All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Is this typical for a duel CPU system?
sebastian97
Member
Ordered a Dual Intel Xeon E5-2683-v4 system from oplink...
I got an error from a program saying it couldn't properly use half of the CPU cores.
root@server:~# numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
node 0 size: 128673 MB
node 0 free: 126948 MB
node 1 cpus: 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
node 1 size: 0 MB
node 1 free: 0 MB
node distances:
node 0 1
0: 10 20
1: 20 10
root@server:~# dmesg | grep -i numa
[ 0.012795] NUMA: Initialized distance table, cnt=2
[ 0.012799] NUMA: Node 0 [mem 0x00000000-0x7fffffff] + [mem 0x100000000-0x207fffffff] -> [mem 0x00000000-0x207fffffff]
[ 0.152623] mempolicy: Enabling automatic NUMA balancing. Configure with numa_balancing= or the kernel.numa_balancing sysctl
[ 0.819385] pci_bus 0000:ff: Unknown NUMA node; performance will be reduced
[ 0.828319] pci_bus 0000:7f: Unknown NUMA node; performance will be reduced
[ 0.860857] pci_bus 0000:00: on NUMA node 0
[ 0.865125] pci_bus 0000:80: on NUMA node 1
That isn't typical for a duel CPU server, is it?
Thanks!
Comments
I don't see anything unusual here. Two CPU system has two NUMA nodes as you would usually expect it to. AMD EPYC can have more NUMA nodes than physical CPU count depending on BIOS configuration.
The RAM is all installed to one CPU, it was an issue. Thankfully support responded quickly and adjusted the billing cycle and will fix it in the morning.
Oh, haven't noticed it, should go sleep already.
Anyway you should be able to use RAM from a different NUMA node, with a performance overhead however.
Thankfully this is for a migration that isn't needed as soon as possible, so it is no trouble to wait. I appreciate the help. Have a good rest.