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 10475

Article: 10475
Subject: Re: Building signal delays inside an FPGA
From: Peter Alfke <peter.alfke@xilinx.com>
Date: Wed, 20 May 1998 17:39:28 -0700
Links: << >>  << T >>  << A >>
Alfredo Rosado wrote:

> Hi,
>     I am trying to build some kind of architecture to delay a signal
> (typical delay ranging from 1 to 3 useconds). I have thought of doing
> it
> with chained flip-flops, but if I have a fast clock signal, it will

You can build a virtual shift register in the synchronous RAM of the
XC4000E or XC4000XL devices, which gives you 32 bits per CLB.

Or you can use a presettable synchronous counter. There are lots of ways
of doing this, depending on your constraints.

You will never get close to microseconds by just concatenating logic
delays of a few nanoseconds each.

Peter Alfke, Xilinx Applications

Article: 10476
Subject: Re: FPGA-based CPUs (was Re: Minimal ALU instruction set)
From: "Lardino@ibm.net" <lardino@ibm.net>
Date: Wed, 20 May 1998 23:56:27 -0400
Links: << >>  << T >>  << A >>
CPUs in FPGAs are already being done at some schools. I
did a project as an undergrad at the University of Winsconsin,
that used Xilinx FPGAs to implement a full student-designed
ISA, from start to finish. There are a lot of limitation with using
FPGAs (as we found... and this was 3 years ago...). Perhaps
the biggest was bus size...

Our implementation used only 8 bit registers (it was a dual,
threaded processor). The whole project (6 FPGAs, external SRAM,
external UART) was wirewrapped (a wiring extravaganza!) and
ended up running at about 8MHz. Quite a technical achievement.

We were also responsible for developing a compiler/assembler
and working applications (the wonder and amazement generated
by Tetris running on a processor built by a group of college students
is probably a whole lot more than any non-EE could realize:).

I'm sure the class is still being taught since it was one of the most
popular in the EE department.

Regards,
Shane
--
>I think that is an excellent point. When I was in grad school, I wanted
>to implement a simulation of a microcoded teaching CPU that they used
>for a couple of classes. Until then it had only existed on paper. But
>the simulation would not have allowed the students to modify or extend
>the design easily. This would allow an EE student to play to his hearts
>content, and he might actually learn something. ;-)
>
>
>> * Satisfaction: Building a small computer from the gates up
>>     through the instruction set architecture up through the
>>     OS and compiler up to apps is a very satisfying exercise,
>>     akin to building one's own small sailboat.  It can be a
>>     potent antidote to the shrink-wrapped proprietary HW and
>>     SW we generally must live with.
>>
>>         --Mike
>
>This is certainly a good reason to do a project in your spare time. The
>education you would receive is not trivial either. It appears that a lot
>of people have done this with a Forth implementation in mind. Do you
>think they are doing this because of the personal satisfaction
>motivation or do you think they have commercial motivations in mind?
>



Article: 10477
Subject: Re: FPGA-based CPUs (was Re: Minimal ALU instruction set)
From: hamish@moffatt.nu (Hamish Moffatt)
Date: 21 May 1998 05:03:58 GMT
Links: << >>  << T >>  << A >>
In comp.arch.fpga Lardino@ibm.net <lardino@ibm.net> wrote:
> CPUs in FPGAs are already being done at some schools. I
> did a project as an undergrad at the University of Winsconsin,
> that used Xilinx FPGAs to implement a full student-designed
> ISA, from start to finish. There are a lot of limitation with using
> FPGAs (as we found... and this was 3 years ago...). Perhaps
> the biggest was bus size...

Here at RMIT (Melbourne, Australia) we did something similar. We implemented
the 6805 at the code-compatible level (not timing compatible), using 
XC3090A parts, including wirewrapping the thing up. This was the data path
only, though; control logic was done by an external Am2901-based sequencer,
with custom software on PCs to control it all. This was the core third
year digital systems subject, and it was a lot of work at the time.

Just FYI,
Hamish
-- 
Hamish Moffatt, StudIEAust              hamish@debian.org, hamish@moffatt.nu
Student, computer science & computer systems engineering.    4th year, RMIT.
http://hamish.home.ml.org/ (PGP key here)             CPOM: [******    ] 66%
Nondeterminism means never having to say you are wrong.
Article: 10478
Subject: Re: Minimal ALU instruction set.
From: Steven Groom <arrownz@ihug.co.nz>
Date: Fri, 22 May 1998 11:20:03 +1200
Links: << >>  << T >>  << A >>


Fitz wrote:

> Mike,
>
> Please keep us informed on your progress.
>
> Thanks,
> Fitz
>
> Mike Butts wrote:
>
> > In fact, a fine poster paper about a small FPGA-based processor
> > was given at FCCM last month.  I'm so inspired by it, and by the
> > new availability of cheap FPGAs and tools, that I'm developing
> > a little minimum-cost FPGA+SRAM board and a similar little
> > processor design for it.  I'm hoping to make it available by
> > the end of the year.  (This is strictly a personal hobby effort
> > on my part, for fun, experimental and educational purposes.)

I have a board already constructed with an Altera part (10K10) for
USD$200 PC104 bus, 512KEPROM 32KSRAM, RS232 LCD Display and manual.

There will be ADC/DAC board options as well, in the next rev.

I have written a micro (actually several) for this.  All public domain.
Currently writing CAN controller, UART and FIR/FFT functions (public
domain when complete)

The software to compile for this part is free from Altera
(www.altera.com)


Steve.

Article: 10479
Subject: Re: XABEL problem
From: Steven Groom <arrownz@ihug.co.nz>
Date: Fri, 22 May 1998 11:24:02 +1200
Links: << >>  << T >>  << A >>
> Just wondered if someone might be able to help before
> I pull my last hair out -- I've a little bit of code
> in ABEL which I'd like to use as a part of a design
> going into one of the Xilinx Spartan things - an XCS30,
> to be exact.  All it's supposed to do is to generate
> even parity over 36 signals.
>
> Anyway, it turns into a netlist OK; 'bit slow' thought
> I, as about 30 minutes passed before the design was fitted.
> 'Good Lord!' thought I, as the report showed it filling 87%
> of my XCS30.  What I'd think to be an equivalent drawn out
> as a schematic behaves much more reasonably.

Would you like me to try this implementation in an Altera CPLD?  The
EPF6016 will tackle this with no problems, and may cost you less than a
comparable Spartan part.

100 up price EPF6016TC144-3 is USD$19.95 (equiv 650 CLB) TQFP144

Regards,

Steve.

Article: 10480
Subject: Re: Xilinx FPGA Configuration Problem
From: Peter Alfke <peter.alfke@xilinx.com>
Date: Thu, 21 May 1998 16:45:11 -0700
Links: << >>  << T >>  << A >>
Yves, I have never heard about such a problem, so I think you are
actually doing something else.

Read the first paragraph of the desciption carefully: "...trailing edge
of the logic AND of all those four pins, three of which are active Low."

What happens in the very beginning, the 40-bit preamble? This is not
protected by CRC and resulting Low on INIT.

I assume you might have some "dirty" levels or spikes on one of the four
control lines ( CS0, CS1, RS and WS ).

Peter Alfke, Xilinx Applications
 

Yves Vandervennet TFE wrote:

> Hi !
>
>         We are using XC4000E FPGA's. According to the
> Data Book, when configurating the FPGA's, the devices sample the data
> on the rising edge of the logical product CS0n and WRn. A
> configuration
> problem (INITn went always done after the first set of datas) lead
> us to observe the DOUT pin. We found out that the data going out on
> this pin was the previous byte present on the data bus BEFORE the
> microprocessor write operation. Does anybody have a solution for
> us ? Do we make a mistake (The setup time of 60ns is respected) ?
>
>         Thank you for your help,
>
>                 Have a nice day,
>
>                                                 Yves Vandervennet.
>
> E-Mail : yves@elmitel.ulb.ac.be

  

Article: 10481
Subject: Re: Minimal ALU instruction set.
From: jhallen@world.std.com (Joseph H Allen)
Date: Fri, 22 May 1998 00:10:55 GMT
Links: << >>  << T >>  << A >>
In article <3564B6A2.51949200@ihug.co.nz>,
Steven Groom  <steven.groom@arrow.co.nz> wrote:

>I have a board already constructed with an Altera part (10K10) for
>USD$200 PC104 bus, 512KEPROM 32KSRAM, RS232 LCD Display and manual.

>There will be ADC/DAC board options as well, in the next rev.

>I have written a micro (actually several) for this.  All public domain.
>Currently writing CAN controller, UART and FIR/FFT functions (public
>domain when complete)

So what's it like?  Where is it available?

-- 
/*  jhallen@world.std.com (192.74.137.5) */               /* Joseph H. Allen */
int a[1817];main(z,p,q,r){for(p=80;q+p-80;p-=2*a[p])for(z=9;z--;)q=3&(r=time(0)
+r*57)/7,q=q?q-1?q-2?1-p%79?-1:0:p%79-77?1:0:p<1659?79:0:p>158?-79:0,q?!a[p+q*2
]?a[p+=a[p+=q]=q]=q:0:0;for(;q++-1817;)printf(q%79?"%c":"%c\n"," #"[!a[q-1]]);}
Article: 10482
Subject: PLA, BLIF, JED files
From: Valentin Serb <vserb@dnt.ro>
Date: Thu, 21 May 1998 23:51:48 -0700
Links: << >>  << T >>  << A >>
Hello everybody

	As a newbie on these newsgroups I decided to ask a question but only 
after I red them very carefully a few weeks. Now the question: Does anybody 
know where can I find a **REFERENCE** regarding the logic interchange format 
files (e.g. Berkeley .PLA files, .BLIF files) and the .JED files used for 
programming devices (not only PAL's but CPLD's too)?

	Thank you


Article: 10483
Subject: graphics processor
From: kevin.brand@gecm.com (kevin brand)
Date: 22 May 1998 08:51:44 GMT
Links: << >>  << T >>  << A >>

I am trying to implement a graphics processor in Vhdl,Are  there any good 
sites which will highlight common architectures etc. and perhaps even a 
more detailed description of simple cores etc.

Article: 10484
Subject: Re: Minimal ALU instruction set.
From: Andrew Veliath <andrewtv@usa.net>
Date: 22 May 1998 14:37:29 GMT
Links: << >>  << T >>  << A >>
 .........----------------==================----
..--==-  Fri, 22 May 1998 11:20:03 +1200,
..--==-  Steven Groom (SG) mentioned:

 SG)  Fitz wrote:

 )) Mike,
 )) 
 )) Please keep us informed on your progress.
 )) 
 )) Thanks, Fitz
 )) 
 )) Mike Butts wrote:
 )) 
 )) > In fact, a fine poster paper about a small FPGA-based processor
 )) > was given at FCCM last month.  I'm so inspired by it, and by the
 )) > new availability of cheap FPGAs and tools, that I'm developing >
 )) a little minimum-cost FPGA+SRAM board and a similar little >
 )) processor design for it.  I'm hoping to make it available by > the
 )) end of the year.  (This is strictly a personal hobby effort > on
 )) my part, for fun, experimental and educational purposes.)

 SG)  I have a board already constructed with an Altera part (10K10)
 SG)  for USD$200 PC104 bus, 512KEPROM 32KSRAM, RS232 LCD Display and
 SG)  manual.

 SG)  There will be ADC/DAC board options as well, in the next rev.

 SG)  I have written a micro (actually several) for this.  All public
 SG)  domain.  Currently writing CAN controller, UART and FIR/FFT
 SG)  functions (public domain when complete)

 SG)  The software to compile for this part is free from Altera
 SG)  (www.altera.com)

..
 ~~~~~~~~~================-------------......---

At Rensselaer last semester for a class project, I designed a small
8-bit register to register RISC processor with a stack and subroutines
and a healthy instruction set, and was able to test it on an XC4010E,
with 16 general purpose registers and a small LUT RAM and some
peripherals.  Total resource count was about 331 of the 400 CLBs.  It
seemed to be stable at a couple tens of MHz.

I had used the Cygnus GNU Win32 tools flex and bison to write an
assembler for it and wrote some bit manipulation to go directly from
assembler to the INIT= statements for the internal RAMS on the XC4010E
which Xilinx uses in its user constraint files.

--

        Andrew Veliath
	andrewtv@usa.net, veliaa@rpi.edu
Article: 10485
Subject: Re: FPGA-based CPUs (was Re: Minimal ALU instruction set)
From: Bernd Paysan <bernd.paysan@remove.muenchen.this.org.junk>
Date: Fri, 22 May 1998 21:57:52 +0200
Links: << >>  << T >>  << A >>
Lardino@ibm.net wrote:
> We were also responsible for developing a compiler/assembler
> and working applications (the wonder and amazement generated
> by Tetris running on a processor built by a group of college students
> is probably a whole lot more than any non-EE could realize:).

Tetris seems to be the standard application for this sort of experience.
About a year ago, I was involved in a FPGA based CPU approach, and my
part was compiler support (it was a Forth compiler, and I spent only a
weekend to do it). But the application to test it was Tetris.

-- 
Bernd Paysan
"Late answers are wrong answers!"
http://www.jwdt.com/~paysan/
Article: 10486
Subject: XC6200
From: mtmason@ix.netcom.com
Date: Sat, 23 May 1998 01:43:17 GMT
Links: << >>  << T >>  << A >>
Taken from EETimes

SAN JOSE, Calif.  Xilinx Inc. has stopped development work on its XC6200
line of partially reconfigurable field-programmable gate arrays (FPGAs), and
the founders of the the company's reconfigurable R&D group in Edinburgh,
Scotland, John Gray and Tom Kean, have both left the company. The remaining
engineering staff at Edinburgh has been reassigned to develop IP cores for
use by Xilinx's customers within the company's FPGAs.

However, Xilinx says it remains committed to the partial reconfigurability
offered by the XC6200 devices and will offer many of the features of the
XC6200 in its next-generation FPGA family, known as Virtex.

"John Gray is still working for Xilinx as a consultant," said Roland
Triffaux, manager of Xilinx Europe. "Tom Kean and another two engineers have
left to start a company in California."

That spin-off company, called Quicksilver, is believed to have some backing
from Xilinx. Quicksilver is looking to apply reconfigurable logic to
multiprotocol handsets for mobile communications, sources said. It is also
believed to be working with systems companies on the application of
reconfigurable logic.

Gray said his departure from the company was amicable. "I am just looking to
do something that's more fun again," he said. "It's time to kick back a bit."

Meanwhile, Kean is reacquiring the name Algotronix from Xilinx. Kean and Gray
led Algotronix Ltd. in the early 1990s before it was acquired by Xilinx in
1993, when it became the basis of the reconfigurable R&D group, and was named
Xilinx Development Corp.

Algotronix developed a reconfigurable FPGA architecture known as CAL
(configurable array logic), which eventually became the XC6200. Kean said the
new Algotronix would act as a consultancy and would advise users on the
application of reconfigurable logic.

Xilinx said the work of the R&D group was largely completed. "The goal of the
reconfigurable group in Edinburgh has been achieved," Triffaux said. XC6200
devices would continue to be available for academic and commercial research
groups, as they have been in the past. "We never really sold it," Triffaux
said.

Peter Cheung, a researcher in the department of electrical and electronic
engineering at Imperial College of Science and Technology (London), has used
XC6200 devices for reconfigurable hardware platforms. "We've heard they are
not developing the XC6200," Cheung said. "Unless there is a real commitment
to it we may have to look at other things. The tools for 6200 are primitive
and not well done.

"In many ways it [the XC6200] was a product ahead of its time," he said. "It
was a beautifully conceived device but not sufficiently well supported."


-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading
Article: 10487
(removed)


Article: 10488
Subject: Evolutionary FPGAs
From: "Vo Cun To (-)" <vto@polymail.calpoly.edu>
Date: Sat, 23 May 1998 01:57:15 -0700
Links: << >>  << T >>  << A >>

On the June issue of Discover Magazine, the cover story reports of two
computer scientists, Inman Harvey and Adrian Thompson, applying evolution
to program a Xilinx XC6216ES. 

One of the things they did was evolved the FGPA to distinguish a 1KHz
input from a 10KHz input. This was done with using only 100 CLBs of the
XC6216, and no clock input. After two weeks and 5000 generations, the chip
evolved to work great --with one drawback, it's not robust. If programmed
to different chips or if input/output pins were reassigned, it wouldn't
work. Also, slightly temperature variations have big effects.

Is there anybody that applied Thompson's method of programming a FPGA? 

More information can be found at http://www.discover.com

Article: 10489
Subject: Re: Xilinx Foundation Student Edition
From: APS <resp@associatedpro.com>
Date: Sat, 23 May 1998 07:14:53 -0400
Links: << >>  << T >>  << A >>
For the XILINX student edition stuff cherck at
http://www.associatedpro.com. We now carry both the boards and books. You
can also get great deals on the Ful;l Foundation kits with the X84 FPGA
board. That includes the LATEST XILINX FOUNDATION kits and support.



Ewan D. Milne wrote:

> Back in April, Peter Alfke wrote:
> >Prentice Hall publishes and sells the "Xilinx Foundation Student
> >Edition" which supports all XC9500 devices and all XC4000-type devices
> >up to 8k gates today, 10k gates in the next upgrade.
> >...
>
> The book and CD appear to be available from amazon.com and also
> at Quantum Books in Cambridge, MA.  I am considering purchasing
> this book and software, but was wondering when the "next upgrade"
> would be available and how one would be able to tell which version
> of the software was included.  I would like to have the capability
> to target a design for the XC4010E device, but would be more inclined
> to go ahead and purchase the current version if the next upgrade
> will not take place within the next year.
>
> Can anyone (Peter?) shed any light on this subject?
>
> Thank you in advance for your replies.
>
> -Ewan



--
__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/

Richard Schwarz, President              EDA & Engineering Tools
Associated Professional Systems (APS)   http://www.associatedpro.com
3003 Latrobe Court                      richard@associatedpro.com
Abingdon, Maryland 21009
Phone: 410.569.5897                     Fax:410.661.2760

__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/


Article: 10490
Subject: Announce: X208 Board Ram pattern Generator VHDL Example
From: APS <resp@associatedpro.com>
Date: Sat, 23 May 1998 07:34:45 -0400
Links: << >>  << T >>  << A >>
The APS -X208 FPGA board is now available with example code which turns
it into a 1/2 meg bit random pattern generator. Using the on board DDS
the bit frequency is programmable to 30 Mhz. The example allows the
board to load bits  from a text file. A sync pulse is generated which
can be offset to allow bit viewing at any offset easily on a standard
scope.

The program example can be used for Pseudo Noise sequence generation,
bit stream simulation etc.

The examples are done in VHDL (FPGAbehavior) ) and C (FPGA control)  and
will come free on the APS CD included with every APS board. The CD
includes schematics and tons of VHDL Examples. The code will also be
made available to current customers via the web upon request.

The X208 board is available with Foundation VHDL software and can be
seen at:

http://www.associatedpro.com/x208/prod/x208_product.html

Other free FPGA/VHDL examples and tips are available on the APS web
including:

APS VHDL Tutorial     http://www.associatedpro.com/aps/x84lab/
APS Technical Solutions Page
http://www.associatedpro.com/Tech_Solutions/aps_tech.html
APS Support Page   http://www.associatedpro.com/aps_sup.html



--
__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/

Richard Schwarz, President              EDA & Engineering Tools
Associated Professional Systems (APS)   http://www.associatedpro.com
3003 Latrobe Court                      richard@associatedpro.com
Abingdon, Maryland 21009
Phone: 410.569.5897                     Fax:410.661.2760

__/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/ __/


Article: 10491
Subject: fpga, pld video interface
From: "Dj" <dejan@versilia.toscana.it>
Date: Sat, 23 May 1998 18:17:25 +0200
Links: << >>  << T >>  << A >>
Does anybody can help me..

I try to implement on my 80186 based board simple video interface.

I try with mc6845 core but I need more than 16 colors 320x200.


Thanks :)

Dejan



Article: 10492
Subject: -SEE SNEAK LOCKER ROOM PIX OF GYMNASTS FREE 95695
From: MelissaGilbert@Nude.Here
Date: Saturday, 23 May 1998 20:52:30 -0600
Links: << >>  << T >>  << A >>
http://members.coolnet.net/~tyler12

    =======ALL FOR FREE=======

YOU HEVER TO SEE THIS TO BELIEVE IT.
4 HOT COLLEGE GIRLS LIVING IN A HOUSE
WITH CANS IN EVERY ROOM INCLUDING
THE SHOWER. THE CAMS ARE ON 24 HOURS
AND THERE IS ALSO A CHAT ROOM SO YOU
CAN TALK TO THEM TOO...

http://members.coolnet.net/~tyler12
http://members.coolnet.net/~tyler12


2k0q

Article: 10493
Subject: Re: FPGA-based CPUs (was Re: Minimal ALU instruction set)
From: Thomas Womack <mert0236@sable.ox.ac.uk>
Date: 24 May 1998 16:23:26 GMT
Links: << >>  << T >>  << A >>
In comp.arch Bernd Paysan <bernd.paysan@remove.muenchen.this.org.junk> wrote:
: Lardino@ibm.net wrote:
: > We were also responsible for developing a compiler/assembler
: > and working applications (the wonder and amazement generated
: > by Tetris running on a processor built by a group of college students
: > is probably a whole lot more than any non-EE could realize:).

Yep; we had that at the Computation open day at Oxford yesterday, though
I think their Tetris was directly hardware-compiled rather than running
on an implemented microprocessor - and the board used was RAMless, which
was pretty impressive.

I was more impressed by 60fps interactive 800x600 Life - which is
non-trivial, and probably not even possible, on Standard Commercial
Hardware (OK, you can use cell lists rather than an array, but the
FPGA approach ran at constant speed independent of population). I
think the 64k of SSRAM attached to the FPGA helped.

Tom
Article: 10494
Subject: [++] Fast Life code (Was:Re: FPGA-based CPUs (was Re: Minimal ALU instruction set))
From: Terje Mathisen <Terje.Mathisen@hda.hydro.com>
Date: Sun, 24 May 1998 21:07:22 +0200
Links: << >>  << T >>  << A >>
Thomas Womack wrote:
> 
> In comp.arch Bernd Paysan <bernd.paysan@remove.muenchen.this.org.junk> wrote:
> : Lardino@ibm.net wrote:
> : > We were also responsible for developing a compiler/assembler
> : > and working applications (the wonder and amazement generated
> : > by Tetris running on a processor built by a group of college students
> : > is probably a whole lot more than any non-EE could realize:).
> 
> Yep; we had that at the Computation open day at Oxford yesterday, though
> I think their Tetris was directly hardware-compiled rather than running
> on an implemented microprocessor - and the board used was RAMless, which
> was pretty impressive.
> 
> I was more impressed by 60fps interactive 800x600 Life - which is
> non-trivial, and probably not even possible, on Standard Commercial
> Hardware (OK, you can use cell lists rather than an array, but the
> FPGA approach ran at constant speed independent of population). I
> think the 64k of SSRAM attached to the FPGA helped.

Mike Abrash used Conway's Game of Life as the target for his second
'Annual Code Optimization Challenge', a few years ago.

The target machine was either a 486 or very early Pentium, i.e. current
hardware is quite a bit faster:

The two joint winners both achieved twice the performance of my entry,
which is still capable of 400 fps in 320x200 resolution on a Pentium
MMX.

With linear scaling, this corresponds to 53 fps in 800x600, and more
than 100 fps for the two winners. On a modern PII even my program would
handle 60fps without breaking a sweat. :-)

One of the winners, David Stafford, used a cell list to reduce the
working set, while the other, a german guy (sorry, I don't remember his
name), used logic operations on 32-bit registers to implement the Life
counting logic directly.

This also means that the german entry probably implemented very nearly
the same algorithm as the FPGA setup above. :-)

Terje

-- 
- <Terje.Mathisen@hda.hydro.com>
Using self-discipline, see http://www.eiffel.com/discipline
"almost all programming can be viewed as an exercise in caching"
Article: 10495
Subject: Partitioning an a large design in Altera's Max+Plus II
From: "Robert L. Hamilton" <rlham1@airmail.net>
Date: Sun, 24 May 1998 16:56:09 -0500
Links: << >>  << T >>  << A >>
Our internal customer desires to use an emulation of our current ASIC
design implemented in programmable logic.  In order to keep the same
hierarchy, assure fidelity, reduce engineering time, ... for the ASIC
and the CPLD implementations, I wish perform partitioning of the design
within the Max+Plus II tool.  (We're using Synopsys (VHDL) and passing
Max+Plus II hierachical EDIF.) I'm in contact with Altera design
support, but I'd like independent verification that the partitioner is
usable. Does anyone have any experience (good or bad) with the tool? 

Robert Hamilton
Article: 10496
Subject: FIRST CALL FOR PAPERS: 1998 Military and Aerospace Applications of Programmable
From: rk <stellare@erols.NOSPAM.com>
Date: Sun, 24 May 1998 23:00:08 -0400
Links: << >>  << T >>  << A >>
                              FIRST CALL FOR PAPERS

                   1998 Military and Aerospace Applications of

                Programmable Devices and Technologies Conference


The conference will address devices, technologies, usage, reliability,
fault tolerance, radiation susceptibility, and applications of
programmable devices and adaptive computing systems in military and
aerospace systems.  The technical program will consist of oral
technical presentations as well as industrial exhibits.  The conference
will be held at the NASA Goddard Space Flight Center in Greenbelt,
Maryland.

Presentations are being solicited in all aspects of the use of
programmable elements, devices, and applications for military and
aerospace applications.  These include: PALs, FPGAs, PROMs, Programmable

Substrates, FPIC, Programmable Analog Circuits, adaptive computing
systems and related technologies.  All presentations are expected to be
technical in nature.  Each presenter will be given approximately 20
minutes with a five minute question and answer session immediately
following.  Submitted abstracts should be approximately 100 words long
and are due by July 3, 1998.  Please include your full correspondence
address including email and fax and a note on your preferred
presentation format.

Notification of acceptance will be posted by 10 July 1998.  Conference
proceedings will published and distributed to all attendees.  At least
one author must be registered for the conference.  Electronic
submissions
are strongly encouraged.

Topics include (but are not limited to) the following:

    - Programmable Technologies and State-of-the-Art Devices
        A. COTS and MIL/AERO
        B. New Technology Development
        C. Adaptive Computing Systems

    - Radiation Effects, Device Reliability and Element Characteristics

    - Device Architecture, Performance, and Capabilities

    - Applications and Novel Techniques for Military and Spaceflight
Circuits.
        A. Signal Processing
        B. High-Speed Designs
        C. System Impact of State-of-the-Art Technologies
        D. Reconfigurable Processing
        E. Low Power Designs
        F. Advanced Packaging

    - Use of COTS Devices in the Military and Spaceflight Environment
        A. PEMS
        B. Shielding, Latchup Protection
        C. SEE including SEFI and Destructive Effects
        D. System Protection

    - Testing and Analysis Techniques

    - Performance Results

    - Use of Programmables in Critical Systems

    - Software Tools for Design/Analysis
        A. Synthesis
        B. Macro Generators
        C. Timing Analysis and Simulation
        D. Redundancy, Fault Tolerance, and SEU-Hardening


PRELIMINARY CONFERENCE SCHEDULE

September 14    NASA/GSFC Tours
                Evening Registration, Happy Hour

September 15    Invited Talk by Janet Barth, Radiation Physics Office
                Day 1 Technical Sessions and Industrial Exhibits
                Dinner

September 16    Invited Talk (to be announced)
                Day 2 Technical Sessions and Industrial Exhibits

September 17    NASA/GSFC Tours


Janet Barth will present: "An Overview of the Radiation
                           Environment for Spaceflight Electronics"




The conference is sponsored by (preliminary list): NASA/GSFC,
JHU/Applied Physics Laboratory, and the NASA Radiation
Effects Program.


For more information on the conference please see
http://rk.gsfc.nasa.gov/richcontent/Ksymposium/kSymposium.htm



Please Send Abstract Submissions to:

Martha O'Bryan
martha.obryan@gsfc.nasa.gov
Tel: (301) 286-1412
Fax: (301) 286-0220


Local Arrangements:

Janet Jew
janet.jew@gsfc.nasa.gov
Tel: (301) 286-5533
     (301) 286-8884


Conference Chair:

Richard Katz
NASA Goddard Space Flight Center
rich.katz@gsfc.nasa.gov
Tel: (301) 286-9705

Conference Co-Chairs:

Alan W. Hunsberger
NSA
awhunsb@alpha.ncsc.mil
Tel: (301) 688-0292

Ann Darrin
Johns Hopkins University/Applied Physics Laboratory
ann.darrin@jhuapl.edu
Tel: (240) 228-4952


Article: 10497
Subject: Re: Partitioning an a large design in Altera's Max+Plus II
From: Georg Diebel <g_diebel@lis.e-technik.tu-muenchen.de>
Date: Mon, 25 May 1998 09:55:57 +0200
Links: << >>  << T >>  << A >>
Robert L. Hamilton wrote:
<snip>
> and the CPLD implementations, I wish perform partitioning of the design
> within the Max+Plus II tool.  (We're using Synopsys (VHDL) and passing
> Max+Plus II hierachical EDIF.) I'm in contact with Altera design
> support, but I'd like independent verification that the partitioner is
> usable. Does anyone have any experience (good or bad) with the tool?
> 
> Robert Hamilton

Hi Robert,

some time ago, I tried the same thing... we wanted to have a prototype
of our ASIC using 5 Altera 10K50. I think I was using Max+PLUS II V7.2 
and 8.0, and in my opinion the partitioner inside these versions suc..
ehrrm, isn't very good.
My problem: of course we had to design our PCBs for the 10K50's before
we had the design ready, so  the connections between the chips were 
fixed. I tried to find a way to tell the partitioner "these are the 
chips, these are the connections, go and do your job" - but I (as well
as Altera support) couldn't find a way save one: to build the connec-
tions between the chips with the graphic editor and to assign internal 
signals to the I/O-pins - and that's manual partitioning, no need for 
a partitioner any more.

I don't know if the partitioner of the current version was improved,
but I don't think so: the trend is to use huge FPGAs if necessary.
And don't forget the penalty for leaving chip #1 and entering chip #2:
unless you have a very moderate clock rate, you won't be able to have
any logic on this line. 

Well, I wish you luck for your prototype - better luck than we had:
we had to rework our VHDL, and to do manual partitioning, but finally 
we had it up and running (must have been 1-2 years ago).

Just my 2c ...

Georg

-- 
All opinions expressed are mine, not my employers'.

######  #   #   #    #  ###  Georg Diebel
  #  #  ## ##   #    # #     Institute for Integrated Circuits
  #  #  # # #   #    #  ##   Technical University of Munich
  #  #  #   #   #    #    #  g_diebel@lis.e-technik.tu-muenchen.de
  #  ####   #   #### # ###   Phone 0049-89-289-28578
Article: 10498
Subject: Re: [++] Fast Life code (Was:Re: FPGA-based CPUs (was Re: Minimal ALU
From: Terje Mathisen <Terje.Mathisen@hda.hydro.com>
Date: Mon, 25 May 1998 10:30:53 +0200
Links: << >>  << T >>  << A >>
Thomas Womack wrote:
> 
> In article <35686FEA.6266@hda.hydro.com> you wrote:
> 
> >I wrote
> 
> : > I was more impressed by 60fps interactive 800x600 Life - which is
> : > non-trivial, and probably not even possible, on Standard Commercial
> : > Hardware (OK, you can use cell lists rather than an array, but the
> : > FPGA approach ran at constant speed independent of population). I
> : > think the 64k of SSRAM attached to the FPGA helped.
> 
> : Mike Abrash used Conway's Game of Life as the target for his second
> : 'Annual Code Optimization Challenge', a few years ago.
> 
> : The target machine was either a 486 or very early Pentium, i.e. current
> : hardware is quite a bit faster:
> 
> : The two joint winners both achieved twice the performance of my entry,
> : which is still capable of 400 fps in 320x200 resolution on a Pentium
> : MMX.
> 
> That seems very interesting; would it be worth putting you in touch with
> the appropriate bit of the Hardware Compilation Group?

Sure, no problem.

> Have you a better reference for this than chapters 17 and 18 of the
> Big Black Book from Abrash? He discusses Stafford's solution in detail,
> ignoring Peter Klerings' one altogether, which seems a pity since Klerings'
> one sounds much closer suited to the FPGA.

Klering's code was actually fairly straightforward, except for a set of
flags used to detect static areas.

Skipping that part would still (most probably) let it run at the
required 60 fps, the code is 'just' a parallel implementation of the
counting logic:

Alive next iteration = (alive now AND (count == 2 OR count == 3)) OR
                       (not alive AND count == 3),

which simplifies to just:

Alive next iteration = (count == 3) OR (alive AND count == 2).

By including the cell itself in the count, then it becomes easier to
reuse the counting logic for multiple rows:

 alive = (iCount == 3) OR (alive AND iCount == 4)

You need 4 bits to count to 8 (or 9), so 4 registers for counting plus
one for the center cells leaves one or two registers for array
addressing on an x86.

Klering did a lot of work to simplify the logic as much as possible,
i.e. he didn't actually implement the full 'count-to-9' bitwise logic,
since it is possible to early-out many of the branches.

Implementing the same logic with MMX-style wide registers should make it
approximately twice as fast.

Terje

-- 
- <Terje.Mathisen@hda.hydro.com>
Using self-discipline, see http://www.eiffel.com/discipline
"almost all programming can be viewed as an exercise in caching"

Article: 10499
Subject: Problem with loading XC4000E configuration from 8051
From: Alexander Sherstuk <Sherstuk@amsd.com>
Date: Mon, 25 May 1998 14:13:25 +0400
Links: << >>  << T >>  << A >>
Hi All,

  I encountered unexpected difficulty, when loading XILINX XC4005E
configuration from ATMEL AT89C52 (in serial slave mode).
  I connected P1.5 pin to configuration clock CCLK pin of XILINX chip,
and connected P1.0 pin to DIN pin of XILINX chip.
  XILINX configuration is loaded, but not with 100% probability -
sometimes (1 attempt of 5) it fails.
  It looks like the problem is with 8051 signals rise time.
When I fed CCLK through 74HC14, everything works fine.
Maybe, somebody knows more about this problem.
How to avoid it?

Thanks,
   Alex Sherstuk
     Sherstuk@amsd.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