NVIDIA Low Latency Mode: Ultra, On, or Off?
Published
On this page
NVIDIA Low Latency Mode (NULL) limits how many frames the CPU queues ahead of the GPU. The short answer: if a game supports NVIDIA Reflex, leave Low Latency Mode Off and use Reflex — it’s better. For older DirectX 9–11 games without Reflex, set Low Latency Mode to Ultra when you’re GPU-bound. It does nothing in DX12/Vulkan titles, where in-engine Reflex is the only real lever.
Low Latency Mode is the driver-level fallback — useful, but Reflex beats it wherever it’s offered.
What Low Latency Mode does
By default the CPU can prepare several frames ahead of the GPU, forming a render queue. That queue is latency: your input is captured into a frame that then waits its turn. Low Latency Mode reins in that queue from the driver side:
- Off — default queue (up to several frames).
- On — limits the queue to one frame.
- Ultra — goes further, timing frame submission so the CPU starts each frame “just in time,” minimizing how long it sits before the GPU.
It only does meaningful work when you are GPU-bound, because that is the only time the queue is deep enough to matter.
Off, On, or Ultra — the decision
| Setting | Use when |
|---|---|
| Off | The game supports Reflex (use Reflex instead), or you’re CPU-bound |
| On | A mild reduction in a DX9–11 game without Reflex |
| Ultra | A DX9–11 game without Reflex, when you’re GPU-bound — the common competitive pick |
For most players the rule is simple: Reflex game → Off (use Reflex). Non-Reflex DX9–11 game, GPU-bound → Ultra.
The big limitation: DX12 and Vulkan
Low Latency Mode only functions in DirectX 9 through 11 games. In DirectX 12 and Vulkan, the game engine controls frame queuing directly, so the driver can’t manage it — the setting simply does nothing. Since most modern competitive titles are DX12/Vulkan, this is exactly why NVIDIA built Reflex into game engines. In those games, the in-engine Reflex option is your only effective queue control.
Why Reflex beats it when available
Both target the render queue, but they operate at different levels:
- Low Latency Mode acts from outside the game, in the driver — coarser control, DX9–11 only.
- Reflex acts inside the engine — finer control, also dynamically caps FPS below your G-Sync ceiling, and works in modern APIs.
Where a game offers Reflex, turn Reflex on and set Low Latency Mode to Off so they don’t conflict. Reflex is simply the more precise, more capable tool.
Don’t forget the FPS cap
Low Latency Mode does not cap your frame rate. In a game without Reflex, you still want an FPS cap a few frames below your refresh rate to stay inside the G-Sync range and avoid V-Sync latency. Low Latency Mode (Ultra) plus a manual cap is the non-Reflex version of what Reflex does automatically. This pairs directly with the G-Sync + V-Sync + Reflex low-latency setup — substitute Ultra + manual cap when Reflex isn’t available.
Set it and verify
Open the NVIDIA Control Panel → Manage 3D Settings, set Low Latency Mode per the table above (globally or per-game), and pair it with a manual FPS cap in non-Reflex titles. Then confirm rather than assume — measure your input lag with the setting On/Ultra/Off in the same GPU-bound scene. And don’t neglect the rest of the chain: a fine Windows timer resolution via Tier1Timer keeps frame pacing steady underneath it all.
Related guides
- NVIDIA Reflex: Should You Turn It On?
- G-Sync + V-Sync + Reflex: The Lowest-Latency Setup
- Best FPS Cap for Low Latency
- How to Minimize Input Delay for Competitive Gaming
- How to Measure Input Lag
Use NVIDIA Low Latency Mode as the fallback it was designed to be: Ultra in older DX9–11 games without Reflex when you’re GPU-bound, Off everywhere Reflex exists. It can’t touch DX12/Vulkan titles, so in modern competitive games Reflex remains the real tool — and a clean FPS cap, G-Sync setup, and timer resolution finish the job.
Frequently asked questions
What is NVIDIA Low Latency Mode (NULL)?
NVIDIA Low Latency Mode, sometimes called NULL, is a setting in the NVIDIA Control Panel that limits how many frames the CPU can queue ahead of the GPU. Off allows the default queue, On limits it to one frame, and Ultra goes further by timing frame submission to start just in time. It is a driver-level way to reduce render-queue latency in games that don't have NVIDIA Reflex.
Should I set Low Latency Mode to On or Ultra?
If a game has no Reflex support and you are GPU-bound, Ultra usually gives the lowest latency and is the common competitive choice. On is a milder version that limits the queue to one frame. If a game supports Reflex, leave Low Latency Mode Off and use Reflex instead, because the two can conflict and Reflex is more effective.
How is Low Latency Mode different from NVIDIA Reflex?
Low Latency Mode works at the driver level and can only manage the frame queue from outside the game. Reflex is integrated inside the game engine, so it has more precise control and reduces latency more effectively. Reflex is the better tool when available; Low Latency Mode is the fallback for games without it.
Does Low Latency Mode work in all games?
It only works in DirectX 9 through 11 titles, and only helps when you are GPU-bound. It does not apply to DirectX 12 or Vulkan games, where the engine controls frame queuing directly — those titles need in-engine Reflex instead. So its usefulness is narrower than it looks, which is another reason Reflex is preferred where offered.
Does Low Latency Mode help if I'm not GPU-bound?
Not much. Like Reflex, it targets the render queue, which is only deep when the GPU is the bottleneck (near 100% usage). If you're CPU-bound or capped well below your GPU's ceiling, there's little queue to trim, so the setting does almost nothing. Pairing it with an FPS cap below your refresh rate is what keeps latency low in those cases.