1 |
On Thursday, 23 September 2021 17:32:46 BST Charlotte Delenk wrote: |
2 |
> On 9/23/21 18:30, Grant Taylor wrote: |
3 |
> |
4 |
> > On 9/23/21 4:39 AM, Miles Malone wrote: |
5 |
> > |
6 |
> >> You'd need NUMA if you had a NUMA machine. In current context, that |
7 |
> >> would be either a) a dual socket system, b) an amd threadripper, or |
8 |
> >> c) some of the really high core xeons. If your motherboard doesnt |
9 |
> >> have certain memory banks allocated to certain processors or cores, |
10 |
> >> you're probably not running a NUMA machine. |
11 |
> > |
12 |
> > |
13 |
> > |
14 |
> > Will a kernel without NUMA support boot and run on a system that has a |
15 |
> > NUMA architecture? |
16 |
> > |
17 |
> > |
18 |
> > |
19 |
> > If it will boot and run, does it simply do so in a sub-optimal way? |
20 |
> |
21 |
> Pure speculation for this one but if it works it's probably only going |
22 |
> to detect part of the memory and some of the devices and one cpu, since |
23 |
> the other resources are physically connected to the other cpus. |
24 |
> |
25 |
> > |
26 |
> > |
27 |
> > Flipping the coin on the other side, is there any negative effect |
28 |
> > (other than kernel size / lines of code / attack surface) for having |
29 |
> > NUMA support enabled on a non-NUMA system? |
30 |
> |
31 |
> There is no meaningful downside to leaving it enabled, it's enabled in |
32 |
> many distribution kernels for a reason. |
33 |
|
34 |
I've booted a kernel with no NUMA config, and it seems to run fine on this |
35 |
single-socket Ryzen motherboard. I just get the one entry in dmesg: |
36 |
|
37 |
$ dmesg | grep -i numa |
38 |
[ 0.297998] pci_bus 0000:00: on NUMA node 0 |
39 |
|
40 |
That's it. I'm running five BOINC projects, some of which run on vbox, so on |
41 |
this motherboard it seems clear that I don't need NUMA. |
42 |
|
43 |
-- |
44 |
Regards, |
45 |
Peter. |