proxmox on arm64

[kbftech]

Really? Not in my world (Europe). Its more or less impossible for most to purchase a stable and powerful arm server which is on the same performance level as widely used amd64 hardware. Seem this is different in China.?

As soon as this change, we are prepared to release Proxmox VE for arm64 (most work is already done).

https://www.ipi.wiki/products/ampere-altra-developer-platform?variant=42970872086690
https://www.ipi.wiki/products/com-hpc-ampere-altra
https://buy.hpe.com/us/en/compute/p...-server/hpe-proliant-rl300-gen11/p/1014682063

How prepared are you?

 

[MasterCATZ]

Did you get this working?
I managed to get Debian 11 latest with Arm64 installed with 24 Gig RAM 4x arm64 using their Ampere CPU.
When I start the apt install proxmox-ve
something seems to disconnect the network connection. I can access it using the cloud shell but no longer with SSH.

yes I did get it working it was something to do with ifupdown versions or something proxmox uses ifupdown2 , so you have to set that up before doing proxmox , going from memory, I thought I had mentioned it on a forum somewhere , I recently cleaned out my test virtualmachines last week so I can not go through my terminal logs now and can not find my cheat document I was copy pasting from

but this sounds like it https://forum.proxmox.com/threads/ifupdown2-install-breaks-network-pve-7-3-4.121414/


and some other links here

https://forum.proxmox.com/threads/proxmox-on-aarch64-arm64.121925/page-2#post-601030

 

[fmotta]

I have an Ampere based system.
It installed (ish) and is 'working'

I encountered one problem in install and cannot seem to get a VM started

- Install problem: It installed and created 3 partitions. Two have EFI data. However; BIOS is not loading it.
It works if I boot to UEFI shell and execute the grub EFI file.

- Creation of a VM: I created the VM and tried to start it and received error

bios: ovmf
boot: order=scsi0;scsi2;net0
cores: 20
cpu: host,flags=+aes
machine: q35
memory: 40960
meta: creation-qemu=8.1.5,ctime=1716178377
name: ubuntu24-vm
net0: virtio=BC:24:11:9E:9A:5E,bridge=vmbr0,firewall=1
numa: 0
ostype: l26
scsi0: ZFS1:vm-100-disk-0,iothread=1,size=32G,ssd=1
scsi2: local:iso/ubuntu-24.04-live-server-arm64.iso,media=cdrom,size=2420396K
scsihw: virtio-scsi-single
serial0: socket
smbios1: uuid=eff3c526-4a7e-4a06-a75f-fedb1fbb80d1
sockets: 1
vga: serial0


root@proxmox:~# qm start 100
WARN: no efidisk configured! Using temporary efivars disk.
kvm: unsupported machine type
Use -machine help to list supported machines
start failed: QEMU exited with code 1
root@proxmox:~# uname -a
Linux proxmox 6.1.73-generic #1 SMP Tue Jan 16 19:00:42 UTC 2024 aarch64 GNU/Linux
root@proxmox:~#

 

[floh8]

Really? Not in my world (Europe). Its more or less impossible for most to purchase a stable and powerful arm server which is on the same performance level as widely used amd64 hardware. Seem this is different in China.?

As soon as this change, we are prepared to release Proxmox VE for arm64 (most work is already done).

I think they are stable enough, now. Sometime, you will not buy a server because of the performance but because of the energy usage/saving. I think its time for proxmox to enter this ship. xcp-ng starts a collaboration already.

 

[fmotta]

I think they are stable enough, now. Sometime, you will not buy a server because of the performance but because of the energy usage/saving. I think its time for proxmox to enter this ship. xcp-ng starts a collaboration already.

My 128 core ARM desktop with a GPU takes far less power than than any of my laptops.

Just in the power for fans this is a power reduction over my traditional 12core desktop.
Occasionally the chassis fans spin up to assist the passive heatsink on the ARM CPU. This means 5 fans are not spinning all the time as in my traditional desktop.

Oh, and... my ARM system (having a few more cores - 128 not 12), actually does more work, allows for more effective workload partitioning.

Simply stated, by getting more cores, more granularity of control, and more work, I also save about 20% power.

Oh, I almost forgot... I also gained 2@10Gb NICs and more DRAM within the power envelope described.

Seems like a win for everyone to move away from transistor bloat of traditional CPUs.

 

[dusatvoj]

Hi everyone.

I can see comments at the beginning of this thread, explaining that a lot of work has been done - and we are waiting for usable HW in Europe.

There are already SMicros, etc. with 96+ core CPUs.

Don't you think it's already time to build a technical preview for ARM? I can understand that there needs to be a support team for it, etc. But the technical preview could be the way to get a huge community driven testing platform.


cc: @tom

 

[Andrey Zentavr]

Hi everyone.

I can see comments at the beginning of this thread, explaining that a lot of work has been done - and we are waiting for usable HW in Europe.

There are already SMicros, etc. with 96+ core CPUs.

Don't you think it's already time to build a technical preview for ARM? I can understand that there needs to be a support team for it, etc. But the technical preview could be the way to get a huge community driven testing platform.


cc: @tom

There are a couple servers available at Hetzner (Finland region). You can also order additional hard drives and switches is needed, so you can build a cluster.

 

[dusatvoj]

@Andrey Zentavr , yup ... Ampere Altra based systems are now available to use in clouds / to buy and host.
So I'm looking forward to PVE@ARM tech preview