1 |
On Friday, 24 September 2021 10:06:49 BST Peter Humphrey wrote: |
2 |
> On Thursday, 23 September 2021 19:20:52 BST Michael wrote: |
3 |
> > Out of interest, have you tried booting a NUMA enabled kernel to see what |
4 |
> > dmesg reports? |
5 |
> |
6 |
> Yes, it's been enabled ever since I had a dual-socket motherboard, years |
7 |
> ago. I didn't understand why I did or didn't need it until I read Miles's |
8 |
> post yesterday (thanks, Miles). I don't know why it hadn't been made clear |
9 |
> in any websites I've visited. |
10 |
> |
11 |
> > On an old laptop, which definitely has only a single AMD |
12 |
> > APU, I get: |
13 |
> > |
14 |
> > $ dmesg | grep -i NUMA -A2 |
15 |
> > [ 0.002078] No NUMA configuration found |
16 |
> > [ 0.002080] Faking a node at [mem |
17 |
> |
18 |
> 0x0000000000000000-0x000000042effffff] |
19 |
> |
20 |
> > [ 0.002085] NODE_DATA(0) allocated [mem 0x42effc000-0x42effffff] |
21 |
> |
22 |
> I had something similar. Oddly, with NUMA configured I get "not found" and |
23 |
> without it I get "pci_bus 0000:00: on NUMA node 0". The system seems to run |
24 |
> happily either way. |
25 |
|
26 |
Sorry I should have made it clear - the above "No NUMA configuration found" |
27 |
message was obtained *with* NUMA enabled in my kernel. |
28 |
|
29 |
I suppose "NUMA on node 0" is the default first socket, which the kernel sets |
30 |
up. If the kernel can't find a second CPU it will be 'faking' a multi-CPU |
31 |
memory allocation setup, when it comes to allocate memory to the only CPU |
32 |
available. If the kernel does not have NUMA enabled then it doesn't need to |
33 |
fake anything. It will treat the hardware as a single socket MoBo and no |
34 |
further tests would be undertaken. All suppositions of course, I haven't |
35 |
looked at the code. ;-) |