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 102275

Article: 102275
Subject: altera cyclone memory example
From: roiavidan@gmail.com
Date: 13 May 2006 07:30:40 -0700
Links: << >>  << T >>  << A >>
Hi Everyone,

I am a student trying to build a project with Altera's Cyclone FPGA
board.
For my project, I need to access a large amount of data, which I would
like to store in the board's memory. I did not find any simple example
which could show me how to simply read and write to the chip's built-in
memory.
Is there any ready made interface I could use?
I would like not to use the Nios implementations, as they seem to be
very large and complex for me at this stage.

Any help would be appreciated !!!

Thanks,
   Roi


Article: 102276
Subject: Re: altera cyclone memory example
From: "Antti" <Antti.Lukats@xilant.com>
Date: 13 May 2006 09:05:59 -0700
Links: << >>  << T >>  << A >>
the best advice dear student is to learn.
learn by doing.

you refer both on-board and on-chip memry, so which one do you want to
use??

FPGAs are not programmed, but rather you write the "description" of the
hardware that you wish the FPGA to implement, this description may
include different memory interfaces to either onchip or off chip
(on-board) memories.

but the way you ask there is no response possible.

you can use Altera tools to create various onchip memories, and connect
your logic to them.
that is if you want to use the onchip memories

the interface to on-board memories depends on the type of the memory,
and that can be very different.

SRAMs are very simple, DDR would be way more complex as example.

antti


Article: 102277
Subject: Re: altera cyclone memory example
From: David Brown <david@westcontrol.removethisbit.com>
Date: 13 May 2006 18:21:20 +0200
Links: << >>  << T >>  << A >>
roiavidan@gmail.com wrote:
> Hi Everyone,
> 
> I am a student trying to build a project with Altera's Cyclone FPGA
> board.
> For my project, I need to access a large amount of data, which I would
> like to store in the board's memory. I did not find any simple example
> which could show me how to simply read and write to the chip's built-in
> memory.
> Is there any ready made interface I could use?
> I would like not to use the Nios implementations, as they seem to be
> very large and complex for me at this stage.
> 
> Any help would be appreciated !!!
> 
> Thanks,
>    Roi
> 

Learn a bit about the Avalon bus, used by SOPC builder.  You don't need 
a Nios II processor on the bus - you can happily link up standard Altera 
SOPC components and your own components (and Wishbone modules from 
opencores).  It's a lot easier than trying to learn how to make DDR 
memory interfaces and the like.

Article: 102278
Subject: Trouble understanding Synplicity timing report
From: Jack Daly <jd1033@yahoo.com>
Date: Sat, 13 May 2006 17:59:20 +0100
Links: << >>  << T >>  << A >>
I'm having a problem understanding a Synplicity timing report. For the
input ports of the design, the report gives column headings of 'port
name', 'falling user constraint', 'rising user constraint', 'arrival
time', 'required time', and 'slack'. The problem is, I can't make any
sense of what 'arrival time' and 'required time' are, and so I can't
work out how they arrived at the slack figure.

Here's an example for input port 'INPA', clocked by 'CLKA'. CLKA is
20ns. At the device pins, INPA has a setup of 8ns and a hold of 1ns
relative to CLKA, so this is how I constrained INPA:

define_input_delay {INPA} -rise 8 -fall 8 -min_rise -1 -min_fall -1
-ref {CLKA:r}

I think this is correct; the rise/fall times are relative to the
*next* clock edge, so in other words INPA will be available >= 8ns
before a clock edge, and will hold for >= 1ns after that clock edge.
This is what the timing report says:

Port clock          cons(F) cons(R) AT     RT     SLACK
INPA CLKA (rising)  8.000   8.000   8.000  5.125  -2.875

So what does this mean? Why is the 'arrival time' always shown as
being equal to the setup constraint (it is in all the reports, and
even the picture in the docs shows this)?

This is what the manual says:

>Port Name                 Port name
>Reference Clock           Source clock
>Falling User Constraint   Specified falling input delay constraint
>Rising User Constraint    Specified rising input delay constraint
>Arrival Time              Clock latency - input delay constraint
>Required time             Requested period - clock delay at destination -
>                          setup of destination
>Slack (max/min analysis)  Required time - arrival time.

I haven't specified any latency on CLKA. My head is spinning - I'll be
eternally grateful to anyone who can explain this!

Jack

Article: 102279
Subject: Re: altera cyclone memory example
From: "Rob" <robnstef@frontiernet.net>
Date: Sat, 13 May 2006 18:08:34 GMT
Links: << >>  << T >>  << A >>
Start a project and open a new block diagram/schematic file.  Once the block 
diagram workspace is open, double click anywhere within that space and the 
Symbol window will pop open.  In the "Name" field type altsyncram and click 
on the "Ok" button, this will start the MegaWizard for Altera's on-chip RAM. 
Once you're in the MegaWizard's setup screens you will see a button labeled 
"Documentation".  Clicking on this will give you documentation on how the 
internal RAM works.


<roiavidan@gmail.com> wrote in message 
news:1147530640.153796.263270@i40g2000cwc.googlegroups.com...
> Hi Everyone,
>
> I am a student trying to build a project with Altera's Cyclone FPGA
> board.
> For my project, I need to access a large amount of data, which I would
> like to store in the board's memory. I did not find any simple example
> which could show me how to simply read and write to the chip's built-in
> memory.
> Is there any ready made interface I could use?
> I would like not to use the Nios implementations, as they seem to be
> very large and complex for me at this stage.
>
> Any help would be appreciated !!!
>
> Thanks,
>   Roi
> 



Article: 102280
Subject: Re: reverse engineering ?
From: fpga_toys@yahoo.com
Date: 13 May 2006 14:18:53 -0700
Links: << >>  << T >>  << A >>

Weng Tianxiang wrote:
> 10 years ago when I was in China, there was a Chinese electronics
> company in Beijing boasting every kind of FPGA chips in the world could
> be decrypted and I really believed it, because they had many products
> copied from other manufactures' products without designing: just copy,
> including copies of PCB, FPGA and software driver.

Yep ... anybody interested in having your PCB's produced in China?


Article: 102281
Subject: Re: Trouble understanding Synplicity timing report
From: Jack Daly <jd1033@yahoo.com>
Date: Sat, 13 May 2006 22:51:29 +0100
Links: << >>  << T >>  << A >>
On Sat, 13 May 2006 17:59:20 +0100, Jack Daly <jd1033@yahoo.com>
wrote:

>Here's an example for input port 'INPA', clocked by 'CLKA'. CLKA is
>20ns. At the device pins, INPA has a setup of 8ns and a hold of 1ns
>relative to CLKA, so this is how I constrained INPA:
>
>define_input_delay {INPA} -rise 8 -fall 8 -min_rise -1 -min_fall -1
>-ref {CLKA:r}

It looks like I got the constraint wrong - define_input_delay seems to
specify the total delay from the previous clock edge to when the
signal is available at the port. This explains why the arrival time is
the same as the constraint.

I've spent hours going through the docs and the website, and I
eventually found this out by checking the documentation on
*Synopsys's* set_input_delay, which Synplicity says is (almost) the
same. This is bizarre - there's not one picture or timing diagram
describing this in the entire manual, and the text describing it was
written by someone who certainly didn't understand it. Has anyone got
any useful pointers to a proper explanation of Synplicity constraints?

Jack

Article: 102282
Subject: filter design
From: "light" <linhaidu840329@163.com>
Date: 13 May 2006 18:27:00 -0700
Links: << >>  << T >>  << A >>
hi everyone!! i designed a fir filter with the Xinlinx ipcore two days
ago. and i am facing a problem! in my design, i have two signal
multiplied, one is sin(2*Pi*240k*t) and the other one is
-sin(2*Pi*240k*t+F).so it is obvious that the result has a signal of
480K,which is undesired in my system. and this is what the filter for!
the system clk of the filter is 48M and the sample rate is 4.8M! the
problem i met is that the output of the filter has a freqency of 4.8M.i
know it may have some relation to the sample rate,cause there is no
more 4.8M clks! in additon ,i simulated the model in Matlab and the
result turned out to be perfect!i got the coefficients of the filter
from the simulation. could anyone help me? any comments would be
appreciated! thanks


Article: 102283
Subject: Re: How to check IOB register packing?
From: "srini" <g.shrinivasan@gmail.com>
Date: 13 May 2006 23:45:06 -0700
Links: << >>  << T >>  << A >>
Hi Phil,
I am new to using these tools and designing with FPGA's. Can you tell
me how to set static timing constraints on the pins that can only be
reached with IOB FFs?

Thanks & Regards,
Srini.


Article: 102284
Subject: Re: How to decide Fanout limit?
From: "srini" <g.shrinivasan@gmail.com>
Date: 14 May 2006 00:00:15 -0700
Links: << >>  << T >>  << A >>
Hi,
My target architecture is ViretxII FPGA. I dont know whether there is
any fanout limitation or not. But I was of the opinion that a driver
having a high fanout might affect the timing in my design and I kept it
to minimum(100). I could see some synthesis notes telling that some of
the nets were replicated 'x' times bcoz of soft fanout limit of 100. I
am meeting my timing constraints with this spevs. Now, if I increase
the fanout limit to the default value of 10,000, will it affect my
timing or will there be any change in the synthesis results?


Article: 102285
Subject: Re: filter design
From: MikeShepherd564@btinternet.com
Date: Sun, 14 May 2006 08:05:01 +0100
Links: << >>  << T >>  << A >>
On 13 May 2006 18:27:00 -0700, "light" <linhaidu840329@163.com> wrote:

>hi everyone!! i designed a fir filter with the Xinlinx ipcore two days
>ago. and i am facing a problem! in my design, i have two signal
>multiplied, one is sin(2*Pi*240k*t) and the other one is
>-sin(2*Pi*240k*t+F).so it is obvious that the result has a signal of
>480K,which is undesired in my system. and this is what the filter for!
>the system clk of the filter is 48M and the sample rate is 4.8M! the
>problem i met is that the output of the filter has a freqency of 4.8M.i
>know it may have some relation to the sample rate,cause there is no
>more 4.8M clks! in additon ,i simulated the model in Matlab and the
>result turned out to be perfect!i got the coefficients of the filter
>from the simulation. could anyone help me? any comments would be
>appreciated! thanks

I'm not an expert on FIRs (sic), but, since you designed this filter
only "two days ago", it's clear that you've spent almost no time
trying to get it right yourself.

Remember that, if anyone here decides to help, they'll be doing it for
nothing, so it would be polite if you made some effort to write
clearly, to express your problem clearly and to press your "shift" key
now and again at the appropriate times.  Instead, we're expected to
respond to a continuous string of text which most of us wouldn't have
written at age 10.

If the style in which you ask for free help reflects how you document
your design (and it makes sense to me that it should be at least as
good), you should give it up now, because no-one will be able to
maintain it.

Article: 102286
Subject: Re: reverse engineering ?
From: MikeShepherd564@btinternet.com
Date: Sun, 14 May 2006 08:14:49 +0100
Links: << >>  << T >>  << A >>
>> 10 years ago...Chinese electronics company...boasting every kind of
>> FPGA chips...could be decrypted...copies of PCB, FPGA and software driver.
>
>Yep ... anybody interested in having your PCB's produced in China?

It's a good point.  My first thought with production in China is "How
do we stop them from copying it?"  Perhaps that's a little unfair,
because it's an important question, no matter to whom you transfer a
design.

Certainly, for any product in which I have a commercial interest, I
wouldn't touch a processor which doesn't offer code protection.  I'm
not clear that the question has yet been addressed for all FPGAs.  I'm
thinking of the Altera mid-range, for which the program load signals
aren't encrypted (as I understand it).  I only mention that
manufacturer because I know some of their products.  Their competitors
may have similar deficiencies.

Article: 102287
Subject: Picture frame
From: "Wiljan" <wiljan@post8.fjern.tele.dk>
Date: Sun, 14 May 2006 15:26:48 +0200
Links: << >>  << T >>  << A >>
Hi

I'm new to fpga..  I do not have any dev-kit yet
I'm looking for a fpga application to build a high res picture frame

Input:  Read a CF or SD card with jpg or gif  or bmp  picture  1280x1024
Output: DVI or LVDS as 1280x1024 60hz
Function: Load a new picture(slide) every X second

Advanced in the future: play video in native resolution (No need now)

Please if anyone can point me in the right direction for what fpga to use?
Any link to a project are welcome.

Regards Wiljan 



Article: 102288
Subject: Re: altera cyclone memory example
From: roiavidan@gmail.com
Date: 14 May 2006 07:28:38 -0700
Links: << >>  << T >>  << A >>
First of all - thanks alot for all of your quick replies.
I will try to ask my question in a more precise manner.

I need to build a piece of hardware ( I already wrote most of it in
vhdl and as schematic designs ) that will manipulate a large amount of
data - data that i believe cannot fit on-chip with the design itself.
thus i would like to use one of the on-board memories available to me.
i believe the cyclone board comes with SRAM (or SDRAM), and with this
board i also have a 32mb flash card.

what i'd like to know is how to use a simple ready-made interface for
either one of these memories ( i don't care which one ). i do not want
to write my own interface.
i want the ability to read and write to memory without needing to know
how the "memory internals" work.

is there such a solution for me?

many thanks again!!

  Roi


Article: 102289
Subject: Re: altera cyclone memory example
From: "Antti" <Antti.Lukats@xilant.com>
Date: 14 May 2006 08:20:14 -0700
Links: << >>  << T >>  << A >>
if you dont know if the memory is SRAM or SDRAM how can anyone really
help you ???

SRAM has virtually no interface needed at all, just supply address, ce,
rd wr and thats it
SDRAM needs special controller. if you dont want to use nios/SOPC then
you need to find some SDRAM controller and adapt it to your own use.

I bet finding the SDRAM controller from opencores should not take long
?

If you ask for simple interface to SDRAM then all the SDRAM IP cores
assume some local bus interface what in turn may not be what you call
simple, but here nobody can really help you, SDRAMs just arent so easy
to handle as SRAMs, sure some SDRAM IP cores are simpler and easier to
use then others. But wich one suites your application the best is up to
you to decide.

Antti


Article: 102290
Subject: Re: How to decide Fanout limit?
From: John_H <johnhandwork@mail.com>
Date: Sun, 14 May 2006 15:27:16 GMT
Links: << >>  << T >>  << A >>
srini wrote:
> Hi,
> My target architecture is ViretxII FPGA. I dont know whether there is
> any fanout limitation or not. But I was of the opinion that a driver
> having a high fanout might affect the timing in my design and I kept it
> to minimum(100). I could see some synthesis notes telling that some of
> the nets were replicated 'x' times [because] of soft fanout limit of 100. I
> am meeting my timing constraints with this spe[c]s. Now, if I increase
> the fanout limit to the default value of 10,000, will it affect my
> timing or will there be any change in the synthesis results?

If there were notes saying the nets were replicated, there will be 
changes in the synthesis results, though probably minor.

The only way to find out if timing is an issue is to rerun your 
synthesis and place & route to see if your timing is worse (and by how 
much) or better!

Article: 102291
Subject: Spartan 3E
From: "Piotr Wyderski" <wyderski@mothers.against.spam-ii.uni.wroc.pl>
Date: Sun, 14 May 2006 17:52:01 +0200
Links: << >>  << T >>  << A >>
I plan to use a Spartan3E device, namely  XC3S500E-4PQG208C,
in my next project, but  there are several questions I would like to ask:

1. what is their performance (relatively to the older Spartan 3 stepping 4)

a) maximal clock frequency;
b) peak and average power consumption at 250MHz (Vcc, Vaux, Vio);
c) average mW/MHz;

2. can I use a multiplier and its "neighbour" BRAM simultaneously,
i.e. is there enough routing?

3. can I clock the device (preferably differential mode) using
a 50MHz sine, extremely pure clock? It's Vpp can be adjusted
to meet the requirements (what are they?).

    Best regards
    Piotr Wyderski





Article: 102292
Subject: Re: reverse engineering ?
From: "dp" <dp@tgi-sci.com>
Date: 14 May 2006 08:55:15 -0700
Links: << >>  << T >>  << A >>
> It's a good point.  My first thought with production in China is "How
> do we stop them from copying it?"

The only way is to keep on deveoping new things faster than the
old ones get copied. Witholding knowledge seems to be a "no-no"
in evolutionary sense, this is how we are designed. The smarter
a society, the more rewards it offers for knowledge creation.
 This is not to say I am all open, of course. The above strategy
is probably optimized at an (inter?) species level, and I also am only
an individual... This is part (not a predominant one, though) of
the reason why I maintain a complete in-house tooling/development
capability, once you do a design on a wintel machine it is
anything but proprietary (can become public on a click
beyond the control of the design owner).

Dimiter

------------------------------------------------------
Dimiter Popoff               Transgalactic Instruments

http://www.tgi-sci.com
------------------------------------------------------

MikeShepherd564@btinternet.com wrote:
> >> 10 years ago...Chinese electronics company...boasting every kind of
> >> FPGA chips...could be decrypted...copies of PCB, FPGA and software driver.
> >
> >Yep ... anybody interested in having your PCB's produced in China?
>
> It's a good point.  My first thought with production in China is "How
> do we stop them from copying it?"  Perhaps that's a little unfair,
> because it's an important question, no matter to whom you transfer a
> design.
>
> Certainly, for any product in which I have a commercial interest, I
> wouldn't touch a processor which doesn't offer code protection.  I'm
> not clear that the question has yet been addressed for all FPGAs.  I'm
> thinking of the Altera mid-range, for which the program load signals
> aren't encrypted (as I understand it).  I only mention that
> manufacturer because I know some of their products.  Their competitors
> may have similar deficiencies.


Article: 102293
Subject: Re: Spartan 3E
From: Uwe Bonnes <bon@hertz.ikp.physik.tu-darmstadt.de>
Date: Sun, 14 May 2006 16:19:45 +0000 (UTC)
Links: << >>  << T >>  << A >>
Piotr Wyderski <wyderski@mothers.against.spam-ii.uni.wroc.pl> wrote:
> I plan to use a Spartan3E device, namely  XC3S500E-4PQG208C,
> in my next project, but  there are several questions I would like to ask:

> 1. what is their performance (relatively to the older Spartan 3 stepping 4)

> a) maximal clock frequency;
> b) peak and average power consumption at 250MHz (Vcc, Vaux, Vio);
> c) average mW/MHz;

> 2. can I use a multiplier and its "neighbour" BRAM simultaneously,
> i.e. is there enough routing?

> 3. can I clock the device (preferably differential mode) using
> a 50MHz sine, extremely pure clock? It's Vpp can be adjusted
> to meet the requirements (what are they?).

First question to ask (if you are no really _big_ customer):

Where to I get XC3SE from?

-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Article: 102294
Subject: Re: Spartan 3E
From: "Piotr Wyderski" <wyderski@mothers.against.spam-ii.uni.wroc.pl>
Date: Sun, 14 May 2006 18:31:18 +0200
Links: << >>  << T >>  << A >>
Uwe Bonnes wrote:

"> First question to ask (if you are no really _big_ customer):
> 
> Where to I get XC3SE from?

There is a small Polish shop which sells them:

http://www.kamami.pl/?id_k1=87&id_k2=76&id_k3=20

1 PLN = 0.26 Euro.

The largest device they have is 3S500, which costs 38,5 Euro @ 1
piece quantity. But I don't know whether they sell the chips outside
the country. Anyway, that's the reason I have decided to divorce
with Altera -- nobody sells their nice chips. :-(

    Best regards
    Piotr Wyderski

Article: 102295
Subject: Raggedstone IO bracket ?
From: "Xavier T" <xavier.tastet@gmail.com>
Date: 14 May 2006 10:02:42 -0700
Links: << >>  << T >>  << A >>
Hi All !
I saw on the web site lots of add-on, but is there "soon available" a
kind of IO rear bracket to fit in the pc slot; to easily access IO from
the rear of the pc ? 
Regards, X.


Article: 102296
Subject: Amontec Komodo board ?
From: "Xavier T" <xavier.tastet@gmail.com>
Date: 14 May 2006 10:04:25 -0700
Links: << >>  << T >>  << A >>
Hi all !
Do you if we can find the schematics and other spec, of the "viewable
version" on the website ? Like, price, Io counts, and so on ?
Regards, Xavier.


Article: 102297
Subject: Re: Spartan 3E
From: Uwe Bonnes <bon@hertz.ikp.physik.tu-darmstadt.de>
Date: Sun, 14 May 2006 17:31:57 +0000 (UTC)
Links: << >>  << T >>  << A >>
Piotr Wyderski <wyderski@mothers.against.spam-ii.uni.wroc.pl> wrote:
> Uwe Bonnes wrote:

> "> First question to ask (if you are no really _big_ customer):
> > 
> > Where to I get XC3SE from?

> There is a small Polish shop which sells them:

> http://www.kamami.pl/?id_k1=87&id_k2=76&id_k3=20

> 1 PLN = 0.26 Euro.

> The largest device they have is 3S500, which costs 38,5 Euro @ 1
> piece quantity. But I don't know whether they sell the chips outside
> the country. Anyway, that's the reason I have decided to divorce
> with Altera -- nobody sells their nice chips. :-(

Make sure they have them on stock. Every other lokation I have seen, that
lists them, state no explicit lead time. 

-- 
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Article: 102298
Subject: Re: Spartan 3E
From: pbdelete@spamnuke.ludd.luthdelete.se.invalid
Date: 14 May 2006 19:03:08 GMT
Links: << >>  << T >>  << A >>
Piotr Wyderski <wyderski@mothers.against.spam-ii.uni.wroc.pl> wrote:
>Uwe Bonnes wrote:

>"> First question to ask (if you are no really _big_ customer):
>> 
>> Where to I get XC3SE from?

>There is a small Polish shop which sells them:

>http://www.kamami.pl/?id_k1=87&id_k2=76&id_k3=20

>1 PLN = 0.26 Euro.

>The largest device they have is 3S500, which costs 38,5 Euro @ 1
>piece quantity. But I don't know whether they sell the chips outside
>the country. Anyway, that's the reason I have decided to divorce
>with Altera -- nobody sells their nice chips. :-(

www.elfa.se does, and they have office in Poland. For over the desk buying
if so required.
  http://www.elfa.se/pl/kontakt.html

Now if altera only had webpack for linux.. ;)


Article: 102299
Subject: Floating point reality check
From: Andreas Ehliar <ehliar@lysator.liu.se>
Date: Sun, 14 May 2006 21:11:15 +0000 (UTC)
Links: << >>  << T >>  << A >>
H have recently been working on a floating point unit for a Virtex 4
SX 35. I have a floating point adder and a floating point multiplier.
The adder has 6 pipeline stages and the multiplier has 3 stages.

The idea behind this project is to find out the kind of floating
point performance that is possible in a modern FPGA.

Our floating point format uses up to 15 bits of mantissa (with an
implicit one) and up to 10 bits of exponent. We have managed to get
a complex butterfly running at up to 250 MHz. So my question now is
if these numbers are reasonable or if anyone knows of a reference to
a faster fpu.


We have already used some tricks to improve performance, for example
by manually instantiating LUTs so that we can build an adder with a
2 to 1 MUX on one of the operands using only one LUT per bit.

We have also tried to build up the design using RLOC:ed modules. This
did not lead to improved performance as compared to a non RLOC:ed design.
This could change once we start to fill up the device though. At the
moment we are only utilizing about 20% of the FPGA.

/Andreas



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