| 1 |
On Saturday 14 May 2005 9:57 pm, Bastian Balthazar Bux wrote: |
| 2 |
> Christian Parpart wrote: |
| 3 |
> > On Saturday 14 May 2005 6:28 pm, Bastian Balthazar Bux wrote: |
| 4 |
> >>Christian Parpart wrote: |
| 5 |
> >>>Hi all, |
| 6 |
> >>> |
| 7 |
> >>>I finally got it; my Operaton workstation with 2x Opteron248 CPUs and |
| 8 |
> >>> 4GB RAM is arrived. However, the BIOS says that I really have 4GB RAM |
| 9 |
> >>> inside, that I have, but the Linux is just telling me something about |
| 10 |
> >>> 3GB. `free -mt` as well as `kinfocenter` are telling this. Although, I |
| 11 |
> >>>(unfortunately) *had* to install windows2003 on a partition, too (for |
| 12 |
> >>>working for my company), and there it is also just telling me something |
| 13 |
> >>>about 3 (instead of 4) GB RAM. |
| 14 |
> >>> |
| 15 |
> >>>Did I miss something? |
| 16 |
> >>> |
| 17 |
> >>>mem=4096M in boot commandline didn't help there much. |
| 18 |
> >>> |
| 19 |
> >>>I'm having a TYAN board, with PhoenixBIOS on it. The handbook is far off |
| 20 |
> >>>being a good handbox, so, did I miss something I should |
| 21 |
> >>>configure/reconfigure in the BIOS then, so, that the OS will use all |
| 22 |
> >>> RAM? |
| 23 |
> >> |
| 24 |
> >>theese are more or less memory related values in my kernel config |
| 25 |
> >> |
| 26 |
> >>CONFIG_MTRR=y |
| 27 |
> >>CONFIG_K8_NUMA=y |
| 28 |
> >>CONFIG_DISCONTIGMEM=y |
| 29 |
> >>CONFIG_GART_IOMMU=y |
| 30 |
> > |
| 31 |
> > trapni@battousai ~ $ zgrep -E 'CONFIG_(MTRR|K8_NUMA|DISCONTIGMEM| |
| 32 |
> > GARTIOMMU)' /proc/config.gz |
| 33 |
> > CONFIG_MTRR=y |
| 34 |
> > CONFIG_K8_NUMA=y |
| 35 |
> > CONFIG_DISCONTIGMEM=y |
| 36 |
> > |
| 37 |
> > though, CONFIG_GART_IOMMU must have been disabled, and I remember I did |
| 38 |
> > not enable gart at all because I'm having Nvidia GeForce 6600GT |
| 39 |
> > (PCI-express) and wanna use the `Option "NvAGP" "1"` line in my xorg.conf |
| 40 |
> > - this is at least what I had with my old GeForce 5200 working before. |
| 41 |
> > |
| 42 |
> >>and this my "free" output: |
| 43 |
> >>free -m |
| 44 |
> >> total used free shared buffers cached |
| 45 |
> >>Mem: 3429 3069 359 0 627 1775 |
| 46 |
> >>-/+ buffers/cache: 667 2762 |
| 47 |
> >>Swap: 0 0 0 |
| 48 |
> > |
| 49 |
> > trapni@battousai ~ $ free -m |
| 50 |
> > total used free shared buffers cached |
| 51 |
> > Mem: 3015 2990 24 0 11 2136 |
| 52 |
> > -/+ buffers/cache: 843 2172 |
| 53 |
> > Swap: 511 19 492 |
| 54 |
> > |
| 55 |
> > that's really borring looking at the Mem-total line, it should be |
| 56 |
> > anything around 4048, not 3015 anyway... |
| 57 |
> > |
| 58 |
> > though, it's unfortunately *not* the same as you. |
| 59 |
> > |
| 60 |
> > but, why duo have a total-mem of 3429M? how much RAM duo have physically |
| 61 |
> > plugged in to your board's slots? |
| 62 |
> |
| 63 |
> (1G+1G) + (1G+1G) |
| 64 |
> |
| 65 |
> and "dmidecode" show this about memory: |
| 66 |
[...] |
| 67 |
|
| 68 |
that's rather interesting - I didn't know about such tools before :) |
| 69 |
|
| 70 |
Though, I must see, that at least dmidecode is aware of my 4GB RAM. |
| 71 |
I've uploaded the full output of dmidecode to [0]. |
| 72 |
|
| 73 |
So, now it's imteresting in why the dmidecode software sees all RAM (at least |
| 74 |
by reading from the boot specs), and why the OS doesn't... |
| 75 |
|
| 76 |
> I've not time/possibility to investigate it further, if you came to a |
| 77 |
> solution ... sharing will be higly apreciated |
| 78 |
|
| 79 |
Of course I'll do so (ASAP I know WTF it isn't using all my RAM) |
| 80 |
|
| 81 |
Regards, |
| 82 |
Christian Parpart. |
| 83 |
|
| 84 |
[0] http://dev.gentoo.org/~trapni/dmidecode.txt |
| 85 |
|
| 86 |
-- |
| 87 |
Netiquette: http://www.ietf.org/rfc/rfc1855.txt |
| 88 |
15:08:20 up 53 days, 4:14, 0 users, load average: 1.53, 1.70, 1.54 |
Archives