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 34350

Article: 34350
Subject: Confusion around BUS LVDS in Virtex-II
From: "Hakon Lislebo" <etohliNOSPAM@eto.ericsson.se>
Date: Wed, 22 Aug 2001 09:45:43 +0200
Links: << >>  << T >>  << A >>
Hi,
I trying to implement an BUS LVDS interface with Virtex-II, but I am getting
a little confused. In the document
http://www.xilinx.com/products/virtex/handbook/ug002_ch2_lvds.pdf it is said
that you still have to use the IOBUF_LVDS primitive (I have used this in
Virtex-E) for blvds. BUT this is not recognized by ngbuild. I have seen some
documentation that points to other primitives like IOBUFDS together with an
IOSTANDARD attribute.

My questions is as follows:
1. What primitive should I use for buslvds applications?
2. Is the voltage swing on the bus still +/- 350mV with 2 x 100 Ohms
terminations using blvds? (The datasheet confuses me and have spec's on
LVDS, not BLVDS.)

Thank you for reading!
Regards Hakon Lislebo



Article: 34351
Subject: Re: hardware damage to a Virtex or Spartan-II?
From: "Tim" <tim@rockylogic.com.nospam.com>
Date: Wed, 22 Aug 2001 11:21:42 +0100
Links: << >>  << T >>  << A >>

"Philip Freidin" <philip@fliptronics.com> wrote in message
news:chu5otc4n394qaklkj21e0kisubphc5ng4@4ax.com...
> On 21 Aug 2001 20:43:17 GMT, gah@ugcs.caltech.edu (glen herrmannsfeldt) wrote:
> >What is the maximum clock speed for Virtex-II, FF, through one LUT,
> >to another FF?
>
> In a V-II , -5 speed grade,  M3.3.08i speed files, with careful placement,
> about 666MHz .
>

The mark of the beast.  See alt.conspiracy.fpga




Article: 34352
Subject: Re: Logic Emulation
From: "Tim" <tim@rockylogic.com.nospam.com>
Date: Wed, 22 Aug 2001 11:30:08 +0100
Links: << >>  << T >>  << A >>

"Michael Boehnel" <boehnel@iti.tu-graz.ac.at> wrote in message
news:3B835877.71635D2A@iti.tu-graz.ac.at...
> Are there any significant differences between a logic emulated version
> (implemented in an FPGA) and an ASIC version.
>
> I could imagine that at least some timing differences should occure. Are
> these timing differences important?

   Raw speed is usually higher in ASIC, at a given process level.

> Are there any other differences?

   ASICs are a generation or more ahead in density, if you can afford it
   No tri-states in ASICs (going away in FPGAs)
   Wierd RAM/ROM configurations in ASICs, regular in FPGAs
   Embedded DRAM in ASICs
   Gated clocks in ASICs, less common in FPGAs
   ASICs are usually lower power

>
> Can anybody tell practical experiences?
>
> Michael
>
>
>



Article: 34353
Subject: Re: FPGA MP3 decoder
From: robin@ivo.co.uk (Robin Kinge)
Date: 22 Aug 2001 04:18:56 -0700
Links: << >>  << T >>  << A >>
I had a look at Celoxica as well, but they want $11,000 for the board...

Robin

> At this point I'm going to use a board from Celoxica (has a Xilinx FPGA)
> pretty nice.. programming it in Handel-C...
> 
> peace,
> - Mark

Article: 34354
Subject: Re: Slowing PCI for FPGA
From: hamish@cloud.net.au
Date: Wed, 22 Aug 2001 12:24:25 GMT
Links: << >>  << T >>  << A >>
Joey Oravec <joey@sun.science.wayne.edu> wrote:
> partitioning. Synplify Pro is great for many projects, but if you're
> doing a serious design you need their higher end tools. They really
> need to integrate Amplify and Certify though; right now they seem to
> be seperate products.

Do you see much advantage with Amplify? Do you see it doing anything
which you couldn't do yourself with Synplify Pro and the Xilinx floorplanner?

I have access to an Amplify license at work and in my brief experience
with it, it just seems to be a way to enter your floorplan and have
it appear in an NCF. So I just enter it into the UCF myself, with
floorplanner or by hand.

I'd like to be proven wrong though. What does the tool itself actually
do with the floorplanning information? What optimisations can
it make at synthesis time given the floorplan?

Hamish
-- 
Hamish Moffatt VK3SB <hamish@debian.org> <hamish@cloud.net.au>

Article: 34355
Subject: Re: I NEED TO BUY A FPGA BOARD
From: "Sanjay Maniku" <Sanjay.Maniku@celoxica.com>
Date: Wed, 22 Aug 2001 12:51:23 +0000 (UTC)
Links: << >>  << T >>  << A >>

Celoxica  have a couple of cool FPGA protoyping boards, a lower end one
(RC100) with a Xilinx SpartanII 200 device on and a load of I/O (video
in/out, ps2 ports, parallel ports, leds, 7 segment displays) which comes
with a great board support package written in Handel-C, and a high end
platform (RC1000) which is a PCI based board designed specifically for data
processing operations. The RC1000 features Xilinx Virtex chips (upto V3200E)
and four independent RAM banks of up to 8 MB each plus  FPGA side and Host
side libraries for DMA to memory, host to FPGA communication and memory
arbitration.

Check them out @  http://www.celoxica.com/products/boards/index.htm

Sanjay Maniku
Senior Applications Engineer
Celoxica Limited
20 Park Gate
Milton Park, Abingdon
Oxon OX14 4RT
Tel: 01235 863656 Fax: 01235 863648
sanjay.maniku@celoxica.com
http://www.celoxica.com
No. 3209209



This email and any files transmitted with it are confidential and 
may be legally privileged. It is intended solely for the use of the 
individual or entity to whom it is addressed. If you have received 
this in error, please contact the sender and delete the material 
immediately. Whilst this email has been swept for viruses, you 
should carry out your own virus check before opening any 
attachment. Celoxica Ltd accepts no liability for any loss or 
damage which may be caused by software viruses or interception 
or interruption of this email.


_____________________________________________________________________
This message has been checked for all known viruses by Star Internet
delivered through the MessageLabs Virus Scanning Service. 

This email and any files transmitted with it are confidential and 
may be legally privileged. It is intended solely for the use of the 
individual or entity to whom it is addressed. If you have received 
this in error, please contact the sender and delete the material 
immediately. Whilst this email has been swept for viruses, you 
should carry out your own virus check before opening any 
attachment. Celoxica Ltd accepts no liability for any loss or 
damage which may be caused by software viruses or interception 
or interruption of this email.


-- 
Posted from mail17.messagelabs.com [62.231.131.67] 
via Mailgate.ORG Server - http://www.Mailgate.ORG

Article: 34356
(removed)


Article: 34357
Subject: Re: FPGA MP3 decoder
From: eng_any@yahoo.com (Lewis)
Date: 22 Aug 2001 05:52:54 -0700
Links: << >>  << T >>  << A >>
"Mark" <m.smulders@philips.com> wrote in message news:<3b827206$0$14114$4d4ebb8e@read-nat.news.nl.uu.net>...
> Alrite!
> thanks for answering my question :-))
> 
> At this point I'm going to use a board from Celoxica (has a Xilinx FPGA)
> pretty nice.. programming it in Handel-C...
> 
> peace,
> - Mark
> 

Surely this is an expensive route to take. Verilog or VHDL would be
better than Handel-C and more portable. Also use a cheaper board -
I've heard the Celoxica gear costs a lot.

L

Article: 34358
(removed)


Article: 34359
Subject: Re: Can't Install Modelsim - Alternatives for Verilog Simulation???
From: Dave Colson <dscolson@rcn.com>
Date: Wed, 22 Aug 2001 09:56:58 -0400
Links: << >>  << T >>  << A >>
Dave,

I have had problems too. I am using the Xilinx version of Modelsim.
The problem is with the Flexlm software recognizing the Ethernet card
MAC
under windows 2000.

There are several programs associated with the licensing software.
lmutil, lmtools, lmgrd and lmgrdxxx.dll where xxx can be 326b for
instance.
Also, every vendor who uses this software protection scheme will provide
these utilities
with there distribution. So you end up with a half a dozen copies of
these utilities
all over your system. The real problem is getting someone to admit it is
their problem.
Frankly IMO it is a Flexlm problem. However, they will not give support
to end users, so
trying to contact them is sort of useless.

I suspect the problem lies in the lmgr326b.dll. Here is what I did to
solve it for myself.

I got real desperate and "rename" the lmgr326b.dll..... It worked!

Although this worked for me, in general it will not work. And I would
not suggest you try it
unless you are really desperate.

I am not sure what the Flexlm software does, and like I said, they will
not support
end users. Oh, they allow you to ask one question, so make it a good
one.

Their (www.globetrotter.com) FAQ and end user stuff gives some common
solutions to this problem.
However, they do not say anything about the dll in their FAQ concerning
this problem.

Maybe you can convince someone at Model tech to resolve this problem.

Hope this helps a little

Dave Colson


Dave Feustel wrote:

> Modelsim licensing refuses to work on my computer.
>
> What alternatives to Modelsim are there for Verilog simulation
> on Windows 2000?
>
> Thanks.


Article: 34360
Subject: Re: Principles of Verifiable RTL Design (2nd ed)
From: Ben Franchuk <bfranchuk@jetnet.ab.ca>
Date: Wed, 22 Aug 2001 08:21:27 -0600
Links: << >>  << T >>  << A >>
James Lee wrote:
> 
> I reviewed the book.  It was quite interesting, I did not fully
> understand it untill I spoke with the authors.  The focus is on
> picking a set of tools and libraries that work consistently with a
> style of verilog.
> 
That sounds bad for the state of hardware development. Bugs in my
code/schematics I can correct, yet with the fact that the final output of logic
synthesizers is totally hidden from the public how do I know if there are no
bugs in the system.
Ben.
- 
Standard Disclaimer : 97% speculation 2% bad grammar 1% facts.
"Pre-historic Cpu's" http://www.jetnet.ab.ca/users/bfranchuk
Now with schematics.

Article: 34361
Subject: Re: hardware damage to a Virtex or Spartan-II?
From: bryan@srccomp.com (Bryan)
Date: 22 Aug 2001 07:44:24 -0700
Links: << >>  << T >>  << A >>
gah@ugcs.caltech.edu (glen herrmannsfeldt) wrote in message news:<9luh55$jak@gap.cco.caltech.edu>...
> bryan@srccomp.com (Bryan) writes:
> 
> >> The path through CLBs and routing to make an oscillator like that
> >> shown should be long enough to keep the frequency relatively low,
> >> compared to discrete logic.
> >> 
> >> -- glen
>  
> >The LUT oscillator that I created had 25% of the chip running at
> >.45ns(lut) + .5ns(route) or ~ 512Mhz and 75% of the chip running at
> >.45ns + .9ns or
> >~ 740Mhz.  That is a Virtex-II 1000 speed 4.  This of course is by 
> >the book delays, which we have found actual routes to be 80% of the
> >delay
> 
> What is the maximum clock speed for Virtex-II, FF, through one LUT,
> to another FF?  Can Virtex-II run a design where most of the logic
> runs at that clock speed?
> 
> -- glen


I didn't say anything about using FFs.  This is a design using nothing
but LUTs.  This design is not clocking latches.  It is simpling
toggling LUTs as fast as they will toggle.

Bryan

Article: 34362
Subject: Re: How does For Loop works in AHDL
From: "Abhimanyu Rastogi" <abhi_rastogi@hotmail.com>
Date: Wed, 22 Aug 2001 14:58:11 GMT
Links: << >>  << T >>  << A >>
yes...    I need DATA
cuz, i have to serailly output data_word[21..0]( just a DFFE) out thru DATA
(output pin & DFFE)
and i really don't wanna put lb_word[] = data_word[] ....  i only did that
cuz DATA = data_word[i] was not working properly...otherwise i would have
made lb_word[i] = LB where LB = DATA ( they are interconected on the synth
chip)

So, how serial output work .....   is there some rule to follow for it??

Abhimanyu


Russell Shaw <rjshaw@iprimus.com.au> wrote in message
news:3B831807.1A01DE25@iprimus.com.au...
> IIRC, this should work:
>
> lb_word[]=data_word[]
>
> Do you really need DATA ?
>
> Abhimanyu Rastogi wrote:
> >
> > hello,
> >
> > How does a for loop executes the statements in AHDL ....
> >
> > for instance...:
> >
> >  FOR i IN 0 to 21 GENERATE
> >   DATA.d = data_word[i];
> >   lb_word[i] = data_word[i]; --assuming DATA is connected to LB
> >  END GENERATE;
> >
> > where...   we have data_word[21..0] DFFE; lb_word[21..0] DFFE and DATA
DFFE;
> >
> > so now on simulation lets say i provide data_word[] = (H"38CCAA")  then
wat
> > should be the result in DATA n lb_word[]
> >
> > Also, does it have sequentially or concurrently..??
> >
> > Thx
> >
> > Abhimanyu rastogi
>
> --
>    ___                                           ___
>   /  /\                                         /  /\
>  /  /__\ Russell Shaw, B.Eng, M.Eng(Research)  /  /\/\
> /__/   / Victoria, Australia, Down-Under      /__/\/\/
> \  \  /  http://home.iprimus.com.au/rjshaw    \  \/\/
>  \__\/                                         \__\/



Article: 34363
(removed)


Article: 34364
Subject: Re: hardware damage to a Virtex or Spartan-II?
From: Austin Lesea <austin.lesea@xilinx.com>
Date: Wed, 22 Aug 2001 08:09:38 -0700
Links: << >>  << T >>  << A >>
John,

Make, model number, and where to buy the IR temp sensor?

I get sooo many questions, and it would be worth its weight in gold to publicize this
info!

Austin

John Larkin wrote:

> On Mon, 20 Aug 2001 19:10:26 +0100, Rick Filipkiewicz
> <rick@algor.co.uk> wrote:
>
> >Peter,
> >
> >On top of all the other useful stuff you contribute to CAF you have just answered a
> >long-standing (or burning ?) question of mine regarding the calibration of heat
> >sensing fingertips.
>
> Peter,
>
> my right forefinger is calibrated as follows:
>
> 50C - ok for infinite duration contact
>
> 52C - 10 seconds to pullaway
>
> 60C - 1 second to pullaway
>
> Interpolate linearly between points. Your digits may vary.
>
> We just got a cheap ($79) infrared temp sensor, which is cool (no pun,
> really) for scanning FPGAs on a board.
>
> John
>
> >
> >To go further and get another data point: Last summer I was wondering why our new
> >board wasn't doing anything from power-on. I, very briefly, put my finger on it &
> >came away with a large &  painful blister that was still very sore 2 days later.
> >
> >Cause = BGA chip pinout mirrored (not by me I hasten to add).
> >
> >What temp do you think the package would have reached ?


Article: 34365
Subject: Re: How does For Loop works in AHDL
From: John_H <johnhandwork@mail.com>
Date: Wed, 22 Aug 2001 16:36:54 GMT
Links: << >>  << T >>  << A >>
The FOR GENERATE does't do anything for you serially.  All you're doing is
implementing 22 different statements by using one FOR GENERATE block rather
than typing 22 different equations by hand (or doing one simple assign in this
case).  If you need serial output you need to think in terms of registers and
how those registers - all clocked at once - can give you your output.


Abhimanyu Rastogi wrote:

> yes...    I need DATA
> cuz, i have to serailly output data_word[21..0]( just a DFFE) out thru DATA
> (output pin & DFFE)
> and i really don't wanna put lb_word[] = data_word[] ....  i only did that
> cuz DATA = data_word[i] was not working properly...otherwise i would have
> made lb_word[i] = LB where LB = DATA ( they are interconected on the synth
> chip)
>
> So, how serial output work .....   is there some rule to follow for it??
>
> Abhimanyu
>
> Russell Shaw <rjshaw@iprimus.com.au> wrote in message
> news:3B831807.1A01DE25@iprimus.com.au...
> > IIRC, this should work:
> >
> > lb_word[]=data_word[]
> >
> > Do you really need DATA ?
> >
> > Abhimanyu Rastogi wrote:
> > >
> > > hello,
> > >
> > > How does a for loop executes the statements in AHDL ....
> > >
> > > for instance...:
> > >
> > >  FOR i IN 0 to 21 GENERATE
> > >   DATA.d = data_word[i];
> > >   lb_word[i] = data_word[i]; --assuming DATA is connected to LB
> > >  END GENERATE;
> > >
> > > where...   we have data_word[21..0] DFFE; lb_word[21..0] DFFE and DATA
> DFFE;
> > >
> > > so now on simulation lets say i provide data_word[] = (H"38CCAA")  then
> wat
> > > should be the result in DATA n lb_word[]
> > >
> > > Also, does it have sequentially or concurrently..??
> > >
> > > Thx
> > >
> > > Abhimanyu rastogi
> >
> > --
> >    ___                                           ___
> >   /  /\                                         /  /\
> >  /  /__\ Russell Shaw, B.Eng, M.Eng(Research)  /  /\/\
> > /__/   / Victoria, Australia, Down-Under      /__/\/\/
> > \  \  /  http://home.iprimus.com.au/rjshaw    \  \/\/
> >  \__\/                                         \__\/


Article: 34366
Subject: Re: Slowing PCI for FPGA
From: Falk Brunner <Falk.Brunner@gmx.de>
Date: Wed, 22 Aug 2001 18:59:49 +0200
Links: << >>  << T >>  << A >>
Rick Filipkiewicz schrieb:
> 
> them [6/5 states resp.]. Only these core machines + the outgoing data pipe
> control should use the raw, unregistered, PCI signals as inputs.

Arnt all inputs defined as registered in the definition of PCI??

-- 
MFG
Falk


Article: 34367
Subject: Re: FPGA MP3 decoder
From: "Andy Peters <andy [@] exponentmedia" <".> com">
Date: Wed, 22 Aug 2001 17:21:07 GMT
Links: << >>  << T >>  << A >>
Mark wrote:
> 
> Hi there!
> 
> I'm pretty new at FPGAs, and what I'm trying to do is
> program an MP3 decoder into an FPGA..
> Does anyone have experience at this?
> Are there VHDL or Handel-C mp3 decoder sources around?
> Or anyone wants to discuss this subject with me? :-))

Seems to me that you're better off implementing the decoder in a DSP.

-a

Article: 34368
Subject: Re: How does For Loop works in AHDL
From: "Abhimanyu Rastogi" <abhi_rastogi@hotmail.com>
Date: Wed, 22 Aug 2001 18:33:51 GMT
Links: << >>  << T >>  << A >>
> case).  If you need serial output you need to think in terms of registers
and
> how those registers - all clocked at once - can give you your output.
>

But how do i do this......    could u give an example of serially outputting
data.....  ??



John_H <johnhandwork@mail.com> wrote in message
news:3B83DFA4.1D9A3AD3@mail.com...
> The FOR GENERATE does't do anything for you serially.  All you're doing is
> implementing 22 different statements by using one FOR GENERATE block
rather
> than typing 22 different equations by hand (or doing one simple assign in
this
> case).  If you need serial output you need to think in terms of registers
and
> how those registers - all clocked at once - can give you your output.
>
>
> Abhimanyu Rastogi wrote:
>
> > yes...    I need DATA
> > cuz, i have to serailly output data_word[21..0]( just a DFFE) out thru
DATA
> > (output pin & DFFE)
> > and i really don't wanna put lb_word[] = data_word[] ....  i only did
that
> > cuz DATA = data_word[i] was not working properly...otherwise i would
have
> > made lb_word[i] = LB where LB = DATA ( they are interconected on the
synth
> > chip)
> >
> > So, how serial output work .....   is there some rule to follow for it??
> >
> > Abhimanyu
> >
> > Russell Shaw <rjshaw@iprimus.com.au> wrote in message
> > news:3B831807.1A01DE25@iprimus.com.au...
> > > IIRC, this should work:
> > >
> > > lb_word[]=data_word[]
> > >
> > > Do you really need DATA ?
> > >
> > > Abhimanyu Rastogi wrote:
> > > >
> > > > hello,
> > > >
> > > > How does a for loop executes the statements in AHDL ....
> > > >
> > > > for instance...:
> > > >
> > > >  FOR i IN 0 to 21 GENERATE
> > > >   DATA.d = data_word[i];
> > > >   lb_word[i] = data_word[i]; --assuming DATA is connected to LB
> > > >  END GENERATE;
> > > >
> > > > where...   we have data_word[21..0] DFFE; lb_word[21..0] DFFE and
DATA
> > DFFE;
> > > >
> > > > so now on simulation lets say i provide data_word[] = (H"38CCAA")
then
> > wat
> > > > should be the result in DATA n lb_word[]
> > > >
> > > > Also, does it have sequentially or concurrently..??
> > > >
> > > > Thx
> > > >
> > > > Abhimanyu rastogi
> > >
> > > --
> > >    ___                                           ___
> > >   /  /\                                         /  /\
> > >  /  /__\ Russell Shaw, B.Eng, M.Eng(Research)  /  /\/\
> > > /__/   / Victoria, Australia, Down-Under      /__/\/\/
> > > \  \  /  http://home.iprimus.com.au/rjshaw    \  \/\/
> > >  \__\/                                         \__\/
>



Article: 34369
Subject: Re: hardware damage to a Virtex or Spartan-II?
From: gah@ugcs.caltech.edu (glen herrmannsfeldt)
Date: 22 Aug 2001 18:42:59 GMT
Links: << >>  << T >>  << A >>

"Philip Freidin" <philip@fliptronics.com> wrote in message
news:chu5otc4n394qaklkj21e0kisubphc5ng4@4ax.com...
> On 21 Aug 2001 20:43:17 GMT, gah@ugcs.caltech.edu wrote:
> >What is the maximum clock speed for Virtex-II, FF, through one LUT,
> >to another FF?
>
> In a V-II , -5 speed grade,  M3.3.08i speed files, with careful placement,
> about 666MHz .
>

So, again with careful design, one could have a design with about half
the gates and FF changing state at 666MHz, or about 333MHz each.
(That is, assuming random data bits.)  About a factor of two from what
was claimed for the oscillation modes.

-- glen

Article: 34370
Subject: Re: Logic Emulation
From: Muzaffer Kal <muzaffer@dspia.com>
Date: Wed, 22 Aug 2001 11:49:24 -0700
Links: << >>  << T >>  << A >>
"Tim" <tim@rockylogic.com.nospam.com> wrote:

>
>"Michael Boehnel" <boehnel@iti.tu-graz.ac.at> wrote in message
>news:3B835877.71635D2A@iti.tu-graz.ac.at...
>> Are there any significant differences between a logic emulated version
>> (implemented in an FPGA) and an ASIC version.
>>
>> I could imagine that at least some timing differences should occure. Are
>> these timing differences important?
>
>   Raw speed is usually higher in ASIC, at a given process level.
>
>> Are there any other differences?
>
>   ASICs are a generation or more ahead in density, if you can afford it
>   No tri-states in ASICs (going away in FPGAs)

I am not sure what you mean by "No tri-states in ASICs". I guess it is
possible that one is discouraged to use tri-states in a standard cell
or gate array targeted design but I have yet to see a cell library
which doesn't have any internal tri-state buffers or inverters.
Especially with a standard cell methodology, you can do what ever you
want, including designing your own cells.


Article: 34371
Subject: Re: Logic Emulation
From: "Tim" <tim@rockylogic.com.nospam.com>
Date: Wed, 22 Aug 2001 20:51:15 +0100
Links: << >>  << T >>  << A >>
"Muzaffer Kal" <muzaffer@dspia.com> wrote in message
news:hgv7ot4plbq621l8au8n2am8lpe4grmtm7@4ax.com...
> "Tim" <tim@rockylogic.com.nospam.com> wrote:
>

<snip>

> >   No tri-states in ASICs (going away in FPGAs)
>
> I am not sure what you mean by "No tri-states in ASICs". I guess it is
> possible that one is discouraged to use tri-states in a standard cell
> or gate array targeted design but I have yet to see a cell library
> which doesn't have any internal tri-state buffers or inverters.
> Especially with a standard cell methodology, you can do what ever you
> want, including designing your own cells.

Well, you have caught me out on this one.  But I have worked with
several teams who have claimed that their ASIC libraries supported
muxes, obviously, but not internal t/s buses.

Perhaps an ASIC specialist can report on the incidence of t/s buffers
in mainline libraries from the big players.  How about SA21 for starters?





Article: 34372
Subject: Re: Slowing PCI for FPGA
From: Richard Iachetta <iachetta@us.ibm.com>
Date: Wed, 22 Aug 2001 15:24:49 -0500
Links: << >>  << T >>  << A >>
In article <fe76d03b.0108210659.668efa4f@posting.google.com>, 
joey@sun.science.wayne.edu says...
> Rick Filipkiewicz <rick@algor.co.uk> wrote in message news:<3B820B62.54174F87@algor.co.uk>...
> > I'm with Austin on this. It *is* possible to get a Virtex-E to meet 33MHz PCI
> > timing. I was in a similar position ~2.5 years ago when (1) I had to proto a PCI
> > i/f destined, ultimately, for an ASIC and (2) a bought in IP solution was not
> > acceptable to the client.
> 
> Yes, I totally agree it could be done, but what I tried to mention in
> my previous post is that the PCI core itself is not the problem at
> all. The problem is the "rest of the chip". 

Its very common for people to emulate ASIC designs in FPGAs and its very 
common that they can only be run at a fraction of the clock speed that 
the ASIC runs at.  To all the people advocating that he use an off the 
shelf PCI interface in the FPGA or giving advice on how to better code 
the state machines for the FPGA, while being good advice for a "from 
scratch" PCI design in an FPGA, I think you're missing the point.  He has 
to keep the FPGA RTL the same as the ASIC RTL or else he is not verifying 
the ASIC design, which is the goal.

-- 
Rich Iachetta
iachetta@us.ibm.com
I do not speak for IBM.

Article: 34373
Subject: Re: Slowing PCI for FPGA
From: joey@sun.science.wayne.edu (Joey Oravec)
Date: 22 Aug 2001 13:31:57 -0700
Links: << >>  << T >>  << A >>
hamish@cloud.net.au wrote in message news:<ZvNg7.21651$A5.65678@news1.eburwd1.vic.optushome.com.au>...
> Do you see much advantage with Amplify? Do you see it doing anything
> which you couldn't do yourself with Synplify Pro and the Xilinx floorplanner?
>
> I have access to an Amplify license at work and in my brief experience
> with it, it just seems to be a way to enter your floorplan and have
> it appear in an NCF. So I just enter it into the UCF myself, with
> floorplanner or by hand.
>
> I'd like to be proven wrong though. What does the tool itself actually
> do with the floorplanning information? What optimisations can
> it make at synthesis time given the floorplan?

Well, amplify feels superior to synplify pro in synthesis. It's kind
of comparing apples to oranges between versions of the software
because I used synpro 7.0b2 and amplify 2.2.4. For me, Amplify's
timing report was far more conservative which was great. Some designs
failed by just a 4-5 ns during PAR (level2) with SynPro that succeeded
under the same conditions with Amplify. Could have just been
dumb-luck, but synthesis time I had those more accurate timing
predictions with Amplify.

As for floorplanning effect on synthesis, I have no idea offhand. You
make a good point that everything could probably be done by hand or
with the floorplanner tool. This tool's strength lies more in its ease
of use. In about an hour using the Amplify floorplanner with little
knowledge of the RTL, I lowered cycle time from 47ns to 37ns post-PAR
meeting the 25mhz constraint with under an hours work. The paths were
25% logic 75% route.

There's not a lot of choice for improving a design besides modifying
the RTL code and improving placement/routing. Since modifying the RTL
is a non-option in an ASIC prototype, Amplify was really the only
route for improvement.

I was very skeptical of the tool, especially because the thought of
rearranging a million gates on a large Virtex is absurd. However it
really was pretty easy working with the RTL structures instead of
gates. My design had a lot of block repetition and wide structures, so
that really helps. Your mileage may vary.

-joey

Article: 34374
Subject: Re: How does For Loop works in AHDL
From: Steve <steve@s-dewey.demon.co.uk>
Date: Wed, 22 Aug 2001 21:38:26 +0100
Links: << >>  << T >>  << A >>
In article <jWSg7.5538$o6.109727@news>, Abhimanyu Rastogi
<abhi_rastogi@hotmail.com> writes
>> case).  If you need serial output you need to think in terms of registers
>and
>> how those registers - all clocked at once - can give you your output.
>>
>
>But how do i do this......    could u give an example of serially outputting
>data.....  ??
>
>

within MAXPLUS (possibly quartus too) look for help on LPM_SHIFTREG .
You will find the information you require there.
-- 
Steve



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