Performance-focused comparisons.
9 articles
Our benchmarks review the metrics that matter when sizing a homelab: throughput (tokens/s for local LLMs, IOPS for storage), latency, power draw, RAM usage, temperatures. Most of our benchmarks are not based on measurements we took ourselves: cross-check their figures against manufacturer spec sheets and official documentation before relying on them. We cover GPUs, mini-PCs, NAS units, VPS and local AI models. No content is sponsored. Use these pages as a starting point to size your homelab or inference server, then head to our shop for the gear mentioned.
Cross-analysis of Wi-Fi 7 and Wi-Fi 6E throughput in real-world conditions. Consolidated benchmark numbers from international tests to help you choose the right router for 2026.
ReadExplore the performance of large language models (LLM) running on a graphics card RTX 4060 Ti, with a detailed comparison of efficiency measured in tokens per second using Ollama.
ReadDiscover our comprehensive benchmark on video transcoding performance with Jellyfin and Intel's integrated GPUs (iGPUs) for 4K content in 2026. Comparative analysis of performance, value-for-money ratio, and expert recommendations.
ReadDiscover our comprehensive comparison between the Raspberry Pi 5 and the Mini-PC N100 in 2026. Detailed performance, power consumption, and real cost are covered to help you make an informed decision.
ReadComparative analysis in 2026 of 3 NVMe SSDs for NAS. Consolidated figures: IOPS, temperature, perf/watt. Which cache is best for your VMs and photos?
ReadComparison of file systems for home NAS in 2026. IOPS, bandwidth and performance per watt. Our consolidated community test results reveal the hidden winner.
ReadDiscover our comprehensive power consumption benchmark for three NAS 4-bay models in 2026, featuring consolidated figures from measurements and manufacturer specifications.
ReadComparison 2026 N100 vs N305: performance, price, power consumption. Our numbers solidify your choice for the perfect Proxmox mini-PC in 2026.
ReadReal-world tests of 17 local LLMs on the NVIDIA GB10 (DGX Spark, 121 GB unified memory). qwen3:30b-a3b dominates at 82.5 tok/s. Data-driven analysis, RAM comparison, and MoE vs. dense verdict.
Read