Hardware Specifications Explorer

The process node (measured in nanometers — 8nm, 5nm, 4nm, 3nm here) is the headline of any spec sheet, but its meaning is mostly relative now: the figure is a marketing name for a transistor library, not a literal feature size. What it reliably tells you is generational density and efficiency — a smaller node from the same foundry generally packs more transistors per square millimeter and delivers better performance-per-watt. That is why filtering for the newest nodes is the fastest way to surface the most power-efficient parts, and why a 5nm GPU and a 5nm CPU can share an efficiency character despite doing completely different work.

For CPUs the spec that decides real responsiveness is the balance of cores, threads, clocks and cache. Cores are physical execution units; threads (via SMT/Hyper-Threading) let one core juggle two instruction streams for throughput. High boost clocks and large per-core caches drive single-thread speed — the thing you feel in games and everyday snappiness — while more cores and threads win multi-threaded rendering, compilation and simulation. A part like a 3D V-Cache chip trades clock for an enormous L3 cache, which is why it can dominate gaming yet sit mid-pack in multi-thread workloads. Read the single- versus multi-thread indices together, never in isolation.

TDP (thermal design power) is a cooling-and-power budget, not a precise wattage meter — it tells you how much heat the cooler must dissipate and roughly how much the part draws under sustained load. It governs your PSU headroom, case airflow and, on laptops and small-form-factor builds, whether a part is even viable. Filter by max TDP when you are power- or thermally-constrained; pair that with the node facet and you can isolate the genuinely efficient silicon. For GPUs, memory bandwidth (GB/s, a product of VRAM bus width and memory clock) is the quiet determinant of high-resolution performance — a card can have ample VRAM capacity yet starve at 4K if its bandwidth is thin, so weigh capacity and bandwidth as a pair.

To run a query like “all 5nm GPUs under 250W released after 2022,” switch the type to GPUs, tick the 5nm node chip, drag the max-TDP slider to 250W and set the release-year-from slider to 2022 — the spec cards and the live summary update instantly. Every figure here is a documented launch specification from a curated reference set. Once you have shortlisted parts, move to the Hardware Benchmark Database for normalized performance ranking and value metrics, or the Silicon Architecture Explorer to see how the underlying designs differ.