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 1075

Article: 1075
Subject: Re: Bugs in Xilinx Prog.Logic Databook -- how to report?
From: tb@xanadu.vingmed.no (Torbjoern Bakke)
Date: 25 Apr 1995 09:19:34 GMT
Links: << >>  << T >>  << A >>

Xilinx has a Web-site at: 

	http://www.xilinx.com/

Somewhere inside their web-pages there is a reference to a support
email-address (I haven't tried it though):

	hotline@xilinx.com

--Torbjørn--

--
----
Torbjorn Bakke (tb@vingmed.no) Tel: +47 33 04 21 32


Article: 1076
Subject: Re: Sunrise ???
From: ast@actcom.co.il (Gideon Amir)
Date: Tue, 25 Apr 1995 13:00:34 GMT
Links: << >>  << T >>  << A >>
In article <D7KLvG.5t4@nntpa.cb.att.com>, pn@anuxt.mv.att.com says...
>
>In article <3ngr7o$5fb@news.uncc.edu>,
>Terry E. Koontz <tkoontz@uncc.edu> wrote:
>>
>>Viewlogic recently add ViewTest to their Powerview product line.
>>
>>ViewTest is based on Sunrise 2.0 software (their words)
>>
>>Does anyone have an address and telephone number for Sunrise.
>>
>>Terry Koontz
>>UNCC
>>
>>tkoontz@uncc.edu
>
>
>If I am not mistaken, ViewLogic has not continued with the addition
>of ViewTest. I was initially making it part of their package,
>however, there were many difficulties with the intergration. So
>they were moving back to the stand alone Sunrise package.
>(ViewLogic purchased Sunrise as it did many other co.s)
>
>Tony
>
Sunrise recently moved in with Viewlogic at their Fremont Ca. plant.
The number there is (510) 440-1000.  You can also try Jean Pierre Braun 
at:  jp@srtest.com

The current TESTGEN rev. is 2.1

Good luck

Gideon





Article: 1077
Subject: Re: Lattice low-cost start kit
From: kugel@mp-sun6.informatik.uni-mannheim.de (Andreas Kugel)
Date: 25 Apr 1995 13:51:49 GMT
Links: << >>  << T >>  << A >>
In article pii@distort.demon.co.uk, herb@case373.demon (Herbert Larbie) writes:
>23/4/95
>
>I thunking about checking out the Lattice low-cost starter kit isp-sk, could
>anyone who has been using it relay their experience to me please.
>
>Thanks in advance.
>
>Herbert Larbie
>

I bought that kit some weeks ago. Although I didn't run a design yet, I implemented
and compiled two designs to see how it works.
It supports isp devices 1016 and 2032 which are not very large (sigh..).
1016 is similar to AMDs MACH210 but has better routing capabilities.
Don't try isp*** if you have different tristate signals in your design, it has no
individual output enables, only blockwise.
The software does not optimize your partitioning.... it will take some time to get 
used to it.
I reached > 90% useage after careful manual optimization, quite nice.
The package comes with complete documentation including download code, which
covers the ispGAL22v10 also.

If you plan a real design, better consider the 1016 to have only 64 registers 
not the 96 mentioned in the manual. The 32 IO registers are of limited use only.

At the time it's the only package for real low-cost designs, for you don't need a
programmer for these chips.


aku
---


--------------------------------------------------------
Andreas Kugel                
Chair for Computer Science V      Phone:(49)621-292-5755
University of Mannheim            Fax:(49)621-292-5756
A5
D-68131 Mannheim
Germany
e-mail:kugel@mp-sun1.informatik.uni-mannheim.de
--------------------------------------------------------



Article: 1078
Subject: Re: BLIF to XNF translator
From: ersheido@cats.ucsc.edu (Ali H Ersheid)
Date: 25 Apr 1995 15:11:10 GMT
Links: << >>  << T >>  << A >>

In article <3mt85v$kru@news.csie.nctu.edu.tw>,
Chuang Hsien-Ho <eea80593@maddux.EE.NCTU.edu.tw> wrote:
>Hi:
>
>I want translate SIS's output(blif/slif/eqn...) to XNF format.
>Does anyone has the translator? Or any other sugesstion to me?
>Thanks a lot!
>
>--
>===============================
>Hsien-Ho Chuang  
>eea80593@yankees.ee.nctu.edu.tw
>===============================

There is a program that was developed at UCSC a while back.  I have to
ask around if it's shareware or not.  I'll let you know.


-- 
==============================================================================
To accomplish great things, we must not only act but also dream, not only
plan but also believe.
			Anatole France


Article: 1079
Subject: How much performance is enough?
From: wirthlim@fpga.ee.byu.edu (Michael J. Wirthlin)
Date: 25 Apr 1995 09:47:54 -0600
Links: << >>  << T >>  << A >>

I just returned from the FCCM conference (FPGAs for Custom Computing
Machines) and was interested in the many different opinions on the
future and direction of custom computing machines (ccms or fccms). I
noticed a wide spectrum of opinion on the question "How much
performance is necessary to justify a FPGA based custom computing
machine?". I would like to paraphrase a number of opinions I heard
from the various attendees. I would appreciate any feedback anyone has
on the same question.


Opinion #1: 

Any speed-up over workstations/PC's using FPGA based computing
machines is sufficient to justify the machine. Don't apologize for
speedups of only 20x. Hardware algorithms that speed up an inner loop
by factors of 2-20 are satisfactory.

Opinion #2:

Only substantial performance improvements can justify the use of an
FPGA based computing machine. FPGAs operate at a clock rate 10x slower
than the processors used today (so they need at least 10x in
parallelism to make up the difference). In addition, FPGA systems can
cost 10x more than the host it was intended to replace (need another
10x to justify the cost). If you consider the extra development time,
the margins are even greater. These systems must provide 100-1000x
speed-up to justify their use.

Opinion #3:

Performance/cost is the metric, not just brute performance. If I can
replace a $10 million of performance X with an FPGA based machine
costing $10 thousand of performance 1/5 X, my performance/cost has
improved by a factor of 200. If I can achieve substantial
performance/cost benefits (100-1000x), I can justify the machine.

Any other opinions?
-- 
Michael J. Wirthlin
Brigham Young University - Electrical Engineering Department
Reconfigurable Logic Laboratory (801) 378-7206


Article: 1080
Subject: Need help about conference chip
From: fengwct@ku.ac.th (Wichai Tang)
Date: 25 Apr 1995 15:55:26 GMT
Links: << >>  << T >>  << A >>
Hi,
	I must design a digital conference chip. But I have no idea about 
this chip.  This chip use to mix several PCM signals together. I know that,
the PCM code is nonlinear. So the first step to do a conference is to 
convert this nonlinear PCM code into linear digital signal system and sum the
whole time slot that need to join a conference together. Then convert 
this signal back to PCM code and put it back to the specific time slot. I 
will use FPGA as my target to implement this design. Could anyone suggest me 
in the following questions ?
	1. I know that PCM is nonlinear digital system so we must convert 
it back to linear digital signal before we can do anything on it. From 8 
bits of PCM code must be convert to 14 bits of linear digital signal. The 
easy solution is to use ROM to convert this signal, but I would like to 
know that is there any other way to do this job ? And dose it appropriate 
for targetting at FPGA ?
	2. When we have many time slot want to join a conference, we must 
attenuate all signal before summing it up. so the question is how can I 
attenuate this linear digital signal ? I think this should not as easy as 
just divide it with number of conferece time slot. Am I correct ?
	Any comment or pointer is very appreciate. Thank you for your concern.

Best Regards,
Wichai Tang


Article: 1081
Subject: Re: fpga design advantages
From: petersr@fpga.ee.byu.edu (Russell Petersen)
Date: 25 Apr 1995 11:18:48 -0600
Links: << >>  << T >>  << A >>

In article <3ngggb$4um@mother.usf.edu>, harihara@sunburn.eng.usf.edu (Shankar Hariharan (EE)) writes:
|> 	hi,
|> 		I am doing a research on the studying the advantages of implementing application specfic archtectures on FPGA's. I would like to hear ur view point on this issue. Any pointers in this direction would be greatly appreciated. 
|> 
|> 	I promise to  post the summary of the reply after compilation.
|> 
|> H.Shankar.
|> -- 
|

	I can give some advantages that I feel are important.  This also
could be seen as an answer to Mike's question about how much performance
is enough.
	The big advantage I see for FPGAs is the ability to amortize the
silicon costs over several problems/architectures.  Thus, in my area of
interest (DSP), one might decide to build a coprocessor type of board based
on reconfigurable logic that can be used for several different types of
high-speed signal processing.  For example, the board may at one time be used
to filter some images at one point and a few minutes later be used to perform a
high speed FFT.  (I pick these two examples since I am most familiar with them. 
I have implemented both on FPGAs and obtained good performance) Thus, the same
hardware can be used to achieve good performance on two separate operations.
I can justify the hardware and even the development cost since the only way
I could get similar performance would be to use ASICs.  Thus, for me the
justification is simply if I can get the performance I desire using FPGAs and I
cannot get it using programmable processors (such as DSP processors), then 
the only choice I have is the FPGA or the ASIC.  I choose the FPGA since I for
most operations cannot justify the ASIC.  Also, the ASIC implementation would
require too many chips to perform all of the operations I desire to implement and
hence it is too costly and complex.

	The other advantage I did not mention that is often cited as a great
advantage to a reconfigurable approach is that of dynamic reconfiguration or the
ability to reconfigure a system while it is operating.  This is usefull since one
can implement at one time exactly what is needed and nothing more.  This is
usefull in many cases.  As an example, for the FFT a basic butterfly element can
be reused over multiple columns of the FFT if one desires.  The only thing that
must change from one column to the next is the addressing.  ASICs have been built
that have this capability.  They allow flexible addressing to handle the data
appropriately in each column of the FFT for which that butterfly element is being
used.  I submit that an advantage an FPGA implementation could have here is the
ability to simply reconfigure the necessary addressing logic.  Thus, the logic
would be smaller and due to its special-purpose nature it could be faster.  In
this case it may not really be a critical path anyways so the point may be a
minor one but it does bring up what I think is a major advantage of the FPGA.
That is, even though they are slower than ASICs in straight comparisons,
implementations using them can still be faster since one can make them more
special purpose through the use of reconfiguration.

	Now for the problems.  Right now I think FPGA systems are very usefull in
many areas as a reconfigurable computing resource but they are also many
problems.  In particular, development time is way to costly.  In my latest
implementation of an FPGA based system the VHDL code for the system took me about
4 hours to write. (It was even correct! :)).  However, due to the difficulty in
seeing what was happening with the FPGA board I was using it took me several days
to write the software interface and get it working correctly.  Thus, most of the
time was spent on debug of something other than the actual hardware.  This is of
course typical of many systems of this type but it could be different.  I think
that before reconfigurable logic can become truly mainstream much, much work must
be done on the tool side.  Ideally a reconfigurable resource could be debugged as
if it was software (somewhat like what is available on Splash-2).  This ability
would greatly cut development costs.  Also, I believe that the devices could
still use more routing.  I read with great interest about HP's new Teramac
computer that is a routing rich device.  It really is frustrating when one is
using a FPGA/PLD device and finds that what should be easily implemented in the
logic (30% utilization) cannot be due to routing constraints imposed by the
pinouts.  This in my opinion should be something a user should have to consider
until one reaches at least 90% utilization.

	As a final note, I believe that reconfigurable logic is becoming ever
more usefull.  There are some exciting devices coming out from several
manufacturers that I see as greatly improving the utility of the reconfigurable
device (especially in my interest area of DSP!).

- Just my thoughts

Russell Petersen
petersr@fpga.ee.byu.edu
Brigham Young University Reconfigurable Logic Laboratory


Article: 1082
Subject: Re: Sunrise ???
From: wolf@aur.alcatel.com (William J. Wolf)
Date: 25 Apr 1995 18:55:36 GMT
Links: << >>  << T >>  << A >>
> Does anyone have an address and telephone number for Sunrise.

>From their web page, 510-440-1000

http://www.viewlogic.com/sunrise/srover.html

or just start from   http://www.viewlogic.com/  and go down.


---
- Bill Wolf, Raleigh NC
- My opinions, NOT my employer's




Article: 1083
Subject: Altera new FLEX 10000 - a worlds first
From: idr@ee.ed.ac.uk (Iain Rankin)
Date: Wed, 26 Apr 1995 08:03:50 GMT
Links: << >>  << T >>  << A >>
Does anyone have any technical details about the forthcoming
FLEX10000 device.

I have heard that the device contains a logic array (similar to the 
FLEX 8000 device) and an embedded hardwired array which contains
useful mega-functions; for example:

	SRAM, ROM, FIFO's and dual-port SRAM
	Multipliers
	ALU's
	Sequencers
	8086, Z80, 6502
	FIR filters, convolvers, wavegenerators

Surely something like this will blow other FPGA's out of the scene.

Iain ;)

idr@ee.ed.ac.uk
http://www.ee.ed.ac.uk/~idr


Article: 1084
Subject: Compression algo's for FPGA's
From: idr@ee.ed.ac.uk (Iain Rankin)
Date: Wed, 26 Apr 1995 08:11:00 GMT
Links: << >>  << T >>  << A >>
Hi,

Does anyone know of, or can anyone suggest, a simple compression (s/w PC) and
decompression (Altera FLEX8000 devices) algo that i could use. 

I am interested in coding byte streams to increase my through put to
an FPGA interface (RIPP10 - supplied by Altera) from a PC.

Cheers,

	Iain

idr@ee.ed.ac.uk
http://www.ee.ed.ac.uk/~idr


Article: 1085
Subject: Re: How much performance is enough?
From: wolf@aur.alcatel.com (William J. Wolf)
Date: 26 Apr 1995 16:01:07 GMT
Links: << >>  << T >>  << A >>
> "How much performance is necessary to justify a FPGA 
> based custom computing machine?"

I think 10-20X performance speed-up, for a few thousand, 
(by focusing on the inner loop of algorithms) is marketable.
The key is support for multiple applications.

Anyone remember the Futurenet simulation card?  Cranked up 
simulation for about $10K.  The problem with this, and most 
other accelerators is that they serve(d) a single application. 

On the other hand, if it could speed up logic simulation, 
transistor level simulation, placement and routing, etc. this 
would be more interesting (and justifiable) for engineering 
applications.  If you expect non-engineers to purchase a 
hardware accelerator, likewise it had better have multiple 
applications.  Otherwise people will spend their $$ on a 
faster computer that speeds up everything they do.

So really, I think that alliances with software vendors is 
the key to make end users bite.  Software vendors in turn will want 
standards so that if one hardware vendor flops they protect their 
investment.  Making a card with a few FPGAs on it seems trivial 
compared to this effort. Perhaps Microsoft will "bless" us with 
a standard for PC accelerators as a way to compete with Unix?
They are trying to get EDA vendors to switch.

In the engineering world, if you can't get Xilinx or Altera to 
speed up FPGA place & route with such an engine, I doubt you 
will convince anyone else to go first. Engineering applications 
have two other complications, 1) the need to support both PC & 
Unix and 2) software vendors may prefer to develop software that 
uses multiple processors on a network instead. 

Multi-media is also an interesting possibility.  Multiple 
applications that could use more horsepower, and just PCs 
to worry about.  Of course the price would need to be low.

Good luck!

---
- Bill Wolf, Raleigh NC
- My opinions, NOT my employer's




Article: 1086
Subject: Market Position of FPGA/CPLD suppliers
From: Ruchir Puri <rpuri@bnr.ca>
Date: 26 Apr 1995 18:17:19 GMT
Links: << >>  << T >>  << A >>
Hi there!

I was wondering if I can get the data regarding the
market position (Annual Revenue) of major FPGA/CPLD
suppliers like : Xilinx, Altera, Actel, Lattice etc.
Thought this will be the most appropriate group.

Thanks...

Best Regards,
--Ruchir


Article: 1087
Subject: Re: Altera new FLEX 10000 - a worlds first
From: Russell Petersen <petersr>
Date: 26 Apr 1995 18:29:08 GMT
Links: << >>  << T >>  << A >>
This is a multi-part message in MIME format.

---------------------------------195971426010166
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=iso-8859-1

idr@ee.ed.ac.uk (Iain Rankin) wrote:
>Does anyone have any technical details about the forthcoming
>FLEX10000 device.
>

Try looking at Altera's web site. They have some info. about the 10k 
series part there. 


http://www.altera.com/

---------------------------------195971426010166
Content-Transfer-Encoding: 7bit
Content-Type: 
---------------------------------195971426010166--


Article: 1088
Subject: Re: Altera new FLEX 10000 - a worlds first
From: Russell Petersen <petersr>
Date: 26 Apr 1995 18:32:40 GMT
Links: << >>  << T >>  << A >>
idr@ee.ed.ac.uk (Iain Rankin) wrote:
>Does anyone have any technical details about the forthcoming
>FLEX10000 device.
>

Look at Altera's web site.  They have info. on the 10k series part
there.

http://www.altera.com/

 ___________________________            ----- 
   Russell Petersen                   || BYU ||
   Brigham Young University           ||     || 
   Reconfigurable Hardware Lab          ----- 
   petersr@fpga.ee.byu.edu            ---------  
   voice: (801) 378-7206             |  -    - | 
                                      ---------
  WEB: http://splish.ee.byu.edu/~petersr/russ.html
  ------------------------------------------------



Article: 1089
Subject: Re: Altera new FLEX 10000 - a worlds first
From: hutch@timp.byu.edu (Brad Hutchings)
Date: 26 Apr 95 12:24:52
Links: << >>  << T >>  << A >>
>>>>> On Wed, 26 Apr 1995 08:03:50 GMT, idr@ee.ed.ac.uk (Iain Rankin) said:

I> Does anyone have any technical details about the forthcoming
I> FLEX10000 device.

I> I have heard that the device contains a logic array (similar to the 
I> FLEX 8000 device) and an embedded hardwired array which contains
I> useful mega-functions; for example:

I> 	SRAM, ROM, FIFO's and dual-port SRAM

The "hard-wired" array is a RAM that can be configured to be any of a:
258x8, 512x4, 1024x2, 2048x1. So, it is a very flexible memory but
it is not an 8086 or Z80, etc. The idea of the megafunction is
that you can use the RAM as lookup tables, etc.

I> 	Multipliers
I> 	ALU's
I> 	Sequencers
I> 	8086, Z80, 6502
I> 	FIR filters, convolvers, wavegenerators

The rest of this stuff is library modules that are being developed
that will likely consume most of the FPGA. As I understand it,
the 8086 was one the designs that was used to test the design.

The 10k part is a very interesting part. The availability of
reasonably sized memories *on-chip* will make for much more
interesting designs.


--
        Brad L. Hutchings (801) 378-2667          Assistant Professor
Brigham Young University - Electrical Eng. Dept. - 459 CB - Provo, UT 84602
                       Reconfigurable Logic Laboratory



Article: 1090
Subject: Re: Lattice low-cost start kit
From: pngai@mv.us.adobe.com (Phil Ngai)
Date: Wed, 26 Apr 1995 20:09:51 GMT
Links: << >>  << T >>  << A >>
In article <3niupl$cs4@trumpet.uni-mannheim.de> kugel@mp-sun6.informatik.uni-mannheim.de writes:
>Don't try isp*** if you have different tristate signals in your design, it has no
>individual output enables, only blockwise.

Isn't it amazing how common this brain damage is? Altera is very limited
in this regard also in their 7000 series parts. You'd think a output enable
product term wouldn't be *that* expensive.

-- 
 Question Authority, but never shoot back.


Article: 1091
Subject: Re: Is anybody using FPGA's to do PCI interfaces?
From: pn@anuxt.mv.att.com (a.palmieri)
Date: Wed, 26 Apr 1995 23:54:46 GMT
Links: << >>  << T >>  << A >>

In reference to fpgas and pci bus. Xilinx about a month ago
came out with a series of app notes about fpgas. There
were a few that discussed pci interfaces in detail and
their implementation in their fpga's.

Call your xilinx rep for a set.

Tony
pn@anuxt.att.com



Article: 1092
Subject: Altera Vs Xilinx
From: matahari@netcom.com (paul chai)
Date: Thu, 27 Apr 1995 03:50:10 GMT
Links: << >>  << T >>  << A >>



XILINX ANNOUNCES LAWSUIT AGAINST ALTERA WILL
PROCEED FIRST 

Source: PR News Wire via DowVision
Date: Apr 25, 1995              Time: 7:46 am

SAN JOSE, Calif., April 25 /PRNewswire/ -- Xilinx, Inc., (Nasdaq: XLNX)
today provided an update of its pending patent litigation with Altera Corp.
(Nasdaq: ALTR). At a recent hearing the judge ruled that the Xilinx suit
against Altera would be tried first and that proceedings in the Altera suit
against Xilinx would be stayed. 

Xilinx sued Altera in June 1993 for infringing certain Xilinx patents. In
response to the Xilinx suit, Altera filed its own suit against Xilinx for patent
infringement. In July 1993 Xilinx asked the court to stop sales of the Altera
FLEX family of devices pending a trial. That matter was referred to a special
master appointed by the court. Based on the findings of the special master the
court said "Xilinx has demonstrated that it is likely to succeed at trial on the
issues of validity and infringement of Carter [patent] claim one [against 
Altera's FLEX product]." However, the court declined to issue the preliminary
injunction to stop sales pending a trial on the basis that continued sales of the
FLEX products were not causing Xilinx irreparable harm in the marketplace.
Both suits were then consolidated before federal judge Robert P. Aguilar in
San Jose, Calif. 

At a recent hearing Judge Aguilar ruled that the Xilinx suit against Altera
would be tried first and that proceedings in the Altera suit against Xilinx
would be stayed. The judge also appointed a new special master to make
factual findings in the case. 

"We are pleased that the court has agreed to move ahead with the Xilinx suit
first and we are quite confident that the court, with the assistance of the
special master, will conclude that the Altera FLEX and MAX product family
infringe our patents," said Chuck Fox, vice-president of Product Marketing of
Xilinx. 

Xilinx also commented on last week's additional suit filed by Altera against
the Xilinx XC5000 family which was filed in Delaware instead of California.
Xilinx said the suit has no merit and is merely a reaction to the court's decision
to stay the current Altera litigation against Xilinx. 

"This is a blatant case of forum shopping where Altera is searching for a way
to avoid the California court's decision," said Mr. Fox. "The XC5000 family is
the first programmable logic family specifically developed as a cost-effective,
high volume production alternative to mask programmed gate arrays. The
XC5000 has been very well received by customers. It will significantly
strengthen our #1 position in the programmable logic market." 

DowVision on the Internet. Brought to you by Dow Jones in collaboration with
WAIS Incorporated. 



Article: 1093
Subject: Re: Altera new FLEX 10000 - a worlds first
From: trev@ss11.wg.icl.co.uk (Trevor Hall)
Date: Thu, 27 Apr 1995 11:04:31 GMT
Links: << >>  << T >>  << A >>
idr@ee.ed.ac.uk (Iain Rankin) wrote:
>Does anyone have any technical details about the forthcoming
>FLEX10000 device.
>
I have a copy of the slides from a press briefing. What in particular do you wish
to know ?

Cheers,
T.H.




Article: 1094
Subject: Re: SIS (where do I find it)
From: rij@miepmiep.HZeeland.nl (Wim Rijkers)
Date: 27 Apr 1995 11:29:56 GMT
Links: << >>  << T >>  << A >>
Niranjan Cooray (niranjan@vlsi45) wrote:

> In article <bjrosen-2204951601120001@bjrosen.tiac.net>, bjrosen@aol.com (Joshua Rosen) writes:

> |>In which package is Berkeley's SIS tool located. I can't find it in the
> |>Octtools package, is it there or somewhere else?
> |>
> |>Thanks
> |>

> Berkeley SIS tool is available for anonymous ftp from:
> ftp://ic.eecs.berkeley.edu/pub/Sis

> It does'nt come with the octtools pkg, but having the octtool pkg when you
> compile SIS, will give you added features in SIS.

> Hope this helps. 


> -- Niranjan

> ------------------------------------------------------------------------------
> Niranjan Cooray	     	     E-mail (Internet): niranjan@nuvlsi.coe.neu.edu
> ECE Department			                niranjan@splinter.coe.neu.edu
> Northeastern University	
> Boston, MA 02115

>       HREF="http://www.ece.neu.edu/personal/niranjan/niranjan.html"
> ------------------------------------------------------------------------------


Article: 1095
Subject: Re: Need help about conference chip
From: randraka@ids.net
Date: Thu, 27 Apr 95 20:16:56 +500
Links: << >>  << T >>  << A >>
In Article <D7pHAy.4u1@bbc.co.uk>
matthew@rd.bbc.co.uk (Matthew Marks) writes:
>Wichai Tang (fengwct@ku.ac.th) wrote:
>: Hi,
>: 	1. I know that PCM is nonlinear digital system so we must convert 
>: it back to linear digital signal before we can do anything on it. From 8 
>: bits of PCM code must be convert to 14 bits of linear digital signal. The 
>: easy solution is to use ROM to convert this signal, but I would like to 
>: know that is there any other way to do this job ? And dose it appropriate 
>: for targetting at FPGA ?
>
>If you use a Xilinx chip (and no doubt many other types as well) you can
>store the look-up tables in the same ROM as the configuration, so no extra
>chips needed.
>
>: 	2. When we have many time slot want to join a conference, we must 
>: attenuate all signal before summing it up. so the question is how can I 
>: attenuate this linear digital signal ? I think this should not as easy as 
>: just divide it with number of conferece time slot. Am I correct ?
>
>As you are not worried about quantising distortion ;-) you won't have to
>worry about adding digital dither, so, yes, you just divide the signals
>by an appropriate amount and add them.  Dividing by a power of 2 is very
>easy of course - just shift the bits.
>
If you can afford the larger accumulator, add the signals first then scale the
result (a simple way of doing this is to just take the 14 msbs of the result). 
Your quantizing noise will be lower by adding first.  
 
A better soultion than simply taking the 14 msbs is to set up a pseudo AGC to
track the average magnitude of the result over a fixed time period and use that
result to select which 14 bits to take from the accumulator.  The selector can
include logic to realize saturating arithmetic so that the signal clips if it
exceeds the range of the 14 selected bits rather than getting the 2's comp wrap
around.  I've used the AGC and saturating arithmetic method for some radar
signal processing work.  The hardware is fairly simple and can be made very
fast.  The AGC value could also be determined as a function of the number of
participants in the conference.

-Ray Andraka
Chairman, the Andraka Consulting Group
401/884-7930    FAX 401/884-7950
email randraka@ids.net
 
The Andraka Consulting Group is a digital hardware design firm specializing in
obtaining the maximum performance from FPGAs.  Services include complete
design, development, simulation, and integration of these devices and the
surrounding circuits.  We also evaluate, troubleshoot, and improve existing
designs.  Please call or write for a free brochure.



Article: 1096
Subject: Re: Need help about conference chip
From: matthew@rd.bbc.co.uk (Matthew Marks)
Date: Thu, 27 Apr 1995 18:13:46 GMT
Links: << >>  << T >>  << A >>
Wichai Tang (fengwct@ku.ac.th) wrote:
: Hi,
: 	1. I know that PCM is nonlinear digital system so we must convert 
: it back to linear digital signal before we can do anything on it. From 8 
: bits of PCM code must be convert to 14 bits of linear digital signal. The 
: easy solution is to use ROM to convert this signal, but I would like to 
: know that is there any other way to do this job ? And dose it appropriate 
: for targetting at FPGA ?

If you use a Xilinx chip (and no doubt many other types as well) you can
store the look-up tables in the same ROM as the configuration, so no extra
chips needed.

: 	2. When we have many time slot want to join a conference, we must 
: attenuate all signal before summing it up. so the question is how can I 
: attenuate this linear digital signal ? I think this should not as easy as 
: just divide it with number of conferece time slot. Am I correct ?

As you are not worried about quantising distortion ;-) you won't have to
worry about adding digital dither, so, yes, you just divide the signals
by an appropriate amount and add them.  Dividing by a power of 2 is very
easy of course - just shift the bits.

Matthew  matthew@rd.bbc.co.uk  My opinions, not Auntie's


Article: 1097
Subject: Re: Altera new FLEX 10000 - a worlds first
From: Leon@lfheller.demon.co.uk (Leon Heller)
Date: Thu, 27 Apr 1995 18:22:38 +0000
Links: << >>  << T >>  << A >>
In article <HUTCH.95Apr26122452@timp.byu.edu>
           hutch@timp.byu.edu "Brad Hutchings" writes:

> The rest of this stuff is library modules that are being developed
> that will likely consume most of the FPGA. As I understand it,
> the 8086 was one the designs that was used to test the design.
> 
> The 10k part is a very interesting part. The availability of
> reasonably sized memories *on-chip* will make for much more
> interesting designs.

I was at a Lattice seminar recently and we were told that they will
be incorporating on-chip RAM soon in their ispLSI devices.

Leon
-- 
Leon Heller, G1HSM                | "Do not adjust your mind, there is
E-mail leon@lfheller.demon.co.uk  |  a fault in reality": on a wall
Phone: +44 (0)1734 266679         |  many years ago in Oxford.


Article: 1098
Subject: Re: Sunrise ???
From: peetj@hp7101.stortek.com (Peet James (peetj))
Date: Thu, 27 Apr 1995 19:24:39 GMT
Links: << >>  << T >>  << A >>
Terry E. Koontz (tkoontz@uncc.edu) wrote:

: Viewlogic recently add ViewTest to their Powerview product line.

: ViewTest is based on Sunrise 2.0 software (their words)

: Does anyone have an address and telephone number for Sunrise.

: Terry Koontz
: UNCC

: tkoontz@uncc.edu

Sunrise
Corp Headquarters
2730 San Tomas Exp.
Suite 200
Santa Clara, CA. 95051
408-980-7600
408-980-7630 fax
Info_request@srtest.com


Article: 1099
Subject: How to use XBLOX librariers in VHDL in Synopsys?
From: oner@paris.usc.edu ()
Date: 27 Apr 1995 13:49:03 -0700
Links: << >>  << T >>  << A >>


Hi Netters,

My subject says it all. I have been using Viewsynthesis to analyze
my VHDL programs and I can use XBLOX libraries in Viewsynthesis
by making special procedure calls. Is there a similar way to do
this in Synopsys tools? If there is, could you explain how I can do it?


An example of my VHDL code written for Viewsynthesis:

-------------------------------------------------------------------------------
--  Data Register & Counter
-------------------------------------------------------------------------------
  data_clk_en <= load_data OR inc_data;
  xb_counter(D_IN => d_in, ASYNC_CTRL => reset, CLOCK => clk,
             LOAD => load_data, CLK_EN => data_clk_en,
             Q_OUT => dout, ATTR => "STYLE=BINARY");
 

Thanks a lot.


Koray Oner




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