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 39625

Article: 39625
Subject: Re: Lean serial communication processor
From: "Mike Johnson" <mikej<NO SPAM>@freeuk.com>
Date: Thu, 14 Feb 2002 22:28:44 -0000
Links: << >>  << T >>  << A >>
maybe use Risc5x at opencores.org as a starting point ?
Mike
"Jim Granville" <jim.granville@designtools.co.nz> wrote in message
news:3C6C37E6.6F8D@designtools.co.nz...
> jetmarc wrote:
> >
> > Hi.
> >
> > For a project I like to invent and implement a lean (co)processor that
handles
> > serial protocols (such as SPI, I2C, UART, etc).  It should be software
> > programmable like a processor, so that new protocols can be added
without
> > changes to the VHDL source / bitstream.  On the other hand, it should be
> > almost as lean and light weight as a simple state machine.  I plan to
fit
> > several such processors (in parallel) into a small 50K gate FPGA.
> >
> > I'm not sure yet about the optimum instruction set, nor about the other
> > characteristics of such a processor.  I think about a structure where
the
> > PC addresses 64 instructions, and all instructions operate on bits (not
bytes).
> > The bits include 8 I/O pins, a byte-buffer register (possibly only MSB
> > and LSB addressable), handshake lines from/to the host, and two temp
bits.
> >
> > My idea is to have two parallel processors per serial channel, one for
TX
> > and one for RX.  The instruction set of them can be biased towards the
data
> > direction (TX processor can read from byte-buffer and write to I/O pin,
while
> > RX processor reads from I/O pin and writes to byte-buffer).
> >
> > What do you think about this?  Is there any code or other work available
that
> > might serve as inspiration for nailing down a good and practical
instruction
> > set?
> >
> > Marc
>
>  Interesting idea.
> Scenix(ubicom) use a PIC 12 bit opcode variant core, and a Soft
> peripheral.
>
> You could also look at Bit-Slice processors - I have seen a number on
> the WEB.
>
> You could see if a subset-opcode of the PIC12 could be used, as even
> that small
> opcode has more 'reach' than you need - and then you can use the PIC or
> SX core
> tools and even silicon, as test vechicle.
>
> A simple 12 Bit Opcode -> Smaller Converter/error catcher pgm, would
> take any
> PIC file, and ready it for your FPGA engine.
>
> Code the tasks first, then see what subset of opcode and RAM coverage
> was needed, and use that subset for your 'state engine'.
>
> Some BYTE operations could be usefull for loop counters, and FIFO
> construction,
> and MACRO blocks can always be used if the PIC opcodes do not map well
> to this.
> ( here two/three/four... opcodes would map onto one of your new opcodes
> )
> eg PIC uses a skip/jump construct, but more efficent is a short jump
> opcode
> as seen in COP8, and Fairchild ACE1101
>
>
> IIRC, Philips also had some sort of communications engine chip, that
> made it to data
> a while ago, using a similar idea of a fast, state engine like core for
> data
> packet pumping ( tho this used underlying silicon as well )
>
> -jg



Article: 39626
Subject: Orca ngdbuild error: could not expand block
From: ikumar@cs.utah.edu (Inder)
Date: 14 Feb 2002 14:29:06 -0800
Links: << >>  << T >>  << A >>
I am trying to simulate a controller circuit on ORCA (OR3T00
architecture) FPGA.
For schematic capture and EDIF file generation I am using Powerview
6.1 tool.
But in the Build operation, which translates and converts the input
EDIF netlist
into ORCA generic database file(*.ngd), I have been getting this
error:

ERROR - ngdbuild: Could not expand block '....' with TYPE='...'

and after lot of looking around havent been able to solve the problem.
Can anyone tell me where and what am I doing wrong..

There was no error while generating EDIF file with Powerview tool and
I used the
Orca3 library in Powerview.
Here is a relevent part of log file generated by Orca.


ORCA Foundry Control Center Log File -- Thursday, February 14, 2002
14:25

----------------------------------------------------------------------
/usr/local/apps/ORCA/bin/sol/edif2ngd -l or3c00 "abcd.edn"
"/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo"
----------------------------------------------------------------------
edif2ngd:  version ORCA Foundry 2000 Production-w/SP-B (14)
Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
     Copyright (c) 1995-2000 Lucent Technologies.  All rights
reserved.
WARNING - edif2ngd: Ampersands will be removed from numeric/underscore
     identifiers.
WARNING - edif2ngd: GLOBAL property on a net other than power or
ground -
     ignoring...
       On or above line 113 in file abcd.edn
WARNING - edif2ngd: GLOBAL property on a net other than power or
ground -
     ignoring...
       On or above line 664 in file abcd.edn
WARNING - edif2ngd: GLOBAL property on a net other than power or
ground -
     ignoring...
       On or above line 3814 in file abcd.edn
Writing the design to /home/ikumar/3700-pv/orca/abcd_1/abcd.ngo...

----------------------------------------------------------------------
/usr/local/apps/ORCA/bin/sol/ngdbuild -a or3t00 -p
"/home/ikumar/3700-pv/orca/sch" -p "/home/ikumar/3700-pv/orca"
"/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo" "abcd_1/abcd_1.ngd"
----------------------------------------------------------------------
ngdbuild:  version ORCA Foundry 2000 Production-w/SP-B (14)
Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
     Copyright (c) 1995-2000 Lucent Technologies.  All rights
reserved.
Reading '/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo' ...
Loading NGL library '/usr/local/apps/ORCA/or3c00/data/orc3clib.ngl'...
Loading NGL library '/usr/local/apps/ORCA/data/neoprims.ngl'...
Loading NGL library '/usr/local/apps/ORCA/data/neomacro.ngl'...
Loading device for application ngdbuild from file 'or3t12x12.nph' in
     environment /usr/local/apps/ORCA.
Loading macro from file "/home/ikumar/3700-pv/orca/sch/all4_3lut.nmc".
Loading macro from file
"/home/ikumar/3700-pv/orca/sch/bottom2_3lut.nmc".
Loading macro from file "/home/ikumar/3700-pv/orca/sch/top3_3lut.nmc".
Loading macro from file
"/home/ikumar/3700-pv/orca/sch/bottom3_3lut.nmc".
ERROR - ngdbuild: Could not expand block '$1I53106/$1I9' with
TYPE='BUF'
ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST998' with
     TYPE='UDFDL'
ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST995' with
     TYPE='PW'
ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST6' with
     TYPE='UDFDL'
ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST34' with
     TYPE='NOT'
ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST27/$1I9' with
     TYPE='BUF'
ERROR - ngdbuild: Could not expand block '$1I99402/INST1' with
TYPE='NOT'
ERROR - ngdbuild: Could not expand block '$1I99428/INST1' with
TYPE='AND3'
ERROR - ngdbuild: Could not expand block '$1I99428/BFI3' with
TYPE='BUF'
ERROR - ngdbuild: Could not expand block '$1I99428/BFI2' with
TYPE='BUF'
 is unexpanded

********* around 10000 such errors ************

WARNING - ngdbuild: logical net 'DPRAM08-11/DO2' has no driver
WARNING - ngdbuild: logical net 'DPRAM08-11/DO3' has no driver
WARNING - ngdbuild: logical net 'DPRAM12-15/A0' has no driver
WARNING - ngdbuild: logical net 'DPRAM12-15/A1' has no driver
WARNING - ngdbuild: logical net 'DPRAM12-15/A2' has no driver
.
.
.

WARNING - ngdbuild: logical net 'TP3' has no driver
WARNING - ngdbuild: logical net 'TP3' has no load
WARNING - ngdbuild: logical net 'TP4' has no driver
WARNING - ngdbuild: logical net 'TP4' has no load
WARNING - ngdbuild: logical net 'TP5' has no driver
WARNING - ngdbuild: logical net 'TP5' has no load
WARNING - ngdbuild: logical net 'TP6' has no driver
WARNING - ngdbuild: logical net 'TP6' has no load
WARNING - ngdbuild: logical net 'TP7' has no driver
WARNING - ngdbuild: logical net 'TP7' has no load
ERROR - ngdbuild: DRC failed with 9865 errors and 492 warnings

Design Results:
     19 blocks expanded
Writing 'abcd_1/abcd_1.ngd' ...

Article: 39627
Subject: Re: Modelsim questions
From: yuryws@optonline.com (Yury)
Date: 14 Feb 2002 14:42:07 -0800
Links: << >>  << T >>  << A >>
Modelsim XE is able to do everything that Modelsim PE can (according
to Xilinx). I have used both. The XE version can be used without ISE
software. You can use Modelsim XE, for example, for compiling
Foundation VHDL libraries and then using them (Simprim, Unisim,
Coregen, etc.). While slower then PE/SE, Modelsim XE will be able to
perform pre-PAR simulation for you. However... if you want to do a
post PAR functional (and possibly include the timing information)
simulation then you are in for a treat. I attempted to use Modelsim XE
for simulating post PAR 60% utilized Spartan-II-200. All I can say is
there is a better way to spend my time. The simulation was taking
forever without any timing info.
 If I new ahead of time that ModelSim XE would be such a poor piece of
software I would never have advised my emploer to purchase it. I guess
you get what you pay for stands ($900 can not buy you a desent
simulator).

Article: 39628
Subject: Re: Lean serial communication processor
From: Goran Bilski <goran@xilinx.com>
Date: Thu, 14 Feb 2002 15:19:37 -0800
Links: << >>  << T >>  << A >>
Maybe you should have a look at KCPSM from Xilinx.
http://www.xilinx.com/xapp/xapp213.pdf

G=F6ran Bilski

"Mike Johnson

> maybe use Risc5x at opencores.org as a starting point ?
> Mike
> "Jim Granville" <jim.granville@designtools.co.nz> wrote in message
> news:3C6C37E6.6F8D@designtools.co.nz...
> > jetmarc wrote:
> > >
> > > Hi.
> > >
> > > For a project I like to invent and implement a lean (co)processor t=
hat
> handles
> > > serial protocols (such as SPI, I2C, UART, etc).  It should be softw=
are
> > > programmable like a processor, so that new protocols can be added
> without
> > > changes to the VHDL source / bitstream.  On the other hand, it shou=
ld be
> > > almost as lean and light weight as a simple state machine.  I plan =
to
> fit
> > > several such processors (in parallel) into a small 50K gate FPGA.
> > >
> > > I'm not sure yet about the optimum instruction set, nor about the o=
ther
> > > characteristics of such a processor.  I think about a structure whe=
re
> the
> > > PC addresses 64 instructions, and all instructions operate on bits =
(not
> bytes).
> > > The bits include 8 I/O pins, a byte-buffer register (possibly only =
MSB
> > > and LSB addressable), handshake lines from/to the host, and two tem=
p
> bits.
> > >
> > > My idea is to have two parallel processors per serial channel, one =
for
> TX
> > > and one for RX.  The instruction set of them can be biased towards =
the
> data
> > > direction (TX processor can read from byte-buffer and write to I/O =
pin,
> while
> > > RX processor reads from I/O pin and writes to byte-buffer).
> > >
> > > What do you think about this?  Is there any code or other work avai=
lable
> that
> > > might serve as inspiration for nailing down a good and practical
> instruction
> > > set?
> > >
> > > Marc
> >
> >  Interesting idea.
> > Scenix(ubicom) use a PIC 12 bit opcode variant core, and a Soft
> > peripheral.
> >
> > You could also look at Bit-Slice processors - I have seen a number on=

> > the WEB.
> >
> > You could see if a subset-opcode of the PIC12 could be used, as even
> > that small
> > opcode has more 'reach' than you need - and then you can use the PIC =
or
> > SX core
> > tools and even silicon, as test vechicle.
> >
> > A simple 12 Bit Opcode -> Smaller Converter/error catcher pgm, would
> > take any
> > PIC file, and ready it for your FPGA engine.
> >
> > Code the tasks first, then see what subset of opcode and RAM coverage=

> > was needed, and use that subset for your 'state engine'.
> >
> > Some BYTE operations could be usefull for loop counters, and FIFO
> > construction,
> > and MACRO blocks can always be used if the PIC opcodes do not map wel=
l
> > to this.
> > ( here two/three/four... opcodes would map onto one of your new opcod=
es
> > )
> > eg PIC uses a skip/jump construct, but more efficent is a short jump
> > opcode
> > as seen in COP8, and Fairchild ACE1101
> >
> >
> > IIRC, Philips also had some sort of communications engine chip, that
> > made it to data
> > a while ago, using a similar idea of a fast, state engine like core f=
or
> > data
> > packet pumping ( tho this used underlying silicon as well )
> >
> > -jg


Article: 39629
Subject: Re: Create a bit stream (BIT file) from an NCD file?
From: "Tim" <tim@rockylogic.com.nooospam.com>
Date: Fri, 15 Feb 2002 00:15:32 -0000
Links: << >>  << T >>  << A >>

"Aaron Eberhart" <aaron.eberhart@mx.iff.navy.mil> wrote in message
news:ee74db4.-1@WebX.sUN8CHnE...
> I have been provided with a schematics based project as well as an HDL based
project, each of which were previously successfully implemented. I am tasked
with re-implementing the project to a new working chip. So far, attempts at
recreating the projects in Foundations 3.1 (for the schematics project) and
ISE4.1 (for the HDL project) have failed. I have been able to recreate the HEX
prom files from the provided bit strems and load these into the system active.
However I have been unable to go from the code or schematic level to working HEX
files. My new thought is to try creating a bit stream from the provided NCD
files which would put me on track to figure out what implementation options were
in use. Any ideas on how to go directly from an NCD to a BIT file for
implementation?
>

From a makefile (the comments come from files in the xilinx
directories, such as bitgen.usg):

dn=foobar

# bitgen generates a programming file, options:
#    Usage: bitgen [-d] [-j] [-b] [-w] [-l] [-m] [-t] [-n] [-u] [-a] {-g
#    <setting_value>} <infile[.ncd]> [<outfile>] [<pcffile[.pcf]>]
#    Where:
#      -d           = Don't Run DRC (Design Rules Checker)
#      -j           = Don't create bit file
#      -b           = Create rawbits file
#      -w           = Overwrite existing output file
#      -f <cmdfile> = Read command line arguments from file <cmdfile>
#      -g <opt:val> = Set option to value, options are (1st is default):
#
# commands in bitgen.ut as follows:
#    -g  ConfigRate     4
#    -g  StartupClk     Cclk
#    -g  CclkPin        Pullup
#    -g  DonePin        Pullup
#    -g  M0Pin          Pullup
#    -g  M1Pin          Pullup
#    -g  M2Pin          Pullup
#    -g  ProgPin        Pullup
#    -g  TckPin         Pullup
#    -g  TdiPin         Pullup
#    -g  TdoPin         Pullup
#    -g  TmsPin         Pullup
#    -g  DONE_cycle     3
#    -g  GSR_cycle      Done
#    -g  GWE_cycle      Done
#    -g  GTS_cycle      Done
#    -g  LCK_cycle      NoWait
#    -g  Persist        No
#    -g  DriveDone      Yes
#    -g  DonePipe       Yes
#
# promgen options:
# -b           Disable bit swapping. Only valid for hex format (-p hex).
#
# -s <size>    PROM size in K bytes (must be power of 2), multiple sizes imply
#              splitting the bitstream(s) into multiple PROMs.
#
# -x <xilinx_prom>
#              Specific Xilinx PROM, multiple PROMs imply splitting the
#              bitstream(s) into multiple PROMs.
#
# -p <format>  PROM format (mcs, exo, hex, or tek)
#
# -o <file>    Output PROM file name (default matches first .bit file),
#              multiple names may be specified when splitting PROMs.
#
# -u <hexaddr> <file[.bit]> {<file[.bit]>}
#              Load .bit file(s) up from address.  Multiple .bit files are
#              daisychained to form a single PROM load.
#
# -d <hexaddr> <file[.bit]> {<file[.bit]>}
#              Load .bit file(s) down from address.  Multiple .bit files are
#              daisychained to form a single PROM load.
#
# -n <file[.bit]> {<file[.bit]>}
#              Load .bit file(s) up or down starting from the next address
#              following previous load.  Multiple .bit files are daisychained
#              to form a single PROM load.  Must follow a -u, -d, or -n option.
#
# -r <promfile>
#              Load the PROM file.  The -r and the -u, -d and -n options are
#              mutually exclusive.
#
# -f <cmdfile>
#              Read command line arguments from file <cmdfile>.
#
#At least one -r, -u, or -d option must appear in the command line.

$(dn).bit : $(dn).ncd bitgen.ut
    bitgen $(dn).ncd  -w -f bitgen.ut
    promgen -p hex -s 512 -o $(dn).hex -u 0 $(dn).bit





Article: 39630
Subject: Re: SpartanXL & VHDL -- free software?
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Date: 14 Feb 2002 17:55:15 -0800
Links: << >>  << T >>  << A >>
"Paul Taylor" <p.taylor@ukonline.co.uk> writes:
> It seems strange to me for Xilinx to not bother supporting XL in this latest
> software, especially if they can't offer a free alternative for XL with
> VHDL?!

That's because they aren't interested in selling the Spartan and Spartan
XL to new customers.  It's now considered a legacy part.  They're happy
to keep selling the chips to existing customers, but they want you
to buy the Spartan II or Spartan IIE for new designs.

Of course, you certainly *can* use the Spartan and Spartan XL for new
designs; just don't expect as much encouragement (e.g., free tools) out
of Xilinx if you do that.

Article: 39631
Subject: Re: Spartan-II becomes Vertex.
From: Kevin Brace <ihatespamkevinbraceusenet@ihatespamhotmail.com>
Date: Thu, 14 Feb 2002 20:09:55 -0600
Links: << >>  << T >>  << A >>
That is not true.
According to Virtex datasheet dated 1/28/2000, the configuration bit
length for XCV100 is 781,216.
However, according to Spartan-II datasheet dated 3/3/2000, the
configuration bit length for XC2S100 is 781,248.




Kevin Brace (Don't respond to me directly, respond within the
newsgroup.)




Ray Andraka wrote:
> 
> The bitstreams for virtex and spartanII for same size device are the same.
> 
> 
> --
> --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: 39632
Subject: Re: Spartan-II becomes Vertex.
From: "X. Q." <qijun@okigrp.com.sg>
Date: Fri, 15 Feb 2002 10:24:22 +0800
Links: << >>  << T >>  << A >>
The diference may come because you create them at different time,
the Time and Device Type in the bit file makes up the difference in file
size.

--
Best Regards,
Qijun.

********************************************************************
Xu Qijun
Design Engineer, RFIC Group
OKI Techno Centre (S) Pte Ltd
20 Science Park Rd, #02-06/10
Teletech Park,Science Park II,
Singapore  117674
DID: 7707049 Fax: 7792382
"Kevin Brace" <ihatespamkevinbraceusenet@ihatespamhotmail.com> wrote in
message news:a4hq5d$l33$1@newsreader.mailgate.org...
> That is not true.
> According to Virtex datasheet dated 1/28/2000, the configuration bit
> length for XCV100 is 781,216.
> However, according to Spartan-II datasheet dated 3/3/2000, the
> configuration bit length for XC2S100 is 781,248.
>
>
>
>
> Kevin Brace (Don't respond to me directly, respond within the
> newsgroup.)
>
>
>
>
> Ray Andraka wrote:
> >
> > The bitstreams for virtex and spartanII for same size device are the
same.
> >
> >
> > --
> > --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: 39633
Subject: Re: Modelsim questions
From: Kevin Brace <ihatespamkevinbraceusenet@ihatespamhotmail.com>
Date: Thu, 14 Feb 2002 20:51:18 -0600
Links: << >>  << T >>  << A >>


Yury wrote:
> 
> Modelsim XE is able to do everything that Modelsim PE can (according
> to Xilinx). I have used both. The XE version can be used without ISE
> software. You can use Modelsim XE, for example, for compiling
> Foundation VHDL libraries and then using them (Simprim, Unisim,
> Coregen, etc.). While slower then PE/SE, Modelsim XE will be able to
> perform pre-PAR simulation for you. However... if you want to do a
> post PAR functional (and possibly include the timing information)
> simulation then you are in for a treat. I attempted to use Modelsim XE
> for simulating post PAR 60% utilized Spartan-II-200. All I can say is
> there is a better way to spend my time. The simulation was taking
> forever without any timing info.


        One thing I noticed when I did a P&R simulation with ModelSim
XE-Starter (Not the paid version ModelSim XE you used.) is that it took
something like 20 to 30 minutes to start responding (Getting data on a
Wave window.).
Perhaps if you leave the computer running for an hour or so, you should
start seeing something on a Wave window.
I simulated a design that occupied about 60% of XC2S150, at 33MHz for
10us.
That took about 4 hours to finish, but I must say that's far better than
paying several thousands of dollars for a full version of ModelSim
(Wasn't it $4,500 for ModelSim PE?) because I cannot afford that.
The design I simulated here was a PCI IP core I developed with Xilinx's
secret PCILOGIC (That mysterious IRDY and TRDY box thing people were
discussing sometime ago.) attached to my PCI IP core.
The PCI IP core simulated absolutely fine with that secret PCILOGIC,
which I was a little surprised since that feature is Xilinx's guarded
secret (unfair advantage).



>  If I new ahead of time that ModelSim XE would be such a poor piece of
> software I would never have advised my emploer to purchase it. I guess
> you get what you pay for stands ($900 can not buy you a desent
> simulator).


        I am poor, so at least I am happy that I can use some kind of
ModelSim without paying anything even though it is slow (I believe Post
P&R simulation is always slow.).
This may not be acceptable to you, but perhaps having a second computer
for the sole purpose of doing Post P&R simulation in ModelSim XE might
not be a bad idea.
After all, a fairly decent computer (Much faster one than the one I use
right now.) nowadays costs $900 or less.



Kevin Brace (Don't respond to me directly, respond within the
newsgroup.)

Article: 39634
Subject: Re: Spartan-II becomes Vertex.
From: Kevin Brace <ihatespamkevinbraceusenet@ihatespamhotmail.com>
Date: Thu, 14 Feb 2002 20:58:43 -0600
Links: << >>  << T >>  << A >>
Nope.
Because some people may disagree with me, I checked both datasheets.
Also see Xilinx Application Note 138 and 176.
That should confirm what I wrote.
        I am pretty sure that I read somewhere that Xilinx doesn't
guarantee both bitstreams to be 100% compatible, but I forgot where I
read that.
Peter Alfke (I haven't seen him for a week or so.) may know more about
that issue.



Kevin Brace (Don't respond to me directly, respond within the
newsgroup.)



"X. Q." wrote:
> 
> The diference may come because you create them at different time,
> the Time and Device Type in the bit file makes up the difference in file
> size.
> 
> --
> Best Regards,
> Qijun.
> 
> ********************************************************************
> Xu Qijun
> Design Engineer, RFIC Group
> OKI Techno Centre (S) Pte Ltd
> 20 Science Park Rd, #02-06/10
> Teletech Park,Science Park II,
> Singapore  117674
> DID: 7707049 Fax: 7792382

Article: 39635
Subject: Re: Spartan-II becomes Vertex.
From: "X. Q." <qijun@okigrp.com.sg>
Date: Fri, 15 Feb 2002 11:34:13 +0800
Links: << >>  << T >>  << A >>
I suspect it was because in the download board they used a Vertex family.
while in the daughter board I put in a Spartan-II chip and daughter board.
I tried to synthesize with spartan-ii chip and program the device, it
displayed
200,123 differences. Could anybody know how to tackle these differneces?

--
Best Regards,
Qijun.

********************************************************************
Xu Qijun
Design Engineer, RFIC Group
OKI Techno Centre (S) Pte Ltd
20 Science Park Rd, #02-06/10
Teletech Park,Science Park II,
Singapore  117674
DID: 7707049 Fax: 7792382
"Kevin Brace" <ihatespamkevinbraceusenet@ihatespamhotmail.com> wrote in
message news:a4ht0v$la0$1@newsreader.mailgate.org...
> Nope.
> Because some people may disagree with me, I checked both datasheets.
> Also see Xilinx Application Note 138 and 176.
> That should confirm what I wrote.
>         I am pretty sure that I read somewhere that Xilinx doesn't
> guarantee both bitstreams to be 100% compatible, but I forgot where I
> read that.
> Peter Alfke (I haven't seen him for a week or so.) may know more about
> that issue.
>
>
>
> Kevin Brace (Don't respond to me directly, respond within the
> newsgroup.)
>
>
>
> "X. Q." wrote:
> >
> > The diference may come because you create them at different time,
> > the Time and Device Type in the bit file makes up the difference in file
> > size.
> >
> > --
> > Best Regards,
> > Qijun.
> >
> > ********************************************************************
> > Xu Qijun
> > Design Engineer, RFIC Group
> > OKI Techno Centre (S) Pte Ltd
> > 20 Science Park Rd, #02-06/10
> > Teletech Park,Science Park II,
> > Singapore  117674
> > DID: 7707049 Fax: 7792382



Article: 39636
Subject: Re: Is Leonardo spectrum OEM version for Altera limited?
From: "Leon de Boer" <ldeboer@iprimus.com.au>
Date: Fri, 15 Feb 2002 03:34:35 GMT
Links: << >>  << T >>  << A >>
You are correct it is a bug.
I worked it out last night that if you have signals that only appear in the
ELSE part of a less than or grater than test (IF > THEN ELSE ENDIF) and are
all zero on one everywhere else Leonardo decides to optimize them away for
some reason. If I invert the test so the test becomes (IF = THEN ELSE END
IF) the signals are in the first equals part then and not in the else part
and Leonardo doesn't optimize them away.

"Jay" <kayrock66@yahoo.com> wrote in message
news:d049f91b.0202131822.6d1deb40@posting.google.com...
> Sounds more like a bug than a gate limit.  Report it as a bug along
> with associated files, that usually gets their attention.
>
>
> "Leon de Boer" <ldeboer@iprimus.com.au> wrote in message
news:<3c6a780f$1_1@news.iprimus.com.au>...
> > I have found while using Leonardo spectrum and Baseline for an Altera
design
> > that when I get to 10-15K gates the design leaves bits out of the
synthesis.
> > I have had this sort of behaviour with Actel licenses where they have a
gate
> > limit. I can't seem to get anyone at Altera or exemplar to be able to
tell
> > me absolutely if the free baseline, Leonardo version is gate limited.
Does
> > anyone know?
> >
> > Regards Leon



Article: 39637
Subject: Re: SpartanXL & VHDL -- free software?
From: "Paul Taylor" <p.taylor@ukonline.co.uk>
Date: Fri, 15 Feb 2002 03:44:06 -0000
Links: << >>  << T >>  << A >>
> That's because they aren't interested in selling the Spartan and Spartan
> XL to new customers.  It's now considered a legacy part.  They're happy
> to keep selling the chips to existing customers, but they want you
> to buy the Spartan II or Spartan IIE for new designs.
>
> Of course, you certainly *can* use the Spartan and Spartan XL for new
> designs; just don't expect as much encouragement (e.g., free tools) out
> of Xilinx if you do that.

OK, but why not add a message "Not recommended for new designs"
when you run the fitter, instead of just not providing support?
Akin to taking the datasheet off the web page? :-)
I understand, and like, the differences between -XL and -II, but I put
a lot of effort into making this experimentation PCB!



Article: 39638
Subject: Re: Orca ngdbuild error: could not expand block
From: rickman <spamgoeshere4@yahoo.com>
Date: Thu, 14 Feb 2002 22:55:22 -0500
Links: << >>  << T >>  << A >>
It has been quite a while since I have worked with the Lucent tools, but
I seem to remember seeing this problem before. I think it was due to
using a component that was not in the library for the chip. Or I may
have had the libraries wrong. I was working with two different chips (2T
and 3T) and had some trouble setting them up the first time. 

Verify that the buffers that are giving you the problem are in the
library. Also verify that you are pointing to the correct library(s). I
do not remember any components called "BFI" anything. 



Inder wrote:
> 
> I am trying to simulate a controller circuit on ORCA (OR3T00
> architecture) FPGA.
> For schematic capture and EDIF file generation I am using Powerview
> 6.1 tool.
> But in the Build operation, which translates and converts the input
> EDIF netlist
> into ORCA generic database file(*.ngd), I have been getting this
> error:
> 
> ERROR - ngdbuild: Could not expand block '....' with TYPE='...'
> 
> and after lot of looking around havent been able to solve the problem.
> Can anyone tell me where and what am I doing wrong..
> 
> There was no error while generating EDIF file with Powerview tool and
> I used the
> Orca3 library in Powerview.
> Here is a relevent part of log file generated by Orca.
> 
> ORCA Foundry Control Center Log File -- Thursday, February 14, 2002
> 14:25
> 
> ----------------------------------------------------------------------
> /usr/local/apps/ORCA/bin/sol/edif2ngd -l or3c00 "abcd.edn"
> "/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo"
> ----------------------------------------------------------------------
> edif2ngd:  version ORCA Foundry 2000 Production-w/SP-B (14)
> Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
>      Copyright (c) 1995-2000 Lucent Technologies.  All rights
> reserved.
> WARNING - edif2ngd: Ampersands will be removed from numeric/underscore
>      identifiers.
> WARNING - edif2ngd: GLOBAL property on a net other than power or
> ground -
>      ignoring...
>        On or above line 113 in file abcd.edn
> WARNING - edif2ngd: GLOBAL property on a net other than power or
> ground -
>      ignoring...
>        On or above line 664 in file abcd.edn
> WARNING - edif2ngd: GLOBAL property on a net other than power or
> ground -
>      ignoring...
>        On or above line 3814 in file abcd.edn
> Writing the design to /home/ikumar/3700-pv/orca/abcd_1/abcd.ngo...
> 
> ----------------------------------------------------------------------
> /usr/local/apps/ORCA/bin/sol/ngdbuild -a or3t00 -p
> "/home/ikumar/3700-pv/orca/sch" -p "/home/ikumar/3700-pv/orca"
> "/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo" "abcd_1/abcd_1.ngd"
> ----------------------------------------------------------------------
> ngdbuild:  version ORCA Foundry 2000 Production-w/SP-B (14)
> Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
>      Copyright (c) 1995-2000 Lucent Technologies.  All rights
> reserved.
> Reading '/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo' ...
> Loading NGL library '/usr/local/apps/ORCA/or3c00/data/orc3clib.ngl'...
> Loading NGL library '/usr/local/apps/ORCA/data/neoprims.ngl'...
> Loading NGL library '/usr/local/apps/ORCA/data/neomacro.ngl'...
> Loading device for application ngdbuild from file 'or3t12x12.nph' in
>      environment /usr/local/apps/ORCA.
> Loading macro from file "/home/ikumar/3700-pv/orca/sch/all4_3lut.nmc".
> Loading macro from file
> "/home/ikumar/3700-pv/orca/sch/bottom2_3lut.nmc".
> Loading macro from file "/home/ikumar/3700-pv/orca/sch/top3_3lut.nmc".
> Loading macro from file
> "/home/ikumar/3700-pv/orca/sch/bottom3_3lut.nmc".
> ERROR - ngdbuild: Could not expand block '$1I53106/$1I9' with
> TYPE='BUF'
> ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST998' with
>      TYPE='UDFDL'
> ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST995' with
>      TYPE='PW'
> ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST6' with
>      TYPE='UDFDL'
> ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST34' with
>      TYPE='NOT'
> ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST27/$1I9' with
>      TYPE='BUF'
> ERROR - ngdbuild: Could not expand block '$1I99402/INST1' with
> TYPE='NOT'
> ERROR - ngdbuild: Could not expand block '$1I99428/INST1' with
> TYPE='AND3'
> ERROR - ngdbuild: Could not expand block '$1I99428/BFI3' with
> TYPE='BUF'
> ERROR - ngdbuild: Could not expand block '$1I99428/BFI2' with
> TYPE='BUF'
>  is unexpanded
> 
> ********* around 10000 such errors ************
> 
> WARNING - ngdbuild: logical net 'DPRAM08-11/DO2' has no driver
> WARNING - ngdbuild: logical net 'DPRAM08-11/DO3' has no driver
> WARNING - ngdbuild: logical net 'DPRAM12-15/A0' has no driver
> WARNING - ngdbuild: logical net 'DPRAM12-15/A1' has no driver
> WARNING - ngdbuild: logical net 'DPRAM12-15/A2' has no driver
> .
> .
> .
> 
> WARNING - ngdbuild: logical net 'TP3' has no driver
> WARNING - ngdbuild: logical net 'TP3' has no load
> WARNING - ngdbuild: logical net 'TP4' has no driver
> WARNING - ngdbuild: logical net 'TP4' has no load
> WARNING - ngdbuild: logical net 'TP5' has no driver
> WARNING - ngdbuild: logical net 'TP5' has no load
> WARNING - ngdbuild: logical net 'TP6' has no driver
> WARNING - ngdbuild: logical net 'TP6' has no load
> WARNING - ngdbuild: logical net 'TP7' has no driver
> WARNING - ngdbuild: logical net 'TP7' has no load
> ERROR - ngdbuild: DRC failed with 9865 errors and 492 warnings
> 
> Design Results:
>      19 blocks expanded
> Writing 'abcd_1/abcd_1.ngd' ...

-- 

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: 39639
Subject: Re: Xilinx synthesis tools
From: "Leon de Boer" <ldeboer@iprimus.com.au>
Date: Fri, 15 Feb 2002 03:58:25 GMT
Links: << >>  << T >>  << A >>
I agree totally having used FPGA Express, Leonardo Spectrum and Xilinx ISE 4
for designs around the 10K-150K range I would rank them
1). ISE 4.x Xilinx
2). FPGA Express
3). Leonardo Spectrum.

Strong Points.

ISE 4.x Xilinx
Gives best predictable and stable synthesis
Picks use of ram/rom blocks from design readily
Usually gives smaller and faster design
All warnings during compilation are coloured Yellow (easy to see amongst the
spool to screen)

FPGA Express
Very intuative to setup and use and has good online help
 Gives good warnings but you have to wade through the report.

Leonardo Spectrum
Good editor with nice colourization of source (easy to read)

Weak Points.

ISE 4.x Xilnx
Exhaustive fit usually results in to many product terms and program crashes
Pin assigning and locking is not intuitive (whats wrong with a graphical
view of chip and place signals)
Gate resource level view of synthesis is hopeless.

FPGA Express
Generally will not pick and use ROM/RAM blocks unless you specify it.
Can give erratic synthesis at times.

Leonardo Spectrum
Never picks RAM/ROM blocks unless you specify it
Has the most bugged and erratic synthesis of the 3
Puts up far to much minor warnings on screen that means you wade through
piles of crap.



"Sanket Bandyopadhyay" <sanket@insight.memec.co.in> wrote in message
news:ee74a6c.1@WebX.sUN8CHnE...
> Well, XST is more stable and has the better QoR than FPGA Express.
> You can make a small design of a PWM and see for yourself.
> You will see that,after synthesising with FPGA express,you may get
slightly different P&R results every time, but with XST you get repeatedly
same P&R results after synthesis.XST is better.Moreover you can actually
feel the difference in a more than million gate design.That is actually the
trial by fire.



Article: 39640
Subject: Configuration in SelectMAP mode and CCLK
From: rickman <spamgoeshere4@yahoo.com>
Date: Thu, 14 Feb 2002 23:19:11 -0500
Links: << >>  << T >>  << A >>
I have been reading about configuration of the SpartanIIE parts in
SelectMAP (Slave Parallel) mode and it is not all that straight forward.
But with XAPP_176, it looks like I have it down. But I am still not
clear about what to do with the CCLK signal. 

I have very few (one) dedicated IO signal to drive the configuration
with. I could come up with another, but I would have to give up a
feature elsewhere. But I have several signals from the DSP that can be
used as general IO during configuration if I can then use them as their
dedicated function after the SpartanIIE is fully configured. It looks
like I can do this for nearly all the required configuration signals.
Other than PROGRAM-, the configuration signals are all inputs or open
driver outputs (open when properly configured) which can be paralleled
with a user IO. This will let me use the DSP signals as they are needed
after configuration. 

The only fly in the ointment is CCLK. As far as I can tell, it is
ignored when CS- is not asserted. I plan to connect CCLK to the WE- bus
signal. This will give me a single CCLK rising edge for each data word
(byte) while configuring. But once configuration is over, is it ok to
continue to toggle CCLK? I seem to remember that in some modes on other
Xilinx parts, CCLK should stop toggling once the part is fully
configured and operational. Will there be a problem in SelectMAP mode if
CCLK continues to toggle after configuration is complete? 

I believe I have all the issues covered for the configuration. CS- is
generated from the DSP decoded Chip Select, Write- is a DSP IO pin,
PROGRAM- comes from a board control output on an MCU and DONE and INIT-
go to DSP IO pins. DONE is paralleled with an FPGA user IO to allow the
DSP IO pin to be used after config and the INIT- signal becomes a user
IO as does the CS-. And of course the 8 data bits are connected to the
low 8 data bits from the DSP and become 8 of the 32 data bus bits once
configured. 

MCU      DSP         FPGA
--------------------------
PRGM- >            > PRGM-
        TINP0   <  < DONE (with a user IO)
        TINP1   <  < INIT- 
        AWE-    >  > CCLK (with a user IO)
        ED[7:0] >  > D[0:7]
        CE3-    >  > CS-
        TOUT0   >  > WRITE-


Now I just need to make sure toggling the CCLK pin won't muck things up
after config is complete. 

-- 

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: 39641
Subject: Re: fifo in coregen? Xilinx (ise4.1) is screwed up!
From: "G" <somewhere@doesnt.matter.invalid>
Date: 15 Feb 2002 05:34:44 GMT
Links: << >>  << T >>  << A >>
Howdy,

Bob Perlman <no-spam@sonic.net> wrote:
> The problems with using clock pulses or gated clocks have been pretty
> much beaten to death in this newsgroup, so I won't belabor the point.

 Do you mean _asynchronous_ clock pulses/gated clocks here?  I was 
just about to rewrite a section of my design to only generate clocks
to the coregen FIFO when necessary, keeping the enable active.

 My current problem is the enable line to the BlockRAMs in the FIFO
is the critical path.  If I can generate pulses only when I need them,
then I can run the system a little faster.

 The idea is to use a FF to divide the main clock by two.  This would
then become the FIFO read clock.  I would limit the pulses by using
the clock enable on that FF.  If I can constrain the path from the
rising/falling edge of the main clock through the FF, through the FIFO,
and to the inputs of a set of FF's, then I don't immediately see
where the problems lie.

Thanks,
G

P.S.  Sorry about the email address.  I have 'misplaced' the slip of
paper with my Yahoo Email password.. :(

Article: 39642
Subject: Xilinx Virtex XCV300
From: "Niv" <niv@ntlworld.com>
Date: Fri, 15 Feb 2002 06:59:41 -0000
Links: << >>  << T >>  << A >>
Hi,
I'm doing a post PAR sim on a Virtex XCV300.  We have an output enable pin
which enables a tristate bus.

When it's enabled, sometimes the bus goes from all 'Z's to the output value,
but transitions thru' a bunch of 'X's for maybe a couple of nanosecs.

There is no bus contention apparently  (using Modelsim & "check contention
add <signame>).

Is this to be expected?  It seems that as the relevent output is enabled, it
turns on in a strange way!

tia,  niv.



Article: 39643
Subject: Re: par and carry chains not allowing manual floorplanning
From: Utku Ozcan <ozcan@netas.com.tr>
Date: Fri, 15 Feb 2002 09:33:39 +0200
Links: << >>  << T >>  << A >>

Ray Andraka wrote:

> I've found 4.1 to be just about useless for anything other than VirtexII designs. [snip]

This "result" proves our opinion which claims that newest version of Xilinx tools
only improves newest Xilinx devices.

So probably a v5.1i, a next major release will be optimized for a new Xilinx device
coming in, for a Xilinx counterpart of Altera's Stratix maybe?

Utku



Article: 39644
Subject: RAM CORE result that I did not understand
From: dottavio@ised.it (Antonio)
Date: 14 Feb 2002 23:39:11 -0800
Links: << >>  << T >>  << A >>
I've symply do the following :
use different configuration of core generator single port blockram
read only of dimension
12x4096 and working on rising edge of the clock, at the address input
I apply a counter that
produce a different address at every clock cycle and a FFD to change
rapidly from rising to
falling edge, my expectation is to have a new output value at every
clock cycle,
but this are the result :

1) only registered output -> a result every 2 clock cycles
2) unregistered input and output -> a result every 2 clock cycles if
the address change on rising
					edge otherwise if change on falling edge I've no output and a lot
of :
		# : WARNING: */RAMB4_S1 SETUP Low VIOLATION ON ADDR(0) WITH RESPECT
TO CLK;
		# :   Expected := 0.01 ns; Observed := 0 ns; At : 1250 ns 

3) only registered input -> no output regardless on the fact that
address change on rising or on falling edge

Article: 39645
Subject: Some doubts on FIFO CORE v.4.0
From: dottavio@ised.it (Antonio)
Date: 14 Feb 2002 23:43:56 -0800
Links: << >>  << T >>  << A >>
My circuit is now really symple, I've the following :

In->FF_fallingEdge-->FIFO_15x1_falling_edge-->Shift_register_SIPO_falling_edge
					      |  |  |  |  |  |  |  |  |						            1 | 2|  |  |  | 
|  |  | 9|
					      |  |  |  |  |  |  |  |  |
                                           
BLOCKRAM_12x4096_rising_edge --> OUT

for the FIFO I use the new Xilinx CORE version 4.0 and also for the
RAM I use the new CORE version 4.0 , I've the following problems :

1) If I choose registered input and output for the blockram I've no
output from it
2) if I choose unregistered input and output I've an output from the
blockram but I've also a big amount of warning like this from ALDEC :

# : WARNING: */RAMB4_S1_S1 SETUP Low VIOLATION ON ADDRB(1) WITH
RESPECT TO CLKB;
# :   Expected := 0.01 ns; Observed := 0 ns; At : 164625 ns
# : Time: 164625 ns,  Iteration: 3,  Instance: /U1/U10/B7.
# : WARNING: */RAMB4_S1_S1 SETUP Low VIOLATION ON ADDRB(0) WITH
RESPECT TO CLKB;
# :   Expected := 0.01 ns; Observed := 0 ns; At : 164625 ns
# : Time: 164625 ns,  Iteration: 3,  Instance: /U1/U10/B7.
# : WARNING: */RAMB4_S1_S1 SETUP Low VIOLATION ON ENB WITH RESPECT TO
CLKB;
# :   Expected := 0.01 ns; Observed := 0 ns; At : 164625 ns

I've read from the Xilinx Problem Solver that this could happen cause
the asynchronous FIFO works with 2 different clock and this could be a
problem for the synchronization logic that It's inside but the output
of the circuit seems to be sensitive to these warnings.

Article: 39646
Subject: Re: fifo in coregen? Xilinx (ise4.1) is screwed up!
From: Jonas Weiss <jweiss@kontronmedical.ch>
Date: Fri, 15 Feb 2002 08:59:27 +0100
Links: << >>  << T >>  << A >>
Hello,
I was just wondering if you got your asynchronous FIFO to work by now. For some
days I'm also trying to get my design using one of these to work, without
success yet (ISE 4.1 & Model Sim XE Starter). I am using continuous read and
write clocks with synchronous enables.
Behavioral simulation works fine, but as soon as I proceed to a post translate
simulation I get 'open' select signals of internal fifo-muxes in the translate
file.
P&R works, just the post P&R simulation is not exactly what I expected it to be,
as if the behavioral model and the coregen netlist were not consistent.
Any new ideas yet?

Thanks

Jonas



Article: 39647
Subject: Re: Orca ngdbuild error: could not expand block
From: ikumar@cs.utah.edu (Inder)
Date: 15 Feb 2002 00:03:11 -0800
Links: << >>  << T >>  << A >>
I have set the viewlogic's powerview to use the orca3 architecture
macro library and there is no error or warnings when i convert the
schematic design to EDIF file... FYI, I am working on schematic files
created by somebody else, also using powerview's viewdraw tool ..

I also think that there is something wrong with the library settings
but am not able to locate it.. i dont think its because of some
component not in the library otherwise powerview should have given the
error ?

i'll appreciate any furthur help in this.. 


rickman <spamgoeshere4@yahoo.com> wrote in message news:<3C6C86AA.9D8A8903@yahoo.com>...
> It has been quite a while since I have worked with the Lucent tools, but
> I seem to remember seeing this problem before. I think it was due to
> using a component that was not in the library for the chip. Or I may
> have had the libraries wrong. I was working with two different chips (2T
> and 3T) and had some trouble setting them up the first time. 
> 
> Verify that the buffers that are giving you the problem are in the
> library. Also verify that you are pointing to the correct library(s). I
> do not remember any components called "BFI" anything. 
> 
> 
> 
> Inder wrote:
> > 
> > I am trying to simulate a controller circuit on ORCA (OR3T00
> > architecture) FPGA.
> > For schematic capture and EDIF file generation I am using Powerview
> > 6.1 tool.
> > But in the Build operation, which translates and converts the input
> > EDIF netlist
> > into ORCA generic database file(*.ngd), I have been getting this
> > error:
> > 
> > ERROR - ngdbuild: Could not expand block '....' with TYPE='...'
> > 
> > and after lot of looking around havent been able to solve the problem.
> > Can anyone tell me where and what am I doing wrong..
> > 
> > There was no error while generating EDIF file with Powerview tool and
> > I used the
> > Orca3 library in Powerview.
> > Here is a relevent part of log file generated by Orca.
> > 
> > ORCA Foundry Control Center Log File -- Thursday, February 14, 2002
> > 14:25
> > 
> > ----------------------------------------------------------------------
> > /usr/local/apps/ORCA/bin/sol/edif2ngd -l or3c00 "abcd.edn"
> > "/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo"
> > ----------------------------------------------------------------------
> > edif2ngd:  version ORCA Foundry 2000 Production-w/SP-B (14)
> > Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
> >      Copyright (c) 1995-2000 Lucent Technologies.  All rights
> > reserved.
> > WARNING - edif2ngd: Ampersands will be removed from numeric/underscore
> >      identifiers.
> > WARNING - edif2ngd: GLOBAL property on a net other than power or
> > ground -
> >      ignoring...
> >        On or above line 113 in file abcd.edn
> > WARNING - edif2ngd: GLOBAL property on a net other than power or
> > ground -
> >      ignoring...
> >        On or above line 664 in file abcd.edn
> > WARNING - edif2ngd: GLOBAL property on a net other than power or
> > ground -
> >      ignoring...
> >        On or above line 3814 in file abcd.edn
> > Writing the design to /home/ikumar/3700-pv/orca/abcd_1/abcd.ngo...
> > 
> > ----------------------------------------------------------------------
> > /usr/local/apps/ORCA/bin/sol/ngdbuild -a or3t00 -p
> > "/home/ikumar/3700-pv/orca/sch" -p "/home/ikumar/3700-pv/orca"
> > "/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo" "abcd_1/abcd_1.ngd"
> > ----------------------------------------------------------------------
> > ngdbuild:  version ORCA Foundry 2000 Production-w/SP-B (14)
> > Copyright 1991-1995 by NeoCAD Inc.  All rights reserved.
> >      Copyright (c) 1995-2000 Lucent Technologies.  All rights
> > reserved.
> > Reading '/home/ikumar/3700-pv/orca/abcd_1/abcd.ngo' ...
> > Loading NGL library '/usr/local/apps/ORCA/or3c00/data/orc3clib.ngl'...
> > Loading NGL library '/usr/local/apps/ORCA/data/neoprims.ngl'...
> > Loading NGL library '/usr/local/apps/ORCA/data/neomacro.ngl'...
> > Loading device for application ngdbuild from file 'or3t12x12.nph' in
> >      environment /usr/local/apps/ORCA.
> > Loading macro from file "/home/ikumar/3700-pv/orca/sch/all4_3lut.nmc".
> > Loading macro from file
> > "/home/ikumar/3700-pv/orca/sch/bottom2_3lut.nmc".
> > Loading macro from file "/home/ikumar/3700-pv/orca/sch/top3_3lut.nmc".
> > Loading macro from file
> > "/home/ikumar/3700-pv/orca/sch/bottom3_3lut.nmc".
> > ERROR - ngdbuild: Could not expand block '$1I53106/$1I9' with
> > TYPE='BUF'
> > ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST998' with
> >      TYPE='UDFDL'
> > ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST995' with
> >      TYPE='PW'
> > ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST6' with
> >      TYPE='UDFDL'
> > ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST34' with
> >      TYPE='NOT'
> > ERROR - ngdbuild: Could not expand block 'GO_READ+13/INST27/$1I9' with
> >      TYPE='BUF'
> > ERROR - ngdbuild: Could not expand block '$1I99402/INST1' with
> > TYPE='NOT'
> > ERROR - ngdbuild: Could not expand block '$1I99428/INST1' with
> > TYPE='AND3'
> > ERROR - ngdbuild: Could not expand block '$1I99428/BFI3' with
> > TYPE='BUF'
> > ERROR - ngdbuild: Could not expand block '$1I99428/BFI2' with
> > TYPE='BUF'
> >  is unexpanded
> > 
> > ********* around 10000 such errors ************
> > 
> > WARNING - ngdbuild: logical net 'DPRAM08-11/DO2' has no driver
> > WARNING - ngdbuild: logical net 'DPRAM08-11/DO3' has no driver
> > WARNING - ngdbuild: logical net 'DPRAM12-15/A0' has no driver
> > WARNING - ngdbuild: logical net 'DPRAM12-15/A1' has no driver
> > WARNING - ngdbuild: logical net 'DPRAM12-15/A2' has no driver
> > .
> > .
> > .
> > 
> > WARNING - ngdbuild: logical net 'TP3' has no driver
> > WARNING - ngdbuild: logical net 'TP3' has no load
> > WARNING - ngdbuild: logical net 'TP4' has no driver
> > WARNING - ngdbuild: logical net 'TP4' has no load
> > WARNING - ngdbuild: logical net 'TP5' has no driver
> > WARNING - ngdbuild: logical net 'TP5' has no load
> > WARNING - ngdbuild: logical net 'TP6' has no driver
> > WARNING - ngdbuild: logical net 'TP6' has no load
> > WARNING - ngdbuild: logical net 'TP7' has no driver
> > WARNING - ngdbuild: logical net 'TP7' has no load
> > ERROR - ngdbuild: DRC failed with 9865 errors and 492 warnings
> > 
> > Design Results:
> >      19 blocks expanded
> > Writing 'abcd_1/abcd_1.ngd' ...
>  
> -- 
> 
> 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: 39648
Subject: Re: Is Leonardo spectrum OEM version for Altera limited?
From: "Victor Schutte" <victors@mweb.co.za>
Date: Fri, 15 Feb 2002 10:13:47 +0200
Links: << >>  << T >>  << A >>
I am using it as part of the NIOS development kit and have used it
successfully for 200K gate devices, with no problems. The only limitation I
know of is that it can only be used for Altera devices.

Maybe the Baseline version is different than the one supplied with NIOS (but
I did compile a NIOS v1.0 design with the original Leonardo early in 2001)

Victor

"Leon de Boer" <ldeboer@iprimus.com.au> wrote in message
news:3c6a780f$1_1@news.iprimus.com.au...
> I have found while using Leonardo spectrum and Baseline for an Altera
design
> that when I get to 10-15K gates the design leaves bits out of the
synthesis.
> I have had this sort of behaviour with Actel licenses where they have a
gate
> limit. I can't seem to get anyone at Altera or exemplar to be able to tell
> me absolutely if the free baseline, Leonardo version is gate limited. Does
> anyone know?
>
> Regards Leon
>
>
>



Article: 39649
Subject: Re: Spartan-II becomes Vertex.
From: Dave.Haynes@sia-mce.co.uk (Dave Haynes)
Date: 15 Feb 2002 01:51:05 -0800
Links: << >>  << T >>  << A >>
"X. Q." <qijun@okigrp.com.sg> wrote in message news:<3c6b8a49@news.starhub.net.sg>...
> Hi, gurus:
> 
> I am programming with Spartan-II chip.
> How come iMPACT indicates vcx100 when initializing the chain
> while my FPGA chip is actually an XC2S100-5PQ208.

I experienced a similar problem a while ago, with Spartan IIEs. Are
you using the Xilinx Webpack? I found that there were some .bsd files
missing from the WebPack installation, and I had to download them from
the Xilinx website. Check if you have the correct .bsd files for your
part.


 Cheers,

                  Dave



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