So I have an Asrock x399 Taichi and an AMD 1950x. I’ve set the memory interleaving mode to channel but numactl still doesn’t see my individual nodes. Has anyone run in to this issue yet?
[timb@arcturus ~]$ numactl --hardware
available: 1 nodes (0)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
node 0 size: 64349 MB
node 0 free: 61991 MB
node distances:
node 0
0: 10
I don’t have Thread Ripper but for clarification, you are saying that you changed the memory mode in the UEFI from UMA to Channel and that your Gnu/Linux install is not reporting the correct NUMA nodes? That is what it looks like.
What kernel are you using? You can find this quickly with uname -a.
Edit
Here is an interesting read. It looks like your system is not recognizing the two nodes though.
I’ll take a look at that link. And yes that’s correct. I changed the memory mode in the UEFI from UMA to Channel and Linux isn’t reporting any numa nodes
[timb@arcturus ~]$ uname -a
Linux arcturus 4.14.13-ARCH #1 SMP PREEMPT Tue Jan 16 09:07:25 PST 2018 x86_64 GNU/Linux
Hmm, you are on the right kernel for support. That is super weird.
You should be seeing at least 2 numa clusters. Very strange that node 0 only sees 11 threads.
Which BIOS version are you on? I have the exact same CPU and mobo combo, memory mode set to channel, and here’s my numactl output:
sudo numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 16 17 18 19 20 21 22 23
node 0 size: 15980 MB
node 0 free: 7227 MB
node 1 cpus: 8 9 10 11 12 13 14 15 24 25 26 27 28 29 30 31
node 1 size: 16115 MB
node 1 free: 7142 MB
node distances:
node 0 1
0: 10 16
1: 16 10
I did have a problem with only one numa node being populated before, but even then the second one showed, albeit empty. I had to switch from using the Asrock recommended layout (putting my two 16GB modules in memory slots A2 and B2) to putting them in A2 and C2 to fix that issue.
How many memory modules do you have, and in which slots did you put them?
Edit: could you also post the output of dmesg|grep -i numa?
Could it be that he needs to use sudo or su because he does not have the correct permissions?
Can’t say for sure, I’m running Fedora 27, @Starlord’s running Arch, but on my rig I can run numactl -H as a non-root user and get accurate results.
4x 16gb in D1, C1 A2, B2.
Installing openSuse Tumbleweed right now
Sorry for the late reply. Things got busy at work.
[timb@arcturus ~]$ dmesg|grep -i numa
[ 0.000000] No NUMA configuration found
[ 1.031169] pci_bus 0000:00: on NUMA node 0
[ 1.077598] pci_bus 0000:40: on NUMA node 1