Good day, I've started testing a dpdk application on a new system with more cores than I usually do. I'm running into an error relating to the number of numa nodes on the system. We use the socket-mem argument to allocate just the right amount of huge pages on each numa node (socket?).
lscpu output looks as follows: lscpu Architecture: x86_64 <snip> CPU(s): 160 On-line CPU(s) list: 0-159 Thread(s) per core: 2 Core(s) per socket: 20 Socket(s): 4 NUMA node(s): 8 Vendor ID: GenuineIntel CPU family: 6 Model: 85 Model name: Intel(R) Xeon(R) Gold 6248 CPU @ 2.50GHz <snip> NUMA node0 CPU(s): 0-9,80-89 NUMA node1 CPU(s): 10-19,90-99 NUMA node2 CPU(s): 20-29,100-109 NUMA node3 CPU(s): 30-39,110-119 NUMA node4 CPU(s): 40-49,120-129 NUMA node5 CPU(s): 50-59,130-139 NUMA node6 CPU(s): 60-69,140-149 NUMA node7 CPU(s): 70-79,150-159 So my current scripting mechanism would use the following --socket-mem argument to start a dpdk instance on the first numa node: app_name <other_args> --socket-mem=570,0,0,0,0,0,0,0 --core 1 On this system, I now get: EAL: Detected 128 lcore(s) EAL: Detected 4 NUMA nodes EAL: invalid parameters for --socket-mem I suppose dkdk is looking at the number of sockets and I've been using number of numa nodes in my scripts. However it's not clear to me how to simply fix the issue since hugepage allocation is done on the level of numa nodes. Also eal init seems to say it detected 4 numa nodes when the system has 8, at least according to lscpu output. I'm on dpdk 20.11 if that's pertinent. Thanks in advance for any suggestions. Pico
