Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Apple's Xserve computer now has 16-GBs of RAM

4 views
Skip to first unread message

Mark Conrad

unread,
Mar 20, 2005, 11:28:08 PM3/20/05
to

According to what a tech' in the "Enterprise" section of Apple told me,
the Apple Xserve computer can now support 16-GB of RAM.

I will believe it when I see it advertised at Apple's Xserve website.

<http://www.apple.com/ca/xserve/specs.html>

The tech' claims the change has been effective for a week now.

Oh, did I fail to mention, the base price for the dual CPU Xserve
machine starts at $4,999


Anyone out there have one of those puppies?

Mark-

antares

unread,
Mar 20, 2005, 11:40:42 PM3/20/05
to

I had to settle for a single processor model. Manual says 8 GB RAM,
though if there is now a higher density DIMM available, that may explain
the upped memory limit. Still a very nice toy, even at the "low end".

antares

Mark Conrad

unread,
Mar 21, 2005, 1:18:23 PM3/21/05
to
In article <Pes%d.148$Cv3.5...@news.sisna.com>, antares
<ant...@email.com> wrote:

> > Anyone out there have one of those puppies?
> >
> > Mark-
>
> I had to settle for a single processor model. Manual says 8 GB RAM,
> though if there is now a higher density DIMM available, that may explain
> the upped memory limit. Still a very nice toy, even at the "low end".

Yes indeed - what I like most is the ability of the special ECC RAM to
effectively "ignore" any minor ram failure.

Supposedly a one-bit failure can be recovered-from, and a two-bit
failure can be "reported" - so a user would be aware of possible data
corruption.

I have had more than my share of ram failures in the past, so I look
for features like that in a computer.

Can your Xserve run a partition with ordinary Panther 10.3.8 - or is
the hardware so different on the Xserve that it does not allow that?

My work demands a combination of "server" and "personal-computer" so I
am considering buying one - - - the negative reports I have heard are
that the Xserve's fans are very noisy, and supposedly it is a real
hassle to get keyboard, mouse, and monitor hooked up to it because it
is more-or-less designed to be a headless server.

I am strongly "addicted" to Macs, however, so would consider buying the
Xserve just for its added ram reliability :)

Mark-

David Magda

unread,
Mar 21, 2005, 4:23:08 PM3/21/05
to
Mark Conrad <NoSpam...@invalid.com> writes:

> Yes indeed - what I like most is the ability of the special ECC RAM
> to effectively "ignore" any minor ram failure.
>
> Supposedly a one-bit failure can be recovered-from, and a two-bit
> failure can be "reported" - so a user would be aware of possible
> data corruption.

Are these reported to the OS in any way?

I know on Sun servers the system will even tell you which slot and
DIMM has the issue.

> I am strongly "addicted" to Macs, however, so would consider buying
> the Xserve just for its added ram reliability :)

Most serious server manufacturers have had this for many years.

--
David Magda <dmagda at ee.ryerson.ca>, http://www.magda.ca/
Because the innovator has for enemies all those who have done well under
the old conditions, and lukewarm defenders in those who may do well
under the new. -- Niccolo Machiavelli, _The Prince_, Chapter VI

Mark Conrad

unread,
Mar 22, 2005, 1:57:10 AM3/22/05
to
In article <86acowy...@number6.magda.ca>, David Magda
<dmagda+tr...@ee.ryerson.ca> wrote:

> > Supposedly a one-bit failure can be recovered-from, and a two-bit
> > failure can be "reported" - so a user would be aware of possible
> > data corruption.
>
> Are these reported to the OS in any way?

The assumption is "yes" - it would make no sense to me that if a
certain bit failed constantly, that it is not reported by the OS.

I could not find anything at the website below that said in so many
words that the bit error was reported, however.

<http://www.anandtech.com/guides/viewfaq.aspx?i=3>


The Linux website below stated that the bit error _was_ reported by
their OS's, also reported which dimm was faulty, like you said about
Sun gear.

<http://www.anime.net/~goemon/linux-ecc/>

As regards my tentative idea to purchase an Apple Xserve computer, and
use it as a personal computer, I have decided that it is not a good
idea for my situation here.

The hardware components of the two different types of Apple computers
are just too different to allow the Xserve to be used as an ordinary
personal computer, without running into problems. <sigh>

What really perturbs me no end is that the one good ram testing utility
that I use presently will not work with new Macs that can't boot OS9.

That good ram testing suite is part of TechTool Pro made by Micromat.

It is _only_ in their older OS9 version of TechTool, not in the newer
OS-X version of TechTool.

The old OS9 version can run these test patterns which take hours:
Rotational
Web
Leap
Arpeggio

...and the below patterns, which take very much longer, depending on
the speed of the computer and the amount of ram being tested:
Minor March
Major March


The Major-March pattern is the most complete, it tests the effect of
varying every possible combination of bits, upon the one bit actually
under test at the time. Major-March is generally only run on
super-computers, because that test takes so long to run.

In my case however, I would run Major-March on the first 256-MB of ram
memory on the dual G5, because that ram is critical for OS X core
processes.

I calculated Major-March would take 9-days to run on 256-MB.<g>

The only alternative we have is the free ram test util' "memtest" run
from single-user mode. It sucks, can't even do the four tests:
Rotational
Web
Leap
Arpeggio

Arpeggio, for example, was the _only_ ram test that failed for an old
Mac Lombard that I had, which was still under warranty. It had
intermittant freezes.

I went round-and-round with Apple for 3-months, finally getting them to
fix the old Lombard for free, based on my test results of the Arpeggio
test. Without that test for "political leverage", I would likely still
be listening to the Apple tech' saying: "The freezes must be because
of something at your end".

Mark-

Bill Lloyd

unread,
Mar 22, 2005, 1:10:41 PM3/22/05
to
On 2005-03-21 13:23:08 -0800, David Magda
<dmagda+tr...@ee.ryerson.ca> said:

> Mark Conrad <NoSpam...@invalid.com> writes:
>
>> Yes indeed - what I like most is the ability of the special ECC RAM
>> to effectively "ignore" any minor ram failure.
>>
>> Supposedly a one-bit failure can be recovered-from, and a two-bit
>> failure can be "reported" - so a user would be aware of possible
>> data corruption.
>
> Are these reported to the OS in any way?

They definitely show up in the server monitor application. I don't
know if it logs it through to the OS or not, but you can pull the logs
in server monitor and probably have them emailed.

0 new messages