EM64T bug

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Looks like Intel decided to reduce the maximum addressable physical memory
on Nocona from 40 bits to 36 bits. But it looks like somebody forgot to tell
the guy in charge of setting up the CPUID instruction, because Nocona still
returns 40 bits when queried. This is apparently what's preventing Linux
kernels from running on Nocona right now (and likely the Windows beta kernel
too).

http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=20289129

Yousuf Khan

--
Humans: contact me at ykhan at rogers dot com
Spambots: just reply to this email address ;-)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Thu, 15 Jul 2004 05:00:33 GMT, "Yousuf Khan" <bbbl67@ezrs.com> wrote:

>Looks like Intel decided to reduce the maximum addressable physical memory
>on Nocona from 40 bits to 36 bits. But it looks like somebody forgot to tell
>the guy in charge of setting up the CPUID instruction, because Nocona still
>returns 40 bits when queried. This is apparently what's preventing Linux
>kernels from running on Nocona right now (and likely the Windows beta kernel
>too).
>
>http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=20289129

Hmm, the AMD on-die memory controller is looking more & more like the way
forward here. Surely the Nocona CPUID result can be corrected with a
Ucode/BIOS upfdate???... but even so, it seems to me that Intel could get
into a whole mess with CPU/chipset matching down the road... admittedly a
while yet but even so.

BTW it's eerily quiet in here just now - everybody on vacation?

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

YK: > Looks like Intel decided to reduce the maximum
> addressable physical memory on Nocona from 40 bits
> to 36 bits.

It doesn't strike me as a reduction. I think it more
likely that despite the new chipset, there are still
some 36-bit restrictions in the infrastructure, a
legacy of PAE (which is also 36-bit on Xeon).

I do agree with the assessment that Intel must have
been in a "tearing hurry" to implement AMD64, as
evidenced by the sloppy paraphrasings in the
documents, and the defective implementation in sand.

After the Illegal Instruction Erratum of some years
back, I'd expected Intel to be much more disciplined
about elementary due-diligence testing. Guess not.

Perhaps some future historian will be able to reveal
to us just what event(s) provoked Intel to adopt
AMD64, and to do it so rashly. Hints are to be found
around the web on this topic, such as:

- Long-standing Microsoft disdain for IA-64, plus
what Intel can see from IA-64 sales trends

- MS probably told Intel to completely forget about
Windows on Yamhill (unless Yamhill became AMD64)

- Clear signs of pervasive 64-bit computing years
before the Intel roadmaps showed it happening - heck
there are already AMD64 laptops

- Tier One OEM adoption of AMD64 (other than Dell)

- Another round of Dell-AMD rumors (which might
suggest Dell leaning on Intel for a 64-bit
solution they can actually sell)

- Mentor, PTC and UGS dropping support for some of
their workstation apps on IA-64 (revealed in the
last 30 days, but Intel likely had indications
earlier)

- Rumors that Longhorn might not support IA-64, yet
might be 64-bit only

Anyway, here we are 17 days after Nocona launch, and
we still have no real info on 64-bit performance. True,
the public beta of Windows XP64 won't run on it, and
apparently the NDA beta has a "no publish" restriction.

But does Linux 64 really not run at all on EM64T?

Has Intel said anything about microcode/BIOS patches
to fix this?

--
Regards, Bob Niland mailto:name@ispname.tld
http://www.access-one.com/rjn email4rjn AT yahoo DOT com
NOT speaking for any employer, client or Internet Service Provider.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

George Macdonald wrote:
> On Thu, 15 Jul 2004 05:00:33 GMT, "Yousuf Khan" <bbbl67@ezrs.com> wrote:
>
>
>>Looks like Intel decided to reduce the maximum addressable physical memory
>>on Nocona from 40 bits to 36 bits. But it looks like somebody forgot to tell
>>the guy in charge of setting up the CPUID instruction, because Nocona still
>>returns 40 bits when queried. This is apparently what's preventing Linux
>>kernels from running on Nocona right now (and likely the Windows beta kernel
>>too).
>>
>>http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=20289129
>
>
> Hmm, the AMD on-die memory controller is looking more & more like the way
> forward here. Surely the Nocona CPUID result can be corrected with a
> Ucode/BIOS upfdate???... but even so, it seems to me that Intel could get
> into a whole mess with CPU/chipset matching down the road... admittedly a
> while yet but even so.
>
> BTW it's eerily quiet in here just now - everybody on vacation?

Not me. I'm still in shock that Intel's validation is so unbelievably
poor that this particular goof slipped through the net. I guess that
they must have been in a tearing hurry to adapt "Yamhill" to comply
with AMD64... It's not going to exactly help sales of the new and
untested Nocona. :/

Cheers,
Rupert

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Rupert Pigott wrote:
> George Macdonald wrote:
>> BTW it's eerily quiet in here just now - everybody on vacation?
>
>
> Not me. I'm still in shock that Intel's validation is so unbelievably
> poor that this particular goof slipped through the net. I guess that
> they must have been in a tearing hurry to adapt "Yamhill" to comply
> with AMD64... It's not going to exactly help sales of the new and
> untested Nocona. :/

I'm here. I'm wondering why this hasn't been shot down. I didn't see
proof and I've talked to designers and validators of that chip who all
say that poster is cracked, Prescott and derivatives support a full 40b
physical space. When I interviewed for the team, that was the first
thing they told me "we have a 40b physical address on the prescott
bus...now using that, solve these problems". Is there a link to
developer.intel.com that shows such an erratum is real? Until I see it
from the source, I'm taking this poster's message as FUD.

Alex
--
My words are my own. They represent no other; they belong to no other.
Don't read anything into them or you may be required to compensate me
for violation of copyright. (I do not speak for my employer.)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

AJ: > I'm wondering why this hasn't been shot down.

If it's not true, I'd expect Intel to respond
promptly to it. Obviously, they can't bother
with every rumor on stock forums, but this
particular story was picked up by the Inq several
days ago, and thus also by all the news robot
sites. It merits debunking if false:
<http://www.theinquirer.net/?article=17181>

It doesn't need a press release. An Intel
employee responding to any of the various
forum threads would suffice.

> Until I see it from the source, I'm taking
> this poster's message as FUD.

That's a reasonable posture. It's estimated that
40% of "facts" posted on stock forums are outright
falsehoods deliberately intended to manipulate
stock prices.

However, I can't see that this specific factoid,
true or false, would have much stock effect. What
is a true fact is that for whatever reason(s), we
have essentially zero reports characterizing 64-bit
operating systems / applications on EM64T, even
though Nocona is supposedly shipping.

--
Regards, Bob Niland mailto:name@ispname.tld
http://www.access-one.com/rjn email4rjn AT yahoo DOT com
NOT speaking for any employer, client or Internet Service Provider.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Sorry about the thread linking, but my newsreading
site updates faster than my newsposting site.

>> ... I can't see that this specific factoid,
>> true or false, would have much stock effect.

AJ: > No significant stock effect?

Well, we could go back and look, I suppose, but I'd
expect this story to have been lost in the noise of
the INTC Q2 announcement.

> You bring up the much more important question I've
> been pondering for weeks. Nobody did 64b benchmarks
> between AMD and intel when Nocona shipped.

Well, what seems to be the case is:

- Win64 public beta won't run, and
Win64 NDA beta has no-publish restriction, so
there's no Windows to test & publish with

- Shipping Noconas may be sub-3GHz, not worth benching,
even if one were inclined to pay real $ to get one

- >3GHz systems may be under 64-bit NDA from Intel

- there may just be very few >3 systems out there
(i.e. this was a paper launch)

- 64-bit PCI-E drivers might be an issue

- I still haven't heard if Linux64 runs on EM64T

All of the above adds up to what we (don't) see, but
all of the above may not be true.

> Lots of sites put that ("64 bit battle" and such)
> in their headlines, but when you read the articles ...

Here's another, from 2004-06-28:
<http://www.amazoninternational.com/html/news/exclusive/quadro3400/tumwater.asp>
"Very quickly over the next few days we shall be
also showing the Windows 64 Bit tests ..."
17 days later and counting, dead silence. Did they
finally read the NDAs? Couldn't get it working?
Someone leaned on them? Who knows.

> It's like someone bribed or bullied everybody out
> there (non-disclosure agreements perhaps) to avoid
> comparisons regarding 64b.

If testers were being strong-armed (more than usual),
I think we'd have heard about it.

> ... total silence? I don't buy it!

So what's your theory?
It's definitely odd.

--
Regards, Bob Niland mailto:name@ispname.tld
http://www.access-one.com/rjn email4rjn AT yahoo DOT com
NOT speaking for any employer, client or Internet Service Provider.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Bob Niland wrote:
> However, I can't see that this specific factoid,
> true or false, would have much stock effect. What
> is a true fact is that for whatever reason(s), we
> have essentially zero reports characterizing 64-bit
> operating systems / applications on EM64T, even
> though Nocona is supposedly shipping.

No significant stock effect? Everyone was waving AMD64 in the air
saying intel was more than a year behind AMD because of the instruction
set direction the two went. Intel stock fell. Everyone grabbed up that
the NX bit was not enabled in the first intel chips to support AMD64.
Intel stock fell. If it were confirmed, or even /failed to deny/, that
intel's 40b addressing was 4 bits short, I'm sure intel's stock would
fall again. Maybe only 2 dollars, but every bit hurts.

You bring up the much more important question I've been pondering for
weeks. Nobody did 64b benchmarks between AMD and intel when Nocona
shipped. Lots of sites put that ("64 bit battle" and such) in their
headlines, but when you read the articles they just say "we loaded win64
and it worked but we didn't run anything, then we loaded win32 and ran
these benchmarks, see?" That puzzles me. GamePC said it was because of
no PCI Express drivers for Win64. Okay, they benchmark with graphical
games...I can see that a dependancy held them back. The rest I can't.
It's like someone bribed or bullied everybody out there (non-disclosure
agreements perhaps) to avoid comparisons regarding 64b. The new clock
speed of some chip comes out and there's a flood of benchmarkers
publishing stuff. A major release like x86-64 and nothing, total
silence? I don't buy it!

Alex
--
My words are my own. They represent no other; they belong to no other.
Don't read anything into them or you may be required to compensate me
for violation of copyright. (I do not speak for my employer.)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Alex Johnson wrote:
> You bring up the much more important question I've been pondering for
> weeks. Nobody did 64b benchmarks between AMD and intel when Nocona
> shipped. Lots of sites put that ("64 bit battle" and such) in their
> headlines, but when you read the articles they just say "we loaded
> win64 and it worked but we didn't run anything, then we loaded win32
> and ran these benchmarks, see?" That puzzles me. GamePC said it was
> because of no PCI Express drivers for Win64. Okay, they benchmark
> with graphical games...I can see that a dependancy held them back.
> The rest I can't. It's like someone bribed or bullied everybody out
> there (non-disclosure agreements perhaps) to avoid comparisons
> regarding 64b. The new clock speed of some chip comes out and
> there's a flood of benchmarkers publishing stuff. A major release
> like x86-64 and nothing, total silence? I don't buy it!

So who do you think could be behind this?

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Thu, 15 Jul 2004 08:53:52 -0400, Alex Johnson <compuwiz@acm.org> wrote:

>Rupert Pigott wrote:
>> George Macdonald wrote:
>>> BTW it's eerily quiet in here just now - everybody on vacation?
>>
>>
>> Not me. I'm still in shock that Intel's validation is so unbelievably
>> poor that this particular goof slipped through the net. I guess that
>> they must have been in a tearing hurry to adapt "Yamhill" to comply
>> with AMD64... It's not going to exactly help sales of the new and
>> untested Nocona. :/
>
>I'm here. I'm wondering why this hasn't been shot down. I didn't see
>proof and I've talked to designers and validators of that chip who all
>say that poster is cracked, Prescott and derivatives support a full 40b
>physical space. When I interviewed for the team, that was the first
>thing they told me "we have a 40b physical address on the prescott
>bus...now using that, solve these problems". Is there a link to
>developer.intel.com that shows such an erratum is real? Until I see it
>from the source, I'm taking this poster's message as FUD.

If you have an address bus which goes up to 40, it does not show on address
lines in the Datasheets for current Xeon/800 EM64T CPUs, where the HA lines
stop at 35, nor the 7525 chipset. Of course there's probably no need for
it right now but is there anywhere it could be mentioned in a doc?

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

>> Win64 NDA beta has no-publish restriction, so
>> there's no Windows to test & publish with

RusH: > So there are real operating systems you know ?

I see that Solaris 10 just booted on AMD64 this
week, but only to command line, and no word on
whether it will run on EM64T, although I can
venture a guess that Scotty won't officially
support it there.

>> - Shipping Noconas may be sub-3GHz, not worth benching,
>> even if one were inclined to pay real $ to get one

> thats a goot one. When exactly did that happened that benchmarking
> new technology from the industry leader is not worth the effort ?

I think you missed my point. If you are interested
in benching a Nocona, but you have to spend your
own real money to do it, and all that's available
are say, 2.8GHz units (and what you want is a 3.6),
and you have a suspicion about the results, are
you going to make that investment?

>> - I still haven't heard if Linux64 runs on EM64T

> I did :
> ---------------
> Red Hat Enterprise Linux 3 Update 2 adds support for ...

Appears to be orderable. I'm surprised we haven't seen
any test results.

--
Regards, Bob Niland mailto:name@ispname.tld
http://www.access-one.com/rjn email4rjn AT yahoo DOT com
NOT speaking for any employer, client or Internet Service Provider.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

RusH wrote:

> email4rjn@yahoo.com (Bob Niland) wrote :
>
>
>>Well, what seems to be the case is:
>>
>> - Win64 public beta won't run
>

Take a look at the thread
"AMD64 bits for Xeon(Nocona) problem"
in microsoft.private.windowsserver_64bit
on server privatenews.microsoft.com


For those that can't be bothered to read the thread:
"A beta version of Windows compatible with Intel's 64-bit
extensions technology is due within weeks
http://www.infoworld.com/article/04/07/06/HNwindowsnocona_1.html"



You will need a user name and password for that
server - I can't remember what they are (I just
have Thunderbird auto-submit them for me) but they
are available from the Win64 pages at MicroSoft.

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Thu, 15 Jul 2004 15:40:04 -0400, Alex Johnson <compuwiz@acm.org> wrote:

>Bob Niland wrote:
>> However, I can't see that this specific factoid,
>> true or false, would have much stock effect. What
>> is a true fact is that for whatever reason(s), we
>> have essentially zero reports characterizing 64-bit
>> operating systems / applications on EM64T, even
>> though Nocona is supposedly shipping.
>
>No significant stock effect? Everyone was waving AMD64 in the air
>saying intel was more than a year behind AMD because of the instruction
>set direction the two went. Intel stock fell. Everyone grabbed up that
>the NX bit was not enabled in the first intel chips to support AMD64.
>Intel stock fell. If it were confirmed, or even /failed to deny/, that
>intel's 40b addressing was 4 bits short, I'm sure intel's stock would
>fall again. Maybe only 2 dollars, but every bit hurts.

Has Intel's stock fallen significantly since AMD released AMD64? IMO AMD's
gains have been pretty modest since then... given the significance of the
improved ISA, memory channel, scalability and of course ASP. In fact many
"analysts" have played AMD64 down to the extent that it's damned near
looking like a conspiracy - take a look at
http://finance.yahoo.com/q/ud?s=AMD - not a lot of "upgrades" and here
http://finance.yahoo.com/q/ud?s=INTC - isn't that different, in fact less
downgrades apart from the last week, which is due to the inventory thingy.

>You bring up the much more important question I've been pondering for
>weeks. Nobody did 64b benchmarks between AMD and intel when Nocona
>shipped. Lots of sites put that ("64 bit battle" and such) in their
>headlines, but when you read the articles they just say "we loaded win64
>and it worked but we didn't run anything, then we loaded win32 and ran
>these benchmarks, see?" That puzzles me. GamePC said it was because of
>no PCI Express drivers for Win64. Okay, they benchmark with graphical
>games...I can see that a dependancy held them back. The rest I can't.
>It's like someone bribed or bullied everybody out there (non-disclosure
>agreements perhaps) to avoid comparisons regarding 64b. The new clock
>speed of some chip comes out and there's a flood of benchmarkers
>publishing stuff. A major release like x86-64 and nothing, total
>silence? I don't buy it!

Yes it *is* *very* weird. Apparently c't has some benchmarks in their most
recent printed edition which they have not published on the Web site. If
there's a NDA, given that Linux is in the vanguard of 64-bit OS, it'd
appear unlikely that M$ is an effective NDA enforcer, which
leaves.....??;-) Many of the gamer sites don't seem to understand the
significance of the 64-bit mode and how it could impact performance... may
not be that interested till they have a repertoire of recoded/recompiled
64-bit games. It couldn't be that the disparity of the results is such
that publication has been absolutely prohibited... could it??:-0

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Bob Niland wrote:
>>> Win64 NDA beta has no-publish restriction, so
>>> there's no Windows to test & publish with
>
> RusH: > So there are real operating systems you know ?
>
> I see that Solaris 10 just booted on AMD64 this
> week, but only to command line, and no word on
> whether it will run on EM64T, although I can
> venture a guess that Scotty won't officially
> support it there.

Scotty has still got some Xeon boxes that he sells, though the Xeons are
definitely playing third fiddle to the Opterons and Sparcs. Eventually those
low-ball Xeons are likely to be coming with EM64T onboard whether Scotty
likes it or not. So he'll eventually have to have Solaris running in 64-bit
on at least all of the boxes that he himself sells at the very least.

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Thu, 15 Jul 2004 08:53:52 -0400, Alex Johnson <compuwiz@acm.org>
wrote:
>I'm here. I'm wondering why this hasn't been shot down. I didn't see
>proof and I've talked to designers and validators of that chip who all
>say that poster is cracked, Prescott and derivatives support a full 40b
>physical space. When I interviewed for the team, that was the first
>thing they told me "we have a 40b physical address on the prescott
>bus...now using that, solve these problems". Is there a link to
>developer.intel.com that shows such an erratum is real? Until I see it
>from the source, I'm taking this poster's message as FUD.

Yup, Errata #23 in Intel's Spec update for Xeon processors with
800MT/s bus speed:

ftp://download.intel.com/design/Xeon/specupdt/30240201.pdf

It's listed as affecting the one and only stepping of these chips that
is available (D0) and with a fix planned for future steppings.

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

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Thu, 15 Jul 2004 15:40:04 -0400, Alex Johnson <compuwiz@acm.org>
wrote:
>Bob Niland wrote:
>> However, I can't see that this specific factoid,
>> true or false, would have much stock effect. What
>> is a true fact is that for whatever reason(s), we
>> have essentially zero reports characterizing 64-bit
>> operating systems / applications on EM64T, even
>> though Nocona is supposedly shipping.
>
>No significant stock effect? Everyone was waving AMD64 in the air
>saying intel was more than a year behind AMD because of the instruction
>set direction the two went. Intel stock fell. Everyone grabbed up that
>the NX bit was not enabled in the first intel chips to support AMD64.
>Intel stock fell. If it were confirmed, or even /failed to deny/, that
>intel's 40b addressing was 4 bits short, I'm sure intel's stock would
>fall again. Maybe only 2 dollars, but every bit hurts.

And so it should! Watching Intel has been like a comedy of errors
lately! They're letting a company that's roughly 1/8th the size trump
it left right and center in the market that is their bread and butter.
The mistakes and failures that Intel has made recently are much more
major than the effect on the marketplace has shown.

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

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Tony Hill wrote:
> On Thu, 15 Jul 2004 08:53:52 -0400, Alex Johnson <compuwiz@acm.org>
> wrote:
>
>>I'm here. I'm wondering why this hasn't been shot down. I didn't see
>>proof and I've talked to designers and validators of that chip who all
>>say that poster is cracked, Prescott and derivatives support a full 40b
>>physical space. When I interviewed for the team, that was the first
>>thing they told me "we have a 40b physical address on the prescott
>>bus...now using that, solve these problems". Is there a link to
>>developer.intel.com that shows such an erratum is real? Until I see it
>
>>from the source, I'm taking this poster's message as FUD.
>
> Yup, Errata #23 in Intel's Spec update for Xeon processors with
> 800MT/s bus speed:
>
> ftp://download.intel.com/design/Xeon/specupdt/30240201.pdf
>
> It's listed as affecting the one and only stepping of these chips that
> is available (D0) and with a fix planned for future steppings.
>
> -------------
> Tony Hill
> hilla <underscore> 20 <at> yahoo <dot> ca

Interesting. It is official. Must be marketting pulling strings behind
the scenes somewhere after the designers and validators put in and
checked 40b worked.

Alex
--
My words are my own. They represent no other; they belong to no other.
Don't read anything into them or you may be required to compensate me
for violation of copyright. (I do not speak for my employer.)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Alex Johnson wrote:
> Interesting. It is official. Must be marketting pulling strings
> behind the scenes somewhere after the designers and validators put in
> and checked 40b worked.

Why would marketing want to reduce the amount of memory addressable? That's
sounds more like anti-marketing.

Yousuf Khan

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Mon, 19 Jul 2004 18:24:53 GMT, "Yousuf Khan" <bbbl67@ezrs.com> wrote:

>Alex Johnson wrote:
>> Interesting. It is official. Must be marketting pulling strings
>> behind the scenes somewhere after the designers and validators put in
>> and checked 40b worked.
>
>Why would marketing want to reduce the amount of memory addressable? That's
>sounds more like anti-marketing.

Well they already had 36-bit addressing - path of least resistance.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Yousuf Khan <bbbl67@ezrs.com> wrote:
>Alex Johnson wrote:
>> Interesting. It is official. Must be marketting pulling strings
>> behind the scenes somewhere after the designers and validators put in
>> and checked 40b worked.
>
>Why would marketing want to reduce the amount of memory addressable?
>That's sounds more like anti-marketing. Yousuf Khan

pro-*ITANIUM* marketing, my son, not pro-earnings marketing

<r&h>