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 34600

Article: 34600
Subject: Re: Urgent Help Needed
From: "Jim E" <jim@e3c.2ndmail.com>
Date: Thu, 30 Aug 2001 07:56:04 -0700
Links: << >>  << T >>  << A >>
I examined the "Place and Route"  and the "Pad" reports. The pad report shows unused as unused. The Place and Route report only describes those instances which were assigned to pin locations. The "Map" report talks about some removed blocks, but describes them as unused VCC blocks. I can't find anything unusual.

Article: 34601
Subject: Re: Principles of Verifiable RTL Design (2nd ed)
From: Nial Stewart <nials@britain.agilent.com>
Date: Thu, 30 Aug 2001 16:18:57 +0100
Links: << >>  << T >>  << A >>
Ben Franchuk wrote:
> 
> "Andy Peters
> >
> > Ben Franchuk wrote:
> > >
> > > > It's not hidden, Ben, it's just in a different form. The way you know it's
> > > > the same is twofold: first, you have a testbench you can run on the
> > > > synthesized result, which should produce the same results as your original
> > > > code. Second, you have formal verification tools, that can compare the
> > > > synthesizer output (gate level verilog, in the case I'm familiar with) with
> > > > your original VHDL or verilog code.
> > >
> > > Since I use free tools I don't have access to the fancy tools.
> >
> > Further proof of the old adage, "you get what you pay for."
> 
> That must mean all the tools are JUNK. You don't buy the tools you RENT them!


That's funny, I'm sure I'd bought the current Altera tool suite.


Nial.

Article: 34602
Subject: Re: Ethernet CRC
From: Mike Treseler <mike_dot_treseler@flukenetworks.com>
Date: Thu, 30 Aug 2001 10:11:25 -0700
Links: << >>  << T >>  << A >>
Tracy Briscoe wrote:
> 
> I'm trying to generate the CRC in hardware.  The problem that I've got is
> that every combination that I've tried to generate the CRC, and place it in
> a frame, comes back as invalid.  I've attached my vhdl testbench, in which I
> doing all this. This is all part of a custom hardware project, that will
> transmit and receive frames to and from Ethernet.

lots of possible bit/byte order problems.
check out:

http://www.erg.abdn.ac.uk/users/gorry/course/dl-pages/crc.html

especially:

   "Bit Order

   "The CRC is the only field which is by convention 
    sent most significant bit first. 
    (This is contrary to all header and payload bytes 
    which are sent least significant bit first.)
   Thus the first bit of the a CRC-32 to be sent 
   is the bit corresponding to X32 and the last, 
   the bit corresponding to X0."

Good luck.

 --Mike Treseler

Article: 34603
(removed)


Article: 34604
Subject: XCV800 : Jbits
From: Dereck <dereckaf@yahoo.com>
Date: Thu, 30 Aug 2001 10:20:24 -0700
Links: << >>  << T >>  << A >>
Hi,
I am Dereck Fernandes, I am a grad student in the University of
Massachuetts, Amherst. I working on a Interconnect Test tool for Xilinx
Virtex FPGAs.

 I have a question regarding JBits.

1. Does the routing in the Tile for XC4000 and XCV800 widely differ. I
saw a Jbits code for XC4000 which allows you to set the routing:
jbits.set(row,col,Y.HORIZ_SINGLE1,Y.ON)
Is this the same in XCV 800.
2. I am basically trying to understand how the routing looks in XCV800, In
the databook for XC4000 you have a figure that gives detail of the
Programmable Interconnect (Figure 27 Page 6-30).     
Do you have a similiar page for XCV 800 or is it the same.
3. I need to know how many single,double,long lines are present in each
tile.
Can you help me in this.

Thank you,
Dereck Fernandes

Article: 34605
Subject: Jbits: more info required
From: Dereck <dereckaf@yahoo.com>
Date: Thu, 30 Aug 2001 10:22:03 -0700
Links: << >>  << T >>  << A >>
1. Is the lastest version of JBits 2.0.1
2. Does the lastest version or the version above support programming of
the Pads, does it have IOB support.
I require this to test the CLBS on the Periphery.

My advisor did not recommend the use of the latest version of Jbits which
I had contacted you for instead he wants to use the JBits 2.0.1.
Please let me know if its a good idea. Incase the new version has fixed
some critical bugs.

3. I guess JBits can be used only with a existing bitstream. Now we don't
have Virtex hardware with us. Can you tell me where can I get a input
bitstream for XCV 800 and other Virtex parts.
I need this to test my tool which is essentially a Interconnect Test tool.

Thank you,
Dereck Fernandes

Article: 34606
Subject: Re: WebPack Con-Game
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Date: 30 Aug 2001 11:31:48 -0700
Links: << >>  << T >>  << A >>
"David Wright" <dwright@srtorque.com> writes:
> The "free" Xilinix Webpack should be classified as a Demo and not a real
> system to do even realistic small designs.

You get what you pay for.  Naturally the free package provides less
functionality than the expensive package; how can you rationally expect
otherwise?  Calling it a "con game" seems entirely uncalled for; the
features and limitations of WebPack are plainly described on the web site.

I've found it to be perfectly usable for a design with a 32-bit RISC
core, memory, UART, Ethernet, and timers.  I did my simulation using
Savant.

Article: 34607
Subject: Re: WebPack Con-Game
From: Dave Colson <dscolson@rcn.com>
Date: Thu, 30 Aug 2001 15:54:14 -0400
Links: << >>  << T >>  << A >>
Eric,

Could you describe you method for running savant with the Web Pack?
I would be interested. Also how about Symphony EDA?

I agree that Web Pack is able to do a lot. I was able to do a Spartan II 150
with it. Only the
simulation was Verrrrrrrrrrrrrry slow. But, however, at least it ran the
simulation. Which
is better then saying: "Sorry design too large to simulate" like some demos do
and exit the simulation.

Also the synthesis and place and route tools are certainly worth the effort to
download.
I was able to do quite a lot with this tool. Actually produced some real
results.

Thanks
Dave Colson

Eric Smith wrote:

> "David Wright" <dwright@srtorque.com> writes:
> > The "free" Xilinix Webpack should be classified as a Demo and not a real
> > system to do even realistic small designs.
>
> You get what you pay for.  Naturally the free package provides less
> functionality than the expensive package; how can you rationally expect
> otherwise?  Calling it a "con game" seems entirely uncalled for; the
> features and limitations of WebPack are plainly described on the web site.
>
> I've found it to be perfectly usable for a design with a 32-bit RISC
> core, memory, UART, Ethernet, and timers.  I did my simulation using
> Savant.


Article: 34608
Subject: Re: Xilinx FPGA Editor - how to route to an internal macro net?
From: 101551.3434@compuserve.com (Mark Taylor)
Date: Thu, 30 Aug 2001 20:04:44 GMT
Links: << >>  << T >>  << A >>
On Wed, 29 Aug 2001 22:22:26 GMT, arast@inficom.com (Alex Rast) wrote:

>In article <3b8b9869.9160001@news.compuserve.com>, 101551.3434@compuserve.com (Mark Taylor) wrote:
>>On Mon, 27 Aug 2001 20:09:17 GMT, arast@inficom.com (Alex Rast) wrote:
>>
>>>In article <3b86ed56.46990212@news.compuserve.com>, 101551.3434@compuserve.com
>> (Mark Taylor) wrote:
>>>>On Fri, 24 Aug 2001 00:19:48 GMT, arast@inficom.com (Alex Rast) wrote:
>>>>
>>>>>This is one I think I've done before, so I probably just need my memory 
>>>>>jogged. I'm sure it's something that happens, and that you need, all the
>> time.
>>>>>
>>>>>I've defined a hard macro, call it custommacro.nmc. .... Now, at least one
>> of the nets connects to an 
>>>>>external pin and an internal route. One common example, for instance, is
>> CLK. 
>>>>>You want the signal to be common to the internal CLB's of the macro and to 
>>>>>connect to external routes (in the case of CLK, to the global clock net). 
>>>>
>>>>As far as I know, nets could never be included in hard macros.
>>>>(despite documentation suggesting otherwise, right back to before EPIC)
>>>
>>>No, I have no problem including a net in a macro. What I have difficulty doing
>>>is routing an external net *to* the macro's internal net.
>>>
>>>>Just use the hard macro to configure slices/CLBs or whatever,
>>>>then embed the macro within a soft macro
>..
>>>I don't think this would work, for 2 reasons. First, I *have* to have a 
>>>specific routing within the hard macro (indeed, the routing itself is a key 
>>>part of the design), so I can't afford to let the software take care of any 
>>>routing at the hardware level. Second, a lot of the things we're doing are 
>>>functions you simply can't enter properly in Schematic Editor...
>>Get rid of any general purpose routing within your hard macro.
>>Keep all interconnections within a CLB. (These doesn't really count as nets at
>>a low level)
>
>I don't think I can do that because some of the nets are both general-purpose 
>to the design and special-purpose to the macro. In other words, we have 
>logic-to-logic connections within the macro that also connect to nets going 
>between macros. And the connections within the macro are routing-sensitive.
>
>>Ensure the number of macro pins is enough to complete all needed routing.
>
>I *hope* I've done that, except if you mean that the number of macro pins must 
>include a separate pin for *each* separate input of a high-fanout net in which 
>case it's a pretty hopeless task.
>
>>Now you will have a hard macro, with perhaps somewhat more pins than you had
>>before.
>>When you instantiate this macro you will be able to complete ALL routing.
>>(there won't be any strange nets for the software to complain about.)
>>This approach has worked for me in the past.
>>Bear in mind that the router is usually pretty good.
>
>IME it actually sucks rather than being pretty good, because it fails to 
>consider routing strategies like multi-drop nets or symmetric "bank-shot" nets 
>among other things. Again, it seems to work via a "one resource, one 
>connection" strategy. Again, for giggles, I experimented with a program-routed 
>net against my hand-routed net for the same design. I tested it with a design 
>occupying 6 CLB's. My manually routed version created a symmetrical route 
>using only the routing resources adjacent to the CLB's in use, while the 
>program's version implemented the routing in an asymmetric net over resources 
>adjacent to CLB's in a 12-row 7-column (84-CLB) block centered on the 6. 
> 
>>It's just the mapping & placement that sometimes needs working on.
>>The mapping & placement is totally defined by the given approach, so
>>theres nothing much to foul up on afterwards. (unlike the M1 software)
>>If you don't like the routing around an instance of the macro you can still 
>>change it.
>>The routing will generally be optimal, unless you have RAMs
>>within your macro.
>
>Given the results of my little test, it was clearly not optimal - and the 
>design I tested it on didn't have RAMs.
>
>> You can pretend (with very great care) that the RAMS are
>>LUTs , which will allow excellent routing. You will then have to patch the RAMS
>>back in later (perhaps using fpga_editor ,  or XDL if you have a large number
>>of instances.)
>>Note that XST (VHDL) sometimes doesn't keep RLOCS.
>>This is a bug that I have recently complained about.
>>If you are not using XST , you should have no problems.
>
>I should make it very clear that the design I have has to be very carefully 
>fitted together like a jigsaw puzzle. I have to design the *shape* and exact 
>configuration of the routing resources for each macro I have, so that I won't 
>be stymied by either lack of an available specific resource or inability to 
>abut a macro against another because of space versus shape restrictions in the 
>CLB, in the overall design.
>
>Does any of this make any sense or am I bringing up needless issues?
>
>Alex Rast
>arast@inficom.com
>arast@qwest.net

The technique which I have mentioned works.
However, I have just proved us both wrong.
The ISE 3.x Fpga_Editor seems to be much improved.
I tried creating a macro with an internal route.
I instanced that and added to the net .
It worked.
I added an undriven clock net to the macro, calling it something else.
I instanced that.
I added a global clock buffer.
I added the global clock buffer output to the 2nd net.
It worked.
Note that all pins used within the macro were entered as macro pins.
So you must be doing something wrong, such as not entering all your pins.
Or perhaps you are attempting to work with Locked routing?
(might be better to unlock it?(just guessing here))
I am assuming that you are using a moderately up-to-date version of
the Xilinx software? If not why not?
I am assuming yiou are using something similar to a Virtex device?
(eg Spartan2 or VirtexE ?)
I'm still fairly sure the Xilinx router is good, when one wishes to have
low net delays. The independent buffering gives good high speed
lines even where fanouts are large.
The clock routing is nicely fast. 

You could try giving me a job, if you have further problems..

Mark Taylor.
Mark_Taylor_Chess@compuserve.com


Article: 34609
Subject: Re: Defending Austin Franklin
From: "Andy Peters <andy [@] exponentmedia" <".> com">
Date: Thu, 30 Aug 2001 20:12:44 GMT
Links: << >>  << T >>  << A >>
Rick Filipkiewicz wrote:

> The *second* key to sane HDL design is to choose a tool that works purely from
> the command line & uses no registry settings. I would put it even more strongly
> than that - if a tool does do this sort of thing then it should be rejected. Env
> variables you can over-ride in the makefile.

In other words, use Unix-based tools, and the wonderful symbolic link,
to go between v8.0 and v8.1 of the tools.

The move towards Windows-based tools is a bad one, and hopefully EDA
vendors will wake up and realize that fact.  We have cheap hardware, and
a free OS (Linux) that works, so support it, already!

-andy

Article: 34610
Subject: Re: Defending Austin Franklin
From: "Austin Franklin" <austin@darkr99oom.com>
Date: Thu, 30 Aug 2001 16:38:10 -0400
Links: << >>  << T >>  << A >>

> > The *second* key to sane HDL design is to choose a tool that works
purely from
> > the command line & uses no registry settings. I would put it even more
strongly
> > than that - if a tool does do this sort of thing then it should be
rejected. Env
> > variables you can over-ride in the makefile.
>
> In other words, use Unix-based tools, and the wonderful symbolic link,
> to go between v8.0 and v8.1 of the tools.
>
> The move towards Windows-based tools is a bad one, and hopefully EDA
> vendors will wake up and realize that fact.  We have cheap hardware, and
> a free OS (Linux) that works, so support it, already!

Yes, but no corporate environments I've ever worked in use Linux.




Article: 34611
Subject: Re: Big SR in Virtex-E
From: Ray Andraka <ray@andraka.com>
Date: Thu, 30 Aug 2001 20:58:04 GMT
Links: << >>  << T >>  << A >>
  The SRL16's are 16 bit serial-serial out, but you can dynamically address
them to access more than one bit per shift clock without actually shifting
them.  If you can provide a multiple of your shift clock, you might be able to
take advantage of that feature.  Also, if you are not using the BRAMs
otherwise, thsoe can be utilized.

Tim wrote:

> "Michael Boehnel" <boehnel@iti.tu-graz.ac.at> wrote
> > I want to implement several big shift-registers with parallel load
> > (parallel to serial converter) in a Virtex-E. Beside these SRs I also
> > want to implement SRs with serial input and (big) parallel output.
> >
> > The number of CLB FFs in the FPGA are limited (Rows x Columns x 4).
> >
> > Is there a way to implement one of these SRs (parallel load or output!)
> > without using the CLBs FF's.
>
> The Xilinx SRLs give you a x16 density gain.  They are serial-in/serial-out
> so you need to play with the algorithm.
>
> BRAMs can give you parallel load (x32) and serial out (x1) by using the
> dual-port feature.
>
> > Can one of the synthesis tools manage this for me automatically(FPGA
> > Express, Synplify, ..)?
>
> Synplify can.  And probably FGPA Express - check with Synopsys.
> For funky stuff, instantiation is probably less trouble to you.

--
-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



Article: 34612
Subject: Re: WebPack Con-Game
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Date: 30 Aug 2001 14:10:56 -0700
Links: << >>  << T >>  << A >>
Dave Colson <dscolson@rcn.com> writes:
> Could you describe you method for running savant with the Web Pack?
> I would be interested.

I just ran Savant normally.  There's no integration with WebPack.

Article: 34613
Subject: Re: Defending Austin Franklin
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Date: 30 Aug 2001 14:15:17 -0700
Links: << >>  << T >>  << A >>
"Austin Franklin" <austin@darkr99oom.com> writes:
> Yes, but no corporate environments I've ever worked in use Linux.

Other than my own employer, nearly every company I've had contact with
here in Silicon Valley that does large ASIC designs is now using Linux
for their simulation farms, since it's so much more cost-effective than
Windows.  And I'm not just talking about the purchase cost, although for
a large farm that may actually be a consideration.  In general, the
Linux-based simulation farms seem to require much less administrative
overhead than the Windows ones they replaced.

Article: 34614
Subject: Re: XCV800 : Jbits
From: "Tim" <tim@rockylogic.com.nospam.com>
Date: Thu, 30 Aug 2001 23:38:16 +0100
Links: << >>  << T >>  << A >>

"Dereck" <dereckaf@yahoo.com> wrote in message news:ee722b3.-1@WebX.sUN8CHnE...
> Hi,
> I am Dereck Fernandes, I am a grad student in the University of
> Massachuetts, Amherst. I working on a Interconnect Test tool for Xilinx
> Virtex FPGAs.
>
>  I have a question regarding JBits.
>
> 1. Does the routing in the Tile for XC4000 and XCV800 widely differ. I
> saw a Jbits code for XC4000 which allows you to set the routing:
> jbits.set(row,col,Y.HORIZ_SINGLE1,Y.ON)
> Is this the same in XCV 800.

Virtex is much simpler and much more powerful.

> 2. I am basically trying to understand how the routing looks in XCV800, In
> the databook for XC4000 you have a figure that gives detail of the
> Programmable Interconnect (Figure 27 Page 6-30).
> Do you have a similiar page for XCV 800 or is it the same.

Write to jbits@xilinx.com

> 3. I need to know how many single,double,long lines are present in each
> tile.

The basics are in the data sheet and visible in FPGA_editor.  More
info comes with JBits.




Article: 34615
Subject: Re: star-wars ascii-animation:)
From: Ron Huizen <rhuizen@bittware.com>
Date: Thu, 30 Aug 2001 18:50:50 -0400
Links: << >>  << T >>  << A >>
Great beer cooler, but I think he should make some use of all that
wasted heat - maybe a combo beer cooler, hot dog roaster ...


Peter Seed wrote:
> 
> Even better
> 
> www.asciimation.co.nz\beer
> 
> Not sure what its got to do with embedded though...
> 
> Russell Shaw wrote in message <3B7B401E.E90A0168@iprimus.com.au>...
> >http://www.asciimation.co.nz

Article: 34616
Subject: Re: WinMe installation
From: Russell Shaw <rjshaw@iprimus.com.au>
Date: Fri, 31 Aug 2001 13:30:22 +1000
Links: << >>  << T >>  << A >>
Many laptops have software that's probing the parallel port for things
like external cd drives etc. Try disabling them.

G E Geiger wrote:
> 
> Has anyone tried the Xilinx download cable "Parallel III" on a Win ME
> machine?
> 
> I have installed the entire tool chain on my Win ME laptop and I can
> get all parts working except this cable.  The Xilinx hotline said that
> it was probably "noise" on the cable since this is a 900MHz machine
> but that ME would not actually be supported until release 4.1 of the
> Alliance tools.
> 
> The cable works fine in the lab on a slower PC running Win 98 SE.
> 
> Could MS have actually messed up a parallel port going from 98 to ME?
> 
> On Tue, 14 Aug 2001 13:50:01 GMT, "Paul Teagle"
> <pteagle@bigpond.net.au> wrote:
> 
> >I had a ME related problem a while ago (look back for some archive
> >material - Xilinx solution 9253 was most appropriate).
> >
> >Basically, it was a problem with the path. Be careful how you attempt to
> >modify the autoexec.bat file. You have to use the system utilities to
> >modifiy, not just a text editor on the autoexec.bat files. There's registry
> >issues involved.
> >
> >I've now got a problem with the ModelSim package not finding the design. Oh
> >well...

--
   ___                                           ___
  /  /\                                         /  /\
 /  /__\ Russell Shaw, B.Eng, M.Eng(Research)  /  /\/\
/__/   / Victoria, Australia, Down-Under      /__/\/\/
\  \  /  http://home.iprimus.com.au/rjshaw    \  \/\/
 \__\/                                         \__\/

Article: 34617
(removed)


Article: 34618
Subject: Timing delay problem
From: "Abhimanyu Rastogi" <abhi_rastogi@hotmail.com>
Date: Fri, 31 Aug 2001 04:02:38 GMT
Links: << >>  << T >>  << A >>
Hello,

I have been working on this project of mine using Altera's AHDL tools andnow
when I all the code is done, i wanted to simulate it to see what i getas my
result.

I don't why i am getting 2 different results :-
1) when i use Functional SNF extractor I get my output exactly as i wanted.
And after simulating thru functional snf i simulate it again without it with
just normal simulation features (compiler netlist extractor, database
builder, logic synth, partitioner, fitter (not quartus fitting), assembler)
to get a TFF file which i include in a dll. this thing work fine but i when
i chk for my clock output on the test pin i don't get any signal ...  why is
that??
2) when i use Timing SNF extractor i don't get my ouput correctly...  like
some are delayed and thus throws out wrong information on simulation.

Now is there any way so that i don't get all these problems with
timing....i'm using FLEX 8000 series chip

Waiting for ur reply as soon as possible....

if u need to look at the code please email me back....thanks




Article: 34619
Subject: ISA(PC/104) BUS DECODE ASYNC or SYNC?
From: diverseg@aol.com (DIVERSEG)
Date: 31 Aug 2001 05:49:00 GMT
Links: << >>  << T >>  << A >>

I have always tried to stick to sync designs.
However lately I have ran into a problem.
While designing some isa cards I read somewhere that I did not need to use BALE
to latch my address lines.

Is this acceptable?

Also I am decoding using a state machine in a FPGA with Sysclk as the clock for
transitions in my state machine.

  Last week I changed CPU and the new CPU card does not drive Sysclk.
Now my boards do not work. I called the board designer and de says that sysclk
now a days is not sync with the I/O signals anyway?

Who do i believe? 

Is async acceptable? 

Are the address lines stable so BALE is not nescessary?

Any comments suggestions?


Xilinx foundation users is it possible to take the VHDL created by the state
diagram editor and make the state machine async and dependant on the input
signal transitions to go from one state to another?


Steven Collins
Macha International, Inc.
713-723-5040x13
scollins@macha.com


Article: 34620
Subject: Xilinx Device Update under Solaris
From: Petter Gustad <newsmailcomp1@gustad.com>
Date: 31 Aug 2001 08:39:26 +0200
Links: << >>  << T >>  << A >>

 Some  time ago I received the CD titled: Device Update Foundation
 Series 3.3i (Contains Virtex-II Device Libraries and Service Pack
 6).
 
 I can't figure out how to install it on my Solaris machine. There
 is nothing about Solaris in the Readme.txt file. There is only  a
 Setup.exe Windows executable.
 
 How  do  I load the libraries under Solaris? I'm primarily inter­
 ested in the Virtex-II device libraries since I have already  in­
 stalled SP8 from the Xilinx Web site.
 
 Petter

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

Article: 34621
Subject: Ugly signal output...
From: "Noddy" <g9731642@campus.ru.ac.za>
Date: Fri, 31 Aug 2001 10:15:09 +0200
Links: << >>  << T >>  << A >>
Hi,
I am using the Spartan XCS10XL just as a clock signal (using as an encode
signal for an A/D). Anyway, I'm reading the clock signal into the FPGA on a
global clock buffer, put it into a toggle to halve the frequency(from 50 to
25MHz), and then output it again on a standard output pad. However, at 25MHz
I would expect the signal from the toggle to be pretty close to square, but
my signal still looks very wobbly.

Can anyone give any suggestions?

Adrian




Article: 34622
Subject: Re: WebPack Con-Game
From: Santiago de Pablo <sanpab@eis.uva.es>
Date: Fri, 31 Aug 2001 11:45:59 +0200
Links: << >>  << T >>  << A >>
David Wright escribió:
> 
> The "free" Xilinix Webpack should be classified as a Demo and not a real
> system to do even realistic small designs.
> 
> By the time you install the design and even a small test vector, you easily
> exceed the 500-line ModelSim Starter Design Limit.  There are even
> limitations on Test Bench and probably other portions of the package.

I agree with the Bencher limitations: are stupid, because you can
generate
two or three vectors and then use notepad to copy-paste-modify.

But I don't agree with the ModelSim (HDL only simulator) 500-line
limitation:
is more than enough to test modules separately, and there is no
slow-down if
you use *.do files.

Please, check "http://www.dte.eis.uva.es/OpenProjects/OpenDSP/index.htm"
and
see how a complete DSP has been synthetized and simulated using WebPack.
 
> Xilinx needs to stop misleading its customers and tell them what they really
> need to know:

At least, it's (now) better than Altera's free tools.

Enjoy, Santiago.

Article: 34623
Subject: Re: ISA(PC/104) BUS DECODE ASYNC or SYNC?
From: "Tim" <tim@rockylogic.com.nospam.com>
Date: Fri, 31 Aug 2001 11:21:11 +0100
Links: << >>  << T >>  << A >>
comp.arch.embedded is probably the best place for this
question.  AFAIK.

"DIVERSEG" <diverseg@aol.com> wrote in message
news:20010831014900.26248.00004289@mb-fy.aol.com...
>
> I have always tried to stick to sync designs.
> However lately I have ran into a problem.
> While designing some isa cards I read somewhere that I did not need to use
BALE
> to latch my address lines.
>
> Is this acceptable?
>
> Also I am decoding using a state machine in a FPGA with Sysclk as the clock
for
> transitions in my state machine.
>
>   Last week I changed CPU and the new CPU card does not drive Sysclk.
> Now my boards do not work. I called the board designer and de says that sysclk
> now a days is not sync with the I/O signals anyway?
>
> Who do i believe?
>
> Is async acceptable?
>
> Are the address lines stable so BALE is not nescessary?
>
> Any comments suggestions?
>
>
> Xilinx foundation users is it possible to take the VHDL created by the state
> diagram editor and make the state machine async and dependant on the input
> signal transitions to go from one state to another?
>
>
> Steven Collins
> Macha International, Inc.
> 713-723-5040x13
> scollins@macha.com
>



Article: 34624
Subject: Re: Xilinx FPGA Editor - how to route to an internal macro net?
From: 101551.3434@compuserve.com (Mark Taylor)
Date: Fri, 31 Aug 2001 13:56:19 GMT
Links: << >>  << T >>  << A >>
A correction:
 I subsequently found problems with respect to internal macro nets.
1) If a net is deleted there is a problem.
2) If multiple instantiations of a macro are made then it might be tough
to connect a common clock line.

So I can only refer you to what I originally said.
DON'T USE INTERNAL MACRO NETS.

ENSURE YOU HAVE INCLUDED ALL MACRO PINS.

I suspect you may have been using the P&R tools at the default
settings: "brain-dead". Turn up the level to "decent".
Ensure you have the delay - based clean up pass selected.

I have in the past only used hard macros when there have
been mapping deficiencies. (ie 4000X series)
I haven't yet found any need to use hard macros for
Spartan2  (Virtex), because the primitives available map
well to the capabilities of the CLBs.


You could try giving me a job, if you have further problems..

Mark Taylor.
Mark_Taylor_Chess@compuserve.com





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