Quick answer

Turn HAGS on for any RTX 30-series or RX 6000-series card and newer; Windows 11 defaults to it and current drivers assume it. It is mandatory for DLSS Frame Generation, which will not activate without it. Frame-rate impact is neutral to slightly positive — the real gain is lower CPU scheduling overhead.

TL;DR · HAGS on or off

  • Modern GPU (RTX 30/40/50, RX 6000+): turn it ON. Current drivers assume it, and it is the default on Windows 11.
  • It is required for frame generation. NVIDIA DLSS Frame Generation will not activate with HAGS off, and AMD's frame-gen path expects it on supported hardware.
  • FPS impact is usually neutral to slightly positive -- the real win is less CPU scheduling overhead and lower latency in some titles, not a big frame-rate jump.
  • Turn it OFF only if you see a regression: documented cases include micro-stutter, high DPC latency, and some VR hitching that clear up when it is disabled.
  • The toggle: Settings > System > Display > Graphics > Change default graphics settings. Restart after.

The short answer for almost everyone reading this: if you are on a modern graphics card, turn hardware-accelerated GPU scheduling on and leave it on. It is the default on Windows 11 for a reason, and it is a hard requirement for the frame generation features people most want from RTX 40/50 and recent Radeon cards. The longer answer – why it exists, what it changes, and the handful of real cases where off still wins – is what decides whether you ever need to touch the switch.

What hardware-accelerated GPU scheduling actually changes

“HAGS” sounds like a performance toggle, but it is really a change to who schedules work onto your GPU. Microsoft introduced it as part of the Windows graphics driver model, and the mechanism is specific.

Before HAGS, Windows scheduled GPU work using a high-priority thread running on your CPU. That thread collected render commands, batched them, and fed them to the GPU. With hardware-accelerated GPU scheduling enabled, Windows “can now offload most of GPU scheduling to a dedicated GPU-based scheduling processor,” as Microsoft’s DirectX team describes it. The GPU’s own scheduling processor takes over the high-frequency work – quanta management and context switching between GPU engines – using its own memory (VRAM) to manage those queues in batches. Windows still decides which application gets priority; only the moment-to-moment scheduling moves down onto the silicon.

Microsoft is blunt that this is “a significant and fundamental change to the driver model,” which is exactly why it shipped as an opt-in feature that was disabled by default, and why it needs a WDDM 2.7 or newer driver plus a GPU built to support it. It is not a tweak layered on top of the old path; it is a different path.

The practical consequence: with the CPU no longer running that scheduling thread, you free up some CPU overhead and can shave latency in the games where that overhead was a factor. That is the upside. It is also why the gains are modest and uneven rather than a universal FPS boost – more on that below.

The honest FPS expectation: neutral-to-positive, not a magic button

This is where a lot of guides oversell it. Hardware-accelerated GPU scheduling does not reliably hand you a big frame-rate increase. The realistic expectation is neutral to slightly positive on average FPS, with the meaningful benefit being reduced CPU scheduling overhead.

That framing tells you where it helps most: CPU-bound situations. If your processor is the limiter, trimming the scheduling work it has to do is more likely to matter. If you are GPU-bound – pushing high resolution and settings where the graphics card is pegged – moving the scheduler off the CPU changes very little, because the CPU was not the bottleneck to begin with. If you are not sure which side you are on, the PC Bottleneck Calculator estimates whether your CPU or GPU is the limiter at your resolution, and that is the variable that decides whether HAGS does anything you can feel.

So the case for turning it on is not “more frames.” It is “modern drivers are built around it, it costs you nothing on average, and it unlocks the feature below.”

The real reason it is non-negotiable: frame generation

Here is the part that turns “probably leave it on” into “leave it on.” NVIDIA’s DLSS Frame Generation requires hardware-accelerated GPU scheduling to be enabled at the operating-system level. This is not a community rumor – it is stated in official game-vendor documentation. CD PROJEKT RED’s support page for DLSS Frame Generation says plainly that “DLSS Frame Generation requires a GeForce RTX 40 Series graphics card and Hardware-Accelerated GPU Scheduling being enabled in operating system,” and walks players through enabling it (CD PROJEKT RED support).

The reason is the mechanism from the first section: frame generation needs the GPU to have direct control over its own scheduling and memory to insert generated frames into the pipeline. If Windows is still running the old CPU-side scheduler, that control is not there, and the feature simply stays unavailable. On Windows 11 this is usually a non-issue because the setting is on by default; on Windows 10 you may have to enable it yourself before Frame Generation appears.

AMD’s frame generation works differently under the hood – its FSR 3 / driver-level Fluid Motion Frames path runs on the GPU’s shaders via asynchronous compute rather than a dedicated optical-flow hardware block like NVIDIA’s RTX 40 cards use, per AMD’s GPUOpen documentation. The takeaway is the same in practice: on a supported modern Radeon, you want HAGS on and your driver current. If you want the full picture of how interpolated frames, latency, and the base-framerate floor interact, that is the subject of frame generation explained – this article is just the OS switch that lets any of it run.

When OFF can actually win

A blanket “always on” would be dishonest, because a minority of setups regress with it enabled. These are real, documented cases – not invented numbers – and they share a pattern: the issue appears with HAGS on and clears when you turn it off.

  • Micro-stutter. Some players report frame-pacing micro-stutter that disappears the moment they disable HAGS, with the game running smooth afterward (ASUS ROG forum thread).
  • High DPC latency. On some Windows 11 systems, high DPC latency tied to the NVIDIA display driver (nvlddmkm.sys) drops sharply after turning HAGS off (NVIDIA GeForce forums). High DPC latency is a classic source of audio crackle and stutter, so this one is worth checking if you have those symptoms.
  • VR. Disabling HAGS is a recommended troubleshooting step for SteamVR stutter on some configurations (Steam community). VR is latency- and pacing-sensitive in a way flat gaming is not, which is why it shows up disproportionately here.
  • Older or weaker GPUs, and some Unreal Engine titles. If your card barely meets the support bar, or a specific game’s engine paces worse with it on, the simpler CPU-scheduled path can feel steadier.

The common thread: none of these are reasons to default to off. They are reasons to test off if you are seeing a specific problem – and if you do not use frame generation, you have nothing to lose by trying it.

A simple decision

Your situationSetting
Modern GPU (RTX 30/40/50, RX 6000+), no issuesOn (it is the default)
You want DLSS / AMD frame generationOn (required)
Micro-stutter, audio crackle, or high DPC latencyTest Off – keep whichever is smoother
VR stutter you cannot otherwise fixTest Off
Older / barely-supported GPU that feels hitchyTest Off

How to change it (and verify)

The toggle lives at Settings > System > Display > Graphics > Change default graphics settings, where you flip the Hardware-accelerated GPU scheduling switch (Microsoft Learn). If you do not see the option at all, your GPU either does not support it or your driver is older than WDDM 2.7 – install the latest driver straight from NVIDIA, AMD, or Intel and check again. Restart after changing it so the display driver reloads cleanly.

To judge whether it helped or hurt your machine, do not trust the average FPS counter. Run a demanding, repeatable scene with a frame-time overlay and compare your 1% lows and frame-time consistency on and off. As with the DX12 vs DX11 question, smoothness lives in the lows, not the average – and if a single game stutters under HAGS, rule out the boring causes first, like a launcher or anti-cheat fault, before blaming the scheduler.

The bottom line

Hardware-accelerated GPU scheduling is best understood as plumbing, not a performance cheat – the same way Resizable BAR / Smart Access Memory is a small, legitimate gain rather than the free-FPS switch it gets sold as. On a modern GPU it belongs on: it is the Windows 11 default, it moves scheduling overhead off your CPU for a neutral-to-small FPS effect, and – the deciding factor – it is a hard requirement for NVIDIA DLSS Frame Generation and the AMD frame-gen path most people on current cards actually want. Turn it off only when you have a concrete regression in front of you: stutter, DPC latency, or VR hitching that demonstrably clears when you disable it. Otherwise, leave the switch where Windows put it and spend your tuning time somewhere it pays off more.

Sources

Frequently asked questions

Should I turn hardware-accelerated GPU scheduling on or off?
On a modern card – NVIDIA RTX 30/40/50 or AMD RX 6000 and newer – leave it on. Current drivers and features assume it, and NVIDIA DLSS Frame Generation will not even activate without it. Its effect on average FPS is usually neutral to slightly positive while trimming CPU scheduling overhead. Turn it off only if you actually see a regression: stutter, high DPC latency, or VR hitching that goes away when you disable it.
Does DLSS Frame Generation require hardware-accelerated GPU scheduling?
Yes. Official game-vendor documentation for DLSS Frame Generation states it requires a GeForce RTX 40 Series card and Hardware-Accelerated GPU Scheduling enabled in the operating system. Frame Generation hands the GPU control of its own scheduling, so the OS-level setting has to be on or the feature stays greyed out. On Windows 11 the setting is on by default; on Windows 10 you may need to enable it yourself.
What does hardware-accelerated GPU scheduling actually do?
Before HAGS, Windows used a high-priority CPU thread to schedule work onto the GPU. With HAGS on, most of that scheduling moves to a dedicated processor on the GPU itself, which manages its own queues and VRAM in batches. Microsoft calls it a fundamental change to the driver model. The practical upside is less CPU scheduling overhead and lower latency in some titles.
Does hardware-accelerated GPU scheduling increase FPS?
Not dramatically, and not in every game. The honest expectation is neutral-to-small. It reduces the CPU’s scheduling workload, so the biggest gains show up in CPU-bound situations where that overhead mattered. When you are GPU-bound, the setting barely moves the average frame rate. Treat it as a plumbing change that enables features, not a free FPS button.
Why would I ever turn hardware-accelerated GPU scheduling off?
A minority of setups regress with it on. Players have reported micro-stutter that disappears when HAGS is disabled, high DPC latency tied to the display driver dropping after turning it off, and SteamVR stutter that improves with it disabled. Some Unreal Engine titles and older or weaker GPUs also feel steadier without it. If you do not use frame generation and you see hitching, toggling it off is a legitimate test.
Where is the hardware-accelerated GPU scheduling toggle in Windows 11?
Settings > System > Display > Graphics > Change default graphics settings, then flip the Hardware-accelerated GPU scheduling switch. The toggle only appears if your GPU supports it and you have a WDDM 2.7 or newer driver, so install the latest driver from NVIDIA, AMD, or Intel first. Restart after changing it so the driver reloads cleanly.

Evidence ledger

Last updated
Methodology
This guide was written and edited by Lowell K. Wood IV in St. Louis County, MO. Specs and prices verified against vendor and project documentation current on the date above. Full editorial standard: methodology.
Update log
  • 2026-07-25 — Last reviewed and updated.
Corrections
Spotted an error or a stale number? Email hello@techfuelhq.com. Confirmed corrections are added to the update log above.

About the author

Written by Lowell K. Wood IV, who builds and runs TechFuelHQ from St. Louis, Missouri.