The HypestKey FPS calculator estimates how many frames per second your PC gets in 24 popular games. Five picks: CPU, graphics card, resolution, game, graphics preset. We show average FPS, 1% low FPS, frame time and whether the CPU or the GPU holds you back.
A few of the 24:
- Cyberpunk 2077.
- Counter-Strike 2.
- Valorant.
- Call of Duty: Black Ops 6.
- Baldur’s Gate 3.
- Elden Ring.
- Red Dead Redemption 2.
Every estimate is built from published benchmark results for 49 graphics cards and 43 processors, taken from Tom’s Hardware and ComputerBase tests.
Use it to check whether a game will run smoothly before you buy a card. Or to compare two upgrade options. Or to see if your CPU keeps up with a new GPU.
Treat it as a planning tool. In most cases the result lands close to a real test, but it doesn’t replace a benchmark on your own PC.
How to use the FPS calculator
CPU and graphics card first. Then resolution, game and settings. Press Calculate FPS and read the right side: average FPS, 1% low, frame time and the part that holds you back. Nothing shows before that press.
- Pick your processor in the CPU list. Exact model missing? Take the closest one from the same generation. Older chips sit in the Derived score group, the one for older tests.
- Pick your graphics card in the GPU list. 8 GB and 16 GB versions of the same card are listed separately because they behave differently in heavy games.
- Choose the resolution you play at: 1080p, 1440p or 4K.
- Choose a game. The Settings list then shows the preset the game was tested with, marked “tested”, plus Low, Medium and High.
- Press Calculate FPS. Read the average FPS, the 1% low, the frame time and the Bottleneck field on the right.
- Look at the bars under the selects. They show the same setup at all three resolutions, and the CPU vs GPU bar shows how many FPS each part could reach on its own. The line under it names the data source.
- Press Copy result to save the numbers. Comparing upgrades? Change one part at a time and press Calculate FPS again, because the old result clears.
| Label in the Settings list | What it tells you |
|---|---|
| Tested | The preset the game was measured with on every card |
| Scaled | The factor comes from a preset test of that same game |
| Estimated | The factor comes from the average of several games |
How the FPS calculator works: where the numbers come from
Short version. We take a measured FPS number for a reference card in your game. We scale it by how fast your card is in the Tom’s Hardware GPU Benchmarks Hierarchy. Then we check it against the CPU limit from the Tom’s Hardware CPU Benchmarks Hierarchy. Your FPS is the lower of the two limits.
| Source | Covers | Date |
|---|---|---|
| Tom’s Hardware GPU Benchmarks Hierarchy | 48 cards, 11 raster and 8 ray tracing games | Updated June 2026 |
| Tom’s Hardware CPU Benchmarks Hierarchy | The CPU limit | October 2026 |
GPU limit. For each game, preset and resolution the tool stores the measured FPS of one reference card per brand. We multiply that number by how fast your card is against the reference at the same resolution. Ray tracing presets use the ray tracing table, not the raster one.
| Brand | Reference cards, for example |
|---|---|
| NVIDIA | RTX 5070 Ti or RTX 4070 |
| AMD | RX 9070 XT or RX 7800 XT |
| Intel | Arc B580, where Intel was tested |
Game data. Three groups of games, three sources.
| Games | Source | Note |
|---|---|---|
| 15 games | Tom’s Hardware hierarchy charts, June 2026 | Tested on a Ryzen 7 9800X3D |
| Baldur’s Gate 3, Diablo IV, Elden Ring, Hogwarts Legacy, Starfield | ComputerBase game tests, 2022 to 2023 | Older drivers and game versions |
| Red Dead Redemption 2 | Tom’s Hardware 2022 to 2024 test suite | Older drivers and game versions |
Older tests ran on older drivers and game versions. So those six games are usually a bit less precise.
Valorant, Escape from Tarkov and Call of Duty. These three have their own CPU and GPU tests. A card or CPU that was tested shows the published number. One that wasn’t is estimated from the tested parts of the same game, and the result says so.
| Game | Source | Test setup |
|---|---|---|
| Valorant | chimolog benchmarks: Valorant, December 2025 | Around 20 graphics cards on a Ryzen 7 9800X3D, and 12 to 14 of our CPUs on an RTX 5090 or RTX 4090 |
| Escape from Tarkov | chimolog benchmarks: Escape from Tarkov, February 2025 | The same setup |
| Call of Duty: Black Ops 6 | Hardware Unboxed and TechSpot GPU reviews of 2025 and their CPU comparisons | Ryzen 7 9800X3D, Extreme, native |
8 GB cards. Some games run out of video memory on 8 GB cards. Tom’s Hardware tests the RTX 5060 Ti and the RX 9060 XT in both 8 GB and 16 GB versions. So for the 2026 games we know how much each game loses on 8 GB. That loss goes onto every 8 GB card of the same brand.
CPU limit. For each game the tool stores how many FPS a Ryzen 7 9800X3D can feed. Where a CPU test exists, the number comes from that test. Otherwise it’s the fastest card’s 1080p result, a safe lower bound. Your processor’s limit is that number scaled by its gaming score. Elden Ring also has its own 60 FPS frame cap.
Games with a CPU test:
- Counter-Strike 2.
- Cyberpunk 2077.
- Doom: The Dark Ages.
- Spider-Man 2.
- Diablo IV.
- Starfield.
Presets. Only one preset per game was measured on every card. Low, Medium and High are calculated from preset tests. ComputerBase measured all presets for five of the games, and Tom’s Hardware measured Medium in Red Dead Redemption 2. For every other game we use the average of those tests and label it “estimated from preset scaling”.
The 1% low uses the same ratio between 1% low and average that the source chart shows for that game. All numbers are native resolution. No DLSS, FSR, XeSS or frame generation.
Every value traces back to a named chart in those articles. The method is the same for every combination, so you can compare parts fairly.
Games in the FPS calculator
24 right now. 15 from the 2026 Tom’s Hardware test suite. 6 older favourites from ComputerBase and Tom’s Hardware tests. And three with their own CPU and GPU tests: Valorant, Escape from Tarkov and Call of Duty: Black Ops 6.
Each game keeps the preset it was tested with. 20 of them also offer Low, Medium and High. The three newest add the presets shown in the table.
| Game | Tested preset | Source and date | Low / Medium / High |
|---|---|---|---|
| Alan Wake 2, Apex Legends, Arc Raiders, Black Myth: Wukong, Clair Obscur: Expedition 33, Cyberpunk 2077, Marvel’s Spider-Man 2, Stalker 2 | Ultra, Very High or Epic (RT presets where tested) | Tom’s Hardware, June 2026 | Estimated |
| Counter-Strike 2, Fortnite, Marvel Rivals | High | Tom’s Hardware, June 2026 | Estimated |
| Assassin’s Creed Shadows, Doom: The Dark Ages, Indiana Jones and the Great Circle | Top preset, ray tracing always on | Tom’s Hardware, June 2026 | Estimated |
| GTA V Enhanced | Maximum RT | Tom’s Hardware, June 2026 | Not offered |
| Baldur’s Gate 3 | Ultra | ComputerBase, August 2023 | Measured |
| Diablo IV | Ultra | ComputerBase, June 2023 | Measured |
| Elden Ring | Maximum (60 FPS cap) | ComputerBase, March 2022 | Measured |
| Hogwarts Legacy | Ultra, Ultra + RT | ComputerBase, February 2023 | Measured |
| Starfield | High | ComputerBase, September 2023 | Measured |
| Red Dead Redemption 2 | Max without MSAA | Tom’s Hardware 2022 to 2024 suite | Medium measured, Low and High estimated |
| Valorant | High + MSAA 4x | chimolog, December 2025 | Low and Medium estimated |
| Escape from Tarkov | Custom medium-high (Streets of Tarkov) | chimolog, February 2025 | Low estimated |
| Call of Duty: Black Ops 6 | Extreme | Hardware Unboxed / TechSpot, 2025 | Basic measured |
Valorant and Tarkov lean hard on the processor. At 1080p almost every modern card in the Valorant test sits at the CPU limit of about 600 FPS. A faster CPU adds more frames there than a faster card. An X3D chip most of all.
Black Ops 6 is the opposite at Extreme. The GPU decides, and Basic roughly doubles the frame rate.
Rust is still missing. We found no published test with several CPUs or graphics cards, only estimator sites. The tool never shows a number without a published source behind it.
Games that aren’t in the calculator yet
We also get asked about League of Legends, Minecraft, Warzone, Roblox and Star Citizen. None of them is in the list today.
The reason is the same as for Rust. We add a game only when a published test covers many cards and several processors.
- League of Legends, Minecraft and Roblox are light games. Most gaming PCs run them far above 60 FPS, and the processor sets the limit.
- Star Citizen is still in alpha. Its frame rate changes from patch to patch.
- Warzone shares its engine with Black Ops 6. That entry gives a rough idea, not a Warzone number.
Need one of them? Search for a benchmark of your exact graphics card in that game.
FPS of popular graphics cards at 1080p, 1440p and 4K
At 1080p an RTX 5090 averages about 204 FPS across the 11 raster games in the Tom’s Hardware suite. An RTX 5070, about 133. An RTX 5060, about 89. At 4K: about 111, 54 and 22.
The table is the raster average the calculator scales from.
| Graphics card | 1080p average FPS | 1440p average FPS | 4K average FPS |
|---|---|---|---|
| GeForce RTX 5090 | 203.8 | 167.3 | 110.8 |
| GeForce RTX 4090 | 183.6 | 143.4 | 89.1 |
| GeForce RTX 5080 | 166.9 | 128.3 | 77.4 |
| Radeon RX 9070 XT | 156.6 | 116.5 | 65.8 |
| GeForce RTX 5070 Ti | 155.4 | 116.8 | 68.6 |
| Radeon RX 9070 | 140.9 | 104.0 | 57.7 |
| GeForce RTX 5070 | 132.6 | 96.4 | 54.3 |
| GeForce RTX 4070 Super | 126.7 | 91.2 | 49.2 |
| Radeon RX 7800 XT | 118.4 | 84.7 | 45.1 |
| GeForce RTX 5060 Ti 16GB | 105.2 | 73.4 | 40.2 |
| Radeon RX 9060 XT 16GB | 98.3 | 67.3 | 35.2 |
| GeForce RTX 5060 | 88.5 | 59.8 | 21.8 |
| Intel Arc B580 | 71.5 | 50.7 | 27.6 |
| GeForce RTX 4060 | 71.5 | 47.6 | 17.4 |
| GeForce RTX 3060 12GB | 61.5 | 41.9 | 22.2 |
Source: Tom’s Hardware GPU Benchmarks Hierarchy 2026, geometric mean of 11 raster games, Ryzen 7 9800X3D test PC.
Single games differ a lot from these averages. Counter-Strike 2 runs past 500 FPS on fast cards. Black Myth: Wukong at 4K stays under 70 FPS even on an RTX 5090.
How much FPS Low and Medium settings add
On average, dropping from Ultra to Medium gives roughly 40% more FPS. Low gives about 55% more. Both numbers rest on six published preset tests.
The spread is wide. In Baldur’s Gate 3, High is only 3% faster than Ultra. In Red Dead Redemption 2, Medium almost doubles the frame rate.
| Game (preset test) | Low vs Ultra | Medium vs Ultra | High vs Ultra |
|---|---|---|---|
| Baldur’s Gate 3 (ComputerBase) | +61% | +25% | +3% |
| Diablo IV (ComputerBase) | +46% | +25% | +11% |
| Elden Ring (ComputerBase, vs Maximum) | +57% | +28% | +12% |
| Hogwarts Legacy (ComputerBase) | +73% | +56% | +23% |
| Starfield (ComputerBase) | +43% | +29% | +15% |
| Red Dead Redemption 2 (Tom’s Hardware, 1080p) | not tested | +99% | not tested |
| Average used for “estimated” presets | +56% | +41% | +13% |
These factors apply to the GPU limit only. If your CPU already caps the frame rate, a lower preset usually adds much less. The processor still has to run the same game logic. So we keep the CPU limit unchanged when you switch presets.
What is an FPS calculator?
An FPS calculator estimates the frame rate of a PC in one game. You pick the processor, the graphics card, the resolution and the preset. It answers with frames per second.
People use it for three things:
- Checking a game before buying it.
- Comparing two graphics cards or processors before an upgrade.
- Finding the part that holds the frame rate back.
It differs from an FPS test. A test measures the PC you own. A calculator predicts from published benchmarks, so it also works for parts you haven’t bought yet.
Some sites use the same name for a frame time converter. That one turns FPS into milliseconds per frame. The formula sits in the frame time section below.
How accurate is an FPS calculator?
We checked the tool against 1,004 published per-game results that it didn’t use as a reference. The median error is 3.6%, and about 91% of results land within 15%.
| Check | Result |
|---|---|
| 1,004 published per-game results | Median error 3.6%; about 91% within 15% |
| 48 CPU results from the Core Ultra 7 270K Plus review | Median error 3.1% |
| Our 17 hand-picked check cases | All within 15% |
Those checks use the same publications. So they show how well the model reproduces real tests, not how your own PC will behave.
The biggest misses:
- 8 GB cards at 4K.
- Ray tracing on entry-level cards.
- The older games, where drivers and patches have changed a lot since the test.
Estimated presets can’t be checked at all, so we’d treat them as a rough guide. Another scene in the same game? Easily 10% more or less.
| Check case | Calculator | Published test | Error |
|---|---|---|---|
| Ryzen 7 9800X3D + RTX 5080, Cyberpunk 2077 Ultra + RT, 1440p | 60 | 60.3 (Tom’s Hardware) | -0.5% |
| Ryzen 7 7800X3D + RTX 5090, Cyberpunk 2077 Ultra, 1080p | 128 | 122.5 (Tom’s Hardware) | +4.5% |
| Ryzen 9 7950X3D + RTX 3080, Baldur’s Gate 3 Ultra, 1440p | 92 | 93.7 (ComputerBase) | -1.8% |
| Ryzen 7 9800X3D + RX 7700 XT, Red Dead Redemption 2 Max, 1440p | 79 | 77.1 (Tom’s Hardware) | +2.5% |
| Ryzen 5 3600 + RTX 4090, Diablo IV Ultra (CPU test) | 196 | 172.9 (ComputerBase) | +13.4% |
How much FPS do you need?
For single-player games 60 FPS usually feels smooth. That’s 16.7 ms per frame. Strategy games often stay playable at 30 to 40 FPS.
Competitive shooters are different. For Counter-Strike 2, Valorant or Marvel Rivals we’d aim for at least 144 FPS on a 144 Hz monitor. 240 FPS or more if you own a 240 Hz panel.
In shooters, more frames bring less delay between your mouse and the screen. You can feel it when tracking a target. Want to see how fast you react? Try our reaction time test.
In story games like Cyberpunk 2077 or Hogwarts Legacy no gunfight is decided by milliseconds. Spend the GPU power on higher settings instead.
| FPS | Frame time | How it feels | Good for |
|---|---|---|---|
| 30 | 33.3 ms | Playable, visible judder | Strategy, city builders |
| 60 | 16.7 ms | Smooth | Single-player games |
| 144 | 6.9 ms | Very smooth, low delay | Online shooters, 144 Hz monitors |
| 240+ | 4.2 ms or less | Lowest delay | Competitive play on 240 Hz monitors |
FPS to frame time: the formula
Frame time in milliseconds = 1000 ÷ FPS. The other way round, FPS = 1000 ÷ frame time. Our calculator shows both numbers for every result.
| FPS | Frame time |
|---|---|
| 30 | 33.33 ms |
| 60 | 16.67 ms |
| 90 | 11.11 ms |
| 120 | 8.33 ms |
| 144 | 6.94 ms |
| 165 | 6.06 ms |
| 240 | 4.17 ms |
| 360 | 2.78 ms |
A steady frame time matters as much as a short one. 60 FPS with even 16.7 ms frames feels smoother than 80 FPS with spikes.
Not sure what your screen can show? The monitor refresh rate test measures its real Hz in the browser. Then you know which FPS target is worth chasing.
What does 1% low FPS mean?
The frame rate of the slowest 1% of frames. A game averages 90 FPS but the 1% low is 45? You’ll feel stutter in heavy moments. The counter still looks fine. A 1% low above 60 FPS usually goes with smooth play in single-player games.
Drops happen when many characters, particles and physics objects appear at once. Big fights, crowded city scenes. Both the CPU and the GPU work harder then.
1% low far below the average? We’d lower crowd density, shadows or particle settings first. Those tend to cost the most in busy scenes.
Does the CPU or the GPU matter more for FPS?
At 4K the graphics card decides almost everything. At 1080p the processor matters much more.
In the Tom’s Hardware data an RTX 5090 in Cyberpunk 2077 reaches 157 FPS at 1080p but only 62 FPS at 4K. At 1080p that’s near the ceiling of a Ryzen 7 9800X3D.
That’s why pairing a top graphics card with an old processor wastes money at 1080p.
| CPU with the same RTX 5090, Cyberpunk 2077 at 1080p | FPS in the CPU test data |
|---|---|
| Ryzen 7 9800X3D | About 143 |
| Core Ultra 5 245K | About 101 |
At 1440p the two limits are often close, so both parts matter. For a closer look at your pair, open the bottleneck calculator.
Checking for a bottleneck on your own PC
Run a game and open an overlay such as MSI Afterburner, or Task Manager on a second screen. Look at GPU usage and CPU usage.
| You see | Verdict |
|---|---|
| GPU usage under 90% while one or more CPU threads sit near 100% | A CPU bottleneck: the graphics card waits for the processor |
| GPU usage at 97 to 99% | The GPU is the limit. That is normal: the card does all it can |
Every PC has a bottleneck somewhere. Only worry about it when you don’t reach your FPS target. Before a GPU upgrade, check the power supply too with the PSU calculator.
How much FPS you lose at 1440p and 4K
| Resolution | Pixels per frame | RTX 5070 Ti in the Tom’s Hardware averages |
|---|---|---|
| 1080p | About 2.1 million | 155 FPS |
| 1440p | About 3.7 million | 117 FPS, minus 25% |
| 4K | About 8.3 million, four times as many as 1080p | 69 FPS, minus 56% |
The loss is smaller when the CPU limits FPS at 1080p, because the card had spare power there. Bigger on 8 GB cards. They can run out of video memory at 4K. The RTX 5060: 88.5 FPS at 1080p, only 21.8 at 4K.
The bars show all three resolutions, so you see the drop for your exact setup.
Do DLSS, FSR and frame generation change FPS?
They change the numbers, yes. And we leave them out on purpose: every number here is native resolution.
DLSS and FSR render the game at a lower resolution and upscale it. Quality mode on a 4K monitor? 2560 x 1440 internally. FPS lands near the 1440p result. The image stays close to 4K.
Frame generation adds generated frames between real ones. It raises the FPS counter but not the responsiveness, because input is still read on the real frames. The gain depends on the game, the mode and the card, so we don’t add a fixed bonus.
Playing at 4K with upscaling? After a 4K result the tool adds a line for it. Quality mode renders at 1440p and Performance mode at 1080p, so we show those two tested numbers. The upscaler costs a few percent on top, so expect slightly less.
DLSS needs an RTX card. FSR 3 also runs on AMD, NVIDIA and Intel cards.
Does ray tracing lower FPS?
It costs a lot. Two examples from the Tom’s Hardware charts, Cyberpunk 2077 at 1440p:
| Card | Without ray tracing | With Ultra ray tracing |
|---|---|---|
| RTX 5070 Ti | 81 fps | 54 fps |
| RX 9070 XT | 82 fps | 45 fps |
DLSS or FSR can win much of that back. Pick the RT preset in the Settings list to see the native numbers.
Does RAM speed change FPS?
It changes FPS a little. The test systems behind these numbers use 32 GB of DDR5, so the estimates assume fast, dual-channel memory. For the 2026 GPU tests that was DDR5-6000 CL30.
Slow or single-channel memory lowers the CPU limit. It shows most at 1080p with a fast graphics card.
For new gaming builds 32 GB is the safe amount in 2026. 16 GB still runs most games but leaves little room for a browser and voice chat in the background. The RAM calculator tells you how much your tasks need.
FPS calculator for gaming laptops
Only as an upper bound. Expect lower numbers than the tool shows. A laptop RTX 4070 or RTX 5070 isn’t the desktop card of the same name. Different chip: smaller and power-limited. The results also depend on the laptop’s power limit and cooling.
All data here comes from desktop cards and desktop processors. Checking a laptop? Pick the desktop equivalent and expect less. Then look for a review of that exact laptop model.
Planning a desktop instead? Put the parts you picked here into the PC price calculator to see the budget.
Why the same hardware gets different FPS
Your friend runs the same parts and gets other numbers? Normal. Lots of things move it.
- Driver version.
- Windows settings.
- RAM speed and channels.
- Cooling and case airflow.
- Power limits.
- Game config files.
Even the test scene matters. Two Tom’s Hardware runs of Cyberpunk 2077 with the same Ryzen 7 9800X3D and RTX 5090 differ by about 10%. 157 against 143 fps.
FPS test: how to check how many FPS you get
The calculator predicts. An FPS test on your own PC measures. We’d do both: estimate before you buy, then check after the parts are in.
The quickest counter is already in Windows. Xbox Game Bar has had an FPS counter in its Performance overlay since its October 2019 update, as Xbox Wire announced.
- Start the game. Press Win + G.
- Open the Performance widget and pick the FPS tab.
- Windows may ask for access to performance data the first time. Allow it, then restart the PC.
- Pin the widget, so it stays on screen while you play.
Other ways to run an FPS test:
- The game’s own benchmark run, if it has one. Same scene every time, so results compare well.
- The performance overlay in your graphics driver app.
Comparing your result with ours? Same game, same resolution, same preset. Play a minute or two in one busy spot and read the average, not the peak. Upscaling off too, since every number in the calculator is native.
Your FPS test lands far below the estimate? Check the Bottleneck field first, then the list in the section above: drivers, power limits, RAM in one channel. On a laptop, read the laptop section above too.
FPS terms
| Word | Short explanation |
|---|---|
| Average FPS | Frames per second over the whole test run |
| 1% low | The frame rate of the slowest 1% of frames |
| Frame time | How long one frame takes, in ms: 16.7 ms at 60 FPS |
| GPU limit | The FPS your graphics card could reach on its own |
| CPU limit | The FPS your processor can feed |
| Native | Rendered at the full resolution, without DLSS, FSR or XeSS |
| Derived score | A score built from older Tom’s Hardware tables |
| Raster | Rendering without ray tracing |