public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Simen Thoresen <simentt@dolphinics.no>
To: gmu 2k6 <gmu2006@gmail.com>
Cc: Joel Jaeggli <joelja@uoregon.edu>, linux-kernel@vger.kernel.org
Subject: Re: Only 3.2G ram out of 4G seen in an i386 box
Date: Thu, 10 Aug 2006 23:09:02 +0200	[thread overview]
Message-ID: <44DBA06E.40104@dolphinics.no> (raw)
In-Reply-To: <f96157c40608090754m1f10e0f2h5fbf3b256d2e55e1@mail.gmail.com>

gmu 2k6 wrote:
...
> so what does it mean that one of Xeons here shows me the full 4GiB as
> total physical memory via `free`?
> 
> btw, the box I'm getting will have the 975X chip and include 4GiB RAM
> and if I understood the problem correctly even with 3GiB there will be
> some memory lost to mapping-issues besides the HI/LO mem
> kernel-reserving issue.
> this is what I get on ia32 P4 with 3GiB
>             total       used       free     shared    buffers     cached
> Mem:       3116108    2608196     507912          0     246652    2039708
> and this is what I get on ia32 Xeon with 4GiB
>             total       used       free     shared    buffers     cached
> Mem:       4087660     268828    3818832          0      57168     103568

Hi Gmu,

This is a bit divergent from the discussion, but will hopefully provide some 
background.

The hardware my employer manufactures asks the BIOS to reserve several 
largeish ranges (16M and 16M to 512M), and it's been used in various HPC 
projects where machines typically have a lot of memory installed.

 From my experience, most pre 750x-P4-Xeon chipsets did not support any kind 
of memory remapping, so often the top 0.5-1.5G of ram would be 'lost'. Great 
fun.
Since then, this functionality has started appearing, more or less randomly. 
I believe it should be present on all later 75xx chipsets (probably the Core 
2 Xeon chipsets as well), but implementation depends on the BIOS (ie the 
motherboard vendors), and their grasp of what people do with their boards is 
often quite limited.

On the P4 boards (and again the Core 2 boards as well), this is probably 
even less ordered as they are 'desktop' boards rather than the more costly 
'server' boards.

I'm pretty sure the Athlon64 and Opteron CPUs should support this, but they 
too depend on BIOS functionality to rearrange the memory tables.

Just out of interest - what chipset does your Xeon system use?

Yours,
-S

-- 
Simen Thoresen, Dolphin ICS

  reply	other threads:[~2006-08-10 21:09 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-08-08  8:24 Only 3.2G ram out of 4G seen in an i386 box Thomas Stewart
2006-08-08 10:15 ` Paul P Komkoff Jr
2006-08-08 19:51   ` David Schwartz
2006-08-09  6:51     ` gmu 2k6
2006-08-09 14:48       ` Joel Jaeggli
2006-08-09 14:54         ` gmu 2k6
2006-08-10 21:09           ` Simen Thoresen [this message]
2006-08-11 11:08             ` gmu 2k6
2006-08-13 10:54               ` Simen Thoresen
2006-08-10 13:04         ` Thomas Stewart
  -- strict thread matches above, loose matches on Subject: below --
2006-08-08  9:13 Mikael Pettersson
2006-08-08 10:01 ` Thomas Stewart
     [not found] <fa.FhtZk3YiH9OyGCt4iXIGt+vYtos@ifi.uio.no>
2006-08-08 14:22 ` Robert Hancock

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=44DBA06E.40104@dolphinics.no \
    --to=simentt@dolphinics.no \
    --cc=gmu2006@gmail.com \
    --cc=joelja@uoregon.edu \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox