Real FPS database from players
Not a prediction and not a review: real FPS measured on players’ computers — from the FPSHQ app and moderator-checked reports. Every number below is computed from the database right now.
How the database grows
- FPSHQ app (auto-measured)
- Site reports (moderated)
Games with the most results
GPUs with the most results
FPSHQ model vs reality
Each dot is a game × GPU × resolution × preset combo (no upscaling): horizontal axis is the FPSHQ model prediction, vertical axis is the median of real results. Dots on the diagonal mean the model got it right.
How contributors play
Resolution
- 720p 3%
- 1080p 70%
- 1440p 18%
- 4K 9%
Graphics preset
- Low 17%
- Medium 17%
- High 17%
- Ultra 47%
- Path Tracing 3%
Mode
- Native 70%
- Upscaling 30%
CPU class
- Entry 9%
- Mid-range 42%
- High-end 26%
- Top 23%
Latest database updates
| Game | GPU | Settings | Median FPS | 1% low | Results |
|---|---|---|---|---|---|
| Serious Sam 4 | GeForce RTX 5060 Ti | 1440p · Ultra | 100 | — | 1 |
| Arma 3 | GeForce GTX 1060 3GB | 1080p · High | 75 | — | 1 |
| Mount & Blade II: Bannerlord | GeForce GTX 1050 | 1080p · Low | 60 | — | 1 |
| The Forest | Palit GameRock GeForce RTX 5060 | 1440p · High | 94 | — | 1 |
| CPM 2 | Radeon Vega 3 | 1080p · Low | 60 | — | 1 |
| OMSI 2: Steam Edition | Radeon RX 580 | 1080p · High | 20 | — | 1 |
| Spark in the Dark | MSI Ventus 2X GeForce GTX 1050 Ti | 1080p · Medium | 60 | — | 1 |
| Raft | GeForce GTX 760 | 720p · Medium Upscaling | 60 | — | 1 |
| Gunner, HEAT, PC! | Gigabyte Windforce GeForce RTX 2060 | 1440p · High | 55 | — | 1 |
| PEAK | ASUS ROG Strix GeForce RTX 3070 | 4K · Path Tracing | 368 | — | 1 |
| PAWSHOT | ASUS ROG Strix GeForce RTX 4080 Super | 1080p · Ultra Upscaling | 75 | — | 1 |
| The Long Drive | GeForce GTX 750 Ti | 720p · Low | 100 | — | 1 |
Anti-fraud: what gets filtered
Every app result goes through automatic checks. Only results that pass every filter reach public statistics; site reports are also reviewed by a moderator.
No app results yet — the filters are on and waiting for the first data.
Methodology
What is measured. The FPSHQ app for Windows records every frame time (like PresentMon) and computes average FPS, 1% and 0.1% lows, the 99th-percentile frame time, plus load and temperatures. Only the summary and a per-second FPS series are sent — no screenshots or personal data.
Checks. We drop results shorter than 30 seconds, with FPS outside the physical range for that GPU and game, with contradictory metrics (frame count, frame time and FPS series disagree with the average), with 1% low above average, duplicates, and runs capped by V-Sync or a limiter. There are per-device and per-IP limits, and every device has a trust level.
Outliers. Within each game × GPU × resolution × preset combo we compute the median and MAD (median absolute deviation) of log FPS. A result more than 3σ away (and at least ±42%) is flagged as an outlier and excluded; the check is repeated as new data arrives.
Aggregates. We publish median FPS, quartiles and median 1% low separately for native rendering, upscaling and frame generation, and by CPU class. A combo becomes public once it has results from at least two different PCs or a moderator-checked report. Individual results and device data are never exposed.
Model calibration. FPSHQ predictions for combos without measurements come from a model. Player medians feed the model only with a sufficient sample: at least 5 results from 3 different PCs while GPU-bound (GPU load ≥ 85%), with moderate spread and no more than 2.5× difference from the model. Lab benchmarks from reviews take priority; every change is logged and can be rolled back.
Privacy. The app sends results only with explicit consent, and you can turn it off in settings. The install ID is stored on the server only as an irreversible hash, used solely for limits and trust.
Coverage map: where real measurements are still missing
13% of “game × GPU class” cells for the 30 most popular games have at least one independent measurement; 104 cells are empty. Tap an empty cell to open a page with the “Be the first” form.
| Game | EntryRTX 3060, RTX 3050 | MidRTX 5060, RTX 5060 Ti | Upper-midRTX 5070, RTX 4070 | High-endRTX 5080, RTX 5070 Ti |
|---|---|---|---|---|
| Cyberpunk 2077 | 3+ | 42 | 93 | 52 |
| Counter-Strike 2 | 75 | 44 | 11 | 21 |
| Dota 2 | 11 | 11 | ++ | 11 |
| Grand Theft Auto V | 11 | 22 | ++ | 11 |
| League of Legends | ++ | ++ | ++ | ++ |
| Binding of Isaac: Rebirth | ++ | ++ | ++ | ++ |
| Hearthstone | ++ | ++ | ++ | ++ |
| Ultimate Chicken Horse | ++ | ++ | ++ | ++ |
| Hades-like Pizza Tower | ++ | ++ | ++ | ++ |
| Stumble Guys | ++ | ++ | ++ | ++ |
| Elden Ring | ++ | ++ | ++ | ++ |
| Human: Fall Flat | ++ | ++ | ++ | ++ |
| Noita | ++ | ++ | ++ | ++ |
| Gang Beasts | ++ | ++ | ++ | ++ |
| Diablo III | ++ | ++ | ++ | ++ |
| Old School RuneScape | ++ | ++ | ++ | ++ |
| Heroes of the Storm | ++ | ++ | ++ | ++ |
| Max Payne 3 | 11 | ++ | ++ | ++ |
| RuneScape | ++ | ++ | ++ | ++ |
| Genshin Impact | ++ | 11 | ++ | ++ |
| Rainbow Six Siege | ++ | ++ | ++ | ++ |
| Albion Online | ++ | ++ | ++ | ++ |
| Elden Ring: Shadow of the Erdtree | ++ | ++ | ++ | ++ |
| Titanfall 2 | ++ | ++ | ++ | ++ |
| Plants vs. Zombies: Battle for Neighborville | ++ | ++ | ++ | ++ |
| Age of Empires II: Definitive Edition | ++ | ++ | ++ | ++ |
| Against the Storm | ++ | ++ | ++ | ++ |
| PowerWash Simulator | ++ | ++ | ++ | ++ |
| Sleeping Dogs | ++ | ++ | ++ | ++ |
| Diablo II: Resurrected | ++ | ++ | ++ | ++ |
- none (+)
- 1–2
- 3+
Where a measurement is needed most
- League of Legends — GeForce RTX 3060+
- Binding of Isaac: Rebirth — GeForce RTX 3060+
- Hearthstone — GeForce RTX 3060+
- Ultimate Chicken Horse — GeForce RTX 3060+
- Hades-like Pizza Tower — GeForce RTX 3060+
- Stumble Guys — GeForce RTX 3060+
- Elden Ring — GeForce RTX 3060+
- Human: Fall Flat — GeForce RTX 3060+
Each cell counts independent measurements across all desktop cards of the class: review tests without aggregated tables and verified player results (app and website), outliers excluded. Classes follow the FPSHQ catalog Bench Score, as on the model accuracy page: entry < 3500, mid < 5000, upper-mid < 7000, high-end ≥ 7000. Games: the 30 most popular released titles (since 2012). Snapshot: 2026-10-14 15:30.
How to contribute
Install FPSHQ for Windows: after a gaming session the app offers to submit the result — anonymously and only with your consent.
Download FPSHQOr do it manually: open a game page and add your result in the “Share your FPS” block — a moderator will check it.
Submit my resultSubmit an FPS measurement
Submit an FPS measurement
Play for 60 seconds in a typical scene, read the average FPS and 1% low from an overlay (FPSHQ app, MSI Afterburner/RTSS or the game benchmark) and submit it — anonymously.
Submit my resultHow to measure: 60 seconds of normal gameplay (not menus or loading), no frame cap or vsync; read average FPS and 1% low from an overlay.
Only aggregates are published: a median over 2+ devices or after review. No name or IP is stored — only a hash to stop abuse. Methodology
How is the player FPS database different from review benchmarks?
Reviews test on a bench with a top CPU and a clean system. The database holds ordinary players’ PCs with their own drivers, background apps and CPUs, so the numbers are closer to what you will see. It currently holds 174 results across 98 games and 137 GPUs.
Frequently asked questions
How is the player FPS database different from review benchmarks?
Can results be faked?
How do I add my result?
How do I submit an FPS measurement without the app?
What do I get for my measurement?
How does FPSHQ filter out abuse and typos?
Key figures
Unique FPSHQ metrics from this page's data
FPSHQ data volume by section
Records in the database by section: games, FPS measurements, GPUs, mobile chips and games, CPUs, emulators.
What is FPSHQ and what data does it run on?
FPSHQ is a games and PC database: 155,728 games with system requirements, 1,965 GPUs and 456 CPUs with performance scores, 165 mobile chips, 49 emulators and an FPS database (measurements: 106 independent measurements and 4,800 rows of aggregated review data; model estimates: 47,950). The single FPSHQ model estimates FPS for any game + GPU + CPU combination at 1080p/1440p/4K; 1,363 users have completed the online PC test. Every number on the pages comes from this database, with an update date and a link to the methodology.
- games in the database
- 155,728
- GPUs in the catalogue (incl. vendor variants)
- 1,965
- CPUs in the catalogue
- 456
- FPS measurement rows in the database
- 52,856
Methodology and sources
This page uses the FPSHQ database: the GPU and CPU catalogue with synthetic benchmark scores, official game system requirements, FPS measurements from users of the FPSHQ app and the FPSHQ performance model. Calculated values are estimates, not a measurement of your exact configuration: the real result depends on drivers, settings and the rest of the hardware. More about the methodology →