >>  << A >>
"Keith R. Williams" wrote:
> 
> On Sun, 1 Oct 2000 20:45:18, Rick Filipkiewicz <rick@algor.co.uk>
> wrote:
> 
> >
> >
> > Ray Andraka wrote:
> >
> > > Like I said, lots of bells and whistles, but nothing that I can see to justify
> > > the extra $$$.
> > >
> >
> > It's the special IPO release. The investors just got a bit freaked by the number
> > of amateurs using the tool.
> 
> Yikes.  I can be considered a newbie here, but at $20K for a
> license, plus $3K a year to keep the tool working, I don't think
> I qualify as an "amateur".  I agree though. I don't see the bux
> for Pro.

Wow! I didn't realize the tool was $20K.  I'll pass, thank you very
much. 

-- a
----------------------------
Andy Peters
Sr. Electrical Engineer
National Optical Astronomy Observatory
950 N Cherry Ave
Tucson, AZ 85719
apeters (at) n o a o [dot] e d u

Article: 26063
Subject: Re: Xilinx Student Edition 2.1i first impressions
From: Andy Peters <"apeters <"@> n o a o [.] e d u>
Date: Mon, 02 Oct 2000 10:48:49 -0700
Links: << >>  << T >>  << A >>
Seems to me that once the "next generation" of Xilinx' tools are
released, the previous generation becomes the Student Edition.

What an interesting way to keep selling the old stuff!

-- a
----------------------------
Andy Peters
Sr. Electrical Engineer
National Optical Astronomy Observatory
950 N Cherry Ave
Tucson, AZ 85719
apeters (at) n o a o [dot] e d u

Article: 26064
Subject: Re: multi-input adders in virtex ?
From: "Jan Gray" <jsgray@acm.org>
Date: Mon, 02 Oct 2000 17:48:51 GMT
Links: << >>  << T >>  << A >>
The efficacy of floorplanning (pipelined) adder trees (a 34% speed
improvement, 78% PAR runtime savings) and other circuit types (constant
coeff multipliers, filters, and butterfly networks), is explored in Satnam
Singh's FCCM'00 paper "Death of the RLOC?", which is available at
www.xilinx.com/labs/satnam.

Jan Gray
Gray Research LLC




Article: 26065
Subject: Re: Programming Cypress Graphics Clock Generator
From: Steve Wiseman <steve@steves-house.org.uk>
Date: Mon, 02 Oct 2000 19:26:16 +0100
Links: << >>  << T >>  << A >>
Sherdyn wrote:
> 
> Hi,
> 
> I have encountered a strange problem with Cypress PLL ICD2061A. 

I've used these successfully several times - they're great, but not
always obvious. Have you used the bitstream generation proglet, or
hand-cranked the calculations?
Things I've got wrong in the past:-
* Bitstream backwards, (the datasheet used to be rather ambigous)
* non-monotonic serial clock, <- this was the real problem, use a
fast, fast scope , or a few pF just to take the edge off anyway. 
* looking at the wrong output pin. (D'oh!)

I'd pay good money for one of these devices, in a regular 14-pin
oscmod case, with a watch LCD and up-down buttons for frequency,
sometimes, just to see how close to the edge a design is running.
Only the non-availability of watch LCDs has stopped me doing this
myself...)  

> So what could FPGA causes the PLL not to work well? Jitter or something
> else. Any clue?

The don't seem too vulnerable to jitter, but I generally never hung
them off anything other than other oscmods, or used their internal
crystal drivers. 

Good luck - 

Steve

Article: 26066
Subject: Re: "Xilinx Adds FPGA Support to Free Web Design Tools"
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Mon, 02 Oct 2000 20:24:03 +0100
Links: << >>  << T >>  << A >>


rickman wrote:

> I also noticed that there are big differences other than just the chips targeted. The existing
> WebPack does not include FPGA Express synthesis, (not sure about XST, whatever that is) but the
> press release indicates that it now includes a full HDL capability.
> --
>

XST, Xilinx's own Verilog & VHDL synth tool, is now just about useable since the 3.1i release has a
decent-ish manual. It also runs nicely from the command line. I'm going to try & benchmark it against
Synplify fairly soon.

Now, Xilinx, how about actually getting some SpartanII's out the door so we can use the new free s/w.


Article: 26067
Subject: Re: Synthesis failures
From: Andy Peters <"apeters <"@> n o a o [.] e d u>
Date: Mon, 02 Oct 2000 13:35:26 -0700
Links: << >>  << T >>  << A >>
Replying to myself:

> Anyways, according to Xilinx tech support, the clock inverter should be
> pushed into the CLB.
> 
> But BOTH synthesis tools invert the clock in an LUT, then BUFG the
> inverted clock.  Therefore, there's no way that the inverter can be
> absorbed into the CLB.

I just got off the phone with Synplicity tech support (kudos to them:
human beings answer the phone, and I was immediately transferred to tech
support, and no one asked me for a User ID or anything!) and confirmed
that it's a bug.  It was reported for Virtex2; I mentioned that I'm
using SpartanXL (and XC4KXLA, which is essentially the same thing).

Falling edge clocks are dealt with by inverting the clock and buffering
it, not by using the CLB clock-polarity mux resources.  How long has
that mux been in the architecture, anyways?  No one uses the falling
edge of the clock?

The workaround is to instantiate the falling-edge triggered flops. 
That's not a solution when it's a counter I'm dealing with.

-- a
----------------------------
Andy Peters
Sr. Electrical Engineer
National Optical Astronomy Observatory
950 N Cherry Ave
Tucson, AZ 85719
apeters (at) n o a o [dot] e d u

Article: 26068
Subject: Re: Synthesis failures
From: rickman <spamgoeshere4@yahoo.com>
Date: Mon, 02 Oct 2000 17:25:50 -0400
Links: << >>  << T >>  << A >>
This seems very odd that both FPGA Express and Synplify have the same bug that I assume worked
correctly at one time. It would be hard to imagine that this problem has been there from the start
and no one noticed it. So how could two vendors develop the same bug at the same time? 



Andy Peters wrote:
> 
> Replying to myself:
> 
> > Anyways, according to Xilinx tech support, the clock inverter should be
> > pushed into the CLB.
> >
> > But BOTH synthesis tools invert the clock in an LUT, then BUFG the
> > inverted clock.  Therefore, there's no way that the inverter can be
> > absorbed into the CLB.
> 
> I just got off the phone with Synplicity tech support (kudos to them:
> human beings answer the phone, and I was immediately transferred to tech
> support, and no one asked me for a User ID or anything!) and confirmed
> that it's a bug.  It was reported for Virtex2; I mentioned that I'm
> using SpartanXL (and XC4KXLA, which is essentially the same thing).
> 
> Falling edge clocks are dealt with by inverting the clock and buffering
> it, not by using the CLB clock-polarity mux resources.  How long has
> that mux been in the architecture, anyways?  No one uses the falling
> edge of the clock?
> 
> The workaround is to instantiate the falling-edge triggered flops.
> That's not a solution when it's a counter I'm dealing with.
> 
> -- a
> ----------------------------
> Andy Peters
> Sr. Electrical Engineer
> National Optical Astronomy Observatory
> 950 N Cherry Ave
> Tucson, AZ 85719
> apeters (at) n o a o [dot] e d u

-- 

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

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

Internet URL http://www.arius.com

Article: 26069
Subject: Re: Synthesis failures
From: Jim Granville <jim.granville@designtools.co.nz>
Date: Tue, 03 Oct 2000 12:05:12 +1300
Links: << >>  << T >>  << A >>
rickman wrote:
> 
> This seems very odd that both FPGA Express and Synplify have the same bug that I assume worked
> correctly at one time. It would be hard to imagine that this problem has been there from the start
> and no one noticed it. So how could two vendors develop the same bug at the same time?

It's not so much a 'bug' that has developed, but a short comming, that
wastes silicon.
Attitude : Since it works the way they do it now, why fix it ?

Falling edge clocks are a blind spot industry wide - they have only just
recently become
'std' on CPLD's

-- 
======= 80x51 Tools & IP Specialists  =========
= http://www.DesignTools.co.nz

Article: 26070
Subject: Re: FPGA Express strikes again! Xilinx response
From: krw@attglobal.net (Keith R. Williams)
Date: Tue, 03 Oct 2000 00:08:02 GMT
Links: << >>  << T >>  << A >>
On Mon, 2 Oct 2000 17:46:50, Andy Peters <"apeters <"@> n o a o 
[.] e d u> wrote:

> "Keith R. Williams" wrote:
> > 
> > On Sun, 1 Oct 2000 20:45:18, Rick Filipkiewicz <rick@algor.co.uk>
> > wrote:
> > 
> > >
> > >
> > > Ray Andraka wrote:
> > >
> > > > Like I said, lots of bells and whistles, but nothing that I can see to justify
> > > > the extra $$$.
> > > >
> > >
> > > It's the special IPO release. The investors just got a bit freaked by the number
> > > of amateurs using the tool.
> > 
> > Yikes.  I can be considered a newbie here, but at $20K for a
> > license, plus $3K a year to keep the tool working, I don't think
> > I qualify as an "amateur".  I agree though. I don't see the bux
> > for Pro.
> 
> Wow! I didn't realize the tool was $20K.  I'll pass, thank you very
> much. 

Be careful here.  I first bought Synplify for DynaChip (no 
laughing out there folks ;-).  It was considerably less than $20K
for a single vendor (though I can't remember).  After Xilinx 
vaporized DynaChip (no, I never saw a DY8000 part, but have some 
pretty DY6000s that I may mount on a plaque, just to keep me 
humble) I had no choice to go for the to the Xilinx license, and 
Synplicity required me to go to the full-vendor license.  

The point being that a single-vendor license may be far cheaper 
than the circutious route I went through.  ,,,I certainly hope 
so.  It's a *great* tool, but it *is* expensive.  

Amplify is even more expen$ive, but it looks to be another 
winner.  I put the $$ in the budget, but we'll see if anyone 
salutes when I try to spend the money.  

That said, I'm still confused about the process, but that's for 
their FAE's to straighten out (at $30K/WinLicense, I think 
they'll talk - when I'm ready ;-).

----
  Keith

P.S.  I'm not threatened by being called an "amateur".  I 
understood the comment.  I am a relative newbie though (25+ years
as a hardware designer, 1+ years doing this *neat* stuff).  Just 
to be clear, I am also rather new to synthesis (good grief, I've 
been dreaming *in* VHDL recently), so treat me kindly.  ;-)

----
  Keith

Article: 26071
Subject: Re: multi-input adders in virtex ?
From: krw@attglobal.net (Keith R. Williams)
Date: Tue, 03 Oct 2000 00:14:01 GMT
Links: << >>  << T >>  << A >>
On Mon, 2 Oct 2000 17:48:51, "Jan Gray" <jsgray@acm.org> wrote:

> The efficacy of floorplanning (pipelined) adder trees (a 34% speed
> improvement, 78% PAR runtime savings) and other circuit types (constant
> coeff multipliers, filters, and butterfly networks), is explored in Satnam
> Singh's FCCM'00 paper "Death of the RLOC?", which is available at
> www.xilinx.com/labs/satnam.

Ok, I've tried to get this discussion going a time or two...  
What about HDL generated floor-planning?  I'm specificly looking 
for *real* information about Synplicity's Amplify.  I can read 
their Power-Point-Ware and I'm impressed.  Anyone that has 
actually used it would be very helpful.  Before I write the check
for this widget ($30K + 20% forever) I'd like to be sure it does 
what I need.

----
  Keith


Article: 26072
Subject: Re: Synthesis failures
From: rickman <spamgoeshere4@yahoo.com>
Date: Mon, 02 Oct 2000 20:49:47 -0400
Links: << >>  << T >>  << A >>
Jim Granville wrote:
> 
> rickman wrote:
> >
> > This seems very odd that both FPGA Express and Synplify have the same bug that I assume worked
> > correctly at one time. It would be hard to imagine that this problem has been there from the start
> > and no one noticed it. So how could two vendors develop the same bug at the same time?
> 
> It's not so much a 'bug' that has developed, but a short comming, that
> wastes silicon.
> Attitude : Since it works the way they do it now, why fix it ?
> 
> Falling edge clocks are a blind spot industry wide - they have only just
> recently become
> 'std' on CPLD's

Are you sure that this is not a new bug. I used both rising and falling
edge clocks in a VHDL design I did using FPGA Express and Xilinx M1.5
about two years ago. There were no inverters in any of my clock paths. 


-- 

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

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

Internet URL http://www.arius.com

Article: 26073
Subject: Re: FPGA Express strikes again! Xilinx response
From: bob_42690@my-deja.com
Date: Tue, 03 Oct 2000 02:17:24 GMT
Links: << >>  << T >>  << A >>


>
> P.S.  I'm not threatened by being called an "amateur".  I
> understood the comment.  I am a relative newbie though (25+ years
> as a hardware designer, 1+ years doing this *neat* stuff).  Just
> to be clear, I am also rather new to synthesis (good grief, I've
> been dreaming *in* VHDL recently), so treat me kindly.  ;-)

I had a dream that I was debugging a VHDL simulation!  Now that's not
right.  Must be some type of recursion thing.

Bob


Sent via Deja.com http://www.deja.com/
Before you buy.

Article: 26074
Subject: Xilinx Demo Board
From: "Qian Zhang" <qianz@cae.wisc.edu>
Date: Mon, 2 Oct 2000 21:26:37 -0500
Links: << >>  << T >>  << A >>
Hi Friends

Here I have Xilinx demo board,
but I dont know how to use it.
Where can I find some information?

Also Who can tell me how to generate bit file
so that I can download to demo board?
Thank you very much!


Qian





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