AVR-Chat

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:

> I need a contact for a semi manufacturer
> who can manufacter 2 devices for me:
> #1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)
> #2 8 bit DDR (I need 10TB for the above cpu's)

Kat, I would recommend TSMC for the CPU.  You will need their 0.13 
micron process with copper interconnect. Even then, you'll need a 
highly parallel design. Once you have the design complete, make sure 
you have a million dollars left for masks and the first fab run.

For the DDR, I think you should save your second million bucks and 
use existing devices. Why do you need an 8-bit device? To get 10 TB 
you will have to stack up an awful lot of chips, so the width of each 
chip shouldn't be a major consideration. I trust you're not thinking 
you can get 10 TB on a single chip.

This is not the sort of post I expected on this group, though I 
recognize you as a regular. Is there any chance you could divulge the 
application? In particular, if you need Gflops why are you 
constrained to an 8-bit CPU? I think you should be looking at 32-bit 
IP cores such as ARM, ARC, MIPS, TenSilica, etc. if not 64-bit.

Graham.

Hi Guys and Gals,

I need a contact for a semi manufacturer who can manufacter 2 devices 
for me:

#1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)

#2 8 bit DDR (I need 10TB for the above cpu's)

Regards,

Kat.

-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

Kathy Quinlan wrote:
> Hi Guys and Gals,
> 
> I need a contact for a semi manufacturer who can manufacter 2 devices 
> for me:
> 
> #1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)
> 
> #2 8 bit DDR (I need 10TB for the above cpu's)

April-fools was a few months ago;)

----- Original Message -----

From: "Kathy Quinlan" <kat-yahoo@kaqelectronics.dyndns.org>
To: <AVR-Chat@yahoogroups.com>
Sent: Wednesday, August 04, 2004 13:28
Subject: [AVR-Chat] Semi manufacturers ?


> Hi Guys and Gals,
>
> I need a contact for a semi manufacturer who can
manufacter 2 devices
> for me:
>
> #1 an 8 bit cpu that is FAST (think Gflops) (able to
address 0.1TB)
>
> #2 8 bit DDR (I need 10TB for the above cpu's)
>
> Regards,
>
> Kat.

Sounds like you're looking for a Bambleweeny 57 SubMeson
Brain with the optional Brownian Motion Generator.  ;-)

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

Yes, I think that the machine is made by "Rockwell Automation" :-)

 

Here is a video explaining the machine: 

 

http://media.ebaumsworld.com/retro.wmv

 

-jr

 

  _____  

From: erikc [mailto:firewevr@airmail.net] 
Sent: Wednesday, August 04, 2004 5:53 PM
To: AVR-Chat@yahoogroups.com
Subject: Re: [AVR-Chat] Semi manufacturers ?

 

----- Original Message -----
From: "Kathy Quinlan" <kat-yahoo@kaqelectronics.dyndns.org>
To: <AVR-Chat@yahoogroups.com>
Sent: Wednesday, August 04, 2004 13:28
Subject: [AVR-Chat] Semi manufacturers ?


> Hi Guys and Gals,
>
> I need a contact for a semi manufacturer who can
manufacter 2 devices
> for me:
>
> #1 an 8 bit cpu that is FAST (think Gflops) (able to
address 0.1TB)
>
> #2 8 bit DDR (I need 10TB for the above cpu's)
>
> Regards,
>
> Kat.

Sounds like you're looking for a Bambleweeny 57 SubMeson
Brain with the optional Brownian Motion Generator.  ;-)

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)








Yahoo! Groups Sponsor



ADVERTISEMENT
 
<http://us.ard.yahoo.com/SIG=1296msnu1/M=295196.4901138.6071305.3001176/D=gr
oups/S=1706554205:HM/EXP=1091747001/A=2128215/R=0/SIG=10se96mf6/*http:/compa
nion.yahoo.com> click here


 
<http://us.adserver.yahoo.com/l?M=295196.4901138.6071305.3001176/D=groups/S=
:HM/A=2128215/rand=168920164> 

 

  _____  

Yahoo! Groups Links

*	To visit your group on the web, go to:
http://groups.yahoo.com/group/AVR-Chat/
  
*	To unsubscribe from this group, send an email to:
AVR-Chat-unsubscribe@yahoogroups.com
<mailto:AVR-Chat-unsubscribe@yahoogroups.com?subject=Unsubscribe> 
  
*	Your use of Yahoo! Groups is subject to the Yahoo!
<http://docs.yahoo.com/info/terms/>  Terms of Service.

On Thursday 05 August 2004 05:52, erikc wrote:
> From: "Kathy Quinlan" <kat-yahoo@kaqelectronics.dyndns.org>

> > Hi Guys and Gals,
> >
> > I need a contact for a semi manufacturer who can manufacter 2 devices
> > for me:
> >
> > #1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)
> > #2 8 bit DDR (I need 10TB for the above cpu's)

See what happens when your cat [sic] runs loose on the keyboard?

> Sounds like you're looking for a Bambleweeny 57 SubMeson
> Brain with the optional Brownian Motion Generator.  ;-)

She has the latter.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

No, what you need is a rockwell.

http://media.ebaumsworld.com/index.php?e=retro.wmv

-Mike

On Wed, 4 Aug 2004 22:52:32 +0100, erikc <firewevr@airmail.net> wrote:
> ----- Original Message -----
> From: "Kathy Quinlan" <kat-yahoo@kaqelectronics.dyndns.org>
> To: <AVR-Chat@yahoogroups.com>
> Sent: Wednesday, August 04, 2004 13:28
> Subject: [AVR-Chat] Semi manufacturers ?
> 
> 
> > Hi Guys and Gals,
> >
> > I need a contact for a semi manufacturer who can
> manufacter 2 devices
> > for me:
> >
> > #1 an 8 bit cpu that is FAST (think Gflops) (able to
> address 0.1TB)
> >
> > #2 8 bit DDR (I need 10TB for the above cpu's)
> >
> > Regards,
> >
> > Kat.
> 
> Sounds like you're looking for a Bambleweeny 57 SubMeson
> Brain with the optional Brownian Motion Generator.  ;-)
> 
> Erikc  - firewevr@airmail.net
> ///
> "An Fhirinne in aghaidh an tSaoil."
> "The Truth against the World."
>         -- Bardic Motto
> ///I did NOT lose my mind!  It's backed up on the server
> (and the network's down....)
> 
> 
> 
> 
> 
> 
> 
> Yahoo! Groups Sponsor
> 
> ADVERTISEMENT
> 
> 
> ________________________________
> Yahoo! Groups Links
> 
> To visit your group on the web, go to:
> http://groups.yahoo.com/group/AVR-Chat/
>   
> To unsubscribe from this group, send an email to:
> AVR-Chat-unsubscribe@yahoogroups.com
>   
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. 
> 
>

Doh! I guess I wasn't the only one to see it then.

 

Differential girdle springs will definitely be needed.Not to mention the
dingle arm.

 

I'm still laughing and I saw it a few days ago.

 

-jr

 

  _____  

From: Michael Haisley [mailto:mhaisley@gmail.com] 
Sent: Wednesday, August 04, 2004 7:14 PM
To: avr-chat@yahoogroups.com
Subject: Re: [AVR-Chat] Semi manufacturers ?

 

No, what you need is a rockwell.

http://media.ebaumsworld.com/index.php?e=retro.wmv

-Mike

On Wed, 4 Aug 2004 22:52:32 +0100, erikc <firewevr@airmail.net> wrote:
> ----- Original Message -----
> From: "Kathy Quinlan" <kat-yahoo@kaqelectronics.dyndns.org>
> To: <AVR-Chat@yahoogroups.com>
> Sent: Wednesday, August 04, 2004 13:28
> Subject: [AVR-Chat] Semi manufacturers ?
> 
> 
> > Hi Guys and Gals,
> >
> > I need a contact for a semi manufacturer who can
> manufacter 2 devices
> > for me:
> >
> > #1 an 8 bit cpu that is FAST (think Gflops) (able to
> address 0.1TB)
> >
> > #2 8 bit DDR (I need 10TB for the above cpu's)
> >
> > Regards,
> >
> > Kat.
> 
> Sounds like you're looking for a Bambleweeny 57 SubMeson
> Brain with the optional Brownian Motion Generator.  ;-)
> 
> Erikc  - firewevr@airmail.net
> ///
> "An Fhirinne in aghaidh an tSaoil."
> "The Truth against the World."
>         -- Bardic Motto
> ///I did NOT lose my mind!  It's backed up on the server
> (and the network's down....)
> 
> 
> 
> 
> 
> 
> 
> Yahoo! Groups Sponsor
> 
> ADVERTISEMENT
> 
> 
> ________________________________
> Yahoo! Groups Links
> 
> To visit your group on the web, go to:
> http://groups.yahoo.com/group/AVR-Chat/
>   
> To unsubscribe from this group, send an email to:
> AVR-Chat-unsubscribe@yahoogroups.com
>   
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. 
> 
>





Yahoo! Groups Sponsor



ADVERTISEMENT
 
<http://us.ard.yahoo.com/SIG=1295ciqgd/M=295196.4901138.6071305.3001176/D=gr
oups/S=1706554205:HM/EXP=1091747627/A=2128215/R=0/SIG=10se96mf6/*http:/compa
nion.yahoo.com> click here


 
<http://us.adserver.yahoo.com/l?M=295196.4901138.6071305.3001176/D=groups/S=
:HM/A=2128215/rand=582081335> 

 

  _____  

Yahoo! Groups Links

*	To visit your group on the web, go to:
http://groups.yahoo.com/group/AVR-Chat/
  
*	To unsubscribe from this group, send an email to:
AVR-Chat-unsubscribe@yahoogroups.com
<mailto:AVR-Chat-unsubscribe@yahoogroups.com?subject=Unsubscribe> 
  
*	Your use of Yahoo! Groups is subject to the Yahoo!
<http://docs.yahoo.com/info/terms/>  Terms of Service.

Graham Davies wrote:

> --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> 
> 
>>I need a contact for a semi manufacturer
>>who can manufacter 2 devices for me:
>>#1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)
>>#2 8 bit DDR (I need 10TB for the above cpu's)
> 
> 
> Kat, I would recommend TSMC for the CPU.  You will need their 0.13 
> micron process with copper interconnect. Even then, you'll need a 
> highly parallel design. Once you have the design complete, make sure 
> you have a million dollars left for masks and the first fab run.
> 
> For the DDR, I think you should save your second million bucks and 
> use existing devices. Why do you need an 8-bit device? To get 10 TB 
> you will have to stack up an awful lot of chips, so the width of each 
> chip shouldn't be a major consideration. I trust you're not thinking 
> you can get 10 TB on a single chip.
> 
> This is not the sort of post I expected on this group, though I 
> recognize you as a regular. Is there any chance you could divulge the 
> application? In particular, if you need Gflops why are you 
> constrained to an 8-bit CPU? I think you should be looking at 32-bit 
> IP cores such as ARM, ARC, MIPS, TenSilica, etc. if not 64-bit.
> 
> Graham.

I have to process a text file that can be up to 10Tb in size, the data 
is only ascii, so I can not see the point for a larger sized CPU. Once a 
  string is found, all it needs to do is hand the string up to the next 
CPU for serving.

This is not a joke Guys, I have someone who is paying me to design this 
system, and speed is essential, so it has to be held in Ram. I do not 
plan to have one CPU to search the whole 10Tb I was planing on using 100 
parallel CPU's each looking after 100Gb of ram.

Kat.


-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

>
>I have to process a text file that can be up to 10Tb in size, the data 
>is only ascii, so I can not see the point for a larger sized CPU. Once a 
>  string is found, all it needs to do is hand the string up to the next 
>CPU for serving.

Sounds like a job for a beowulf cluster.

Sounds intriguing, but the development costs are
likely to be astronomical if you're only making 1.

Have you considered a Linux cluster?  Sure, there's no
AVR content, however you should be able to simply
assemble and configure a solution.

$0.02,

Jeff Engel
--- Kathy Quinlan
<kat-yahoo@kaqelectronics.dyndns.org> wrote:

> Graham Davies wrote:
> 
> > --- In AVR-Chat@yahoogroups.com, Kathy Quinlan
> <kat-yahoo@k...> wrote:
> > 
> > 
> >>I need a contact for a semi manufacturer
> >>who can manufacter 2 devices for me:
> >>#1 an 8 bit cpu that is FAST (think Gflops) (able
> to address 0.1TB)
> >>#2 8 bit DDR (I need 10TB for the above cpu's)
> > 
> > 
> > Kat, I would recommend TSMC for the CPU.  You will
> need their 0.13 
> > micron process with copper interconnect. Even
> then, you'll need a 
> > highly parallel design. Once you have the design
> complete, make sure 
> > you have a million dollars left for masks and the
> first fab run.
> > 
> > For the DDR, I think you should save your second
> million bucks and 
> > use existing devices. Why do you need an 8-bit
> device? To get 10 TB 
> > you will have to stack up an awful lot of chips,
> so the width of each 
> > chip shouldn't be a major consideration. I trust
> you're not thinking 
> > you can get 10 TB on a single chip.
> > 
> > This is not the sort of post I expected on this
> group, though I 
> > recognize you as a regular. Is there any chance
> you could divulge the 
> > application? In particular, if you need Gflops why
> are you 
> > constrained to an 8-bit CPU? I think you should be
> looking at 32-bit 
> > IP cores such as ARM, ARC, MIPS, TenSilica, etc.
> if not 64-bit.
> > 
> > Graham.
> 
> I have to process a text file that can be up to 10Tb
> in size, the data 
> is only ascii, so I can not see the point for a
> larger sized CPU. Once a 
>   string is found, all it needs to do is hand the
> string up to the next 
> CPU for serving.
> 
> This is not a joke Guys, I have someone who is
> paying me to design this 
> system, and speed is essential, so it has to be held
> in Ram. I do not 
> plan to have one CPU to search the whole 10Tb I was
> planing on using 100 
> parallel CPU's each looking after 100Gb of ram.
> 
> Kat.
> 
> 
> -- 
>
---------------------------------------------------------------
> K.A.Q. Electronics	Website:
> www.kaqelectronics.dyndns.org
> IM: Yahoo: PinkyDwaggy  MSN:
> katinka@kaqelectronics.dyndns.org
> For Everything Electronics     Phone: 0419 923 731
>
---------------------------------------------------------------
> 
> 
> 
> ------------------------ Yahoo! Groups Sponsor
> --------------------~--> 
> Yahoo! Domains - Claim yours for only $14.70
>
http://us.click.yahoo.com/Z1wmxD/DREIAA/yQLSAA/dN_tlB/TM
>
--------------------------------------------------------------------~->
> 
> 
>  
> Yahoo! Groups Links
> 
> 
>     AVR-Chat-unsubscribe@yahoogroups.com
> 
>  
> 
> 


=====
Happiness is - positive intake manifold pressure.


	
		
__________________________________
Do you Yahoo!?
New and Improved Yahoo! Mail - 100MB free storage!
http://promotions.yahoo.com/new_mail

Jeffrey Engel wrote:

> Sounds intriguing, but the development costs are
> likely to be astronomical if you're only making 1.
> 
> Have you considered a Linux cluster?  Sure, there's no
> AVR content, however you should be able to simply
> assemble and configure a solution.
> 
> $0.02,
> 
> Jeff Engel

Yeap, that is one option (using FreeBSD)

It will not be just one system, I plan to have atleast 3 working in 
tandom for fault tolerance.

Here is a link to the project:
http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html

Regards,

Kat.

-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

> I have to process a text file that can be up to 10Tb in size, the data
> is only ascii, so I can not see the point for a larger sized CPU. Once 
> a
>   string is found, all it needs to do is hand the string up to the next
> CPU for serving.
>
> This is not a joke Guys, I have someone who is paying me to design this
> system, and speed is essential, so it has to be held in Ram. I do not
> plan to have one CPU to search the whole 10Tb I was planing on using 
> 100
> parallel CPU's each looking after 100Gb of ram.

Stop...  Don't think CPUs, think FPGAs.

As much parallelism as you care to create...

Mike

I am just wondering where and how you managed to get funding for a 
project this?
Also what sort in what sort of budget are you planing to build this system?

Chad

Kathy Quinlan wrote:

> Jeffrey Engel wrote:
>
> > Sounds intriguing, but the development costs are
> > likely to be astronomical if you're only making 1.
> >
> > Have you considered a Linux cluster?  Sure, there's no
> > AVR content, however you should be able to simply
> > assemble and configure a solution.
> >
> > $0.02,
> >
> > Jeff Engel
>
> Yeap, that is one option (using FreeBSD)
>
> It will not be just one system, I plan to have atleast 3 working in
> tandom for fault tolerance.
>
> Here is a link to the project:
> http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html
>
> Regards,
>
> Kat.
>
> -- 
> ---------------------------------------------------------------
> K.A.Q. Electronics      Website: www.kaqelectronics.dyndns.org
> IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
> For Everything Electronics     Phone: 0419 923 731
> ---------------------------------------------------------------     
>
> *Yahoo! Groups Sponsor*
> ADVERTISEMENT
> click here 
> <http://us.ard.yahoo.com/SIG=129a42o7a/M=295196.4901138.6071305.3001176/D=groups/S=1706554205:HM/EXP=1091753332/A=2128215/R=0/SIG=10se96mf6/*http://companion.yahoo.com> 
>
>
>
> ------------------------------------------------------------------------
> *Yahoo! Groups Links*
>
>     * To visit your group on the web, go to:
>       http://groups.yahoo.com/group/AVR-Chat/
>        
>     * To unsubscribe from this group, send an email to:
>       AVR-Chat-unsubscribe@yahoogroups.com
>       <mailto:AVR-Chat-unsubscribe@yahoogroups.com?subject=Unsubscribe>
>        
>     * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
>       Service <http://docs.yahoo.com/info/terms/>.
>
>

Kathy Quinlan wrote:
> Jeffrey Engel wrote:
> 
>>Sounds intriguing, but the development costs are
>>likely to be astronomical if you're only making 1.
>>
>>Have you considered a Linux cluster?  Sure, there's no
>>AVR content, however you should be able to simply
>>assemble and configure a solution.
>>
>>$0.02,
>>
>>Jeff Engel
> 
> Yeap, that is one option (using FreeBSD)
> 
> It will not be just one system, I plan to have atleast 3 working in 
> tandom for fault tolerance.
> 
> Here is a link to the project:
> http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html

Ok, 10TB is ~10^13 bytes i assume.

Do with hash tables. First table can be 1024 point hash that
indexes 1024 1024-point tables in the 2nd stage. 3rd stage is
1048576 1024-point tables. 4th stage is 1073741824 arrays, each
containing 10.24 kBytes. Can do it easy with an AVR and a few
hard-disks.

How many users will a single database device have?

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:

> I have to process a text file
> that can be up to 10Tb in size,
> the data is only ascii, so I
> can not see the point for a
> larger sized CPU. Once a string
> is found, all it needs to do is
> hand the string up to the next 
> CPU for serving.

If you're processing text, why are you concerned with Gflops? Why do 
you need floating point? Your line of thinking that you should use an 
8-bit CPU to process text is broken. Any decent algorithm you choose 
will need memory pointers and you want to hold them in registers so 
you need a bigger CPU. You're not thinking of a linear search, are 
you? Also, before we all put more effort into helping you, could you 
confirm that you have a multi-million dollar budget for this? If you 
don't, you can forget a custom CPU right now.

Graham.

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> Graham Davies wrote:
> 
> > --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> 
wrote:
> > 
> > 
> >>I need a contact for a semi manufacturer
> >>who can manufacter 2 devices for me:
> >>#1 an 8 bit cpu that is FAST (think Gflops) (able to address 
0.1TB)
> >>#2 8 bit DDR (I need 10TB for the above cpu's)
> > 
> > 
> > Kat, I would recommend TSMC for the CPU.  You will need their 
0.13 
> > micron process with copper interconnect. Even then, you'll need a 
> > highly parallel design. Once you have the design complete, make 
sure 
> > you have a million dollars left for masks and the first fab run.
> > 
> > For the DDR, I think you should save your second million bucks 
and 
> > use existing devices. Why do you need an 8-bit device? To get 10 
TB 
> > you will have to stack up an awful lot of chips, so the width of 
each 
> > chip shouldn't be a major consideration. I trust you're not 
thinking 
> > you can get 10 TB on a single chip.
> > 
> > This is not the sort of post I expected on this group, though I 
> > recognize you as a regular. Is there any chance you could divulge 
the 
> > application? In particular, if you need Gflops why are you 
> > constrained to an 8-bit CPU? I think you should be looking at 32-
bit 
> > IP cores such as ARM, ARC, MIPS, TenSilica, etc. if not 64-bit.
> > 
> > Graham.
> 
> I have to process a text file that can be up to 10Tb in size, the 
data 
> is only ascii, so I can not see the point for a larger sized CPU. 
Once a 
>   string is found, all it needs to do is hand the string up to the 
next 
> CPU for serving.
> 
> This is not a joke Guys, I have someone who is paying me to design 
this 
> system, and speed is essential, so it has to be held in Ram. I do 
not 
> plan to have one CPU to search the whole 10Tb I was planing on 
using 100 
> parallel CPU's each looking after 100Gb of ram.
> 
> Kat.

Could you just tell us what this thing is going to be used for?

erikc

--- In AVR-Chat@yahoogroups.com, "Chad O'neill" <chad@c...> wrote:
> You do realise it costs over a million dollars just for the mask 
set 
> when making a 0.13u chip,
> let alone the development costs, packaging etc etc.
> 
> Chad


Probably, and if one of the Semi houses like the idea and see the 
sales potential, they may make it a product.

Of course, if the project flies, that small sum would be written off 
in start-up costs and patents and all that.

THINK BIG !

Personally, I thought the internet was supposed to be all that !

Dave

----- Original Message -----

From: "David VanHorn" <dvanhorn@cedar.net>
To: <AVR-Chat@yahoogroups.com>; <AVR-Chat@yahoogroups.com>
Sent: Thursday, August 05, 2004 01:38
Subject: Re: [AVR-Chat] Re: Semi manufacturers ?


>
> >
> >I have to process a text file that can be up to 10Tb in
size, the data
> >is only ascii, so I can not see the point for a larger
sized CPU. Once a
> >  string is found, all it needs to do is hand the string
up to the next
> >CPU for serving.
>
> Sounds like a job for a beowulf cluster.

Well, if you want to develop semiconductors, why not a RAM
chip with a simple processor built in?  The processor would
have its own small RAM for holding its programmes and a
means of entering those programmes, causing them to run, and
getting results off-chip.   You'd load all the little CPU's
in parallel and let them loose, each one looking at the data
on its own chip.  Now I'm not talking a CPU with some RAM,
I'm talking 256 megabytes of RAM and maybe 4K of entirely
separate RAM for the CPU's code.  Since you are basically
using it to search for patterns, the CPU could be optimised
for that.

I actually think there might be a market for that kind of
thing.

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

At least.

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)


----- Original Message -----
From: "Chad O'neill" <chad@chadoneill.com>
To: <AVR-Chat@yahoogroups.com>
Sent: Friday, August 06, 2004 01:28
Subject: Re: [AVR-Chat] Re: Semi manufacturers ?


> You do realise it costs over a million dollars just for
the mask set

> when making a 0.13u chip,
> let alone the development costs, packaging etc etc.
>
> Chad
>
> erikc wrote:

----- Original Message -----

From: "Bernd Felsche" <bernie@innovative.iinet.net.au>
To: <AVR-Chat@yahoogroups.com>
Sent: Friday, August 06, 2004 01:39
Subject: Re: [AVR-Chat] Re: Semi manufacturers ?


> On Friday 06 August 2004 07:06, erikc wrote:
> > Well, if you want to develop semiconductors, why not a
RAM
> > chip with a simple processor built in?  The processor
would
> > have its own small RAM for holding its programmes and a
> > means of entering those programmes, causing them to run,
and
> > getting results off-chip.   You'd load all the little
CPU's
> > in parallel and let them loose, each one looking at the
data
> > on its own chip.  Now I'm not talking a CPU with some
RAM,
> > I'm talking 256 megabytes of RAM and maybe 4K of
entirely
> > separate RAM for the CPU's code.  Since you are
basically
> > using it to search for patterns, the CPU could be
optimised
> > for that.
> >
> > I actually think there might be a market for that kind
of
> > thing.
>
> You mean a large, secondary data cache on a simple
processor,
> or a CPU with mega-registers?

No, you're still stuck in the mindset of "CPU with RAM".
I'm having it the other way around.  A big DRAM with a
little CPU in one corner.  It's address pointers would be
whatever size needed to address the memory on that one chip.
Someone not knowing that there was a CPU inside could still
use the part as a "dumb" DRAM.

> There's still matter of partial matches spanning
boundaries.


Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

You do realise it costs over a million dollars just for the mask set 
when making a 0.13u chip,
let alone the development costs, packaging etc etc.

Chad

erikc wrote:

>
> ----- Original Message -----
> From: "David VanHorn" <dvanhorn@cedar.net>
> To: <AVR-Chat@yahoogroups.com>; <AVR-Chat@yahoogroups.com>
> Sent: Thursday, August 05, 2004 01:38
> Subject: Re: [AVR-Chat] Re: Semi manufacturers ?
>
>
> >
> > >
> > >I have to process a text file that can be up to 10Tb in
> size, the data
> > >is only ascii, so I can not see the point for a larger
> sized CPU. Once a
> > >  string is found, all it needs to do is hand the string
> up to the next
> > >CPU for serving.
> >
> > Sounds like a job for a beowulf cluster.
>
> Well, if you want to develop semiconductors, why not a RAM
> chip with a simple processor built in?  The processor would
> have its own small RAM for holding its programmes and a
> means of entering those programmes, causing them to run, and
> getting results off-chip.   You'd load all the little CPU's
> in parallel and let them loose, each one looking at the data
> on its own chip.  Now I'm not talking a CPU with some RAM,
> I'm talking 256 megabytes of RAM and maybe 4K of entirely
> separate RAM for the CPU's code.  Since you are basically
> using it to search for patterns, the CPU could be optimised
> for that.
>
> I actually think there might be a market for that kind of
> thing.
>
> Erikc  - firewevr@airmail.net
> ///
> "An Fhirinne in aghaidh an tSaoil."
> "The Truth against the World."
>         -- Bardic Motto
> ///I did NOT lose my mind!  It's backed up on the server
> (and the network's down....)
>
>
>
> *Yahoo! Groups Sponsor*
> ADVERTISEMENT
> click here 
> <http://us.ard.yahoo.com/SIG=129n2c6mf/M=295196.4901138.6071305.3001176/D=groups/S=1706554205:HM/EXP=1091837883/A=2128215/R=0/SIG=10se96mf6/*http://companion.yahoo.com> 
>
>
>
> ------------------------------------------------------------------------
> *Yahoo! Groups Links*
>
>     * To visit your group on the web, go to:
>       http://groups.yahoo.com/group/AVR-Chat/
>        
>     * To unsubscribe from this group, send an email to:
>       AVR-Chat-unsubscribe@yahoogroups.com
>       <mailto:AVR-Chat-unsubscribe@yahoogroups.com?subject=Unsubscribe>
>        
>     * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
>       Service <http://docs.yahoo.com/info/terms/>.
>
>

On Thursday 05 August 2004 08:14, Kathy Quinlan wrote:
> Graham Davies wrote:
> > --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> >>I need a contact for a semi manufacturer
> >>who can manufacter 2 devices for me:
> >>#1 an 8 bit cpu that is FAST (think Gflops) (able to address 0.1TB)
> >>#2 8 bit DDR (I need 10TB for the above cpu's)

> > Kat, I would recommend TSMC for the CPU.  You will need their 0.13
> > micron process with copper interconnect. Even then, you'll need a
> > highly parallel design. Once you have the design complete, make sure
> > you have a million dollars left for masks and the first fab run.

[snip]

> I have to process a text file that can be up to 10Tb in size, the

File? Does that mean that it's stored somewhere?

How long is the string to which you want to match?

> data is only ascii, so I can not see the point for a larger sized
> CPU. Once a string is found, all it needs to do is hand the string
> up to the next CPU for serving.

Matching even a non-trivial string in that volume of "random" data
is always probable.

> This is not a joke Guys, I have someone who is paying me to design
> this system, and speed is essential, so it has to be held in Ram.

How does it get into RAM? A byte at a time? Is the string byte-sized
or are the characters larger? I think you'll find that for any
reasonable volume of data, that you'll be I/O-bound.

Matching a string in un-indexed data is a matter of comparing each
character starting from the first and repeating that until a match
is found, checking the next character for match, etc...

It doesn't appear to require much RAM at all; not unless you're
matching multiple strings in the same data. If multiple is many,
you should investigate indexes; hashing for example.

> I do not plan to have one CPU to search the whole 10Tb I was
> planing on using 100 parallel CPU's each looking after 100Gb of
> ram.

Divide and conquer is a good approach. 
Perhaps you're byte-ing off more than you can chew.

Using 1000 Opterons with 10GB each will cost significantly less...
you'll get a good volume discount from AMD as well. If you use the
quad-processor version, then you can reduce the number of
"computers" to 250; with 4 CPU and 64GB RAM each; they can talk to
each other at very high speed.

Comms between computers at 10Gbit at least, using off-the-shelf
hardware. expect a total power draw of about 100kW. At that comms
rate, you'll be able to process the 10Tbytes in under an hour with
250 parallel links of that speed ... but if you're peeling the data
off high-speed hard drives at 100Mbytes/second, it's 10 seconds for
a gigabyte or 10,000 seconds for a terrabyte. 

Software gets more complex as you go multi-processor. Dividing the
data set is not enough; you have to have overlapping data sets;
overlapping by at least the size of the matching string to ensure
that you find all possible matches...

Final words; talk to Cray or Fujitsu.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

On Friday 06 August 2004 07:06, erikc wrote:
> Well, if you want to develop semiconductors, why not a RAM
> chip with a simple processor built in?  The processor would
> have its own small RAM for holding its programmes and a
> means of entering those programmes, causing them to run, and
> getting results off-chip.   You'd load all the little CPU's
> in parallel and let them loose, each one looking at the data
> on its own chip.  Now I'm not talking a CPU with some RAM,
> I'm talking 256 megabytes of RAM and maybe 4K of entirely
> separate RAM for the CPU's code.  Since you are basically
> using it to search for patterns, the CPU could be optimised
> for that.
>
> I actually think there might be a market for that kind of
> thing.

You mean a large, secondary data cache on a simple processor,
or a CPU with mega-registers? 

There's still matter of partial matches spanning boundaries.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

I used to work on projects which did string matching on big databases. BIG.
The techique was to create an inverted index in real time as the data is
stored, coming in. Then the searches are fast.
Essentially, this how Google works.

-----Original Message-----
From: Bernd Felsche [mailto:bernie@innovative.iinet.net.au]
Sent: Thursday, August 05, 2004 5:39 PM
To: AVR-Chat@yahoogroups.com
Subject: Re: [AVR-Chat] Re: Semi manufacturers ?


On Friday 06 August 2004 07:06, erikc wrote:
> Well, if you want to develop semiconductors, why not a RAM
> chip with a simple processor built in?  The processor would
> have its own small RAM for holding its programmes and a
> means of entering those programmes, causing them to run, and
> getting results off-chip.   You'd load all the little CPU's
> in parallel and let them loose, each one looking at the data
> on its own chip.  Now I'm not talking a CPU with some RAM,
> I'm talking 256 megabytes of RAM and maybe 4K of entirely
> separate RAM for the CPU's code.  Since you are basically
> using it to search for patterns, the CPU could be optimised
> for that.
>
> I actually think there might be a market for that kind of
> thing.

You mean a large, secondary data cache on a simple processor,
or a CPU with mega-registers?

There's still matter of partial matches spanning boundaries.

--
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!






Yahoo! Groups Links

--- In AVR-Chat@yahoogroups.com, "erikc" <firewevr@a...> wrote:

> ... A big DRAM with a little
> CPU in one corner ...

Mitsubishi tried this but nobody bought it. It makes a lot of sense 
to me. The DRAM was 128 bits wide so the CPU could clean or fill a 
cache line in one cycle. I benchmarked their chip against a much more 
powerful MIPS it ran at half the speed with the caches on, but five 
times as fast if you turned the caches off. For any processing that 
involved a lot of data it beat the MIPS plus it was smaller, cheaper 
and used less power. As I said, nobody bought it and it faded away.

You have to bear in mind that the semiconductor fab process used to 
make DRAM is different from that used to make fast logic.  Mixing 
them on a chip is non-trivial. Plus, you sacrifice generality and 
limit your market.

Graham.

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:

> Here is a link to the project:
> http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html

I think Kat is basically the victim of a hoax. If you have a little 
time, read the second (two page) document she points us to. It gives 
an example of a useful database operation and then an example of an 
operation on "Perfect Data", which is what Kat thinks she's working 
on. The "Perfect Data" operation reads like the Monty Python "Free 
Association" sketch. How anyone could think this is useful, let alone 
possible, beats the heck out of me.

The first document is longer (eight pages), but worth a read if 
you've got the time. It raises a bunch of sensible points and fails 
to address any of them except with vague assurances and quotes from 
Shakespeare. The inventors of modern database architectures are 
characterized as American geeks who messed up because Alan Turin was 
dead and they couldn't ask him what to do.

On the other hand, if Kat *does* have the couple of million dollars 
she needs to fund this project, I take it all back. Go Kat! Spend, 
spend, spend!

Graham.

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> Graham Davies wrote:
> 
> > --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> 
wrote:
> > 
> > 
> >>Here is a link to the project:
> >>http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html
> > 
> > 
> > I think Kat is basically the victim of a hoax. If you have a 
little 
> > time, read the second (two page) document she points us to. It 
gives 
> > an example of a useful database operation and then an example of 
an 
> > operation on "Perfect Data", which is what Kat thinks she's 
working 
> > on. The "Perfect Data" operation reads like the Monty 
Python "Free 
> > Association" sketch. How anyone could think this is useful, let 
alone 
> > possible, beats the heck out of me.
> > 
> > The first document is longer (eight pages), but worth a read if 
> > you've got the time. It raises a bunch of sensible points and 
fails 
> > to address any of them except with vague assurances and quotes 
from 
> > Shakespeare. The inventors of modern database architectures are 
> > characterized as American geeks who messed up because Alan Turin 
was 
> > dead and they couldn't ask him what to do.
> > 
> > On the other hand, if Kat *does* have the couple of million 
dollars 
> > she needs to fund this project, I take it all back. Go Kat! 
Spend, 
> > spend, spend!
> > 
> > Graham.
> 
> Graham, Mate where can I start, THIS IS NO HOAX, the customer is 
paying, 
> the investors are onboard. David Lawrence is actually a respected 
Dr in 
> the area of Neurophysiology (ie the workings of the brain)
> 
> The money will be available when I get to Level 5 (right now I am 
doing 
> the Level 3 work which is implimenting the system on a PC with only 
1 
> user, Level 4 is a server based system with maybe 100 users all 
within 
> one organisation, Level 5 is 1000's of users across many 
organisations)
> 
> Regards,
> 
> Kat.

Is this guy trying to develop some sort of associative memory and 
have it operate something like a human thought process?

> >>
> >>Graham, Mate where can I start, 
THIS IS NO HOAX, the customer is paying, 
> > 
> >>the investors are onboard. 
David Lawrence is actually a respected 
> > Dr in the area of Neurophysiology 
(ie the workings of the brain)
> >>The money will be available 
when I get to Level 5 (right now I am 
> > doing the Level 3 work which 
is implimenting the system on a 
PC with only 1 user, Level 4 is 
a server based system with maybe 
100 users all within one 
organisation, Level 5 is 1000's 
of users across many organisations)
> > 
> >>Regards,
> >>
> >>Kat.
> > 
> > 
> > Is this guy trying to develop 
some sort of associative memory and 
> > have it operate something 
like a human thought process?
> 
> No he isn't, I AM ;)
> 
> Kat.
> 

It sounds to me like the Holy Grail.
if you do all the research and then 
figure out how to pull this off it seems
you bridged the technological gap
that will make someone else very wealthy.

Can you patient things as you discover them ?

being totally from the outside and not really
understanding all the technicalities it just 
seems on the outside that if Edison had level
5 as a working lightbulb, and had offered
to pay at that level, he could have had 
thousands of peoples on the project at no charge.

Just my 2 cents

Dave

Graham Davies wrote:

> --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> 
> 
>>Here is a link to the project:
>>http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html
> 
> 
> I think Kat is basically the victim of a hoax. If you have a little 
> time, read the second (two page) document she points us to. It gives 
> an example of a useful database operation and then an example of an 
> operation on "Perfect Data", which is what Kat thinks she's working 
> on. The "Perfect Data" operation reads like the Monty Python "Free 
> Association" sketch. How anyone could think this is useful, let alone 
> possible, beats the heck out of me.
> 
> The first document is longer (eight pages), but worth a read if 
> you've got the time. It raises a bunch of sensible points and fails 
> to address any of them except with vague assurances and quotes from 
> Shakespeare. The inventors of modern database architectures are 
> characterized as American geeks who messed up because Alan Turin was 
> dead and they couldn't ask him what to do.
> 
> On the other hand, if Kat *does* have the couple of million dollars 
> she needs to fund this project, I take it all back. Go Kat! Spend, 
> spend, spend!
> 
> Graham.

Graham, Mate where can I start, THIS IS NO HOAX, the customer is paying, 
the investors are onboard. David Lawrence is actually a respected Dr in 
the area of Neurophysiology (ie the workings of the brain)

The money will be available when I get to Level 5 (right now I am doing 
the Level 3 work which is implimenting the system on a PC with only 1 
user, Level 4 is a server based system with maybe 100 users all within 
one organisation, Level 5 is 1000's of users across many organisations)

Regards,

Kat.


-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

erikc77099 wrote:

> --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> 
>>Graham Davies wrote:
>>
>>
>>>--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> 
> 
> wrote:
> 
>>>
>>>>Here is a link to the project:
>>>>http://www.kaqelectronics.dyndns.org/the_buzz/the_buzz.html
>>>
>>>
>>>I think Kat is basically the victim of a hoax. If you have a 
> 
> little 
> 
>>>time, read the second (two page) document she points us to. It 
> 
> gives 
> 
>>>an example of a useful database operation and then an example of 
> 
> an 
> 
>>>operation on "Perfect Data", which is what Kat thinks she's 
> 
> working 
> 
>>>on. The "Perfect Data" operation reads like the Monty 
> 
> Python "Free 
> 
>>>Association" sketch. How anyone could think this is useful, let 
> 
> alone 
> 
>>>possible, beats the heck out of me.
>>>
>>>The first document is longer (eight pages), but worth a read if 
>>>you've got the time. It raises a bunch of sensible points and 
> 
> fails 
> 
>>>to address any of them except with vague assurances and quotes 
> 
> from 
> 
>>>Shakespeare. The inventors of modern database architectures are 
>>>characterized as American geeks who messed up because Alan Turin 
> 
> was 
> 
>>>dead and they couldn't ask him what to do.
>>>
>>>On the other hand, if Kat *does* have the couple of million 
> 
> dollars 
> 
>>>she needs to fund this project, I take it all back. Go Kat! 
> 
> Spend, 
> 
>>>spend, spend!
>>>
>>>Graham.
>>
>>Graham, Mate where can I start, THIS IS NO HOAX, the customer is 
> 
> paying, 
> 
>>the investors are onboard. David Lawrence is actually a respected 
> 
> Dr in 
> 
>>the area of Neurophysiology (ie the workings of the brain)
>>
>>The money will be available when I get to Level 5 (right now I am 
> 
> doing 
> 
>>the Level 3 work which is implimenting the system on a PC with only 
> 
> 1 
> 
>>user, Level 4 is a server based system with maybe 100 users all 
> 
> within 
> 
>>one organisation, Level 5 is 1000's of users across many 
> 
> organisations)
> 
>>Regards,
>>
>>Kat.
> 
> 
> Is this guy trying to develop some sort of associative memory and 
> have it operate something like a human thought process?

No he isn't, I AM ;)

Kat.


-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

--- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:

> Graham, Mate where can I start ...

I'm more concerned about where it will end.

> ... the customer is paying, 
> the investors are onboard ...

But, apparently, they're not paying you. Yet. You will get paid if 
you can do something that has eluded the planet's best educational 
institutions (the MIT Artificial Inteligence Research Lab springs to 
mind) for decades.

> David Lawrence is actually a respected
> Dr in the area of Neurophysiology (ie
> the workings of the brain) ...

Fair enough. But read what he has written on the subject of "Perfect 
Data". Note that he had to have "a few wines" before he could come up 
with the subject material. All he is doing is brainstorming (while 
drunk) on how "biological thinking" is different from the mechanistic 
processes of computers. He isn't saying it's desirable, useful or 
possible in the way John Dowling wants to suggest. He probably thinks 
John Dowling is a total jerk or doesn't know enough about computer 
technology to have an opinion. He certainly does't seem to 
endorse "Perfect Data" in any way.

In John Dowling's own paper, he quotes a representative of 
the "Computer Industry Establishment" as saying "This is really just 
a Star Schema Database. I suggest you read a textbook John". He seems 
to think that the scorn of the establishment proves that he is onto 
something. Who does he think he is, Galileo?

Bail now, Kat, before you waste any more time! (But, as I said, if 
they are forking over millions, shut up and take it. I'll write you a 
paper extolling the virtues of "Perfect Data" myself.)

Graham.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=2228&item=4214982610&rd=1&ssPageName=WD1V

Please tell me that this is horrendously overpriced!
Alternatively, I have a copy that I would part with.. :)

Given that its $69.62 at Amazon, new, and they are asking
$327.96 for immediate sale, I'd say waaaay overpriced.

Jim

On Fri, 06 Aug 2004 11:40:25 -0500
 David VanHorn <dvanhorn@cedar.net> wrote:
> 
> 
> 
> 
> 
>
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=2228&item=4214982610&rd=1&ssPageName=WD1V
> 
> Please tell me that this is horrendously overpriced!
> Alternatively, I have a copy that I would part with.. :)
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Yahoo! Groups Sponsor
> 
> 
>   ADVERTISEMENT 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Yahoo! Groups Links
> 
> To visit your group on the web, go
> to:http://groups.yahoo.com/group/AVR-Chat/ 
> To unsubscribe from this group, send an email
> to:AVR-Chat-unsubscribe@yahoogroups.com 
> Your use of Yahoo! Groups is subject to the Yahoo! Terms
> of Service.
> 
> 
> 
> 
> 
> 
> 
> 
> 

---------------------------------------------------------------
The Think Different Store
http://www.thinkdifferentstore.com/
For All Your Mac Gear
---------------------------------------------------------------

Uau..... There is one digit more than he knows.. :-)

If anyone want's mine I will let it go for half price...

best regards,
Alexandre Guimaraes

----- Original Message -----
From: "David VanHorn" <dvanhorn@cedar.net>
To: <AVR-Chat@yahoogroups.com>; <AVR-Chat@yahoogroups.com>
Sent: Friday, August 06, 2004 1:40 PM
Subject: [AVR-Chat] Embedded programming book


>
>
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=2228&item=4214982610&
rd=1&ssPageName=WD1V

>
> Please tell me that this is horrendously overpriced!
> Alternatively, I have a copy that I would part with.. :)
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>

At 12:42 PM 8/6/2004, Alexandre Guimaraes wrote:

>Uau..... There is one digit more than he knows.. :-)
>
>If anyone want's mine I will let it go for half price...

It IS a very good book.  The guys aren't that far away from me, just on the other side of indiana. Never have run into them though.

Hi, Dave

    It is indeed a great book but I already have read it and for us$160 I
can buy one more and still keep the other us$80 ;-)

    I recommend it to anyone getting involved in using C for embedded
systems. It really helped me a lot when I started using Codevision C for
some projects. I never used C before and was a "assembler only" guy before
Codevision. I still use assembler a lot, even in C projects but to make fast
prototypes codevision is much faster to code than assembler..

best regards,
Alexandre Guimaraes

----- Original Message -----
From: "David VanHorn" <dvanhorn@cedar.net>
To: <AVR-Chat@yahoogroups.com>; <AVR-Chat@yahoogroups.com>
Sent: Friday, August 06, 2004 2:57 PM
Subject: Re: [AVR-Chat] Embedded programming book


> At 12:42 PM 8/6/2004, Alexandre Guimaraes wrote:
>
> >Uau..... There is one digit more than he knows.. :-)
> >
> >If anyone want's mine I will let it go for half price...
>
> It IS a very good book.  The guys aren't that far away from me, just on
the other side of indiana. Never have run into them though.

>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>

----- Original Message -----

From: "Graham Davies" <YahooGroups@ecrostech.com>
To: <AVR-Chat@yahoogroups.com>
Sent: Friday, August 06, 2004 14:24
Subject: [AVR-Chat] DRAM with CPU (was Re: Semi
manufacturers ?)


> --- In AVR-Chat@yahoogroups.com, "erikc" <firewevr@a...>
wrote:
>
> > ... A big DRAM with a little
> > CPU in one corner ...
>
> Mitsubishi tried this but nobody bought it. It makes a lot
of sense
> to me. The DRAM was 128 bits wide so the CPU could clean
or fill a
> cache line in one cycle. I benchmarked their chip against
a much more
> powerful MIPS it ran at half the speed with the caches on,
but five
> times as fast if you turned the caches off. For any
processing that
> involved a lot of data it beat the MIPS plus it was
smaller, cheaper
> and used less power. As I said, nobody bought it and it
faded away.

Perhaps it was an idea ahead of its time.  Considering all
the databases in use these days and the demands made on
them, it seems that having memory with integral search
capability would be a moneymaker.  Especially if someone
came out with a line of machines specifically for managing
databases.  (Something like SQL in hardware).

>
> You have to bear in mind that the semiconductor fab
process used to
> make DRAM is different from that used to make fast logic.
Mixing
> them on a chip is non-trivial. Plus, you sacrifice
generality and
> limit your market.

Umm, isn't there some kind of logic on the chip for decoding
and timing?  Is it somehow different (other than
arrangement) than logic used for computation?  Or am I
missing something here?

> Graham.

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

----- Original Message ----- >
>     I recommend it to anyone getting involved in using C
for embedded
> systems. It really helped me a lot when I started using
Codevision C for
> some projects. I never used C before and was a "assembler
only" guy before
> Codevision. I still use assembler a lot, even in C
projects but to make fast
> prototypes codevision is much faster to code than
assembler..
>
> best regards,
> Alexandre Guimaraes

Is C the only high-level language for the AVR?  I would be
especially interested in Pascal, Modula or any of the other
Wirth-style languages out there.

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

----- Original Message -----

From: "Alexandre Guimaraes" <listas@logikos.com.br>
To: <AVR-Chat@yahoogroups.com>
Sent: Saturday, August 07, 2004 04:12
Subject: Re: [AVR-Chat] Embedded programming book


> Hi,
>
> > Is C the only high-level language for the AVR?  I would
be
> > especially interested in Pascal, Modula or any of the
other
> > Wirth-style languages out there.
>
>     I also bought embedded pascal wich is fine but not as
"complete" as
> codevision and the code generator is much, much worse..

Hmm, that kind of bites.  But then again, C is a kind of
lingua franca.

>     I searched for many options and for the first time in
a 20 year carrer I
> really had to learn C and live with it's weird sintax.
There are not better
> options for real world projects that I have seen... It is
either assembler
> or C or both together as I do.

I am constantly criticied that my C code looks like pascal
(after I've declared a few macros, etc.)  ;-)

> best regards,
> Alexandre Guimaraes



Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

----- Original Message -----

From: "Alexandre Guimaraes" <listas@logikos.com.br>
To: <AVR-Chat@yahoogroups.com>
Sent: Saturday, August 07, 2004 04:21
Subject: [AVR-Chat] Timing Profiler.


> Hi,
>
>     Does anyone have a ready made structure to make a
"profiler" for a
> program that has the usual embedded structure of some
timer routines that
> take quite some time to execute and a small main loop ??
>
>     I have an application where I maybe pretty close to
use all CPU time
> available and I would love to find a way to check how
close I am without
> having to count all the cycles in all possible paths of
the program. That
> get's even harder with the timer interrupts.
>
>     Any ideas ??
>
> best regards,
> Alexandre Guimaraes

If you have a spare I/O pin, you could set that pin on entry
to a function and clear it on exit.  You can then either
observe the pulse width on a scope or make an integrator and
read the duty cycle directly on a volt meter.

Erikc  - firewevr@airmail.net
///
"An Fhirinne in aghaidh an tSaoil."
"The Truth against the World."
        -- Bardic Motto
///I did NOT lose my mind!  It's backed up on the server
(and the network's down....)

Dave Mucha wrote:

>>>>Graham, Mate where can I start, 
> 
> THIS IS NO HOAX, the customer is paying, 
> 
>>>>the investors are onboard. 
> 
> David Lawrence is actually a respected 
> 
>>>Dr in the area of Neurophysiology 
> 
> (ie the workings of the brain)
> 
>>>>The money will be available 
> 
> when I get to Level 5 (right now I am 
> 
>>>doing the Level 3 work which 
> 
> is implimenting the system on a 
> PC with only 1 user, Level 4 is 
> a server based system with maybe 
> 100 users all within one 
> organisation, Level 5 is 1000's 
> of users across many organisations)
> 
>>>>Regards,
>>>>
>>>>Kat.
>>>
>>>
>>>Is this guy trying to develop 
> 
> some sort of associative memory and 
> 
>>>have it operate something 
> 
> like a human thought process?
> 
>>No he isn't, I AM ;)
>>
>>Kat.
>>
> 
> 
> It sounds to me like the Holy Grail.
> if you do all the research and then 
> figure out how to pull this off it seems
> you bridged the technological gap
> that will make someone else very wealthy.
> 
> Can you patient things as you discover them ?
> 
> being totally from the outside and not really
> understanding all the technicalities it just 
> seems on the outside that if Edison had level
> 5 as a working lightbulb, and had offered
> to pay at that level, he could have had 
> thousands of peoples on the project at no charge.
> 
> Just my 2 cents
> 
> Dave

Yes Dave, my company will hold the copyright on the technology :) but I 
am not going to patent it and will make it avaliable for the good of 
human kind (or if HAL wins the downfall of human kind ;)


-- 
---------------------------------------------------------------
K.A.Q. Electronics	Website: www.kaqelectronics.dyndns.org
IM: Yahoo: PinkyDwaggy  MSN: katinka@kaqelectronics.dyndns.org
For Everything Electronics     Phone: 0419 923 731
---------------------------------------------------------------

Hi,

> Is C the only high-level language for the AVR?  I would be
> especially interested in Pascal, Modula or any of the other
> Wirth-style languages out there.

    I also bought embedded pascal wich is fine but not as "complete" as
codevision and the code generator is much, much worse..

    I searched for many options and for the first time in a 20 year carrer I
really had to learn C and live with it's weird sintax. There are not better
options for real world projects that I have seen... It is either assembler
or C or both together as I do.

best regards,
Alexandre Guimaraes

At 10:12 PM 8/6/2004, Alexandre Guimaraes wrote:

>Hi,
>
>> Is C the only high-level language for the AVR?  I would be
>> especially interested in Pascal, Modula or any of the other
>> Wirth-style languages out there.
>
>    I also bought embedded pascal wich is fine but not as "complete" as
>codevision and the code generator is much, much worse..
>
>    I searched for many options and for the first time in a 20 year carrer I
>really had to learn C and live with it's weird sintax. There are not better
>options for real world projects that I have seen... It is either assembler
>or C or both together as I do.

My man! 

That's exactly how I see it.

Hi,

    Does anyone have a ready made structure to make a "profiler" for a
program that has the usual embedded structure of some timer routines that
take quite some time to execute and a small main loop ??

    I have an application where I maybe pretty close to use all CPU time
available and I would love to find a way to check how close I am without
having to count all the cycles in all possible paths of the program. That
get's even harder with the timer interrupts.

    Any ideas ??

best regards,
Alexandre Guimaraes

At 12:21 AM 8/7/04 -0300, you wrote:
>     Does anyone have a ready made structure to make a "profiler" for a
>program that has the usual embedded structure of some timer routines that
>take quite some time to execute and a small main loop ??
>
>     I have an application where I maybe pretty close to use all CPU time
>available and I would love to find a way to check how close I am without
>having to count all the cycles in all possible paths of the program. That
>get's even harder with the timer interrupts.
>
>     Any ideas ??

Do you want an absolute maximum usage number of just a good estimate?

If you've got a spare pin or two you can get a pretty good idea with a 
logic analyzer,  a oscilloscope. or even a DMM.  Just set the pin high 
during the period you want to measure.  With multiple pins and an 
oscilloscope or logic analyser you can even get a good indication of 
latency sources and the effects of various interrupts.  Has the advantage 
of being simple and relatively non-intrusive (not much of a "Heisenburg" 
effect).

More complicated, read one of your timers on entry to and exit from code 
regions and accumulate the reading for later analysis (maybe min/max/average).

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,
be they legal, genetic, or physical.  If you don't believe me, try to
chew a radio signal. "

                         Kelvin Throop, III

>
>If you've got a spare pin or two you can get a pretty good idea with a 
>logic analyzer,  a oscilloscope. or even a DMM.  Just set the pin high 
>during the period you want to measure. 

I was just about to suggest that. I've used this a fair bit.

On Saturday 07 August 2004 10:25, Kathy Quinlan wrote:
> Dave Mucha wrote:
> >
> > some sort of associative memory and

Smart storage node; yes... with "infinite" cross-links to other
nodes.

Synaptic network sort of thing. Each synapse matches a particular
pattern and "fires". When you get plenty of synapses firing, there's
a recognition that something important is happening.  As a
particular data set passes through the synaptic network,
correlations are formed between firing synapses. 

It's plausible (IMHO necessary) for synapses to be generated
automatically for the on a significant firing. One that recognizes
the larger pattern immediately without having to wait for the
correlation in others to be recognized.

Data sets which don't automatically relate to 'significant' synaptic
response, are placed in memory. The human brain regularly "discards"
huge amounts of data because there's insignificant synaptic
response... some of that usually occurs during "down-time", when the
brain replays remembered events through the low-priority synapses.
Synaptic connections can be made during that phase.

Such "dream states", when the brain is "disconnected" from the real
world, are the foundation of creativity.

The dream state is also where the synaptic connections are likely
"tidied up", removing "unimportant", remembered correlations and
those for which specific synapses now exist.

It's a huge task to try to emulate that sort of thing because the
whole thing is so vague; perhaps beyond simple implementation in
binary states.

[snip]

> > Can you patient things as you discover them ?

> Yes Dave, my company will hold the copyright on the technology :) but I
> am not going to patent it and will make it avaliable for the good of
> human kind (or if HAL wins the downfall of human kind ;)

You must Patent the invention to give you the right to dedicate it
to the public. The existence of the Patent record establish the
prior art so that even USPTO will find it before approving a
spurious application for a similar "invention".

The alternative is to widely publish everything; but still leaves
the possibility of a Patent being granted to other parties. It then
becomes a case for litigation for others to show prior art to the
satisfaction of a court.

Classic example is the XOR cursor Patent which appears to have been
applied for and granted well after the prior art existed. Leaving
out the matter of it being bleeding obvious to anybody skilled in
the art, the Patent was granted because the examiners found nothing
conflicting in other Patents.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

Just to check it out... A free pin a 1K resistor and a 100nf cap to
ground. VCC at the junction is 100% of the time and vcc/2 is around 50%.
Sounds nice.... All the other times that I needed this I used a 32bit timer
summing up a cpu timer but that disturbs the results a little.. That should
do the trick...

> >
> >If you've got a spare pin or two you can get a pretty good idea with a
> >logic analyzer,  a oscilloscope. or even a DMM.  Just set the pin high
> >during the period you want to measure.
>
> I was just about to suggest that. I've used this a fair bit.
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>

On Saturday 07 August 2004 02:33, Graham Davies wrote:
> --- In AVR-Chat@yahoogroups.com, Kathy Quinlan <kat-yahoo@k...> wrote:
> > Graham, Mate where can I start ...
>
> I'm more concerned about where it will end.
>
> > ... the customer is paying,
> > the investors are onboard ...
>
> But, apparently, they're not paying you. Yet. You will get paid if
> you can do something that has eluded the planet's best educational
> institutions (the MIT Artificial Inteligence Research Lab springs to
> mind) for decades.

Their minds may be too highly trained.

Seriously.

Prior knowledge and abilities can be a handicap as much as an aid
because all that "experience" prejudices what is perceived.

> > David Lawrence is actually a respected
> > Dr in the area of Neurophysiology (ie
> > the workings of the brain) ...

> Fair enough. But read what he has written on the subject of "Perfect
> Data". Note that he had to have "a few wines" before he could come up
> with the subject material. All he is doing is brainstorming (while
> drunk) on how "biological thinking" is different from the mechanistic
> processes of computers. He isn't saying it's desirable, useful or
> possible in the way John Dowling wants to suggest. He probably thinks
> John Dowling is a total jerk or doesn't know enough about computer
> technology to have an opinion. He certainly does't seem to
> endorse "Perfect Data" in any way.

> In John Dowling's own paper, he quotes a representative of
> the "Computer Industry Establishment" as saying "This is really just
> a Star Schema Database. I suggest you read a textbook John". He seems
> to think that the scorn of the establishment proves that he is onto
> something. Who does he think he is, Galileo?

Maybe he has trouble in expressing his concepts in language that
you, and others skilled in the art can understand. That's the real
problem created by too many specialists and jargon; and not enough
individuals who _understand_ the concepts of disparate disciplines.

NB: I don't wish to add to the credibility of the "Perfect Data"
thingy. Just venturing a guess as to why it may look like a hoax.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

On Saturday 07 August 2004 09:36, erikc wrote:
>
> Perhaps it was an idea ahead of its time.  Considering all
> the databases in use these days and the demands made on
> them, it seems that having memory with integral search
> capability would be a moneymaker.  Especially if someone
> came out with a line of machines specifically for managing
> databases.  (Something like SQL in hardware).

Experts in the field tried that in the 1980's, IIRC. The architects
of SQL, if you want. Instead, they realized that it was more
realistic and practical to try to do it well in software _first_ on
generic hardware that was rapidly growing to be more capable by the
month.

You may have heard of a company called Sybase.

Implementing new features is software is still trivial compared to
doing it in hardware.

Triggers and stored procedures are two relevant concepts.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

The problem/strength of C is that it is pretty close to the metal.  It
neither holds your hand nor gets in the way (for the most part), which, is
why C and assembly is such a kick-ass combination and finds widespread use
across *all* architectures.  Of course, one can write incomprehensible C,
or, one can write lucid code.   Skill and style make the difference.

It just so happens that the AVR is particularly suited for C and embedded
style programs.  No wonder since the AVR was the last 8 bit architecture
developed.  Nothing like 20 years of hindsight to make a tight processor.  A
classic CISC, like the Hitachi H8/300, produces amazingly tight C code, but
some of those instructions take forever to execute...

Cheers!

-----Original Message-----
From: David VanHorn

At 10:12 PM 8/6/2004, Alexandre Guimaraes wrote:

>Hi,
>
>> Is C the only high-level language for the AVR?  I would be
>> especially interested in Pascal, Modula or any of the other
>> Wirth-style languages out there.
>
>    I also bought embedded pascal wich is fine but not as "complete" as
>codevision and the code generator is much, much worse..
>
>    I searched for many options and for the first time in a 20 year carrer
I
>really had to learn C and live with it's weird sintax. There are not better
>options for real world projects that I have seen... It is either assembler
>or C or both together as I do.

My man!

That's exactly how I see it.





Yahoo! Groups Links

--- In AVR-Chat@yahoogroups.com, Bernd Felsche <bernie@i...> wrote:

> Their minds [at the MIT AI Lab]
> may be too highly trained.

I hear you. And, so do they. The AI people are the most active in 
cross-discipline brainstorming and just about every other way of 
looking at problems from new angles.

> Maybe he has trouble in expressing
> his concepts in language ...

I don't think you've read the paper.

Graham.

>Experts in the field tried that in the 1980's, IIRC. The architects
>of SQL, if you want. Instead, they realized that it was more
>realistic and practical to try to do it well in software _first_ on
>generic hardware that was rapidly growing to be more capable by the
>month.
>
>You may have heard of a company called Sybase.
>
>Implementing new features is software is still trivial compared to
>doing it in hardware.
>
>Triggers and stored procedures are two relevant concepts.

Chip manufacturing today, is entirely different from what it was "in the
80's". The battle for hardware vs. software is shifting as (re-)programmable
chip-fabric is dropping in price, and really speeding up... I've noted
several FPGA-based computers arising, and paralell processing is now beeing
put to a different level.
I believe that tomorrow's computers will be able to rearrange their internal
structure, or parts of it, to suite their immediate needs. I also believe
that this is comparable to organic neural networks...

However, this kind of structure requires a new generation operating-system.
AFAIK no currently available OS can handle this kind of parallell
processing...

So, Kat, I'll say. As long as you get reasonably payed (and secure your
copyrights), and believe - go for it!

However - this discussion kind'of get quite OT (for AVR-Chatting, that is
:), so I'll second Graham Davies' suggestion on creating a new
yahoo-group...

- Øystein Fallo

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.732 / Virus Database: 486 - Release Date: 29.07.2004

On Saturday 07 August 2004 21:31, Graham Davies wrote:
> --- In AVR-Chat@yahoogroups.com, Bernd Felsche <bernie@i...> wrote:
> > Their minds [at the MIT AI Lab]
> > may be too highly trained.

> I hear you. And, so do they. The AI people are the most active in
> cross-discipline brainstorming and just about every other way of
> looking at problems from new angles.

> > Maybe he has trouble in expressing
> > his concepts in language ...

> I don't think you've read the paper.

You reckon?

Confusing 3NF with chaos is a something only an amateur would make.
1NF is the most-verbose means of storing data, and the one in which
relationships are least-obvious. 

Yet that is the "Perfect Data" model. 20 out of 35 table entries are
undefined in the example, producing very low information density.
What aggravates the situation is that every detail will include a
great deal of redundant data; data which are naturally "compressed"
in a well-designed relational database.

The RDBMS model of data is just one way of managing data. The closer
the model to the real world, the more complex it gets.

Jumbling all data into a single "table" isn't going to solve any
technological problems; maybe problems of perception and prejudice.
What it will do is create huge computing overheads because there's
no relationship to entities established in Perfect Data. If for
example, an entity changes its name, then every record containing
the old name has to be changed to the new. 

In a well-designed RDBMS, one changes one row that contains that
entity's "name" property. All future data retrievals will then
produce the current company name... a change record may be
desirable; but that means at most 3 records changed/added per entity
property change.

In the past decade and a half, I've gone through a number of poorly
designed RDBMS that were closer to the "Perfect Data" model; and
every time they had to change an entity attribute, it's been a
mountain-moving exercise to implement because all the transactional
histories required for audit purposes have had to reflect the
change. That sort of thing is so ugly; and prone to unleashing all
sorts of gremlins when it takes a production database offline for 2
days or more to do the data change in tens of millions of records.
Of course; they always want an audit trail!


BTW: Implying that the Orbital Engine has changed the way in which
people "view the future of the combustion motor car" is simply
ludicrous.  It demonstrates a great lack of knowledge and
understanding in that realm. If he'd said that the Orbital Engine
changed the way in which the Orbital Engine Company viewed its
future, then that would be spot-on.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

At 01:22 AM 8/7/04 -0300, you wrote:
>     Just to check it out... A free pin a 1K resistor and a 100nf cap to
>ground. VCC at the junction is 100% of the time and vcc/2 is around 50%.
>Sounds nice.... All the other times that I needed this I used a 32bit timer
>summing up a cpu timer but that disturbs the results a little.. That should
>do the trick...

That'll do it, although I'd use a larger resistor (5-10K or more) to reduce 
any loading effects.  I'd also sink rather than source but either should 
work.  I've used this a lot for verification and debugging myself.  It's 
really useful when verifying responses to an event or faster inner loops 
for jitter and elapsed time.

I also wouldn't bother with the cap.  If you are using a DMM let it do the 
input filtering and if you are using an oscilloscope then you can get a 
direct reading of usage and period.  Of course if you are using a poor DMM 
or want to read the average directly on an oscilloscope without averaging......

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,
be they legal, genetic, or physical.  If you don't believe me, try to
chew a radio signal. "

                         Kelvin Throop, III

Another useful tidbit, is to change it around a little, and have it pulse N times in ISR-X and N+1 times in ISR-Y. This way, you can see exactly which ISR is firing. 

Also, as you gain confidence in the code, take out the pulses on the active ISRs, but leave the pulses in for the inactive vectors. Now, if you ever have an ISR active that shouldn't be, you'll start getting pulses. 

The one pin pulse trick can really help.

I did a routine called "pong" that outputs a byte to a single pin, with wide pulses for 1, and narrow for 0, but the byte is always the same width. I've gone as far as thirty-odd bytes this way, since I have an excellent DSO and logic analyzer to read the pulses with.

> I do not plan to have one CPU to search the whole
10Tb > I was planing on using 100 parallel CPU's each
looking > after 100Gb of ram.

Speaking from experience (In a past life, I configured
and tuned IBM SP/2 machines...) a data set that is
this large could only be processed using a massively
parallel super computer. Something along the line of
an IBM SP/2 (If speed is your need and price is no
option, you'll want 128 bit RISC cpu's...) or a
beowulf cluster composed of off the shelf hardware (If
speed is good, but economy is better...). 

Specifically, the type of data sets you have described
cannot even be measured (Which is the starting point
for processing any data set this large...), much less
processed by any currently available cpu that is
operating as a distinctly autonomous computational
device. Well, I take that back, it could be done, but
even the accuracy of the measurements of a data set
that large would be suspect and the time it took to
realize an answer would most likely render the data
usesless (The measurement would be suspect because the
time it took to measure it would prohibit redundancy).
Further, even if you were only calculating the MD5
checksum of a data set this large on a single cpu, the
process would run for a very_long_time: Months
(Plural).

Regards

Marvin Dickens
Alpharetta, Georgia  USA


=====
Registered Linux User No. 80253
If you use linux, get counted at: 
http://www.linuxcounter.org


		
__________________________________
Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages!
http://promotions.yahoo.com/new_mail

I have to maintain some equipment which uses the AT90S4414 (40 pin DIP) 
which has been obsolete for a few years. The usual UK suppliers (Farnell, 
CPC, RS) no longer have them (nor the AT90S8515 which might be an 
alternative).

Is anyone aware of any (new) old stock lying in the UK - just want ten or 
twelve to last me until I retire :-)    I see that Digikey is still listing a few 8515s 
but their 12GBP delivery charge is a turn-off... 




George Smith

----- Original Message ----- 

From: George Smith
To: AVR-Chat@yahoogroups.com
Sent: Wednesday, January 12, 2005 1:48 PM
Subject: [AVR-Chat] AT90S4414 old stock in UK?


I have to maintain some equipment which uses the AT90S4414 (40 pin DIP)
which has been obsolete for a few years. The usual UK suppliers (Farnell,
CPC, RS) no longer have them (nor the AT90S8515 which might be an
alternative).

Is anyone aware of any (new) old stock lying in the UK - just want ten or
twelve to last me until I retire :-)    I see that Digikey is still listing 
a few 8515s
but their 12GBP delivery charge is a turn-off...


What about the ATmega8515?

Leon
--
Leon Heller, G1HSM
http://www.geocities.com/leon_heller



-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.300 / Virus Database: 265.6.9 - Release Date: 06/01/2005

Thanks for the suggestions about the atmega8515 - unfortunately it has to be a straight 
plug-in replacement. I am not allowed to touch the code (even assuming the correct 
source version could be located...).
As an aside, there are actually two completely different pieces of hardware, in two 
different companies both using this chip. The first has not reported any failures - but I it 
would not hurt to have a couple of spares available. I have a unit on the bench from the 
second site, where the program has been corrupted, and even re-programming and 
verifying, it does not work in the pcb. I suspect it might have been damaged by static - 
there is a smartcard socket on the pcb so there is a possibility that this is an entry point 
for static charge.


George Smith

On Wed, 12 Jan 2005 13:48:39 -0000, you wrote:

>
>I have to maintain some equipment which uses the AT90S4414 (40 pin DIP) 
>which has been obsolete for a few years. The usual UK suppliers (Farnell, 
>CPC, RS) no longer have them (nor the AT90S8515 which might be an 
>alternative).
>
>Is anyone aware of any (new) old stock lying in the UK - just want ten or 
>twelve to last me until I retire :-)    I see that Digikey is still listing a few 8515s 
>but their 12GBP delivery charge is a turn-off... 

Have you looked at the ATMega8515 - this is mostly compatible. 
Note however that if you use EEPROM, the mega8515 is NOT 100% compatible as it has an eeprom
high-order adddress register (not present on 4414) which is not initialised to a known state at
power-up, so you will need to add code to initialise it. 

Atmel chips do appear on ebay quite often.
How about this : 
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=50915&item=3865385533&rd=1

(Most  sellers listing 'US only' will ship to UK if asked)

OK I bought a couple of ATMEG8515 and programmed them with the hex file (and after 
setting the fuses correctly) it seems to work properly. So many thanks for the 
suggestion - I was really expecting a lot more problems...

regards

George Smith

On Fri, 14 Jan 2005 17:02:56 -0000, you wrote:

>
>OK I bought a couple of ATMEG8515 and programmed them with the hex file (and after 
>setting the fuses correctly) it seems to work properly. So many thanks for the 
>suggestion - I was really expecting a lot more problems...

WARNING!!! 

Does your product use the internal eeprom ? 

If so, you WILL have problems, which may well only surface later (like when the product is in the
field, on th eother side of the world...!) 

This is because the high-order eeprom address register (not present on the 4414) is not initialised
on reset, so what you effectively get is  a random page-swap. You can fire up the device and write
to the eeprom, and it will l read back OK, but on the next power-up here is a 50% changce it will
power up in the wrong bank, and accessthe wrong eeprom location.

On 14 Jan 2005, at 17:12, Mike Harrison wrote:

> WARNING!!!  
> Does your product use the internal eeprom ?

Thanks for that Mike - since your post(s) I have been able to examine the 
source for both pieces of equipment. The one with the current failure uses the 
EEPROM only in an undocumented (as far as the customer is concerned) 
debug mode which requires soldering a link to activate it. I will have to write a 
warning in the internal docs about this.
The second piece of equipment can optionally use the EEPROM to identify 
each unit and /may/ be used by the customer, since the commands to 
read/write the ID are in the handbook. Again I will update the internal docs to 
highlight that a code mod will be required to init the EEARH reg if there are 
any future failures.
Again, sincere thanks for the heads-up - I would definitely have missed that ... 

regards,


George Smith

We are working up a Mega128 based board that will have both 5V and 3V 
parts. The 128 will run at 5V.  Anyone have suggestions for converting the 
5V logic levels to/from the 128 to the 3V levels needed for some of the 
other components.  Space is at a premium.

Jim

Hi, Jim.

You wrote:

> We are working up a Mega128 based board that will have both 5V and 3V
> parts. The 128 will run at 5V.  Anyone have suggestions for converting the
> 5V logic levels to/from the 128 to the 3V levels needed for some of the
> other components.  Space is at a premium.

> Jim

You can read an Atmel Application note "Interfacing the 3-volt
DataFlash with a 5-volt System" in section Memory/DataFlash/Application
Notes on te atmel website.

-- 

 .Valerij Avrealij

Jameco has a 74ALVC164245 level translator.

Mike


--- In AVR-Chat@yahoogroups.com, Jim White <jim@c...> wrote:
> We are working up a Mega128 based board that will have both 5V and 
3V 
> parts. The 128 will run at 5V.  Anyone have suggestions for 
converting the 
> 5V logic levels to/from the 128 to the 3V levels needed for some of 
the 

> other components.  Space is at a premium.
> 
> Jim

At 08:56 AM 1/16/05 -0700, you wrote:
>We are working up a Mega128 based board that will have both 5V and 3V
>parts. The 128 will run at 5V.  Anyone have suggestions for converting the
>5V logic levels to/from the 128 to the 3V levels needed for some of the
>other components.  Space is at a premium.

Analog Devices has a nice bidirectional level translator (8 lines in a 4 mm 
square package) but it doesn't look like it's available yet.

http://www.analog.com/en/prod/0%2C2877%2CADG3308%2C00.html

Robert


" 'Freedom' has no meaning of itself.  There are always restrictions,
be they legal, genetic, or physical.  If you don't believe me, try to
chew a radio signal. "

                         Kelvin Throop, III

Jim White wrote:
> We are working up a Mega128 based board that will have both 5V and 3V 
> parts. The 128 will run at 5V.  Anyone have suggestions for converting the 
> 5V logic levels to/from the 128 to the 3V levels needed for some of the 
> other components.  Space is at a premium.
> 
> Jim

For 5V_cpu->3.3V_ttl use a series resistor (less resistance for more speed).
for 3.3V_ttl->5V_cpu use pullup resistors to 3.3 or 5V.

For more speed and less power consumption, use a 5/3.3V interface chip.

I have similar design. I'm using SN74LVTH244 to interface Atmel 5V 
part to 3.3V comm device.
-- D.G.

--- In AVR-Chat@yahoogroups.com, Jim White <jim@c...> wrote:
> We are working up a Mega128 based board that will have both 5V and 
> Jim

On Jan 16, 2005, at 10:40 PM, DG (dee-gee) wrote:

>
>
> I have similar design. I'm using SN74LVTH244 to interface Atmel 5V
> part to 3.3V comm device.

Another useful characteristic of these BiCMOS family parts is that they 
can be attached to a live bus while powered off without any ill 
effects.  They do not contain the standard ESD diode to the positive 
rail and remain high impedance even when off.  ABT parts are in the 
same family as these.

Mike

You can do it cheaply by using two of the non inverting hex buffers 
and pull up resistors. The sn74lvc1g07 is a single gate, runs on 
3vdc and has open drain. depends on how many signals you have to 
translate. For something like SPI signals across the two nodes you 
could use the low voltage version of the hex open drain buffer and 
the standard ttl on the other side with pull ups. I think Maxim Semi 
makes a single part for both sides of the fence but they cost more.

Al Welch

--- In AVR-Chat@yahoogroups.com, Mike Murphree <mike@t...> wrote:
> 
> On Jan 16, 2005, at 10:40 PM, DG (dee-gee) wrote:
> 
> >
> >
> > I have similar design. I'm using SN74LVTH244 to interface Atmel 
5V
> > part to 3.3V comm device.
> 
> Another useful characteristic of these BiCMOS family parts is that 
they 
> can be attached to a live bus while powered off without any ill 
> effects.  They do not contain the standard ESD diode to the 
positive 
> rail and remain high impedance even when off.  ABT parts are in 
the 

> same family as these.
> 
> Mike

Thanks for all the good ideas on this.  We are looking at the Maxim parts.

Jim