slowly but surely, the funnest (ja, synack) hardware will be rescued from the storage space and returned to service. this pair of lovely Nvidia T40 24GB GPUs (Turing architecture) will be swapped into another Xeon server for a variety of ML/HPC workloads.
a new state, a new house, and all electrical costs included... well it's time to build out the latest iteration of my Homelab. over two decades of technical shenanigans, formally starting in 2001 with a new ebay account and my first online credit card purchase as a college undergrad (who doesn't want their apartment bedroom full of servers?).
the full architecture plan is still floating around in the brain undergoing spatial/comparative analysis, but in strong consideration for the GPU workload box is a 12-18U silent server rack with Tripplite AC connected via duct and blower with MERV14+ air filtration. most of the rest will be embedded systems installed on DIN rails with a 48V—240W PoE++ core network and standardizing on 5-12V via DC/gigE and DC/2.5ge PoE+ splitters — that gear is all ready to go, just need to decide which rack.
what kind of rack would you prefer, if you were rebuilding your lab?
#homelab #gpu #hpc #machinelearning #nvidia #tesla #rtx #turing #virtualization #virtualisierung #freebsd #linux #oss #foss #WomenInTech #WomenInSTEM
Alt...
Eva kneeling down, balancing a GPU, model "Nvidia T40 24GB" datacenter GPU (Turing architecture)
Alt...
Eva's left hand is balancing a GPU, model "Nvidia T40 24GB" for use in datacenters.
* Used Resources: Rebalances VMs for the optimal resource usage by their currently used resources for efficiency
* Assigned Resources: Rebalances VMs by their max. assigned resources to guarantee the resources (avoid overprovisioning)
#proxmox #virtualmachines #hosting #kvm #qemu #esx #esxi #drs #cluster #vm #vms #virtualisation #virtualization #virtualisierung #opensource
#kvm #qemu #vmware #virt #vm #virtualization #virtualisation #virtualisierung #opensource
