Microsoft Task Manager creator Dave Plummer powers a Stirling engine with AMD Threadripper chipset heat
Server platform hardware - directly relevant to the H12SSL-i / EPYC stack.
Independent publication // local-first AI // field reports from owned systems
Nodehome
A running publication about local models, private inference, self-hosted agents, weird hardware, research sweeps, and the builders wiring their own AI stack together.
AI is getting physical again. It shows up in terminals, racks, side projects, and ugly little workflows people actually control.
Latest
Daily Sweep - Jul 05, 2026The Log is the Agent
Agent or tool-use pattern - relevant to local AI workflow automation.
BareMetal RAM Dumper - Bare-metal x86 tool for Cold Boot Attack experiments
French retailer lists ASUS ROG Astral RTX 5090 Edition 20 at €6,499, and it’s not even the most expensive RTX 5090
GPU hardware mention - directly relevant to the build or resale market.
Silicon Motion Shows Off Enterprise SM8008 and SM8388 SSD Controllers at Computex 2026
ServeTheHome coverage - server hardware reviews and benchmarks from a trusted source.
Jensen Huang’s signed leather jacket may sell for the price of 10 RTX 5090 cards
GPU hardware mention - directly relevant to the build or resale market.
RTX 5090 backplate melts PCIe riser insulation in vertical GPU setup
Bus or storage bandwidth topic - relevant to the H12SSL-i's PCIe 4.0 layout.
Samsung 990 PCIe 4.0 SSD surfaces with up to 7,250 MB/s read speed
Bus or storage bandwidth topic - relevant to the H12SSL-i's PCIe 4.0 layout.
Field Reports
builds, experiments, notesThree RTX 3090s, One 32B Model: A Pipeline-Parallel Canary
A current field note on why the 3x3090 serving path moved through pipeline parallelism, not tensor parallelism, for the tested 32B AWQ model.
Gemma 4 12B And The Sensory Agent Lane
A public-safe read on Gemma 4 12B as a local sensory preprocessor: useful for seeing, hearing, and structuring observations without turning into an action system.
Hardware
machines, thermals, economicsPower Caps On Three RTX 3090s: Bursts Versus Sustained Load
A measured note on 300W bursty inference, lower caps for sustained runs, and why power-cap sweet spots are workload-specific.
Parallel Agent Serving Is A Hardware Shape Now
A field-report read on 14x RTX 3090 agent serving, EXL3, FP8 KV cache, Aphrodite, and why concurrency is becoming the local hardware metric.