Refresh Rate & Frame Cap Calculator

Work out the frame-rate cap for a VRR monitor, where Low Framerate Compensation has a legal multiple, and how many pixels of motion blur a frame interval costs.

Frame cap & VRR Motion clarity

Your display

Maximum refresh rate

Off the panel's spec sheet. Decimals are accepted: 239.76 and 165.06 are real modes.

VRR window floor

The bottom of the panel's G-SYNC or FreeSync range, off the same spec sheet.

Your frame rate

Frame rate you sustain

fps

1% low frame rate

fps

From a benchmark run of the game you are tuning, not its average.

Sync and limiter

Frame limiter

V-SYNC on V-SYNC off

Cap rule — 3 fps below refresh

Fixed fps below refresh Fixed frametime margin

fps below max refresh

fps

G-SYNC 101 publishes 3 as a minimum, not as a target.

Cap your frame rate at

141 fps

7.092 ms per frame, 0.148 ms clear of one refresh

One refresh

6.944 ms

at 144 Hz, and the scanout time at any frame rate

One rendered frame

5.000 ms

at 200 fps, before any cap

200 fps sustained is above the 141 fps cap, so the limiter is the binding constraint. That is what it is for.

What the cap costs, and where LFC covers you

Cap frametime

7.092 ms

Clear of one refresh

0.148 ms · 2.1%

Delay avoided by capping

13.9 ms

Limiter's own delay

0.0 ms

VRR window ratio

×3.00

1% low frametime spread

4.09 ms

LFC multiplier at your 1% low

×1

Panel refresh under LFC

110.0 Hz

Highest legal multiplier

×1

Where LFC has a legal multiple, and where it has none

Frame rate	Frames duplicated	Panel refresh	Verdict
48 to 144 fps	none	48 to 144 Hz	Inside the window as rendered
24 to 48 fps	×2	48 to 96 Hz	Covered: each frame shown 2 times
16 to 24 fps	×3	48 to 72 Hz	Covered: each frame shown 3 times
12 to 16 fps	×4	48 to 64 Hz	Covered: each frame shown 4 times
9.6 to 12 fps	×5	48 to 60 Hz	Covered: each frame shown 5 times

At ×3.00 this window is ×2 or wider, and a window that wide has no dead bands at all: every frame rate down to 9.6 fps has a legal multiple.

Every refresh rate: what −3 fps costs, and what 0.30 ms holds

Refresh	One refresh	Cap at −3 fps	Its margin	Cap at 0.30 ms
60 Hz	16.667 ms	57	0.877 ms	59
75 Hz	13.333 ms	72	0.556 ms	73
100 Hz	10.000 ms	97	0.309 ms	97
120 Hz	8.333 ms	117	0.214 ms	116
144 Hz	6.944 ms	141	0.148 ms	138
165 Hz	6.061 ms	162	0.112 ms	157
180 Hz	5.556 ms	177	0.094 ms	171
240 Hz	4.167 ms	237	0.053 ms	224
280 Hz	3.571 ms	277	0.039 ms	258
360 Hz	2.778 ms	357	0.023 ms	325
480 Hz	2.083 ms	477	0.013 ms	420
540 Hz	1.852 ms	537	0.010 ms	465

The −3 fps rule is published as a minimum, and its frametime cushion falls from 0.877 ms at 60 Hz to 0.010 ms at 540 Hz — a factor of about 85 — while a fixed 0.30 ms margin holds the same cushion at every rung. The last column reproduces the five caps Reflex settles on with G-SYNC and NVCP V-SYNC: 59, 97, 116, 138 and 224 fps.

The cap, step by step

one refresh = 1000 / 144 = 6.944 ms

cap = 144 - 3 fps = 141

cap frametime = 1000 / 141 = 7.092 ms

clear of one refresh = 7.092 - 6.944 = 0.148 ms

Why there are two cap rules

G-SYNC 101 sets the cap "a minimum of 3 FPS below display's maximum refresh rate", and publishes 57 at 60 Hz, 97 at 100 Hz, 117 at 120 Hz and 141 at 144 Hz. A minimum, not a target.

A constant fps step is a shrinking frametime cushion: the same −3 fps is worth 0.877 ms at 60 Hz and 0.010 ms at 540 Hz. What the panel actually needs is time, not frames.

The driver's own limiter behaves like a fixed frametime margin instead, holding about 0.30 ms at every refresh rate. That figure is derived here from the caps G-SYNC 101 publishes for Reflex; it is not an NVIDIA specification.

Blur at every frame rate, at this motion speed

Frame rate	Persistence	Blur	Share of width	MCR
30 fps	33.333 ms	32.00 px	1.67%	30 Hz
60 fps	16.667 ms	16.00 px	0.83%	60 Hz
75 fps	13.333 ms	12.80 px	0.67%	75 Hz
100 fps	10.000 ms	9.60 px	0.5%	100 Hz
120 fps	8.333 ms	8.00 px	0.42%	120 Hz
144 fps	6.944 ms	6.67 px	0.35%	144 Hz
165 fps	6.061 ms	5.82 px	0.3%	165 Hz
180 fps	5.556 ms	5.33 px	0.28%	180 Hz
240 fps	4.167 ms	4.00 px	0.21%	240 Hz
360 fps	2.778 ms	2.67 px	0.14%	360 Hz
480 fps	2.083 ms	2.00 px	0.1%	480 Hz
540 fps	1.852 ms	1.78 px	0.09%	540 Hz
1,000 fps	1.000 ms	0.96 px	0.05%	1,000 Hz

On a sample-and-hold panel persistence is the displayed frame interval, so the blur column only halves when the frame rate doubles. At 960 px/s a 4.0 px budget first holds at 240 fps.

The 0.30 ms margin is derived here from the five caps G-SYNC 101 publishes for Reflex with G-SYNC and NVCP V-SYNC on. It is not an NVIDIA specification. Blur Busters Law is a floor: it assumes 0 ms GtG.

Frame cap calculator for VRR panels. The cap to set, and the blur a frame interval costs.

On a variable-refresh panel the cap goes a fixed margin under the ceiling, and this calculator turns that margin into the fps number your limiter takes. It also reports whether Low Framerate Compensation has a legal multiple for your 1% low, and how many pixels of motion blur one frame interval smears at a given motion speed.

What a frame cap does on a G-SYNC or FreeSync panel

A frame-rate cap on a variable-refresh panel exists to keep the frame rate off the panel's ceiling, where variable refresh stops varying. G-SYNC 101 measures 2 to 6 frames of delay once the frame rate reaches the maximum refresh rate with V-SYNC on; with V-SYNC off the same ceiling costs tearing instead. At 144 Hz one refresh lasts 6.944 ms, and a 141 fps cap holds each frame for 7.092 ms — 0.148 ms clear of a refresh, or 2.1% of the frame. That 0.148 ms is the entire margin, and it is what a fixed fps step spends as the refresh rate climbs.

Two rules answer the same question. jorimt's G-SYNC 101 publishes the community one, setting the limit "a minimum of 3 FPS below display's maximum refresh rate", with 57 at 60 Hz, 97 at 100 Hz, 117 at 120 Hz and 141 at 144 Hz. The second rule fixes the margin in milliseconds instead, and a 0.30 ms margin reproduces every cap the driver settles on with Reflex set to On or On + Boost alongside G-SYNC and NVIDIA Control Panel V-SYNC: 59 at 60 Hz, 97 at 100, 116 at 120, 138 at 144 and 224 at 240. That 0.30 ms is derived on this page from those five published caps; it is not an NVIDIA specification.

Underneath the window's floor the panel cannot slow down any further, so Low Framerate Compensation shows each rendered frame more than once until the refresh rate lands back inside the range. AMD's own worked example is a 60 to 144 Hz display running a game at 40 fps: doubling the frames puts the panel at 80 Hz, inside its range. The multiplier is a whole number, and that is the part that bites — a frame rate has a legal multiple only when some whole number times it lands between the floor and the ceiling.

That constraint has a closed form, so the bands can be located instead of guessed at. Dead bands exist if and only if the ceiling is below twice the floor, and in a window that narrow they can be listed exactly. A 48 to 60 Hz window has three: between 30 and 48 fps, between 20 and 24, and between 15 and 16. Both ends of each band are legal — 30 fps doubles to 60 Hz, the ceiling, and 48 fps needs no multiplier at all — so it is the inside of the interval that has nothing to pick. A 48 to 144 Hz window is ×3.00 and has none: every frame rate down to 9.6 fps has a legal multiple.

The second mode answers what the frame interval you settled on looks like in pixels. Blur Busters Law states that 1 ms of persistence smears 1 pixel of motion blur per 1000 px/s of eye-tracked motion, so a frame interval converts straight into a distance on the glass. On a sample-and-hold panel persistence is the displayed frame interval, which makes the panel's refresh rate irrelevant to blur whenever the game runs slower than it: 60 fps smears 16.00 px at 960 px/s on a 60 Hz panel and on a 480 Hz panel alike.

The cap rules, the dead-band condition and Blur Busters Law

\text{cap} = \text{round}\left(\frac{1000}{1000/\text{Hz} + m}\right) \qquad \text{blur (px)} = \frac{\text{MPRT} \times \text{speed}}{1000}

$\text{Hz}$ = The panel's maximum refresh rate. One refresh lasts 1000/Hz ms, which is also the scanout time at any frame rate.
$m$ = The frametime margin in milliseconds. 0.30 ms reproduces the five caps G-SYNC 101 publishes for Reflex, and is derived from them rather than specified by NVIDIA.
$f$ = The fps step of the other rule, where cap = Hz minus f. G-SYNC 101 publishes 3 as a minimum.
$\text{MPRT}$ = Persistence in milliseconds: the displayed frame interval on a sample-and-hold panel, the strobe pulse length on a strobed one.
$\text{speed}$ = Eye-tracked motion speed in pixels per second. TestUFO's default is 960 px/s.
$k$ = The LFC multiplier: the smallest whole number that puts k times the frame rate at or above the window floor.

The two cap rules differ in what they hold constant. Subtracting a fixed number of frames from the ceiling keeps the fps step identical at every rung, so its cushion in time shrinks as the refresh rate rises: −3 fps is worth 0.877 ms at 60 Hz and 0.010 ms at 540 Hz, a factor of about 85. The frametime rule holds the cushion and lets the fps step grow instead, which is why a fixed 0.30 ms lands 16 fps under the ceiling at 240 Hz and 75 fps under it at 540 Hz.

The dead bands follow from the whole-number multiplier. For each multiplier k the covered band runs from floor/k up to ceiling/k, and the gap between two neighboring covered bands is an open interval:

\text{gap}(k) = \left(\frac{\text{ceiling}}{k+1},\; \frac{\text{floor}}{k}\right) \quad \text{non-empty when} \quad \frac{\text{ceiling}}{\text{floor}} < \frac{k+1}{k}

The first gap therefore opens as soon as the ceiling drops below twice the floor, and the narrower gaps follow at ×1.5, ×1.33 and ×1.25. A 48 to 75 Hz window is ×1.56, so it has one dead band and no more; a 48 to 60 Hz window is ×1.25, which sits below the first three thresholds and produces three. Every endpoint stays legal, which is why the page writes these bands as "between 30 and 48 fps" rather than as a range from 30 to 48.

Blur Busters Law turns the frame interval into pixels: 1 ms of persistence smears 1 pixel per 1000 px/s of motion, so blur in pixels is MPRT times speed divided by 1000. At TestUFO's default 960 px/s a 144 fps frame interval of 6.944 ms smears 6.67 px, which is 0.35% of a 1920 px width. The Motion Clarity Ratio inverts the same figure, 1000 divided by MPRT, and TestUFO describes it as a simulated refresh rate for eye-tracking a moving object. On a strobed panel the pulse length replaces the frame interval, so a 1 ms pulse holds persistence at 1.000 ms whatever the frame rate does.

How to drive the calculator

1. Pick the mode. Frame cap & VRR answers what to cap at and where LFC covers you; Motion clarity answers what the frame interval smears.

2. Maximum refresh rate and VRR window floor both come off the panel's spec sheet, and both accept decimals — 239.76 Hz and 165.06 Hz are real modes, not typos.

3. Frame rate you sustain and 1% low frame rate come from a benchmark run of the game you are tuning, not from an average across a review's test suite. The 1% low is the figure that decides whether LFC has anything to do.

4. Frame limiter changes what the cap costs you. An engine-level cap in the game or its config file adds nothing; RTSS or NVIDIA Max Frame Rate caps at CPU level and costs about a frame; a driver-level limiter costs about two. Reflex on either setting caps by itself, so the page switches to the frametime rule and tells you to leave the manual limiter alone.

5. Cap rule sits in a fold that ships closed, because the default rule is the one most people want. Open it to swap between a fixed fps step and a fixed frametime margin, or to change either margin.

6. In Motion clarity, choose sample-and-hold or strobed, give the motion speed in pixels per second or in screen widths per second, set the screen width, and set the blur budget you are willing to accept in pixels. The budget is what turns the mode into a target frame rate.

7. The refresh-rate ladder under the results also ships folded. Its rows are in the page either way, so find-in-page opens it, and it is the fastest way to see what −3 fps is worth at your rung.

Worked cases

144 Hz, a 48 Hz floor, 200 fps sustained

The −3 fps rule puts the cap at 141 fps, which is 7.092 ms per frame and 0.148 ms clear of the panel's 6.944 ms refresh, or 2.1% of the frame. Capping avoids 13.9 ms of buffered delay at the ceiling: two refreshes, the floor of the 2 to 6 frames G-SYNC 101 measures with V-SYNC on. The window ratio is ×3.00, so the LFC ladder comes back covered at every frame rate down to 9.6 fps. A 110 fps 1% low sits inside the range as rendered, needs no multiplier, and leaves a 4.09 ms frametime spread against the 200 fps average. Switch the cap rule to a 0.30 ms margin and the cap moves to 138 fps, which is where Reflex would land it.

A 48 to 60 Hz window, and a 35 fps 1% low

Set the floor to 48 Hz and the ceiling to 60 Hz and the ladder grows from five rows to eight, three of them dead: between 30 and 48 fps, between 20 and 24, and between 15 and 16. Enter a 35 fps 1% low and the page reports no legal multiple, with the highest legal multiplier at ×1 against the ×2 the floor demands. Showing each frame once leaves the panel at 35 Hz, under the 48 Hz floor; showing it twice asks for 70 Hz, over the 60 Hz ceiling. LFC has nothing to pick, so the panel leaves the variable-refresh range at the one frame rate you most needed it to hold. Both endpoints of that band stay legal: 30 fps doubles to 60 Hz, and 48 fps is inside the window as rendered.

What −3 fps is worth at 60 Hz and at 540 Hz

Open the refresh ladder and the two rules separate. At 60 Hz the −3 fps rule caps at 57 fps and leaves 0.877 ms of frametime cushion, while a 0.30 ms margin caps at 59. At 144 Hz they land 3 fps apart, 141 against 138. By 540 Hz the −3 fps cap is 537 fps and its cushion is 0.010 ms, about 85 times thinner than at 60 Hz, while the frametime rule caps at 465 fps and holds the same 0.30 ms it held at the bottom of the ladder. G-SYNC 101 publishes 3 as a minimum, and the 480 and 540 Hz rungs are where you can read how much of a minimum it is.

144 fps at 960 px/s, and a 4-pixel budget

In Motion clarity, 200 fps rendered on a 144 Hz panel displays at 144 fps, because persistence cannot be shorter than one refresh. That gives 6.944 ms of persistence, 6.67 px of blur at TestUFO's default 960 px/s, 0.35% of a 1920 px width, a Motion Clarity Ratio of 144 Hz and an equivalent camera shutter of 1/144 s. The blur budget is where the mode turns into a purchase decision: 4 px first holds at 240 fps. The rest of the ladder at the same speed reads 16.00 px at 60 fps, 8.00 px at 120, 4.00 px at 240, 2.67 px at 360, 2.00 px at 480 and 0.96 px at 1000. Each halving of the blur costs a doubling of the frame rate.

The same panel strobed with a 1 ms pulse

Switch persistence to strobed and enter a 1 ms pulse and the answer stops depending on the frame rate. Persistence reads 1.000 ms, blur 0.96 px at 960 px/s, and the duty cycle 20.0% of the 5.000 ms rendered frame. The clarity ladder goes flat, because the pulse sets persistence: a 60 fps frame and a 500 fps frame are each visible for the same millisecond. What the frame rate decides on a strobed panel is how often the same frame is flashed again, which the eye reads as a double image at a fixed separation rather than as smear. Blur Busters documents pulses down to 0.25 ms; the bill is brightness, and panels raise peak luminance to cover it.

Dead bands by VRR window

VRR window	Ratio	Dead bands	Frame rates with no legal multiple
48 to 60 Hz	×1.25	3	Between 30 and 48, 20 and 24, and 15 and 16 fps
60 to 75 Hz	×1.25	3	Between 37.5 and 60, 25 and 30, and 18.75 and 20 fps
40 to 60 Hz	×1.50	1	Between 30 and 40 fps
48 to 75 Hz	×1.56	1	Between 37.5 and 48 fps
48 to 100 Hz	×2.08	0	None: the ceiling is already at least twice the floor
48 to 144 Hz	×3.00	0	None: covered at every frame rate down to 9.6 fps

Where tuning this goes wrong

Capping at the refresh rate instead of under it. A limiter set to the panel's own number leaves no margin, so a frametime spike reaches the ceiling. With V-SYNC on that ceiling costs the 2 to 6 frames G-SYNC 101 measures; with V-SYNC off it costs tearing.
Treating 2.5:1 as the LFC requirement. The threshold that decides whether dead bands exist is 2:1, and it is exact. AMD's FreeSync page describes the mechanism and gives its own example — a 60 to 144 Hz display doubling a 40 fps game to 80 Hz — and states no ratio between a panel's floor and its ceiling anywhere. A 48 to 100 Hz window is ×2.08 and has no dead bands at all.
Reading −3 fps as a fixed amount of safety. It is a fixed number of frames, so what it buys in time collapses from 0.877 ms at 60 Hz to 0.010 ms at 540 Hz. The frametime rule is the same idea expressed in milliseconds.
Expecting a faster panel to fix blur at an unchanged frame rate. Persistence on a sample-and-hold panel is the displayed frame interval, so 60 fps smears 16.00 px at 960 px/s on a 60 Hz panel and on a 480 Hz panel alike. A higher refresh rate shortens persistence once the frame rate rises to meet it.
Comparing a vendor's MPRT figure against the one here. These are MPRT(100%) figures, while the original paper defines MPRT(90%), so a spec sheet quoting 1 ms MPRT is measuring a different window. GtG is a third quantity again: a panel with a 1 ms GtG transition displaying 60 fps still holds each frame for 16.667 ms.
Stacking a manual limiter on top of Reflex. With Reflex set to On or On + Boost alongside G-SYNC and NVIDIA Control Panel V-SYNC, the driver already settles near 138 fps on a 144 Hz panel. A second cap underneath only lowers the frame rate further.
Setting a margin wide enough to push the cap under the floor. The panel then sits in LFC permanently, duplicating frames at a frame rate it could have displayed directly.

What these numbers do not include

Blur Busters Law is a floor. It assumes 0 ms GtG and square-wave persistence, so the real blur on every panel ever made is worse than the pixel figure here. Blur Busters publishes no rule for adding GtG to it, which is why the clarity mode has no GtG field: a number with no documented way to combine it would make the answer look more precise while making it less true.

The persistence figures are MPRT(100%), not the MPRT(90%) of the original paper, so a vendor spec sheet quoting MPRT will not match them.

The Motion Clarity Ratio is a simulated refresh rate for eye-tracking, not a rate the panel runs at, and TestUFO states plainly that it will not fix stroboscopic effects caused by finite refresh rates. A high MCR from a short strobe pulse sharpens a moving object; the stepping of a low frame rate survives it untouched.

The 0.30 ms frametime margin is derived here from the five caps G-SYNC 101 publishes for Reflex with G-SYNC and NVIDIA Control Panel V-SYNC on. It reproduces all five, which is why the page offers it as a rule, and it remains this page's derivation rather than an NVIDIA specification.

The LFC column reports the smallest legal multiplier. A variable-refresh module may pick a higher one, and no arithmetic here can tell you which: two published data points do not pin a vendor's choice.

Strobing costs brightness, and the page reports duty cycle rather than a percentage of luminance lost. Panels raise peak luminance to compensate, so no single percentage covers it.

Terms on this page

VRR window

The range of refresh rates a panel varies between, quoted on its spec sheet as a floor and a ceiling. Inside the window the refresh period is the frame interval.

LFC (Low Framerate Compensation)

AMD's name for showing a rendered frame more than once when the frame rate falls under the window floor, so the panel's refresh rate stays inside its range. AMD's example: a 60 to 144 Hz display doubling a 40 fps game to 80 Hz.

Dead band

An open interval of frame rates for which no whole-number multiple lands inside the VRR window. Dead bands exist only where the ceiling is below twice the floor, and both endpoints of a band are legal.

MPRT (persistence)

How long a pixel stays visible: the displayed frame interval on a sample-and-hold panel, the strobe pulse length on a strobed one.

GtG

Grey-to-grey transition time, how long a pixel takes to change color. Close to independent of frame rate, and not what eye-tracked motion blur is made of.

Motion Clarity Ratio (MCR)

1000 divided by MPRT. TestUFO describes it as a simulated refresh rate for eye-tracking a moving object, and states that it will not fix stroboscopic effects caused by finite refresh rates.

Duty cycle

On a strobed panel, the pulse length as a share of one rendered frame. A 1 ms pulse inside a 5.000 ms frame is 20.0%.

Scanout

The time a panel takes to paint one refresh. It is one refresh period at the maximum refresh rate whatever the frame rate.

1% low

The frame rate of the slowest 1% of frames in a benchmark run — the 99th-percentile frametime, read as a frame rate. It is the figure that decides whether LFC has anything to do.

Frequently asked questions about frame caps, LFC and motion blur

What FPS should I cap at on a 144 Hz monitor?

141 fps under the rule G-SYNC 101 publishes, which sets the limit at least 3 fps below the maximum refresh rate. 138 fps under a fixed 0.30 ms frametime margin, which is where Reflex settles with G-SYNC and NVIDIA Control Panel V-SYNC on.

Is the −3 fps rule still enough at 480 Hz?

It is still a valid cap, and G-SYNC 101 publishes 3 as a minimum rather than a target. What changes is the cushion: −3 fps leaves 0.877 ms at 60 Hz and 0.013 ms at 480 Hz, so a single frametime spike is far more likely to reach the ceiling. Holding a fixed 0.30 ms instead caps a 480 Hz panel at 420 fps and a 540 Hz panel at 465 fps.

Does LFC require the maximum refresh rate to be 2.5 times the minimum?

No. A frame rate loses its legal multiple only where the ceiling is below twice the floor, and that threshold is exact. AMD's FreeSync page states no ratio between floor and ceiling. A 48 to 100 Hz window is ×2.08 and has no dead bands.

Why does my FreeSync monitor stutter at around 35 fps?

On a narrow window that frame rate can have no legal multiple. In a 48 to 60 Hz window, 35 fps shown once puts the panel at 35 Hz, under the floor, and shown twice asks for 70 Hz, over the ceiling. LFC has nothing to pick, so the panel leaves the variable-refresh range at the frame rate where the game is already struggling. Anything between 30 and 48 fps behaves the same way in that window.

What is a dead band in a VRR window?

A dead band is a range of frame rates for which no whole-number multiple of the frame rate lands inside the panel's variable-refresh window. The intervals are open, so both endpoints are legal: in a 48 to 60 Hz window, 30 fps doubles to 60 Hz and 48 fps needs no multiplier, while every frame rate strictly between them is uncovered.

Does a 240 Hz panel halve the motion blur of a 120 Hz panel?

Only if the frame rate doubles with it. Persistence on a sample-and-hold panel is the displayed frame interval, so at 960 px/s the blur falls from 8.00 px to 4.00 px when the frame rate goes from 120 to 240. Run 120 fps on both panels and both smear 8.00 px.

Is MPRT the same as GtG response time?

No. MPRT is how long a pixel stays visible; GtG is how long it takes to change color. A panel with a 1 ms GtG transition displaying 60 fps still holds each frame for 16.667 ms, which is its MPRT.

How many pixels of motion blur should I aim for?

The blur budget field exists because the answer is a preference, and the page turns whatever you pick into a frame rate. At 960 px/s, 4 px first holds at 240 fps and 2 px at 480 fps. Set the motion speed to at least four times the refresh rate you are testing, which is the floor Blur Busters gives for telling sample-and-hold panels apart, and to 1920 px/s above 240 Hz.

Which frame limiter should I use: the in-game one, RTSS, or the driver?

The in-game or config-file limiter, when the game has one, because it caps at engine level and adds nothing. RTSS and NVIDIA Max Frame Rate cap at CPU level and cost about a frame; a driver-level limiter costs about two. The page reports that delay for whichever you pick, so the cost sits next to the cap itself.

My VRR window is narrower than ×2. What can I do about the dead bands?

Two things remove them and the arithmetic allows nothing else. Keep the 1% low above the floor, which is a settings decision: a dead band only matters if the game reaches it. Or move to a window whose ceiling is at least twice its floor, at which point every frame rate has a legal multiple. The range is a panel property, so the practical lever is usually the first one.

Is this calculator free to use?

Yes. It runs in the browser with no signup, and the numbers you type stay on your machine. Both modes and all three tables work without an account.

How accurate are these caps and blur figures?

The fps rule reproduces the caps G-SYNC 101 publishes at 60, 100, 120 and 144 Hz, and the 0.30 ms frametime rule reproduces all five of the caps it publishes for Reflex: 59, 97, 116, 138 and 224 fps. The blur figures are Blur Busters Law applied to the displayed frame interval, so they are exact against the law and optimistic against any real panel, because the law assumes 0 ms GtG. The 0.30 ms margin itself is derived here from those five published caps and is not an NVIDIA specification.

Sources & References

Content verified by the Smart Calculators Team