Dual-core Intels by 2005

In article ,
says...
On Sun, 21 Sep 2003 21:24:53 -0400, Keith R. Williams
wrote:

In article ,
says...
On Sat, 20 Sep 2003 01:22:57 GMT, "Felger Carbon" wrote:

"Keith R. Williams" wrote in message
...

Oh, good grief yes. Ancient mainframes had ducting in the 60's,
as did my parents house in the 50s. Even the Romans had (aqua)
ducting.

(haughty sniff) And _I_ have two full rolls of duct tape and a tad
remaining on a third roll. No, not for the reason you're thinking. ;-)

"Ancient mainframes"...Ouch...

Well... If you know what I'm talking about...

The first semiconductor mainframe memory design I did (1972) used 256 *bit*
bipolar ram. As memories these were so-so, but as tiny hibachis they worked
damned well. In free air you could easily roast flesh on the metal package
lids. The blowers we used were humongous (and loud!) and barely kept up across
a class A environment (glassed computer room).

You *ARE* an old fart! (I just turned 33, BTW .

I am indeed an old fart.
The "ouch" was because you made me remember that fact ;-)

Well, I didn't know you were *THAT*...

Sprague Electric bought a one megabyte box for their 370/145 located in a huge
old mill building in North Adams, MA. One megabyte of bipolar memory made for
a pretty big pile of silicon.

Hmmm, the /145 (and /155 and /165) had core memory. Perhaps they
upgraded it to a /148? The timing was perhaps early, but about
right.

The 370/155 and 165 used core, the follow-on 370/158 and 168 used nmos ram.
But the 370/145 was semiconductor from the start (as was the 370/135). The
370/148 (and /138) were incremental upgrades (cache size, memory capacity,
processor speed bump, etc).

Ok, I sit corrected. The 31x5 series was before my time (man you
are an antique! ;-). My reason for living was FS, though that
didn't last long.

I designed memories for all of those but the /155 & /165 (those were done by
another group). I then designed memories for the 308X and then 309X machines
(all nmos memories from the start)....

Memories? ...as in chips, or sub-system?. I worked on the 308x,
303x, and 309x, before doing time in CCP (playing with micros).
I then went back to mainframe-land to work on the 309x and ES9000
crypto feature (then escaped to the north ;-).


/daytripper (nearly decrepit ;-)

I guess, maybe! ;-)

--
Keith

In article et,
says...
"daytripper" wrote:

/daytripper (nearly decrepit ;-)

Don't take any guff off that kid Keith! He's still wet behind the ears,
only 3x years old (modulo 16).

Hey, hey! (for values of x=3)

Me? The older I get, the better I am. I'm approaching magnificent!
;-)

You'd better clean out your virus folder, ol' bean. It's full!

(Also testing new fake return address... sigh)

Fake? Why bother. Simply filter out anything but To: and CC:
and you have 99% of it licked. ...until next time. :-(

--
Keith

"Keith R. Williams" wrote in message
. ..
In article et,
says...
"daytripper" wrote:

/daytripper (nearly decrepit ;-)

Don't take any guff off that kid Keith! He's still wet behind the ears,
only 3x years old (modulo 16).

Hey, hey! (for values of x=3)


Damn! I feel positively youthful. (not even 30 yet!!)

Regards,
Dean

On Mon, 22 Sep 2003 21:45:33 -0400, Keith R. Williams
wrote:

In article ,
says...
On Sun, 21 Sep 2003 21:24:53 -0400, Keith R. Williams
wrote:
[snipped]
The 370/155 and 165 used core, the follow-on 370/158 and 168 used nmos ram.
But the 370/145 was semiconductor from the start (as was the 370/135). The
370/148 (and /138) were incremental upgrades (cache size, memory capacity,
processor speed bump, etc).

Ok, I sit corrected. The 31x5 series was before my time (man you
are an antique! ;-). My reason for living was FS, though that
didn't last long.

I designed memories for all of those but the /155 & /165 (those were done by
another group). I then designed memories for the 308X and then 309X machines
(all nmos memories from the start)....

Memories? ...as in chips, or sub-system?. I worked on the 308x,
303x, and 309x, before doing time in CCP (playing with micros).
I then went back to mainframe-land to work on the 309x and ES9000
crypto feature (then escaped to the north ;-).

Subsystem. I designed the memory interface card, the companion array cards,
the backplane, the power system, a power control/environmental monitoring unit
(8086 based, coded in assembly ;-) that interfaced to the Service
Processor...and hacked the service processor's configuration data tables so
it'd use the added memory. Cool project, involved multiple disciplines outside
of my regular haunts, did a lot of reading and learning in a hurry...

Hit the fast forward button through a few decades...having joined The Dark
Side of IA32 architecture...

[Warning: Unabashed Plug Alert!]

....and another project hits the street:

http://www.stratus.com/news/2003/20030922.htm

/daytripper (sniffle my baby! :-)

"daytripper" wrote in message
...

....

...and another project hits the street:

http://www.stratus.com/news/2003/20030922.htm

/daytripper (sniffle my baby! :-)

Congratulations! I've always thought that Stratus had an interesting and
respectable product line. But, since you brought it up, I have a question:
while your new product doubtless stands up to its predecessors, I find it
difficult to believe that it can limit unplanned downtime to 0.0002% over a
6-month period while using Win2K as its OS - a single reboot per system
would blow that figure away. Or do your customers perform frequent
'planned' reboots to avoid such problems (which, IMO, would somewhat dilute
the relevance of the uptime figure)?

- bill

On Tue, 23 Sep 2003 00:32:01 -0400, "Bill Todd"
wrote:


"daytripper" wrote in message
.. .

...

...and another project hits the street:

http://www.stratus.com/news/2003/20030922.htm

/daytripper (sniffle my baby! :-)

Congratulations! I've always thought that Stratus had an interesting and
respectable product line. But, since you brought it up, I have a question:
while your new product doubtless stands up to its predecessors, I find it
difficult to believe that it can limit unplanned downtime to 0.0002% over a
6-month period while using Win2K as its OS - a single reboot per system
would blow that figure away. Or do your customers perform frequent
'planned' reboots to avoid such problems (which, IMO, would somewhat dilute
the relevance of the uptime figure)?

Not trying to weasel on this, but it's an important question, and I don't know
what the answer actually is. On the one hand we have multitudes of QA and beta
platforms that have been running with 100% availability for almost a year,
many of them in the face of a wide variety of injected faults. On the other
hand the implication of Windows Update would seem to bound continuous
availability, as sooner or later you're going to hit an update that requires a
reboot to complete.

I'm going off on a much needed vacation tomorrow (I'd have left tonite but I
pulled two all-niters back-to-back and I'm totally flogged ;-) but when I
return I'll try to scope out the corporate line on this.

My expectation, for whatever it's worth, is any loss of availability due to OS
updates is not counted against *our* "nines". But I could easily be wrong, and
there's a neat feature of our architecture that may explain an alternate
answer, but I don't know whether the software types are enabling it yet...

cheers

/daytripper

On Tue, 23 Sep 2003 02:21:52 GMT, "Dean Kent"
wrote:
"Keith R. Williams" wrote in message
...
In article et,
says...
"daytripper" wrote:

/daytripper (nearly decrepit ;-)

Don't take any guff off that kid Keith! He's still wet behind the ears,
only 3x years old (modulo 16).

Hey, hey! (for values of x=3)


Damn! I feel positively youthful. (not even 30 yet!!)

Dammit! You guys have GOT to stop talking ages like that, there is NO
WAY that I want to be a teenager again! :

-------------
Tony Hill
hilla underscore 20 at yahoo dot ca

On Sat, 20 Sep 2003 04:48:04 GMT, daytripper
wrote:

Sprague Electric bought a one megabyte box for their 370/145 located in a huge
old mill building in North Adams, MA. One megabyte of bipolar memory made for
a pretty big pile of silicon. Picture three upright full size freezers lashed
side by side for the size of the memory cabinet. It used 16x 1000w switchers
running around 85%. The mains cable was as big around as a fire hose. You
could not keep your hand above the top of the vent stacks in a fully loaded
cabinet. A few months of constant operation actually changed the color of the
pcb material.

Man, that's incredible. I wonder if the tech they will have 30 years
from now will make what we are using today look so archaic. Hard to
imagine...



----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! 100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---

On Tue, 23 Sep 2003 05:25:53 GMT, daytripper
wrote:

On the other
hand the implication of Windows Update would seem to bound continuous
availability, as sooner or later you're going to hit an update that requires a
reboot to complete.

Heh. I know our IT guys get to work a lot of Sundays, applying
updates and resetting the servers...



----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! 100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---

"daytripper" wrote in message
...

Not trying to weasel on this, but it's an important question, and I don't
know
what the answer actually is. On the one hand we have multitudes of QA and
beta
platforms that have been running with 100% availability for almost a year,
many of them in the face of a wide variety of injected faults. On the
other
hand the implication of Windows Update would seem to bound continuous
availability, as sooner or later you're going to hit an update that
requires a
reboot to complete.

I know you won't answer this due to the vacation mentioned below - but I am
wondering if '100% continuous availability' means 'sans scheduled outages'
or if it means that there have been absolutely no scheduled or unscheduled
outages for almost a year. My understanding of 'availability' implies the
former.


I'm going off on a much needed vacation tomorrow (I'd have left tonite but
I
pulled two all-niters back-to-back and I'm totally flogged ;-) but when I
return I'll try to scope out the corporate line on this.

My expectation, for whatever it's worth, is any loss of availability due
to OS
updates is not counted against *our* "nines". But I could easily be wrong,
and
there's a neat feature of our architecture that may explain an alternate
answer, but I don't know whether the software types are enabling it yet...

This sounds more like what I am familiar with. Generally, zero down time
is called 'continuous 24/7 operation', with a disclaimer about planned
outages every few weeks, months or quarters - depending upon the situation -
for maintenance.

Regards,
Dean


cheers

/daytripper