hog

J

John Larkin

Guest
http://www.theregister.co.uk/2005/01/19/intel_smithfield_power/

John
 
On Thu, 20 Jan 2005 01:50:31 -0000, "john jardine"
<john@jjdesigns.fsnet.co.uk> wrote:


Wonder which weasel dreamed up the dual idea?. Why stop at two-on-a-chip,
lets go the whole hog and bung 32 in there. Talks can then start with
microsoft, who'll wet themselves at the potential profits of a yet to come,
'fully parallel' windows2010 (tm), C++++++(tm), MultiWord (tm), 3D Excel
(tm), etc, etc. Also no end of joy for the commodity markets of memory,
PSUs, hard discs and motherboards.

A processor-per-process makes a lot of sense. There's really no reason
not to have, say, 256 register-rich CPUs on a chip, with a nice
central rondezvous/semaphore thing in the middle to organize them.
That might allow a good, secure, roadblock-free OS to finally be
created. But not by Microsoft, of course, who would just use the
architecture to make a bigger mess, and license it per CPU.

John
 
In article <jamvu0trhcq3v121mhfr1bbch09bhj3p0m@4ax.com>,
John Larkin <jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:
[...]
A processor-per-process makes a lot of sense. There's really no reason
not to have, say, 256 register-rich CPUs on a chip, with a nice
central rondezvous/semaphore thing in the middle
The "central" and "middle" may be the downfall. It may be a lot better to
keep the inter-processor communications spread around the chip and
allocate tasks that interact to CPUs that are nearer each other. It does
make for sort of a "traveling salesman" problem to assign the tasks but it
would remove the need for a 256 ported memory.


Anyone who really wants to can make more than one CPU on a chip since the
VHDL model for an 8051 takes up much less than 1/2 a modest FPGA.



--
--
kensmith@rahul.net forging knowledge
 
"Keith Williams" <krw@att.bizzzz> wrote in message
news:MPG.1c5985a66f54080998987a@news.individual.net...
In article <csn2b3$v1o$1@newsg4.svr.pol.co.uk>,


Wrong side of the bed this mornin'? Why don't you think the consumer
will benefit? If they don't Intel surely won't either.

--
Keith
Like sheep to slaughter, consumers will do as wintel bids them.
I'm feeling ratty as some POS virus scumware has locked me out of IE. It's
loading the browser, dialling out and showing adverts every 2 minutes.
Nothing can kill it. There seems about 4 POSs working together to regenerate
any deletions and to constantly randomise their .dll .exe filenames.
regards
john
 
In article <csp5cu$120$1@news7.svr.pol.co.uk>,
john@jjdesigns.fsnet.co.uk says...
"Keith Williams" <krw@att.bizzzz> wrote in message
news:MPG.1c5985a66f54080998987a@news.individual.net...
In article <csn2b3$v1o$1@newsg4.svr.pol.co.uk>,


Wrong side of the bed this mornin'? Why don't you think the consumer
will benefit? If they don't Intel surely won't either.

--
Keith

Like sheep to slaughter, consumers will do as wintel bids them.
Nonsense. In case you hadn't noticed, there are alternatives. ...also
going dual (or better) processors.

I'm feeling ratty as some POS virus scumware has locked me out of IE. It's
loading the browser, dialling out and showing adverts every 2 minutes.
Nothing can kill it. There seems about 4 POSs working together to regenerate
any deletions and to constantly randomise their .dll .exe filenames.
regards
AH, so you're careless enough to use Windows, IE, *and* Lookout
Express, then blame Intel for your virus woes because they're
_following_ everyone else on the path to dual processors. Gotcha. ;-)

--
Keith
 
On Thu, 20 Jan 2005 20:55:00 -0000, "john jardine"
<john@jjdesigns.fsnet.co.uk> wrote:

"Keith Williams" <krw@att.bizzzz> wrote in message
news:MPG.1c5985a66f54080998987a@news.individual.net...
In article <csn2b3$v1o$1@newsg4.svr.pol.co.uk>,


Wrong side of the bed this mornin'? Why don't you think the consumer
will benefit? If they don't Intel surely won't either.

--
Keith

Like sheep to slaughter, consumers will do as wintel bids them.
I'm feeling ratty as some POS virus scumware has locked me out of IE. It's
loading the browser, dialling out and showing adverts every 2 minutes.
Nothing can kill it. There seems about 4 POSs working together to regenerate
any deletions and to constantly randomise their .dll .exe filenames.
regards
john
Something like that happened to me a few weeks ago.

Boot up in Safe Mode, run NAV and follow the directions. I also found
"Scan Spyware" a useful tool.

It took me several passes to get rid of everything.

Helpful hint: Clear Recycle Bin each pass, or you'll go nuts ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
 
On Thu, 20 Jan 2005 23:15:02 -0000, "john jardine"
<john@jjdesigns.fsnet.co.uk> wrote:

"Jim Thompson" <thegreatone@example.com> wrote in message
news:8680v01ue1gmluof0va7oigt44skc0qnf1@4ax.com...

Something like that happened to me a few weeks ago.

Boot up in Safe Mode, run NAV and follow the directions. I also found
"Scan Spyware" a useful tool.

It took me several passes to get rid of everything.

Helpful hint: Clear Recycle Bin each pass, or you'll go nuts ;-)

...Jim Thompson

Went into DOS and deleted a bunch of suspect filenames and used Spybot to
delete some more. Think I manually got the ringleader (only 35kB!). Been
clear for the past hour (yippee!). Trust my luck to find I've amputated
*too* much.
I've saved your note for the event :)
regards
john
Good luck!

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
 
In article <h970v01h47pposfisp9nvl7kl06iqe7oh6@4ax.com>,
John Larkin <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:
[...]
But gaah! Who would want multiple 8051's?
Me.

My company used to sell a product that had 7 8051s in it. Each thing that
needed doing got its own micro.

--
--
kensmith@rahul.net forging knowledge
 
On Fri, 21 Jan 2005 01:47:07 +0000 (UTC), kensmith@green.rahul.net
(Ken Smith) wrote:

In article <h970v01h47pposfisp9nvl7kl06iqe7oh6@4ax.com>,
John Larkin <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:
[...]
But gaah! Who would want multiple 8051's?

Me.

My company used to sell a product that had 7 8051s in it. Each thing that
needed doing got its own micro.
That's the idea: a processor per process.

Can you run a multitasking RTOS on an 8051 anyhow? I wrote one once
for the 6800 (that's *not* 68000) and it was a moderate nuisance. A
6800 won't even push the index register. I did it on loose-leaf paper
in Juneau, Alaska... took almost two weeks, but, er, some other stuff
was going on. I mailed it back to work, they keyed it in, and it
worked, with one bug.

John
 
On Thu, 20 Jan 2005 18:11:26 -0700, uvcceet wrote:

In <pan.2005.01.21.00.52.45.405635@example.net>, on 01/21/05
at 12:47 AM, Rich Grise <richgrise@example.net> said:

What the world needs is a Linux distribution that's Aunt-Tillie-Friendly.

Amen.

With all the wonderful resources out there spending time creating yet
another text editor, or partition tool, writing a new browser that is a
Mozilla ripoff, and writing file managers that just leave me shaking my
head, I truly lament the fact that they continue to splinter off in
different directions, making it more and more complex, instead of
hunkering down and working on some kind of useful desktop interface that
doesn't mimic that awful M$ thing, and that doesn't require editing a text
file to create an icon.

It would appear that goal is not to come together, but to continue to fly
apart, and that is too bad, as the potential was there, but is going away
as things get more and more diversified. Too many egos?
It's just another manifestation of the eternal battle between the poles of
the duality. Management/labor, order/chaos, Iron-fisted dictatorship/Free
Will, that sort of thing.

And, nobody gets paid to develop Linux stuff; it's a labor of love, and
I'm surmising there's a (an?) "I'm gonna do it _my_ way" kind of vibe
around it.

I'm using KDE here, and it's pretty OK, but still has little things about
it. And I get an eerie feeling every time I'm reading some kind of Linux
instructions for this-n-that, and it says, "Ask your system
administrator". Crap! I _am_ the system administrator! ;-)

I'd do it myself, if I had somebody to subsidize a year-long development
project.

Cheers!
Rich
 
On Thu, 20 Jan 2005 20:50:41 -0800, the renowned John Larkin
<jjSNIPlarkin@highTHISlandPLEASEtechnology.XXX> wrote:

On Fri, 21 Jan 2005 01:47:07 +0000 (UTC), kensmith@green.rahul.net
(Ken Smith) wrote:

In article <h970v01h47pposfisp9nvl7kl06iqe7oh6@4ax.com>,
John Larkin <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:
[...]
But gaah! Who would want multiple 8051's?

Me.

My company used to sell a product that had 7 8051s in it. Each thing that
needed doing got its own micro.


That's the idea: a processor per process.

Can you run a multitasking RTOS on an 8051 anyhow?
Is this an RTOS?
http://www.keil.com/rtx51/

I wrote one once
for the 6800 (that's *not* 68000) and it was a moderate nuisance. A
6800 won't even push the index register.
You can do self-modifying code on those Von Neumann machines, so the
instruction set limitations are not so important. That's how I got
around some limitations of the 68HC05. A bit ugly, but it worked, and
only a few instructions had to be in RAM.


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
In article <fOydnRzrF5u1v2zcRVn-iQ@buckeye-express.com>,
Mark Jones <abuse@127.0.0.1> wrote:
Ken Smith wrote:
[...]
Her: There are 16 thousand processors!

Me: What type of processors are they?

Her: Little bitty ones.

So now you know.



Was it one of these ladies?
http://mission.base.com/tamiko/cm/going/tadtbc.html
No, she was a blond.


--
--
kensmith@rahul.net forging knowledge
 
On Wed, 19 Jan 2005 11:48:16 -0800, John Larkin
<jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:

http://www.theregister.co.uk/2005/01/19/intel_smithfield_power/
I may be nuts, but 125A draw for a microprocessor is just plain silly.
Why don't these chip manufacturers up the signal/supply voltages these
devices can handle and balance VA out a bit more evenly in the
interests of better overall efficiency?
--

"What is now proved was once only imagin'd." - William Blake, 1793.
 
In article <f4d2v0d7i7ov0rcscmc83rqj7t4flpkjh9@4ax.com>,
pb@notthisbit.osiris1.co.uk says...
On Wed, 19 Jan 2005 11:48:16 -0800, John Larkin
jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote:


http://www.theregister.co.uk/2005/01/19/intel_smithfield_power/

I may be nuts, but 125A draw for a microprocessor is just plain silly.
Why don't these chip manufacturers up the signal/supply voltages these
devices can handle and balance VA out a bit more evenly in the
interests of better overall efficiency?
Why not indeed...

1. Have you ever considered that P~CV^^2? That V^^2 part is a killer.
Indeed the voltage is being forced down to reduce power.

2. Now that we've gone into the deep-deep, sub-micron geometries,
leakage becomes significant. Again, the current is a high order
function of voltage. Indeed, the voltage is being forced down to
reduce power.

3. Gate oxides are only a few atoms thick (see leakage #2 above). At
higher voltages the gates will break down. ...not good.

3. You haven't invented a DC integrated transformer for us to "balance
VA out a bit more").

--
Keith
 
In article <pja2v0dsh4pe6r6d1e6o2ne2cn90ehkc40@4ax.com>,
Spehro Pefhany <speffSNIP@interlogDOTyou.knowwhat> wrote:
[... me ...]
In the 8051, if you want you can make self modifying code posible.

Broadly speaking, that's true, but not with the single-chip variants.
You need to play tricks externally to map code memory into the program
memory space
It just takes an NOR between PSEN/ and RD/ to do the simplest version.
In most 8051 like devices that have under 64K of PROM, code accesses
beyond the PROM space are external.


I don't recall the exact issue with the 6805, but it was something
where the processor had an instruction (say a call or a jump) that was
required, but it only accepted a constant (or a constant plus a single
signed byte offset). To make it a variable that could point to
anywhere in memory required self-modifying code.
Or you could put that address on the stack
--
--
kensmith@rahul.net forging knowledge
 
nobody gets paid to develop Linux stuff
Rich Grise
Well, not entirely true:
http://www.google.com/search?&q=OSDL-hires-*-Torvalds
You already mentioned RedHat; plenty of developers there.
 
Rich Grise wrote...
What the world needs is a Linux distribution that's Aunt-Tillie-Friendly.
I think Redmond^H^H^H^HHat is trying to do just that - I installed RH
once, but didn't like it, because there was _too much_ automatic stuff,
and it was terribly nonobvious how to get to my configuration files. I
guess I'm just a born Slacker. ;-)
Zandros www.xandros.com


--
Thanks,
- Win
 
Rich Grise wrote...
... nobody gets paid to develop Linux stuff; it's a labor of love,
and I'm surmising there's a (an?) "I'm gonna do it _my_ way" kind
of vibe around it.
My hunch is most linux programmers these days are in fact paid
to do their programming. People at IBM, Novell, and elsewhere.
One of the fellows here at the Institute works on fixing Linux
USB problems. He does this mostly at work, while getting paid,
and is proud of his contributions now residing in the kernel.




--
Thanks,
- Win
 
It's a microprocessor *and* a toaster oven! :) :)

"John Larkin" <jjlarkin@highSNIPlandTHIStechPLEASEnology.com> wrote in
message news:kbetu01tt15csq9dmnh3k9gnihfm0mb7hr@4ax.com...
http://www.theregister.co.uk/2005/01/19/intel_smithfield_power/

John
 
32 CPUs on one chip?

That's one to run the OS, 1 to run the app, and 30 to run the spyware
processes...
 

Welcome to EDABoard.com

Sponsor

Back
Top