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 38000

Article: 38000
Subject: Re: How to set block ram contents ?
From: brimdavis@aol.com (Brian Davis)
Date: 29 Dec 2001 19:05:19 -0800
Links: << >>  << T >>  << A >>
> I generate code from an assembler that's then run through a 
> little utility to create a series of "INIT_xx=" attributes.
> Is it possible to script the update so that I don't have to
> key anything ? I think I have seen posted before that it's
> possible to modify the bitstream (using a program) to do this.
> I'd like to automate the whole thing if possible. (My coding
> skills are expert C/C++ preferred and I'm not afraid to hack
> things as long as I have the info..).

 Another method, besides FPGA Editor, would be the XDL ASCII design
language; alas, it is also missing from the Webpack tools.

 XDL is a nifty ASCII <-> NCD conversion utility that was supplied
and documented in the xilinx/userware/doc directory of the 2.1i
release; the executable is still supplied sans documentation
( other than its' built in help ) in the later versions.

 Also note, if you try the FPGA Editor trick with 2.1i ( i.e. the
student edition tools), the INIT attributes can not be edited
properly within the FPGA Editor, see Xilinx solution record 7167
in the answers archive ( I don't think they ever fixed that
in any 2.1i service packs ).

 Your best bet may be to use Jbits or hack your own bitstream 
editor as you mention; I think the Xilinx app notes on Virtex
bitstreams give enough information to do this. 

 When I started using block RAMs for my own processor tinkering,
I came up with a mental list of ways to get the bits into their
proper places, which went something like this:

 1) VHDL file generator, constant array for simulation & inferred RAMs
 2) VHDL file generator, INIT attributes for instantiated RAMs
 3) FPGA Editor command script
 4) XDL based NCD file editing
 5) homebrew bitstream mangler

 In the real world, once I had #2 working, the first 'real' program I
wrote was a tiny debug monitor, which lets me download over an RS232 link.
Once that was working, I never got around to writing any of the others...

 A small 'copy N bytes of data from the serial port into RAM & execute'
bootstrap routine would fit in even less space than a tiny monitor; a
small ROM built with a single bank of 32 x instruction_width CLB ROM's
would probably hold it with room to spare, leaving the entire block RAM
free for user code.

Brian

Article: 38001
Subject: Re: How to set block ram contents ?
From: rashidk@home.net (Rashid Karimov)
Date: 29 Dec 2001 20:16:15 -0800
Links: << >>  << T >>  << A >>
What is an "FPGA editor" ? :)

Anyway, since you know that RAM block content initialization is
accomplished (in Xilinx FPGA tools) via INIT_XX parameters, and
understand that nice ISE
GUI is just a wrapper to accomplish the same, I would advise to 

a) learn Perl (shouldn't take more than a week, assuming you know
C/C++).
You can accomplish the same thing in C/C++, but this is kinda thing
Perl
is ideal for, plus you get an incentive to learn a new language :)

b) learn how to use makefile mechanish to drive ISE/Webpack to regen
the
bit file.

Perl would parse your  ASM file(s) into hex string(s) and put them
where they belong in VHD source. Makefile would allow to automate .bit
regeneration.

There might be a way to do a surgery directly on .bit file, but I'd
imagine
it is safeguarded via some CRC mechanism that would make it rather
hard, especially since you wouldnt save a whole lotta time ... unless
you have to (re)PAR a 3M+ gate design on a 386 PC :)

Gotta thank Xilinx for giving us BlockRAM. They tried to stay within
HDL semantics as far as initialization goes ... but for large ROM-like
designs,
an easier way to init it would be a better alternative, even at
expense of
being proprietory.

My $.02

Article: 38002
Subject: Re: ALTERA's Mercury CDR
From: Kevin Brace <nospamformekevinbraceusenet@hotmail.comnospamforme>
Date: Sun, 30 Dec 2001 01:34:26 -0600
Links: << >>  << T >>  << A >>
Sounds like a posting from an Altera employee.



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

Article: 38003
Subject: CRC-32 48bit(width)
From: Kenily <aiurh@iuehr.erug>
Date: Sun, 30 Dec 2001 00:27:24 -0800
Links: << >>  << T >>  << A >>
who can post CRC-32 48bit(width) verilog source code?

Article: 38004
Subject: Re: CRC-32 48bit(width)
From: Petter Gustad <newsmailcomp1@gustad.com>
Date: 30 Dec 2001 12:49:18 +0100
Links: << >>  << T >>  << A >>
Kenily <aiurh@iuehr.erug> writes:

> who can post CRC-32 48bit(width) verilog source code?

http://www.easics.com/webtools/crctool

Petter
-- 
________________________________________________________________________
Petter Gustad   8'h2B | (~8'h2B) - Hamlet in Verilog   http://gustad.com

Article: 38005
Subject: Re: CRC-32 verilog source code
From: Petter Gustad <newsmailcomp1@gustad.com>
Date: 30 Dec 2001 12:54:35 +0100
Links: << >>  << T >>  << A >>
Mardin <chens_w@yahoo.com.cn> writes:

> who can post it?

http://www.easics.com/webtools/crctool

Petter
-- 
________________________________________________________________________
Petter Gustad   8'h2B | (~8'h2B) - Hamlet in Verilog   http://gustad.com

Article: 38006
Subject: Re: Innoveda Speedwave vs. Modelsim?
From: hamish@cloud.net.au
Date: 30 Dec 2001 12:47:07 GMT
Links: << >>  << T >>  << A >>
rickman <spamgoeshere4@yahoo.com> wrote:
> My problem surfaced when I reloaded simulations from scratch repeatedly
> without restarting Modelsim. Because there was no way to position and
> size the windows on startup and I found it irritating to have to save
> the waveform setups just because I added a couple of signals, I would
> just reload the simulation and keep using my current window setups.
> After 5 or so reloads Modelsim would crap out under NT. It never crashed
> the machine, so it could be restarted. But it took a couple of minutes
> to get everything organized again so I could see what I was looking for. 

> This was reported to Mentor, so it may be fixed by now. 

It isn't fixed in (PE) 5.5d; possibly in 5.5e but I'd guess not.


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

Article: 38007
Subject: Re: Innoveda Speedwave vs. Modelsim?
From: "Patrick Muller" <patrick at scs dot ch>
Date: Sun, 30 Dec 2001 14:27:34 +0100
Links: << >>  << T >>  << A >>

<hamish@cloud.net.au> wrote in message
news:3c2f0ccb$0$32687$afc38c87@news.optusnet.com.au...
> It isn't fixed in (PE) 5.5d; possibly in 5.5e but I'd guess not.

No, not fixed yet! Probabely it will be fixed in PE 5.6............

Patrick



Article: 38008
Subject: Re: How to generate .edn in Webpack ?
From: "Tim Boescke" <tboescke@engmail.uwaterloo.ca>
Date: Sun, 30 Dec 2001 18:01:13 -0500
Links: << >>  << T >>  << A >>

> > the inconvenience, I'm told that the optimization that comes from the
> > elimination of the netlist is a huge advantage that XST now has.
>
> I'm really not ready to buy the ``huge advantage ...''. Running edif2ngd
to
> create a .ngo file from a .edf takes hardly any time at all.

Interestingly it seems that also ispDesignStarter (Lattice), which comes
with
Synplify, switched to encrypted netlists recently. I might be mistaken,
but I was just able to locate a file which looks very similar to an edif
file instead of the edf file that was generated in earlier version.

It seems hiding the EDIF output became some kind of general fashion for
"free" tools. Why do the vendors not want us to access the netlist ?





Article: 38009
Subject: Re: How to generate .edn in Webpack ?
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Sun, 30 Dec 2001 23:26:47 +0000
Links: << >>  << T >>  << A >>


Tim Boescke wrote:

> > > the inconvenience, I'm told that the optimization that comes from the
> > > elimination of the netlist is a huge advantage that XST now has.
> >
> > I'm really not ready to buy the ``huge advantage ...''. Running edif2ngd
> to
> > create a .ngo file from a .edf takes hardly any time at all.
>
> Interestingly it seems that also ispDesignStarter (Lattice), which comes
> with
> Synplify, switched to encrypted netlists recently. I might be mistaken,
> but I was just able to locate a file which looks very similar to an edif
> file instead of the edf file that was generated in earlier version.
>
> It seems hiding the EDIF output became some kind of general fashion for
> "free" tools. Why do the vendors not want us to access the netlist ?

I'd suggest that its an outburst of paranoia that it might be possible to
re-target e.g. CoreGen netlists to another technology. IIRC Leonardo Spectrum
has some facility like this.Yet another round in the seemingly endless,
depressing, fight against marketing's desperate desire to lock us all into
proprietary standards.

By all means let the Vendors use encrypted netlists or binary formats for
their own IP but but I'll thank you to *leave my designs alone!* . There have
been at least 3 occasions in the past where the only way I could work around a
tool bug was to use the last backstop and Perl hack the netlists directly (one
XNF & two EDIF).


Article: 38010
Subject: Re: Innoveda Speedwave vs. Modelsim?
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Sun, 30 Dec 2001 23:35:56 +0000
Links: << >>  << T >>  << A >>


Patrick Muller wrote:

> <hamish@cloud.net.au> wrote in message
> news:3c2f0ccb$0$32687$afc38c87@news.optusnet.com.au...
> > It isn't fixed in (PE) 5.5d; possibly in 5.5e but I'd guess not.
>
> No, not fixed yet! Probabely it will be fixed in PE 5.6............
>
> Patrick

Oh dear! I've been extolling the virtues of ModelSim and now I find that
upgrading from 5.4c to 5.5e is causing problems. Specifically almost all
the sims run o.k. but in one case where I use a PLI based CPU bus model
a restart+run causes ModelSim to go total SPLAAT. It could be a bug in
the PLI  code (its not mine) that was being hidden by 5.4c ... Anyone
know how to debug a PC based PLI application ?


Article: 38011
Subject: Stupid Foundation Question (Hey Peter, some Kindergarden stuff for you)
From: nweaver@CSUA.Berkeley.EDU (Nicholas Weaver)
Date: Mon, 31 Dec 2001 00:10:16 +0000 (UTC)
Links: << >>  << T >>  << A >>
Damn, a stupid question, but I can't figure out how.

I have a design in foundation 4.1 (schematics, yeah, call me geriatric
but I THINK in schematics, as tedious as it is), targeting a Virtex
E.

How can I retarget the design to a Spartan II, as painlessly as
possible.  Can I just force a compile saying "Compile Spartan II", or
is there a convert tool?  I'm not using any virtex E specific
features.


-- 
Nicholas C. Weaver                                 nweaver@cs.berkeley.edu

Article: 38012
Subject: exclude a path in TRACE timing
From: khtsoi@cse.cuhk.edu.hk
Date: 31 Dec 2001 00:46:56 GMT
Links: << >>  << T >>  << A >>
Hi,

I have a path with large combinational logic and long route but the result
should only be used after 2 clock cycles which means there is sufficient
time for the signals to pass through it. How can I force the TRACE program
not to think this is the critical path? I am using Xilinx XCV1000E and with
Synopsys DC/FPGA Express and Xilinx Alliance 3.1i. Any hints or pointers to
documents are welcome. Thanks in advance!

---- Brittle

Article: 38013
Subject: Re: Innoveda Speedwave vs. Modelsim?
From: Rick Filipkiewicz <rick@algor.co.uk>
Date: Mon, 31 Dec 2001 01:50:18 +0000
Links: << >>  << T >>  << A >>


I wrote:

Oh dear!  ...

It looks like ModelSim is o.k. (except for the memory leak!) & the problem
was the PLI stuff. It was relying on a couple of global variables being
maintained across a restart. Since they were pointers all hell broke loose
when the PLI app tried to free the memory they ``pointed at''.



Article: 38014
Subject: Re: Innoveda Speedwave vs. Modelsim?
From: Duane Clark <junkmail@junkmail.com>
Date: Sun, 30 Dec 2001 18:55:31 -0800
Links: << >>  << T >>  << A >>
rickman wrote:

> ...
> My problem surfaced when I reloaded simulations from scratch repeatedly
> without restarting Modelsim. Because there was no way to position and
> size the windows on startup...


Well, I guess that is one advantage of using the Linux version then. The 
Gnome window manager in Linux makes this very easy.





Article: 38015
Subject: Re: Stupid Foundation Question (Hey Peter, some Kindergarden stuff for
From: Ray Andraka <ray@andraka.com>
Date: Mon, 31 Dec 2001 03:11:21 GMT
Links: << >>  << T >>  << A >>
As long as you aren't using any virtexE specific stuff, like the LVDS or too
many DLLs, you should be able to convert just by changing the target in the
xilinx tools.  I am assuming the schematic was done with the xilinx unified
library.

Nicholas Weaver wrote:

> Damn, a stupid question, but I can't figure out how.
>
> I have a design in foundation 4.1 (schematics, yeah, call me geriatric
> but I THINK in schematics, as tedious as it is), targeting a Virtex
> E.
>
> How can I retarget the design to a Spartan II, as painlessly as
> possible.  Can I just force a compile saying "Compile Spartan II", or
> is there a convert tool?  I'm not using any virtex E specific
> features.
>
> --
> Nicholas C. Weaver                                 nweaver@cs.berkeley.edu

--
--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: 38016
Subject: Re: exclude a path in TRACE timing
From: hamish@cloud.net.au
Date: 31 Dec 2001 05:32:41 GMT
Links: << >>  << T >>  << A >>
khtsoi@cse.cuhk.edu.hk wrote:
> I have a path with large combinational logic and long route but the result
> should only be used after 2 clock cycles which means there is sufficient
> time for the signals to pass through it. How can I force the TRACE program
> not to think this is the critical path? I am using Xilinx XCV1000E and with
> Synopsys DC/FPGA Express and Xilinx Alliance 3.1i. Any hints or pointers to
> documents are welcome. Thanks in advance!

You need a FROM:TO constraint in your UCF (user constraints file)
to indicate that the maximum delay FROM the source flip flop to
the TO flip-flop is twice the normal clock period.

See the constraints guide (in 4.1i) or libraries guide(?) (in 3.x)
for details.

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

Article: 38017
Subject: Re: Stupid Foundation Question (Hey Peter, some Kindergarden stuff for
From: Kevin Brace <nospamformekevinbraceusenet@hotmail.comnospamforme>
Date: Mon, 31 Dec 2001 00:19:30 -0600
Links: << >>  << T >>  << A >>
Isn't it better to target recently released Spartan-IIE (0.18u part)
rather than older Spartan-II (0.22u part) unless there are special
reasons on doing so?
Other than connecting 5V I/O stuff like 5V PCI, I think there are very
little reasons using Spartan-II instead of Spartan-IIE.
I heard somewhere that compared to Virtex-E, Spartan-IIE has half of
BlockRAM and DLL (4 instead of 8).
That is somewhat different from what Xilinx did when they released
Spartan-II (kept the same amount of BlockRAM and DLL that Virtex had).




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




Nicholas Weaver wrote:
> 
> Damn, a stupid question, but I can't figure out how.
> 
> I have a design in foundation 4.1 (schematics, yeah, call me geriatric
> but I THINK in schematics, as tedious as it is), targeting a Virtex
> E.
> 
> How can I retarget the design to a Spartan II, as painlessly as
> possible.  Can I just force a compile saying "Compile Spartan II", or
> is there a convert tool?  I'm not using any virtex E specific
> features.
> 
> --
> Nicholas C. Weaver                                 nweaver@cs.berkeley.edu

Article: 38018
Subject: Re: instruction processor
From: Kevin Brace <nospamformekevinbraceusenet@hotmail.comnospamforme>
Date: Mon, 31 Dec 2001 00:48:13 -0600
Links: << >>  << T >>  << A >>
I am not sure what you are asking, but don't all microprocessors process
some kind of instruction codes?
Perhaps, readers of news:comp.arch might be able to do a better job
answering your question (news:comp.arch.fpga is for pretty much for the
users of FPGAs).



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




"M. Sachemo" wrote:
> 
> can some body help me understand what an instruction processor is?
> thanks for your reply
> newbie

Article: 38019
Subject: Re: CRC-32 48bit(width)
From: Kenily <aiurh@iuehr.erug>
Date: Mon, 31 Dec 2001 00:00:03 -0800
Links: << >>  << T >>  << A >>
Thanks!
But there is not 48bit(width)

Article: 38020
Subject: Re: exclude a path in TRACE timing
From: khtsoi@cse.cuhk.edu.hk
Date: 31 Dec 2001 09:27:12 GMT
Links: << >>  << T >>  << A >>
Thanks!

---- Brittle

Article: 38021
Subject: Re: Stupid Foundation Question (Hey Peter, some Kindergarden stuff for you)
From: "Carl Brannen" <carl.brannen@terabeam.com>
Date: Mon, 31 Dec 2001 14:19:07 +0000 (UTC)
Links: << >>  << T >>  << A >>
To change target in Foundation 4.1 between (more or less) compatible chip
types:

At the "Project Manager" window (the one with the little boxes on it for
design entry etc.), select the "File" pull down menu, and choose "Project
Type".
Or you can just type ^T, but I can never remember that.

A box labeled "Change Project Type" will come up, and you can use the bottom
left pull down menu to select between the various architectures.  Of course
this will only work if you included the architecture you want in your install.
That's why I always install them all.

I've noticed that some of the schematics tend to be a bit messed up when you
do this.  In particular, the "IPAD" component tends to be different sizes
between certain Xilinx families.  Changing the project type can cause them
to become disattached.  To fix this, you have to go through all the schematic
pages and move the IPADs back or forth one 'tic' or whatever the unit of
measurement in schematics is.  I'd report this little fact as a bug, but it
isn't much of a show stopper.

Carl


-- 
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

Article: 38022
Subject: Re: THE SIGNAL LIST IS NOT AVAILABLE TO SIMULATE
From: "Carl Brannen" <carl.brannen@terabeam.com>
Date: Mon, 31 Dec 2001 14:25:41 +0000 (UTC)
Links: << >>  << T >>  << A >>
Freny, the only time I've experienced something like what you're talking about
is when I've screwed up the Project Manager image.  So here's the way of doing
that:
(1) Make PM too small.  Then the flowchart gets hidden and you can't click it.
(2) Move the bar that separates the "Files / Versions" side from the "Flow /
Contents / Reports / Synthesis" side too far to the right.  Then there's
not enough room to show the "Simulation" and "Verification" buttons.
(3) Move the vertical bar that separates the top half of the screen from
the bottom half too far up.
(4) Choose "Contents", "Reports", or "Synthesis" instead of "Flow" over where
the flowchart is usually shown.

Other than that, the tool has worked great for me for years.  For the sake of
your self esteem, I hope that your problem is more complicated, and requires
repair by reinstalling Foundation.

Carl


-- 
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

Article: 38023
Subject: Re: CRC-32 48bit(width)
From: Ray Andraka <ray@andraka.com>
Date: Mon, 31 Dec 2001 16:05:48 GMT
Links: << >>  << T >>  << A >>
Sure there is, but it is not obvious.  check the 32 AND the 16 bit
boxes.  That gives you the 48 bit variant.

Kenily wrote:

> Thanks!
> But there is not 48bit(width)

--
--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: 38024
Subject: Synplicity to Xilinx hierarchical net names
From: gsr_1976@yahoo.com (gajju)
Date: 31 Dec 2001 10:19:38 -0800
Links: << >>  << T >>  << A >>
Hi
I am using synplicity for synthesizing my code and targetting the edif
to xilinx par. In par i am giving area constraint to for hierarchical
design.
INST "U1/U2/U3" AREA_GROUP=g0;
AREA_GROUP g0 .....
I have found that some of my hierarchical block nets are not
synthesized with name U1/U2/U3 etc. Instead they are named as G_900..
Bcas of this par tool is not able to place them in specified area.
These nets are some combo logic which has been synthesized and given
some interim name by synplicity. Is there any way i can tell
synplicity tool to prepand all of them with hierarchy name??



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