View Single Post
Old 2021-07-16, 06:51   #35
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

26·7 Posts
Default

Quote:
Originally Posted by sdbardwick View Post
A quick read of the MAN page for NUMACTL makes me think that
Code:
$ numactl --physcpubind=0 mprime0 -m
attempts to assign mprime to core 0, not socket 0. I'd investigate cpunodebind:

--cpunodebind=nodes, -N nodes
Only execute process on the CPUs of nodes. Note that nodes may consist of multiple CPUs.

Thank you. I looked in the man page and see the -H command shows the hardware. I see there are two nodes, and each node has 52 cpus, which I assume is due to the hyperthreading.

Code:
drkirkby@canary:~$ numactl -H
available: 2 nodes (0-1)
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 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
node 0 size: 192100 MB
node 0 free: 167392 MB
node 1 cpus: 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
node 1 size: 193498 MB
node 1 free: 175122 MB
node distances:
node   0   1
  0:  10  21
  1:  21  10
drkirkby@canary:~$
Maybe for physical CPU 0 I need to specify the cpus as 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 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 and 77. I wonder if 0 to 25 correspond to physical cores and 52-77 are due to the hyperthreading.

I assume the 192100 MB and 193498 MB correspond to the RAM connected to the nodes. I'm wondering why they are slightly different. i would expect 196,608 MB on each.

Thank you for your help - I will look into this more today.
drkirkby is offline   Reply With Quote