Recent content by T.Herrmann

@Q3tNHn you mean S7150 x2. In my Opionen it could be a bioas problem with Dell. Dell and HPE are very unfriendly for extension by users.

Dear @t.lamprecht , is there any timeline for zfs 2.3 implementation?

What is your Hardware Config and Software (Linux Kernel)?

Dear @Q3tNHn this works for the latest PVE Kernel https://github.com/nicman23/MxGPU-Virtualization

Dear @kevinlels , I testet Windows 10 and it works fine. Config: Windows 10 VM Driver: https://drivers.amd.com/drivers/prographics/win10-radeon-pro-software-enterprise-20.q2.2-dec4.exe 1x GPU Best Tim

Dear @kevinlels , thank you now it works. We will test now a Linux Ubuntu 24 VM and a Windows 10 VM. I will inform you if we are have a stable running for both operation systems.

Dear @kevinlels, this repo works with PVE 8.3 and Kernel 6.8. Now the problem is to load the gim.ko by modprobe gim. INSTALL /lib/modules/6.8.12-8-pve/updates/gim.ko even if I copy the gim.ko modul direct to /lib/modules/6.8.12-8-pve I get error massage. >>> modprobe: FATAL: Module gim...

Dear @kevinlels , I test you tutorial. Here is the error massages I get. root@diz-ummd-s01:/home/TH_2025/V2/MxGPU-Virtualization# ./gim.sh Current User is root check: /usr/src/linux-headers-6.8.12-8-pve make -C /lib/modules/6.8.12-8-pve/build M=/home/TH_2025/V2/MxGPU-Virtualization/drv modules...

Yes, but especially for LLM with large models that not fit into the GPU RAM a LXC with big EPYC 512 Core platform would be a very efficient way to use this as a side application instead of bind this resources by a VM which can't be used for other LXC/VM. LXC are really nice for over provisioning.

Top shows the same problem. If I adapt 8 cores it shows all cores of the two socket CPU's in the docker container but not in the LXC it self.

Is there any way to fix cores of one socket from dual socket system to a LXC? If have this problem with Ollama https://github.com/ollama/ollama/issues/5554 Thank you to all supporters.

Dear @unholyhumorousunratedoat thank you. I extend it for a better understanding. proxmox-backup-client list --repository pbsIP-pbsdatastore export PBS_REPOSITORY='pbsIP-pbsdatastore' proxmox-backup-client list Then proxmox-file-restore list vm/142/2024-10-03T01:23:15Z / is working.

Dear @Chris, I have the problem when I use proxmox-backup-client list or proxmox-file-restore list vm/142/2024-10-03T01:23:15Z / I get always the error massage: Error: unable to get (default) repository What is wrong? Best Tim

Dear George, yes I try it also on the newest Kernel and QEMU Version. There is on interesting thing by the ACPI PCI path between Windows an Other OS Config. If change the OSTyp. VM OSType Other: IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/SE0@1C/IOPP/GFX0@0...

Thank you, very interesting, that PCI Path is changing PciRoot a little bit if change the OSTyp. VM OSType Other: IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/SE0@1C/IOPP/GFX0@0 PciRoot(0x1)/Pci(0x1C,0x0)/Pci(0x0,0x0) VM OSType: Win10...