Hardware Benchmark Database Pro
A searchable, sortable engineering database of CPUs and GPUs — specs, a normalized performance index, perf-per-dollar and perf-per-watt, tiers, process nodes and generations. Filter, rank and drill into any part.
| Part | Type | Arch / Node | Perf | Tier | TDP | MSRP | Perf/$ |
|---|---|---|---|---|---|---|---|
| GeForce RTX 5090 NVIDIA · 2025 | GPU | Blackwell 4nm | 100 | S | 575W | $1,999 | 5.0 |
| Ryzen 9 9950X AMD · 2024 | CPU | Zen 5 4nm | 99 | S | 170W | $649 | 15.3 |
| Core i9-14900K Intel · 2023 | CPU | Raptor Lake Refresh 10nm | 97 | S | 125W | $589 | 16.5 |
| Core Ultra 9 285K Intel · 2024 | CPU | Arrow Lake 3nm | 95 | S | 125W | $589 | 16.1 |
| Ryzen 9 7950X AMD · 2022 | CPU | Zen 4 5nm | 95 | S | 170W | $699 | 13.6 |
| Core i9-13900K Intel · 2022 | CPU | Raptor Lake 10nm | 93 | S | 125W | $589 | 15.8 |
| Apple M3 Max Apple · 2023 | CPU | M3 (3nm) 3nm | 92 | S | 78W | N/A | — |
| GeForce RTX 4090 NVIDIA · 2022 | GPU | Ada Lovelace 4nm | 91 | S | 450W | $1,599 | 5.7 |
| Ryzen 9 7900X AMD · 2022 | CPU | Zen 4 5nm | 88 | A | 170W | $549 | 16.0 |
| Core i7-14700K Intel · 2023 | CPU | Raptor Lake Refresh 10nm | 87 | A | 125W | $409 | 21.3 |
| Ryzen 9 5950X AMD · 2020 | CPU | Zen 3 7nm | 85 | A | 105W | $799 | 10.6 |
| GeForce RTX 5080 NVIDIA · 2025 | GPU | Blackwell 4nm | 85 | A | 360W | $999 | 8.5 |
| Ryzen 7 7800X3D AMD · 2023 | CPU | Zen 4 (3D V-Cache) 5nm | 82 | A | 120W | $449 | 18.3 |
| Core i9-12900K Intel · 2021 | CPU | Alder Lake 10nm | 81 | A | 125W | $589 | 13.8 |
| Radeon RX 7900 XTX AMD · 2022 | GPU | RDNA 3 5nm | 80 | A | 355W | $999 | 8.0 |
| GeForce RTX 4080 Super NVIDIA · 2024 | GPU | Ada Lovelace 4nm | 80 | A | 320W | $999 | 8.0 |
| Core i5-14600K Intel · 2023 | CPU | Raptor Lake Refresh 10nm | 79 | A | 125W | $319 | 24.8 |
| GeForce RTX 4080 NVIDIA · 2022 | GPU | Ada Lovelace 4nm | 78 | A | 320W | $1,199 | 6.5 |
| Core i5-13600K Intel · 2022 | CPU | Raptor Lake 10nm | 77 | A | 125W | $319 | 24.1 |
| Radeon RX 9070 XT AMD · 2025 | GPU | RDNA 4 4nm | 74 | B | 304W | $599 | 12.4 |
| GeForce RTX 3090 Ti NVIDIA · 2022 | GPU | Ampere 8nm | 74 | B | 450W | $1,999 | 3.7 |
| Radeon RX 7900 XT AMD · 2022 | GPU | RDNA 3 5nm | 73 | B | 315W | $899 | 8.1 |
| GeForce RTX 4070 Ti Super NVIDIA · 2024 | GPU | Ada Lovelace 4nm | 73 | B | 285W | $799 | 9.1 |
| Ryzen 7 5800X3D AMD · 2022 | CPU | Zen 3 (3D V-Cache) 7nm | 72 | B | 105W | $449 | 16.0 |
| Ryzen 5 7600X AMD · 2022 | CPU | Zen 4 5nm | 71 | B | 105W | $299 | 23.7 |
| Apple M2 Apple · 2022 | CPU | M2 (5nm) 5nm | 67 | B | 20W | N/A | — |
| GeForce RTX 3080 NVIDIA · 2020 | GPU | Ampere 8nm | 65 | B | 320W | $699 | 9.3 |
| Ryzen 5 5600X AMD · 2020 | CPU | Zen 3 7nm | 63 | B | 65W | $299 | 21.1 |
| GeForce RTX 4070 NVIDIA · 2023 | GPU | Ada Lovelace 4nm | 63 | B | 200W | $599 | 10.5 |
| Radeon RX 7800 XT AMD · 2023 | GPU | RDNA 3 5nm | 62 | B | 263W | $499 | 12.4 |
| Radeon RX 6800 XT AMD · 2020 | GPU | RDNA 2 7nm | 60 | B | 300W | $649 | 9.2 |
| GeForce RTX 3070 NVIDIA · 2020 | GPU | Ampere 8nm | 56 | C | 220W | $499 | 11.2 |
| Radeon RX 7700 XT AMD · 2023 | GPU | RDNA 3 5nm | 55 | C | 245W | $449 | 12.2 |
| GeForce RTX 4060 Ti NVIDIA · 2023 | GPU | Ada Lovelace 4nm | 51 | C | 160W | $399 | 12.8 |
| Intel Arc B580 Intel · 2024 | GPU | Battlemage 5nm | 49 | C | 190W | $249 | 19.7 |
| Radeon RX 6700 XT AMD · 2021 | GPU | RDNA 2 7nm | 48 | C | 230W | $479 | 10.0 |
| Intel Arc A770 Intel · 2022 | GPU | Alchemist 6nm | 46 | C | 225W | $329 | 14.0 |
| GeForce RTX 4060 NVIDIA · 2023 | GPU | Ada Lovelace 4nm | 44 | C | 115W | $299 | 14.7 |
| GeForce RTX 3060 NVIDIA · 2021 | GPU | Ampere 8nm | 41 | C | 170W | $329 | 12.5 |
| Radeon RX 7600 AMD · 2023 | GPU | RDNA 3 6nm | 41 | C | 165W | $269 | 15.2 |
How to read the database
Raw benchmark numbers are noisy and vendor-specific, so this database normalizes performance to a single 0–100 relative index per part — tuned to reflect well-established relative ordering across generations rather than any one suite's score. It lets you line a 2020 flagship up against a 2025 mid-ranger and see, at a glance, where each really lands. CPUs additionally carry single- and multi-thread indices (a gaming chip like the 7800X3D ranks high single-thread but lower multi-thread), and GPUs split gaming versus compute/AI.
The two columns that change buying decisions are derived, not raw: perf-per-dollar (index per $100 of launch MSRP) and perf-per-watt (index per 100W of TDP). The fastest part is rarely the smartest purchase — sort by perf-per-dollar and the mid-range almost always wins; sort by perf-per-watt and efficiency-first designs (Apple silicon, the newest nodes) jump to the top. Tiers (S→D) bucket the index into Flagship / Enthusiast / High-end / Mainstream / Entry so you can shop by class.
Process node and release year tell the architectural story: smaller nodes (5nm → 4nm → 3nm) generally bring better perf-per-watt, and pairing them with the index reveals generational efficiency gains. Click any row to open the full spec sheet — cores, clocks, cache, memory subsystem and every derived metric.
This is a curated reference dataset (launch specs and MSRP), not a live price feed — use it for relative comparison, upgrade planning and class shopping. Then jump to the CPU/GPU Compare tool for head-to-head detail or the Tier List for the full ranking.
Trusted by Builders, Reviewers & Engineers
“The single normalized index plus perf-per-dollar in one sortable table is exactly how I shortlist parts for build guides. Sorting GPUs by perf-per-dollar instantly shows which mid-range card to recommend instead of eyeballing five review sites.”
“Perf-per-watt sorting is what I came for — for our edge deployments TDP matters more than peak, and seeing efficiency normalized next to the node makes the trade-offs obvious. Clean, fast, and it runs entirely client-side.”
“Love the tier buckets and the per-part spec sheet. I'd like even more obscure parts in the list, but the relative ordering of what's here matches my own benchmarking well. The detail panel's cache and bandwidth fields are a nice touch.”
“We use the perf-per-dollar and tier columns to justify standardized build specs to finance. Being able to filter by brand and sort by value turns a messy debate into a one-screen decision. The N/A handling for integrated parts is thoughtful.”
Love using our calculator?
More hardware tools
Related Articles
Dive deeper with our expert guides and tutorials related to Hardware Benchmark Database Pro
curated reference dataset · normalized index · in-browser · Last reviewed: 2026-06