Audited ·Last updated 28 Jul 2026·4 citations·Tier 3·0 uses

Refresh Rate to Frame Time Calculator

Convert Hz to milliseconds per frame and back, check how much of the frame budget your renderer uses, and see the frame rate hard V-Sync actually locks you to.

Refresh Rate to Frame Time Calculator

What do you want to work out?
Used by the first mode. 60 and 75 for office panels, 120, 144, 165, 240 or 360 for gaming monitors, 90 or 120 for headsets.
Hz
Rate basis
Used by the second mode. The time from one vertical blank to the next: 16.67 ms is 60 Hz, 6.94 ms is 144 Hz.
ms
Used by both modes. How long the CPU and GPU actually take to produce one frame. Enter the 99th-percentile frame time rather than the average if you are chasing stutter — the worst frames are the ones you see.
ms
Result
A 144 Hz display refreshes every 6.94 ms (6944 µs). A frame that takes 8 ms uses 115.2% of that budget — 1.06 ms over budget. With hard V-Sync each frame is held for 2 refresh intervals, delivering 72 fps against the 125 fps the renderer could sustain uncapped.
The refresh interval, how much of it your renderer consumes, and the frame rate hard V-Sync delivers as a result.
Frame time
6.9444 ms
Refresh rate
144 Hz
V-Sync locked frame rate
72 fps
Frame budget used
115.20
Headroom
-1.0556 ms
Uncapped frame rate
125 fps
Refresh intervals per frame
2
Frame time
6,944.4444 µs
Frames per minute
8,640

Background.

A refresh rate and a frame time are the same fact stated twice. The hertz is an SI derived unit whose expression in base units is s⁻¹, so a panel running at 144 Hz simply has a period of 1 ÷ 144 seconds — 6.944 milliseconds — between one vertical blank and the next. That millisecond figure is the more useful of the two, because it is a budget: every frame your machine draws has to be finished inside it, and profilers report the cost of shadow maps, post-processing and physics in milliseconds, not in frames per second.

The reason frame time beats frame rate as a working unit is that frames per second is a reciprocal, and reciprocals compress at the top of the scale. Going from 30 to 60 fps buys 16.7 ms per frame; going from 120 to 144 fps buys only 1.4 ms. A change of "30 fps" means something completely different at each end of the range, which is why optimisation targets are set in milliseconds.

The third output on this page is the one people are usually surprised by. With hard V-Sync, a finished frame cannot be shown the moment it is ready — it has to wait for a vertical blanking boundary. Microsoft's DXGI documentation states the rule plainly: the SyncInterval parameter of Present is an integer, where 1 through 4 means synchronise presentation after the nth vertical blank. Because that count is a whole number, a renderer that misses its window by a fraction of a millisecond does not lose a fraction of its frame rate — it loses half of it. On the default inputs here a renderer taking 8 ms on a 144 Hz panel is only 1.06 ms over budget and could sustain 125 fps unsynchronised, yet V-Sync holds every frame for two intervals and delivers exactly 72.

The rate-basis option handles the other thing that catches people out: broadcast rates are not round. Recommendation ITU-R BT.709-6 lists picture rates of 60, 60/1.001, 30, 30/1.001, 50, 25, 24 and 24/1.001, and ITU-R BT.2020-2 carries the same 1.001 divisor into 120 Hz and 100 Hz UHD. Selecting the NTSC basis applies the exact divisor rather than the rounded 59.94 that most tools show, so 60 Hz becomes 59.940059940… Hz and a 16.6667 ms interval becomes 16.6833 ms. Over an hour that 0.0167 ms per frame is the entire reason drop-frame timecode exists.

What is refresh rate to frame time calculator?

Refresh rate is how many times per second a display redraws its image, measured in hertz. Frame time — also called the refresh interval or the frame budget — is the elapsed time between two consecutive refreshes, and it is simply the reciprocal: 1000 divided by the refresh rate gives milliseconds per frame. 60 Hz is 16.67 ms, 120 Hz is 8.33 ms, 144 Hz is 6.94 ms and 240 Hz is 4.17 ms.

The distinction between the display's frame time and the renderer's frame time matters. The first is fixed by the panel and is a deadline; the second is measured and is a cost. When the cost exceeds the deadline the frame is late, and what happens next depends on how presentation is synchronised: with tearing allowed the partial frame is shown immediately, and with hard V-Sync the frame waits for the next blanking boundary, which quantises the delivered frame rate into a staircase of the refresh rate divided by 1, 2, 3 and so on.

How to use this calculator.

  1. Choose whether you are starting from the refresh rate in hertz or from a measured interval in milliseconds.
  2. Enter the panel's refresh rate — 60, 120, 144, 240 — or the interval you measured, whichever matches your mode.
  3. Leave the rate basis on "exactly as entered" for a computer display; switch to the 1.001 divisor for broadcast and film rates, where 60 means 59.940 and 24 means 23.976.
  4. Enter your renderer's actual frame time in milliseconds, taken from a profiler or a frame-time overlay. Use the 99th-percentile figure rather than the average if you are investigating stutter.
  5. Read the frame budget used and the headroom to see how close you are to the deadline, and the V-Sync locked frame rate to see what the staircase costs you when you miss it.

The formula.

T = 1000 ÷ f; f = 1000 ÷ T; n = ⌈t ÷ T⌉ (min 1); fps_vsync = f ÷ n

The hertz is defined in SI as a reciprocal second, so period and frequency are exact reciprocals of one another: T = 1 ÷ f seconds, or T = 1000 ÷ f milliseconds. At 144 Hz that is 1000 ÷ 144 = 6.9444… ms, and inverting it returns 144 Hz exactly. The microsecond figure is the same number multiplied by 1000, and frames per minute is the rate multiplied by 60.

The broadcast basis divides the entered rate by exactly 1.001 before anything else happens. This is not an approximation of 59.94 — it is the value the standards themselves specify. ITU-R BT.709-6 item 5.5 lists picture rates as 60, 60/1.001, 30, 30/1.001, 50, 25, 24, 24/1.001, and ITU-R BT.2020-2 Table 2 lists 120, 120/1.001, 100, 60, 60/1.001, 50, 30, 30/1.001, 25, 24, 24/1.001. Sixty divided by 1.001 is 59.940059940059940… Hz, whose interval is 1000 × 1.001 ÷ 60 = 16.68333… ms.

The budget figures are straightforward subtraction and division. Headroom is the interval minus the render time and is deliberately allowed to go negative, because a negative headroom is the informative case: it is the number of milliseconds that must be found. Budget used is the render time divided by the interval, expressed as a percentage, so anything above 100 per cent means the renderer cannot keep pace with the panel. Uncapped frame rate is 1000 divided by the render time — what the machine would deliver with synchronisation switched off entirely.

The V-Sync figure is where the whole numbers bite. A frame can only be flipped to the screen on a blanking boundary, so it occupies a whole number of intervals: n = ⌈t ÷ T⌉, floored at 1 because even an instantly-rendered frame still waits for one boundary. The delivered rate is then f ÷ n. On a 60 Hz panel that produces the familiar 60, 30, 20, 15 staircase as the render time crosses 16.67, 33.3 and 50 ms; on a 144 Hz panel it is 144, 72, 48, 36. This models hard, double-buffered V-Sync only — triple buffering, frame pacing and variable-refresh displays each break the staircase differently, and Microsoft notes that variable-refresh operation actually requires tearing to be enabled.

A worked example.

Example

A game is being tuned for a 144 Hz monitor, and the frame-time overlay reads a steady 8.0 ms. The panel's budget is 1000 ÷ 144 = 6.9444 ms per frame, which is 6944.4 microseconds and 8 640 frames every minute. At 8 ms the renderer is using 8 ÷ 6.9444 = 115.2 per cent of that budget, leaving a headroom of −1.06 ms; it is over the deadline by just over a millisecond. Uncapped, 1000 ÷ 8 = 125 frames per second would reach the screen — respectable, but short of 144. Switch hard V-Sync on and the picture changes sharply: each frame must be held for ⌈8 ÷ 6.9444⌉ = ⌈1.152⌉ = 2 whole refresh intervals, so the delivered rate is 144 ÷ 2 = 72 fps. Missing the window by 1.06 ms costs 53 of the 125 frames per second the machine was capable of. Finding 1.1 ms of render time — one post-processing pass, or a shadow cascade — would take the frame rate straight back to 144.

refresh Rate Hz144
render Time Ms8
frame Time Ms10
rate Basisexact
solve Forframe-time

Frequently asked questions.

How many milliseconds is one frame at 144 Hz?
6.944 milliseconds, or 6 944 microseconds. The refresh interval is the reciprocal of the refresh rate: 1000 ÷ 144 = 6.9444. For comparison, 60 Hz is 16.667 ms, 75 Hz is 13.333 ms, 120 Hz is 8.333 ms, 165 Hz is 6.061 ms and 240 Hz is 4.167 ms. Frame time is the more useful unit for optimisation work because it is linear: shaving 1 ms off a frame is worth the same amount of work whatever frame rate you are at, whereas "10 more fps" means something completely different at 30 fps and at 200 fps.
Why does my frame rate drop straight from 144 to 72 with V-Sync on?
Because presentation is quantised to whole vertical blanks. Microsoft's DXGI documentation defines the SyncInterval parameter of Present as an integer where "1 through 4 — Synchronize presentation after the nth vertical blank". If your frame needs even a fraction more than one refresh interval, it has to wait for the second one, so the delivered rate becomes the refresh rate divided by two. On a 144 Hz panel that is 72 fps; on 60 Hz it is 30 fps. The staircase continues: 48 and 36 fps on 144 Hz, 20 and 15 fps on 60 Hz.
What is the difference between 60 Hz and 59.94 Hz?
Exactly a factor of 1.001. Recommendation ITU-R BT.709-6 lists the permitted picture rates as 60, 60/1.001, 30, 30/1.001, 50, 25, 24 and 24/1.001, and ITU-R BT.2020-2 carries the same divisor into 120 Hz and 100 Hz ultra-high-definition. Sixty divided by 1.001 is 59.940059940… Hz, an interval of 16.6833 ms rather than 16.6667 ms — 0.0167 ms longer per frame. Across an hour that adds up to about 3.6 seconds, which is precisely the discrepancy drop-frame timecode was invented to absorb. Select the 1.001 basis to work with the exact value rather than the rounded 59.94.
Should I optimise for average frame time or worst-case frame time?
Worst case, usually expressed as the 99th or 99.9th percentile. Stutter is perceived from the frames that arrive late, not from the mean, and a single 30 ms frame in an otherwise smooth second is far more visible than a uniform two-millisecond increase everywhere. Enter your 99th-percentile frame time in the render-time field to see how often the V-Sync staircase is actually being triggered, then compare it against the average to see how much variance you are carrying.
Does variable refresh rate remove the frame rate staircase?
Yes, that is essentially what it is for. Adaptive-sync displays vary their own refresh interval to match when the frame is actually ready, so there is no fixed boundary to miss and no whole-number division. Microsoft's documentation notes that "it is a requirement of variable refresh rate displays that tearing is enabled" — the fixed-boundary constraint that produces the staircase is exactly the thing being switched off. This calculator models hard, double-buffered V-Sync, so treat the locked frame rate as the worst case that variable refresh is designed to avoid.
How do I convert a frame time back into a refresh rate?
Divide 1000 by the frame time in milliseconds. A measured 10 ms interval is 100 Hz, 8.33 ms is 120 Hz, 6.94 ms is 144 Hz and 4.17 ms is 240 Hz. Use the second mode of this calculator to do it, along with the budget figures. Bear in mind that a frame time read off a game's overlay is usually the renderer's time, not the display's interval — the two only coincide when the renderer is exactly keeping pace with the panel.

In this category

Embed

Quanta Pro

Paid features are coming later.

  • All 590 calculators remain free
  • No billing is enabled
Coming soon