| 1 |
On 4/1/06, kashani <kashani-list@××××××××.net> wrote: |
| 2 |
> |
| 3 |
> I've made my problem go away, but I'm a bit curious about why it |
| 4 |
> was |
| 5 |
> happening. |
| 6 |
> |
| 7 |
> I've got nine or so Dell 1850's with 2 x 2GB chips. When I compile |
| 8 |
> gentoo-source 2.6.15-r1 (and 2.6.14-r5) with support for 4GB I see only |
| 9 |
> 3GB. 1GB lowmem and 2GB highmem. |
| 10 |
> |
| 11 |
> ws04 ~ # cat /proc/meminfo |
| 12 |
> MemTotal: 3107408 kB |
| 13 |
> MemFree: 1731660 kB |
| 14 |
> Buffers: 219720 kB |
| 15 |
> Cached: 937980 kB |
| 16 |
> SwapCached: 0 kB |
| 17 |
> Active: 455100 kB |
| 18 |
> Inactive: 717924 kB |
| 19 |
> HighTotal: 2227968 kB |
| 20 |
> HighFree: 1271588 kB |
| 21 |
> |
| 22 |
> When I recompile on the same hardware with support for 64GB RAM I see |
| 23 |
> the full 4GB in the usual 1GB/3GB split. |
| 24 |
> |
| 25 |
> nms02 ~ # cat /proc/meminfo |
| 26 |
> MemTotal: 4147776 kB |
| 27 |
> MemFree: 3762712 kB |
| 28 |
> Buffers: 151404 kB |
| 29 |
> Cached: 45116 kB |
| 30 |
> SwapCached: 0 kB |
| 31 |
> Active: 56724 kB |
| 32 |
> Inactive: 145000 kB |
| 33 |
> HighTotal: 3276544 kB |
| 34 |
> HighFree: 3222936 kB |
| 35 |
> |
| 36 |
> However on my HP DL360 with 4 x 1GB RAM I see the 4GB, though it's a bit |
| 37 |
> smaller, with 4GB enabled in the kernel. Same kernel version as the |
| 38 |
> Dell's. |
| 39 |
> |
| 40 |
> ramin@azul ~ $ cat /proc/meminfo |
| 41 |
> MemTotal: 3977744 kB |
| 42 |
> MemFree: 116792 kB |
| 43 |
> Buffers: 296108 kB |
| 44 |
> Cached: 3157952 kB |
| 45 |
> SwapCached: 0 kB |
| 46 |
> Active: 2725152 kB |
| 47 |
> Inactive: 782956 kB |
| 48 |
> HighTotal: 3096552 kB |
| 49 |
> HighFree: 8680 kB |
| 50 |
> |
| 51 |
> I've been Googling around for an explantion, but am not really sure what |
| 52 |
> I'm looking for. I'd assume that I might lose a few hundred MB if the 2 |
| 53 |
> x 2GB chips are slightly bigger than the 4GB limit rather than the full |
| 54 |
> 1GB that seems to disappear. I'm guessing some wacky motherboard |
| 55 |
> interleaving thing where highmem ends up being one chip and the lowmem |
| 56 |
> the 896MB on the other chip. |
| 57 |
> |
| 58 |
> Anyone have a decent theory on this with a nice link? |
| 59 |
> |
| 60 |
> kashani |
| 61 |
> -- |
| 62 |
> gentoo-user@g.o mailing list |
| 63 |
> |
| 64 |
> |
| 65 |
|
| 66 |
this is the same thing that happens at the 800-1G spot... a machine with 1G |
| 67 |
need to enable 4G supprt to see it all |
Archives