Hey guys, i noticed on my Intel N5105/ 4 intel i226v ports mini pc with 16GB memory, which is running Openwrt router VM and AGH LXC, while doing download speedtest or downloading a big file i get very high cpu usage, almost 85% total cpu load while only downloading at around 900mbps, shouldn't cpu load be much less at around 900mbps?
I'm not even running SQM on download, only on upload side..and turning off SQM makes no difference also, tested Opnsense and cpu load was even worse.
Shouldn't N5105 be able to handle 2gb hypervisor routing without any issue? i'm a doing anything wrong?
1st screenshot from proxmox host with openwrt using virtios for wan lan
then later i tested NICs passthrough: (2nd screenshot)
enabled PCIE pass through for WAN/LAN nics and kept 3rd port as virtio inside openwrt lan bridge for management and proxmox internet access, this setup dropped cpu load by almost 40%!!
Does it make sense that virtio have such huge overhead? almost double the load at the same speed, i already tried lowering multique on virtios, while it helped a bit it didn't fix the high load.
root@proxmox:~# qm config 100
agent: 1
balloon: 0
boot: order=scsi0
cores: 4
cpu: host
ide2: none,media=cdrom
machine: q35
memory: 4096
meta: creation-qemu=11.0.0,ctime=1787008241
name: Openwrt
net0: virtio=BC:24:11:37:7B:7E,bridge=vmbr0,queues=4
net1: virtio=BC:24:11:09:00:09,bridge=vmbr1,queues=4
numa: 0
onboot: 1
ostype: l26
parent: backup1
scsi0: local-lvm:vm-100-disk-0,iothread=1,size=20604M
scsihw: virtio-scsi-single
smbios1: uuid=215c1bb6-a8e5-488d-b361-d9d82564ce35
sockets: 1
vmgenid: f8d313b1-0c29-44e6-8638-fc8389496443
qm config 100
agent: 1
allow-ksm: 0
balloon: 0
boot: order=scsi0
cores: 4
cpu: host
hostpci0: 0000:03:00,pcie=1
hostpci1: 0000:02:00,pcie=1
ide2: none,media=cdrom
machine: q35
memory: 4096
meta: creation-qemu=11.0.0,ctime=1789602894
name: Openwrt
net1: virtio=BC:24:11
A:3E:A4,bridge=vmbr2
numa: 0
onboot: 1
ostype: l26
scsi0: local-lvm:vm-100-disk-0,iothread=1,size=20604M,ssd=1
scsihw: virtio-scsi-single
smbios1: uuid=d5f1ed33-cc47-4436-8ef7-91e41ebefcc4
sockets: 1
vmgenid: 87e7032d-8dae-4fcb-a8a6-4c8a1ee86eaf
root@prxmox:~#
I'm not even running SQM on download, only on upload side..and turning off SQM makes no difference also, tested Opnsense and cpu load was even worse.
Shouldn't N5105 be able to handle 2gb hypervisor routing without any issue? i'm a doing anything wrong?
1st screenshot from proxmox host with openwrt using virtios for wan lan
then later i tested NICs passthrough: (2nd screenshot)
enabled PCIE pass through for WAN/LAN nics and kept 3rd port as virtio inside openwrt lan bridge for management and proxmox internet access, this setup dropped cpu load by almost 40%!!
Does it make sense that virtio have such huge overhead? almost double the load at the same speed, i already tried lowering multique on virtios, while it helped a bit it didn't fix the high load.
root@proxmox:~# qm config 100
agent: 1
balloon: 0
boot: order=scsi0
cores: 4
cpu: host
ide2: none,media=cdrom
machine: q35
memory: 4096
meta: creation-qemu=11.0.0,ctime=1787008241
name: Openwrt
net0: virtio=BC:24:11:37:7B:7E,bridge=vmbr0,queues=4
net1: virtio=BC:24:11:09:00:09,bridge=vmbr1,queues=4
numa: 0
onboot: 1
ostype: l26
parent: backup1
scsi0: local-lvm:vm-100-disk-0,iothread=1,size=20604M
scsihw: virtio-scsi-single
smbios1: uuid=215c1bb6-a8e5-488d-b361-d9d82564ce35
sockets: 1
vmgenid: f8d313b1-0c29-44e6-8638-fc8389496443
qm config 100
agent: 1
allow-ksm: 0
balloon: 0
boot: order=scsi0
cores: 4
cpu: host
hostpci0: 0000:03:00,pcie=1
hostpci1: 0000:02:00,pcie=1
ide2: none,media=cdrom
machine: q35
memory: 4096
meta: creation-qemu=11.0.0,ctime=1789602894
name: Openwrt
net1: virtio=BC:24:11
A:3E:A4,bridge=vmbr2numa: 0
onboot: 1
ostype: l26
scsi0: local-lvm:vm-100-disk-0,iothread=1,size=20604M,ssd=1
scsihw: virtio-scsi-single
smbios1: uuid=d5f1ed33-cc47-4436-8ef7-91e41ebefcc4
sockets: 1
vmgenid: 87e7032d-8dae-4fcb-a8a6-4c8a1ee86eaf
root@prxmox:~#
Attachments
Last edited: