A missed first bullet in Valorant or a late peek in CS2 rarely comes down to aim alone. Somewhere between your mouse click and the pixel changing on screen, a stack of small delays adds up, and in 2026 that stack is finally something you can measure and cut down piece by piece. NVIDIA’s Reflex 2, AMD’s Anti-Lag 2, and a new generation of 360-500Hz monitors have turned input lag from a vague complaint into a number you can test before and after every change.

This tutorial walks through a full low-latency setup for competitive PC gaming: measuring your baseline, turning on the right GPU features, tuning your monitor and mouse, and building a repeatable PowerShell profile so you don’t have to redo this every time you reinstall Windows. By the end you’ll have a script you can run on any new rig and a reference table for exactly where your milliseconds are going. It’s a companion piece to our broader esports coverage, including our guide on climbing Valorant ranks, where every millisecond of reaction time matters just as much as game sense.

Why Input Lag Still Decides Rounds in 2026

Frame rate gets all the attention in benchmark charts, but frame rate and input lag are not the same thing. A game can render 400 frames per second and still feel sluggish if the render queue, the display pipeline, or a chatty background process is adding delay between your click and what shows up on screen. That gap is what players mean when they say a game “feels” slow even at a high FPS counter.

NVIDIA put real numbers behind this at its CES 2025 Reflex 2 announcement. In an NVIDIA-run demo of The Finals on an RTX 5070 at 4K with settings maxed out, total PC latency measured 56ms with no latency-reduction tech enabled. Turning on the original Reflex Low Latency mode cut that to 27ms. Reflex 2 with its new Frame Warp feature took it down again to 14ms, a 75% reduction from the untouched baseline. NVIDIA also reported that on an RTX 5090 running Valorant at over 800 FPS, Reflex 2 with Frame Warp pushed average PC latency under 3ms in its own testing, among the lowest figures the company says it has measured in a first-person shooter.

Those are best-case, vendor-reported numbers from controlled demos, not a guarantee for your exact rig. But they show where the ceiling sits, and that ceiling is a lot lower than it was two years ago. This guide focuses on the settings and hardware choices that get a normal gaming PC closer to that ceiling without spending a paycheck on new gear.

The Full Latency Chain: Where Every Millisecond Comes From

Input lag is not one number, it’s a chain of six or seven separate delays stacked on top of each other. Your mouse samples movement at a fixed polling rate, that sample travels over USB to Windows, the game engine picks it up on its next simulation tick, the GPU renders and queues the frame, the frame travels down a cable to your monitor, and the panel itself takes time to scan and change pixel color. Fix one link and the others still matter.

Render Queue vs Display Latency vs Peripheral Latency

It helps to split the chain into three buckets. Peripheral latency covers your mouse and keyboard, measured in fractions of a millisecond at modern polling rates. Render latency covers everything between the game engine deciding what to draw and the GPU handing a finished frame to the display, which is where Reflex, Reflex 2, and Anti-Lag 2 do their work. Display latency covers the cable and the monitor panel itself, including scan-out time and pixel response.

StageTypical ContributionBest-Case With Optimization
Mouse sensor + USB polling (1000Hz)~1.0ms~0.125ms at 8000Hz
Windows input pipeline + background load1-4ms<1ms with Game Mode, no overlays
Game engine tick + CPU-side render prep2-8msCut by Reflex’s render queue trim
GPU render queue and frame submission5-20msSub-3ms with Reflex 2 + Frame Warp (Valorant, NVIDIA demo)
Cable transmission (DisplayPort/HDMI)<1msNegligible on a direct GPU-to-monitor run
Monitor scan-out + pixel response2-4.2ms at 240Hz~2ms or below near 500Hz panels

The render queue and GPU stage is usually the biggest single chunk, which is why Reflex, Reflex 2, and Anti-Lag 2 target it first. Everything downstream of that, cabling, monitor response, and mouse polling, adds up to a few more milliseconds you can claw back once the render side is under control.

Prerequisites: What You Need Before You Start

You don’t need a full upgrade to see gains, but a few pieces of software and access matter before you touch any settings.

  • Windows 11 with an administrator account (this guide uses registry and power-plan changes that require elevation)
  • PowerShell 5.1 or later, included by default on current Windows 11 builds
  • An NVIDIA GPU on the latest Game Ready driver for Reflex, or an AMD RDNA 3/4 card on a current Adrenalin driver for Anti-Lag 2
  • A game that supports Reflex or Anti-Lag natively, such as Valorant, CS2, or Fortnite
  • A monitor capable of at least 144Hz, ideally 240Hz or higher, connected via DisplayPort 1.4 or newer, or HDMI 2.1
  • A wired or high-polling wireless mouse, set to 1000Hz or higher
  • 15-20 minutes for the driver and settings pass, plus another 20 minutes if you build and test the PowerShell profile in Step 11

None of this requires new hardware. The biggest single-setting gains, Reflex and Anti-Lag 2, are free features already built into drivers most players have installed.

Step 1: Measure Your Baseline Latency First

Changing five settings at once and calling it faster is guesswork. Log where you’re starting from first, so you can prove each change actually helped instead of just assuming it did.

NVIDIA GPUs report an in-game PC Latency number through the GeForce overlay (Alt+R) when Reflex is supported by the title, even before you turn Reflex on. Run three matches or practice sessions and note the average and peak numbers before making any changes. If you’re on AMD, Radeon Software’s performance overlay reports frame time and frame time variance, which is a reasonable stand-in since AMD does not expose a system-wide PC latency counter the way NVIDIA does.

Alongside the in-game number, log your system state with a short script so you have a record of exactly what was configured when you took the reading.

# baseline-latency-check.ps1
# Run in an elevated PowerShell window before changing any settings.

$report = [ordered]@{
    Timestamp       = Get-Date -Format "yyyy-MM-dd HH:mm:ss"
    GPUDriver       = (Get-CimInstance Win32_VideoController | Select-Object -First 1 -ExpandProperty DriverVersion)
    GPUName         = (Get-CimInstance Win32_VideoController | Select-Object -First 1 -ExpandProperty Name)
    ActivePowerPlan = (powercfg /getactivescheme)
    GameModeOn      = (Get-ItemProperty -Path "HKCU:\Software\Microsoft\GameBar" -Name "AutoGameModeEnabled" -ErrorAction SilentlyContinue).AutoGameModeEnabled
    MouseAccelOff   = ((Get-ItemProperty -Path "HKCU:\Control Panel\Mouse").MouseSpeed -eq 0)
}

$report | Format-List
$report | Export-Csv -Path "$env:USERPROFILE\Desktop\latency-baseline.csv" -NoTypeInformation

Running that after each major change in this guide gives you a paper trail. It also catches the most common mistake in latency tuning: a driver update or Windows update silently reverting a setting you turned off weeks ago.

Step 2: Update Drivers and Confirm Reflex or Anti-Lag Support

Reflex 2’s Frame Warp and Anti-Lag 2 are both driver-and-game features, meaning an outdated driver or an older game build can silently disable them even if the toggle looks on. Open the NVIDIA app (the successor to GeForce Experience) or AMD Software: Adrenalin Edition and check for the current driver before doing anything else.

Not every Reflex-supported game has Frame Warp yet. NVIDIA rolled Reflex 2 out title by title starting with The Finals, and support has expanded since, but you should treat Frame Warp as a per-game feature to verify rather than assume. Check your specific game’s in-game video settings menu for a Reflex or “Reflex + Frame Warp” option rather than relying on the driver alone.

AMD’s Anti-Lag 2 is integrated per-game rather than as a global driver toggle, which is a deliberate design choice. Unlike the original driver-level Anti-Lag, Anti-Lag 2 hooks into a specific game’s rendering pipeline, so it only appears in games the developer has added support for, including Valorant.

Step 3: Turn On NVIDIA Reflex and Reflex 2 Frame Warp

Inside the game’s video or graphics settings menu, look for “NVIDIA Reflex Low Latency” and set it to On, or to Boost if the game offers it. Boost forces the GPU clock higher during CPU-bound moments to keep the render queue from backing up, at the cost of slightly higher power draw and fan noise.

Where Frame Warp is available, it appears as a separate toggle alongside standard Reflex, not a replacement for it. Frame Warp works by re-projecting an already-rendered frame using your latest mouse input right before it’s sent to the display, effectively shaving off close to a full frame’s worth of delay at your current frame rate. Independent testing of a Reflex 2 Frame Warp demo running at 240 FPS showed latency drop from roughly 7.8ms with Frame Warp off to the 1-2ms range with it on, in a controlled test environment rather than live competitive play.

Reflex and Frame Warp are separate from V-Sync, and this is worth calling out because it trips people up: leave V-Sync off inside the game. Reflex’s own frame-pacing logic already prevents the render queue buildup that V-Sync is meant to fix, and stacking both adds latency instead of removing it.

Step 4: Configure AMD Anti-Lag 2 Without Stacking the Old Anti-Lag

If you’re on an AMD card, enable Anti-Lag 2 inside the supported game’s settings menu the same way you would enable Reflex on NVIDIA. In Valorant specifically, 2026 latency tests logging over 150 clicks per run found Anti-Lag 2 off measured about 10.00ms of input-to-display latency at roughly 934 FPS, and Anti-Lag 2 on measured about 9.84ms at roughly 921 FPS, a 0.16ms reduction that testers described as within normal measurement variance rather than a dramatic jump.

The more important finding from that same testing: turning on the legacy, driver-level Anti-Lag alongside the newer in-game Anti-Lag 2 pushed measured latency up slightly, to around 10.03ms, worse than leaving Anti-Lag 2 on by itself. The two features aren’t meant to run together. Open Radeon Software, go to the driver-level Anti-Lag setting for the game’s profile, and make sure it’s off if Anti-Lag 2 is already active inside the game.

In practice, Anti-Lag 2’s real value in a game like Valorant, which is already CPU-bound and running at very high frame rates, is small. Its bigger impact shows up in GPU-bound titles where the render queue has more room to back up. Don’t expect it alone to close the gap with an NVIDIA Reflex 2 setup, treat it as one part of the stack rather than the whole fix.

Step 5: Set Your Monitor’s Refresh Rate and Overdrive Preset

Open Windows Display Settings and confirm your monitor is running at its maximum refresh rate, not the 60Hz default some monitors fall back to after a cable swap or driver update. Set it again in your GPU control panel too, since the two aren’t always in sync. NVIDIA’s own system latency optimization guide walks through the same refresh-rate and driver checks from the GPU vendor’s side.

Confirm the refresh rate Windows is actually driving with a quick PowerShell check, since the number shown in Display Settings can lag behind what’s applied after a sleep/wake cycle.

PS C:\> Get-CimInstance Win32_VideoController | Select-Object Name, CurrentRefreshRate, VideoModeDescription

Name                                  CurrentRefreshRate VideoModeDescription
----                                  ------------------ --------------------
NVIDIA GeForce RTX 4070                              240 1920 x 1080 x 4294967296 colors

Refresh rate itself changes how often the panel can show a new frame, called scan-out time. Reporting on 2025-2026 esports monitors puts scan-out time at roughly 4.2ms at 240Hz, dropping to about 2.78ms at 360Hz, and down near 2ms or below on panels approaching 500Hz. The jump from 240Hz to 360-500Hz is real but small in absolute terms, typically landing in a 1-3ms range of total latency difference once your GPU can actually hit those frame rates, which is why plenty of pros still compete comfortably on 240Hz panels. For a closer look at how much refresh rate alone changes competitive feel, see our 144Hz vs 240Hz breakdown.

Response time is the other half of the equation, and it’s controlled by your monitor’s overdrive setting, usually labeled Response Time, Overdrive, or a similar name in the on-screen menu. Testing from display specialists at outlets like Blur Busters and TFTCentral consistently finds the “fast” or “normal” overdrive preset gives the best real gray-to-gray response, often in the 1-2ms range, while “extreme” settings frequently introduce visible overshoot and ghosting despite a marginally lower number on paper. Set it to fast or normal, not extreme, and check for pale trailing edges behind moving objects in a scrolling test pattern to confirm you haven’t overshot.

Step 6: Pick a Cable and Connection That Won’t Bottleneck You

Cable transmission itself adds under a millisecond and isn’t usually your bottleneck, but the wrong cable or a hub in the chain can force your monitor to a lower refresh rate without you noticing, which then costs you the scan-out gains from Step 5.

Run DisplayPort 1.4 or newer if your monitor and GPU support it, since it has the bandwidth headroom for 240Hz-500Hz at your monitor’s native resolution without compression. HDMI 2.1 is a fine alternative on monitors that support it. Connect the monitor directly to the GPU’s output rather than through a USB-C dock, KVM switch, or capture card in the signal path, all of which can silently cap refresh rate or introduce a frame or two of buffering. If your monitor’s on-screen menu shows a refresh rate lower than what you set in Windows, the cable or an adapter in the chain is almost always the reason.

Step 7: Tune Mouse Polling Rate for Your System

Polling rate controls how often your mouse reports its position to the PC. At 1000Hz, the standard for most gaming mice, that’s once every 1ms. Mice released in 2025-2026 with 8000Hz polling cut that interval to 0.125ms per report, a real but small reduction on its own.

The gains are not linear. Going from 1000Hz to 2000Hz saves roughly 0.5ms in mouse-side latency, and going all the way from 1000Hz to 8000Hz saves up to about 0.875ms, most of which is captured well before you reach the top of the range. That’s why current pro settings databases and interviews show most Valorant and CS2 pros still running 1000Hz or 2000-4000Hz rather than maxing out at 8000Hz. Higher polling rates also increase CPU overhead from constant USB interrupts, which can introduce stutter on older or budget CPUs, the opposite of what you’re trying to fix.

Set your polling rate in your mouse manufacturer’s software, not Windows, since Windows has no native control for USB polling rate. Start at 1000Hz, confirm the game feels stable with no stutter or micro-freezes during fast flicks, then step up to 2000Hz or higher only if your system stays smooth. If you notice occasional hitching after raising polling rate, that’s your CPU struggling with the interrupt load, not a mouse problem, and dropping back down is the fix.

Step 8: Enable Windows Game Mode and Cut Background Load

Windows Game Mode prioritizes your foreground game’s CPU and GPU scheduling over background tasks, which matters more than it sounds like on a system running Discord, a browser, and an overlay or two at the same time. Turn it on in Settings under Gaming, or force it through the registry key used in Step 11’s script.

Alongside Game Mode, disable Fullscreen Optimizations for your competitive game and run it in exclusive fullscreen rather than borderless windowed, if the game supports it. Fullscreen Optimizations is Windows’ compositing layer for borderless windowed games, and it adds a compositing step that true exclusive fullscreen skips entirely. You can toggle this per-executable from the .exe’s Properties > Compatibility tab, checking “Disable fullscreen optimizations.”

Close third-party overlays you don’t need during a match, including recording software, chat overlays, and FPS counters that hook into the render pipeline rather than reading from outside it. Each one adds a small amount of overhead, and they stack.

Step 9: Strip In-Game Settings That Add Latency

Some visual settings cost you frames directly, and every frame you’re not rendering is time added to the queue. Turn off motion blur, film grain, depth of field, chromatic aberration, and vignette. None of them help you see or react faster, and several of them exist specifically to blur fast motion, working against the exact thing you’re trying to fix.

Drop shadow quality, reflections, and post-processing effects to low or medium. These settings are usually GPU-render heavy without meaningfully changing your ability to spot an enemy player model. The goal is pushing your achievable frame rate as high above your monitor’s refresh rate as your GPU allows, since Reflex and Anti-Lag 2 both perform best when there’s frame rate headroom for them to work with. If you play CS2 alongside Valorant, our CS2 FPS boost guide and CS2 launch options and console setup guide cover the same ground for that engine specifically.

Step 10: Cap FPS the Right Way for Your Refresh Rate

It sounds backward to intentionally cap your frame rate when you’re chasing lower latency, but an uncapped frame rate that spikes and dips unevenly can hurt consistency more than a slightly lower, rock-steady frame rate helps. If your GPU is bouncing between 350 and 550 FPS in busy scenes, that inconsistency shows up as uneven frame pacing, which feels worse than it measures on paper.

A reasonable rule: cap your frame rate a small amount below the maximum your GPU can sustain in the busiest moments of a match, not your average FPS in an empty lobby. This keeps the GPU from bouncing between power states and keeps frame time variance low. Set the cap in-game rather than through the NVIDIA app or a third-party tool where possible, since an in-game cap sits closer to the render pipeline that Reflex is managing.

Step 11: Build a Repeatable Latency Profile (Complete Working Script)

Once you’ve found settings that work, the last problem is keeping them. A Windows Update, a fresh driver install, or a new PC build resets half of this without warning. The script below bundles the reversible, system-level pieces from this guide, the power plan, mouse acceleration, Game Mode, Game DVR, and fullscreen optimizations for a target executable, into one script you can rerun any time.

# apply-latency-profile.ps1
# Run as Administrator. Use -Revert to undo the power plan and mouse changes.

param(
    [switch]$Revert,
    [string]$TargetExe = "C:\Riot Games\VALORANT\live\VALORANT-Win64-Shipping.exe"
)

$highPerfGuid = "8c5e7fda-e8bf-4a96-9a85-a6e23a8c635c"

if ($Revert) {
    powercfg /setactive SCHEME_BALANCED
    Set-ItemProperty -Path "HKCU:\Control Panel\Mouse" -Name "MouseSpeed" -Value "1"
    Write-Host "Reverted to the balanced power plan and default mouse curve."
    exit
}

# 1. High performance power plan, so the CPU doesn't downclock between inputs
powercfg /setactive $highPerfGuid

# 2. Disable Enhance Pointer Precision (mouse acceleration)
Set-ItemProperty -Path "HKCU:\Control Panel\Mouse" -Name "MouseSpeed" -Value "0"
Set-ItemProperty -Path "HKCU:\Control Panel\Mouse" -Name "MouseThreshold1" -Value "0"
Set-ItemProperty -Path "HKCU:\Control Panel\Mouse" -Name "MouseThreshold2" -Value "0"

# 3. Force Windows Game Mode on
New-Item -Path "HKCU:\Software\Microsoft\GameBar" -Force | Out-Null
Set-ItemProperty -Path "HKCU:\Software\Microsoft\GameBar" -Name "AutoGameModeEnabled" -Value 1

# 4. Disable background Game DVR capture hooks
New-Item -Path "HKCU:\System\GameConfigStore" -Force | Out-Null
Set-ItemProperty -Path "HKCU:\System\GameConfigStore" -Name "GameDVR_Enabled" -Value 0

# 5. Disable fullscreen optimizations for the target game
if (Test-Path $TargetExe) {
    $layerKey = "HKCU:\Software\Microsoft\Windows NT\CurrentVersion\AppCompatFlags\Layers"
    New-Item -Path $layerKey -Force | Out-Null
    Set-ItemProperty -Path $layerKey -Name $TargetExe -Value "~ DISABLEDXMAXIMIZEDWINDOWEDMODE"
    Write-Host "Fullscreen optimizations disabled for $TargetExe"
} else {
    Write-Host "Target executable not found at $TargetExe, skipping that step. Pass -TargetExe to point at your game."
}

Write-Host "Latency profile applied. Reboot, then rerun baseline-latency-check.ps1 to compare against your original numbers."

Save both this script and the baseline logger from Step 1 in the same folder, ideally on a USB drive or synced cloud folder, so a fresh Windows install or a new PC build is a two-command fix instead of an hour of clicking through menus. The -TargetExe parameter means you can point it at CS2’s or Fortnite’s executable path just as easily as Valorant’s.

PS C:\Windows\system32> .\apply-latency-profile.ps1
Fullscreen optimizations disabled for C:\Riot Games\VALORANT\live\VALORANT-Win64-Shipping.exe
Latency profile applied. Reboot, then rerun baseline-latency-check.ps1 to compare against your original numbers.

Step 12: Re-Measure and Verify Your Gains

Reboot after running the profile script, then repeat the same process from Step 1: run the baseline logger again, and pull up your game’s PC Latency overlay for a few matches. Compare the new CSV against the original one on your desktop.

You’re looking for two things, not one. The average latency number should drop, but the spread between your best and worst readings matters just as much. A setup that averages 8ms but spikes to 25ms during fights is worse to play on than one that holds a steady 11ms, even though the second number looks worse on paper. If your average improved but spikes are still showing up, go back to Step 8 and check for a background process or overlay you missed.

Common Pitfalls When Chasing Lower Input Lag

  • Stacking Anti-Lag with Anti-Lag 2. As covered in Step 4, running AMD’s legacy driver-level Anti-Lag alongside the newer in-game Anti-Lag 2 measured worse, not better, in 2026 Valorant testing. Turn one off.
  • Leaving V-Sync on “just in case.” V-Sync and Reflex solve overlapping problems in different ways, and running both adds a buffering step Reflex is specifically designed to remove.
  • Maxing out mouse polling rate on an aging CPU. 8000Hz sounds like a free win, but the interrupt overhead can cause stutter on weaker or older systems, undoing any latency gain with new inconsistency.
  • Cranking monitor overdrive to “extreme.” The most aggressive overdrive preset often introduces visible ghosting and overshoot for a marginal, sometimes fake, response-time gain over the “fast” or “normal” preset.
  • Routing your monitor through a dock, KVM, or capture card. Any device between the GPU and monitor can silently cap your refresh rate, quietly erasing the gains from Step 5 without any error message.
  • Changing five settings at once. Without a baseline measurement from Step 1, you won’t know which change actually helped, and you won’t catch a setting that made things worse.

Troubleshooting: Input Lag Fixes for Specific Symptoms

Most latency complaints trace back to one of a small handful of causes. Here’s a quick-reference table for the most common ones, followed by more detail on the trickier cases.

SymptomLikely CauseFix
Reflex toggle is on but latency overlay shows no changeOutdated GPU driver or an unsupported build of the gameUpdate to the current Game Ready driver, verify Reflex support in that specific title’s patch notes
Monitor stuck at 60Hz after a fresh cable or driver installWindows or the GPU control panel reset the refresh rateRe-set refresh rate in both Windows Display Settings and the GPU control panel
Mouse feels “floaty” or inconsistent after raising polling rateCPU can’t keep up with USB interrupt load at 4000-8000HzDrop back to 1000-2000Hz and confirm stability
Frame rate is high but the game still feels delayedUncapped FPS is spiking and dipping, hurting frame pacingSet an in-game FPS cap slightly below your worst-case sustained frame rate
Ghosting or smearing trails behind fast-moving objectsMonitor overdrive set to “extreme” or too aggressiveDrop overdrive to the “fast” or “normal” preset
Anti-Lag 2 barely changes anything in your latency testGame is already CPU-bound at a very high frame rateExpect a small gain here, focus effort on Steps 8-10 instead
Latency profile script settings revert after a Windows UpdateWindows Update can reset Game Mode and GameDVR registry keysRerun apply-latency-profile.ps1 after major Windows updates
Full-screen game keeps minimizing or stuttering on alt-tabFullscreen Optimizations disabled incorrectly for the wrong executable pathConfirm the exact .exe path in Step 11’s -TargetExe parameter matches your install

If your latency numbers still look high after working through every step, check for a wireless USB dongle sharing a hub with other high-bandwidth devices, since USB bandwidth contention can add jitter to a wireless mouse’s polling even at a fixed rate. Plugging the dongle directly into a rear motherboard USB port, rather than a front-panel hub or a USB extension cable, resolves this in most cases.

Advanced Tips: Squeezing Out the Last Milliseconds

Once the core setup is in place, a few smaller tweaks can shave off fractions of a millisecond more, with diminishing but real returns for players chasing the absolute floor.

Network-side latency is a separate problem from render latency, but it affects how responsive a match feels overall. Disabling Nagle’s algorithm on your network adapter can reduce small packet delays in some setups. This is a network-jitter tweak, not a render-pipeline fix, so test it carefully since it can also destabilize slower or less stable connections.

# disable-nagle.ps1
# Run as Administrator. Test carefully, this affects network stack behavior.

$adapters = Get-NetAdapter | Where-Object { $_.Status -eq "Up" }
foreach ($adapter in $adapters) {
    $ifIndex = $adapter.ifIndex
    $regPath = "HKLM:\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces\Tcpip_$ifIndex"
    if (Test-Path $regPath) {
        Set-ItemProperty -Path $regPath -Name "TcpAckFrequency" -Value 1 -Type DWord -ErrorAction SilentlyContinue
        Set-ItemProperty -Path $regPath -Name "TCPNoDelay" -Value 1 -Type DWord -ErrorAction SilentlyContinue
    }
}
Write-Host "Nagle's algorithm disabled on active adapters. Reboot to apply."

On the render side, keep an eye on driver release notes for expanded Frame Warp support since NVIDIA has been adding titles to the list rather than shipping it everywhere at once. If your main competitive game doesn’t have it yet, standard Reflex still delivers the bulk of the render-queue reduction on its own, Frame Warp adds an incremental layer on top rather than being the whole story.

For monitor shopping decisions specifically, cross-reference independent response time and input lag databases like the ones maintained by RTINGS and TFTCentral rather than trusting a manufacturer’s marketing spec sheet alone, since panel-to-panel variance within the same model is common and marketing numbers tend to reflect best-case lab conditions.

Reflex, Reflex 2, and Anti-Lag 2 Head to Head

Pulling the numbers from earlier steps into one table makes the relative scale of each optimization easier to compare, though keep in mind these come from different test scenarios and games rather than one controlled apples-to-apples run.

ConfigurationTest ScenarioMeasured Latency
No latency reductionThe Finals, RTX 5070, 4K max settings (NVIDIA CES 2025 demo)56ms
Original Reflex Low LatencySame scenario as above27ms
Reflex 2 + Frame WarpSame scenario as above14ms (75% reduction from baseline)
Reflex 2 + Frame WarpValorant, RTX 5090, 800+ FPS (NVIDIA-reported)Under 3ms average
AMD Anti-Lag 2 offValorant, ~934 FPS (2026 independent testing)10.00ms
AMD Anti-Lag 2 onValorant, ~921 FPS (2026 independent testing)9.84ms
Anti-Lag 2 + legacy driver Anti-Lag stackedValorant, same test series10.03ms (worse than Anti-Lag 2 alone)

The gap between vendor demo conditions and your own PC will always be there, different game, different GPU, different resolution. Treat these as a directional guide to where the biggest opportunities sit, GPU-side render latency first, then monitor and mouse, rather than as numbers you should expect to hit exactly.

Building the Complete Setup: What Success Looks Like

Put together, the finished project from this tutorial is two PowerShell scripts and a short checklist, not a pile of one-off tweaks you’ll forget in six months. The baseline logger from Step 1 gives you a before-and-after record. The profile script from Step 11 reapplies the system-level settings in seconds after any Windows reinstall or new PC build. The in-game and driver-level settings from Steps 3, 4, 5, and 9 live inside the game and GPU software, so they need a five-minute manual check after major updates but nothing more.

A realistic outcome for a mid-range setup, an RTX 4070-class or equivalent GPU with a 240Hz monitor and a 1000-2000Hz mouse, is shaving several milliseconds off total system latency through Reflex or Anti-Lag 2 alone, with another one to three milliseconds available from monitor overdrive tuning and mouse polling rate once the render side is handled. It won’t match NVIDIA’s sub-3ms Valorant demo number on an RTX 5090, but the relative improvement, and the consistency from removing frame time spikes, is what actually shows up in how a match feels.

Frequently Asked Questions

What’s a good input lag number for competitive gaming?

There’s no single agreed threshold, but anything in the single digits to low teens of milliseconds for total PC latency, as reported by NVIDIA’s in-game overlay, is considered strong for a CPU-bound esports title in 2026. What matters more day to day is consistency: a stable 12ms plays better than an average of 8ms with regular spikes into the 20s.

Does NVIDIA Reflex work on AMD GPUs?

No. Reflex and Reflex 2 are NVIDIA-specific technologies tied to GeForce hardware and drivers. AMD’s equivalent feature is Anti-Lag 2, which is integrated per-game rather than as a global driver setting, and it’s the one to look for if you’re on an RDNA 3 or RDNA 4 card.

Is Reflex 2 Frame Warp available in every game?

No. Frame Warp launched with a small set of supported titles starting with The Finals and has expanded since, but support is added game by game rather than applied automatically across every Reflex-enabled title. Check your specific game’s video settings menu for a Frame Warp option before assuming it’s active.

Will Anti-Lag 2 alone fix bad input lag?

Not by itself in a CPU-bound, already-high-FPS title like Valorant, where 2026 testing measured only a 0.16ms improvement. Anti-Lag 2 has more impact in GPU-bound games where the render queue has more room to build up. Treat it as one part of a full setup, alongside monitor tuning, mouse polling, and background process cleanup, rather than a single fix.

Does higher mouse polling rate always mean lower latency?

Not linearly, and not without cost. Going from 1000Hz to 8000Hz saves up to roughly 0.875ms of mouse-side latency, most of which is captured well before 8000Hz, while adding real USB interrupt overhead that can cause stutter on weaker CPUs. Most pros still run 1000-4000Hz rather than maxing out the range.

Should I still use G-Sync or FreeSync if I care about latency?

Yes, for tear-free visuals, but pair it with Reflex or Anti-Lag 2 rather than relying on adaptive sync alone, and avoid stacking it with driver-level Fast Sync or in-game V-Sync, which reintroduce the buffering these technologies are designed to avoid.

Does capping FPS below my max hurt latency?

A small, well-chosen cap below your worst-case sustained frame rate typically helps more than it hurts, because it prevents the GPU from bouncing between power states and keeps frame time variance low. An uncapped frame rate that swings widely between busy and quiet scenes tends to feel less consistent than a steady, slightly lower number.

Do wireless mice add meaningful input lag in 2026?

Current-generation wireless gaming mice using a dedicated 2.4GHz dongle are generally considered latency-equivalent to wired at the same polling rate. The bigger risk is USB bandwidth contention if the dongle shares a hub with other high-traffic devices, which shows up as inconsistent jitter rather than a fixed added delay. Plugging the dongle into a rear motherboard port directly avoids most of this.