Anti-Lag 2 vs Reflex in CS2: how many milliseconds does each save?
About 1 to 2.5 ms for Anti-Lag 2 and about 1 ms for Reflex, in one German lab's tests at 300 fps and above. The two finish level. With the same 300 fps cap, the Radeon and the GeForce landed 0.1 ms apart. Turn on whichever one your graphics card has.
Anti-Lag 2 saves about 1 to 2.5 ms in CS2 and Reflex saves about 1 ms, in the only published side-by-side click-to-photon test we found. With the same 300 fps cap, an AMD and an NVIDIA card ended up 0.1 ms apart: 18.0 ms against 18.1 ms. So neither one is faster. Turn on the one your graphics card has: Anti-Lag 2 on a Radeon RX 5000 or newer, Reflex on a GeForce GTX 900 or newer.
What the two settings are
Both settings fight the same delay. A game's CPU work can run ahead of the graphics card, and finished frames then wait in line for the GPU in what is called the render queue. Each frame in that line was built from input that is already old by the time you see it. Our render queue explainer shows how big that wait can get.
NVIDIA Reflex is built into the game. NVIDIA's latency guide says it makes sure "your CPU is submitting work to the GPU just in time for it to be rendered". It works on GTX 900 series cards and newer, and its Boost mode, in PCGH's words, holds GPU clocks higher, among other things.
AMD Anti-Lag 2 is AMD's version of the same idea, also built into the game, while AMD's older Anti-Lag lives in the driver instead. AMD's SDK guide says driver Anti-Lag adds "a carefully calculated delay" inside the driver. Anti-Lag 2 "does a similar thing", but the delay sits inside the game, "just before the user controls (mouse/gamepad/keyboard) are sampled". That later sampling is why it can save more.
The number that matters is click-to-photon latency: the time from your click to the moment the screen changes. Our click-to-photon chain breaks it into stages.
One case worked through: a 300 fps cap at 4K
PC Games Hardware (PCGH) tested both settings in CS2. It used NVIDIA's LDAT, a light sensor on the screen that times each shot from click to muzzle flash, and each result is the average of about 600 shots. Both 4K tests used the same Intel Core i7-12700K, the same 120 Hz 4K monitor, the Very High preset with 4x MSAA, the Vertigo map and a fps_max 300 cap.
Walk the numbers through. The RX 7900 XTX started at 20.5 ms with nothing on, and Anti-Lag 2 brought it to 18.0 ms, a saving of 2.5 ms. The RTX 4080 Super started at 19.0 ms, and Reflex with Boost brought it to 18.1 ms, a saving of 0.9 ms.
Read only the savings and Anti-Lag 2 looks almost three times better. Read where each card finished and they are level. PCGH's own conclusion, in our translation, is that there is "no notable difference" despite 600 measuring points, and that AMD and NVIDIA "are level here". The Radeon saved more because it had more queue to remove: it started 1.5 ms higher.
Now put that in frames. At 300 fps one frame takes 3.33 ms, so a 2.5 ms saving is three quarters of a frame and a 0.9 ms saving is about a quarter of one. That is the scale in CS2 at high frame rates. The 10 to 40 ms you may have read about belongs to games running far slower, where each frame takes much longer.
Every result PCGH published
| Card and setup | Off | Driver Anti-Lag | Anti-Lag 2 or Reflex + Boost | Saved |
|---|---|---|---|---|
| RX 6700 XT, 1080p, 540 Hz, no cap (351 fps) | 12.7 ms | 12.2 ms | 11.6 ms (Anti-Lag 2) | 1.1 ms (8.7%) |
| RX 7900 XTX, 4K, 120 Hz, 300 fps cap | 20.5 ms | 18.7 ms | 18.0 ms (Anti-Lag 2) | 2.5 ms (12.2%) |
| RTX 4080 Super, 4K, 120 Hz, 300 fps cap | 19.0 ms | Not applicable | 18.1 ms (Reflex + Boost) | 0.9 ms (4.7%) |
| RX 7900 XTX, 4K, no cap | Not published | Not published | 16.4 ms (Anti-Lag 2) | Not published |
| RTX 4080 Super, 4K, no cap | Not published | Not applicable | 16.8 ms (Reflex + Boost) | Not published |
Two more things fall out of the table. First, Anti-Lag 2 beat AMD's own driver Anti-Lag on both Radeons, by 0.6 ms on the 6700 XT and 0.7 ms on the 7900 XTX. So in CS2, the in-game version is worth using over the driver toggle. Second, both cards were faster with no cap at all: the 7900 XTX ran about 326 fps uncapped, per PCGH, and dropped to 16.4 ms. Uncapped, PCGH found the Radeon "just under 0.4 ms" ahead of the GeForce. That gap is small and comes from two different cards, so do not read it as a ruling on the settings.
Why two answers circulate
AMD's 24.6.1 driver notes promise "a decrease in latency in Counter-Strike 2 of up to 40%" at 4K on an RX 7900 GRE, and up to 27% at 1080p on a Ryzen 7 8700G's built-in graphics. Those are AMD lab figures, from the Ancient benchmark at Very High in DirectX 11, and AMD's footnote compares Anti-Lag and Anti-Lag 2 both off against both on. PCGH's independent figures are 8.7% and 12.2%. Both can be true, because a percentage depends on how deep the queue was in that test, and "up to" names the best case.
The other mistake is comparing savings across brands, because a cross-GPU result is not one system with one setting flipped. The Radeon and the GeForce have different render times and different drivers, so they start from different places. The fair comparison is where each one finishes on the same CPU, screen and cap, and on that measure they tie.
Which setting to use
| Your card | Turn on | Why |
|---|---|---|
| Radeon RX 5000 series or newer (RDNA) | Anti-Lag 2, in CS2's Advanced Video menu | AMD lists CS2 as supported. It saved 1.1 and 2.5 ms in PCGH's tests, more than driver Anti-Lag. |
| Radeon older than RX 5000 (RX 400 series on) | Driver Anti-Lag | Anti-Lag 2 needs RDNA. Driver Anti-Lag supports RX 400 and newer. |
| GeForce GTX 900 series or newer | Reflex, On + Boost | NVIDIA's support floor. Boost is the mode PCGH measured, 0.9 ms saved. |
AMD's notes say Anti-Lag 2 "will be enabled by default" in CS2's "ADVANCED VIDEO" menu, so check it is still on rather than assuming. AMD's Anti-Lag page lists CS2, Dota 2 and Ghost of Tsushima Director's Cut as Anti-Lag 2 games, and says it "requires developer integration". Driver Anti-Lag covers DirectX 9, 11 and 12 on RX 400 cards and newer. AMD's GPUOpen page sets the Anti-Lag 2 floor at RDNA 1, the RX 5000 series, with driver 24.6.1 or newer for DirectX 11 games. To check the result on your own PC, follow our latency measuring guide.
The verdict
- Turn on whichever one your card has. Anti-Lag 2 on a Radeon RX 5000 or newer, Reflex on a GeForce.
- In CS2, expect about 1 to 2.5 ms from Anti-Lag 2 and about 1 ms from Reflex with Boost, at 300 fps and up.
- With the same 300 fps cap, the two finished 0.1 ms apart: 18.0 ms against 18.1 ms. Neither is faster.
- On a Radeon, use Anti-Lag 2 over driver Anti-Lag. It saved 0.6 to 0.7 ms more.
- AMD's "up to 40%" is a best case from AMD's lab, not what one independent test found.
What this does not cover
All the figures come from one lab, one CPU, one map and one CS2 build in June 2024, when Anti-Lag 2 was still a preview driver. Newer drivers and game updates may change them. Both settings work on the GPU-bound case, which is what AMD's GPUOpen page names. If your CPU limits your frame rate, expect less, as our guide to when these settings do nothing explains. PCGH did not publish Reflex On without Boost, so we give no number for it. The player bans of 2023 were tied to Anti-Lag+, a driver feature missing from AMD's software since then, per Club386. We found no AMD statement on driver Anti-Lag and anti-cheat. If you use a G-Sync or FreeSync cap below your refresh rate, the trade is different, and our G-Sync cap guide covers it.
Sources
- PC Games Hardware, Richard Engel, "AMD Radeon Anti-Lag 2" in Counter-Strike 2, page 1 (Radeon results), 4 June 2024. German.
- PC Games Hardware, page 2: NVIDIA Reflex compared (RTX 4080 Super results), 4 June 2024. German.
- AMD GPUOpen, AMD Radeon Anti-Lag 2 (requirements, supported APIs).
- AMD GPUOpen, Integrating the AMD Radeon Anti-Lag 2 SDK in your game (driver Anti-Lag against Anti-Lag 2).
- AMD, AMD Radeon Anti-Lag product page (supported games and hardware).
- AMD, AMD Software: Adrenalin Edition 24.6.1 release notes (40% and 27% claims, footnote RS-647, CS2 default).
- NVIDIA, System Latency Optimization Guide (Reflex and supported GPUs).
- Club386, Samuel Willetts, AMD launches Anti-Lag 2 with native support for Counter-Strike 2, 23 May 2024 (Anti-Lag+ removal).
Frequently asked questions
Is Anti-Lag 2 better than Reflex in CS2?
No. They end up in the same place. In PCGH's tests with a 300 fps cap at 4K, an RX 7900 XTX with Anti-Lag 2 measured 18.0 ms and an RTX 4080 Super with Reflex and Boost measured 18.1 ms. PCGH found no notable difference across about 600 measurements each. Anti-Lag 2 saved more (2.5 ms against 0.9 ms) because the Radeon started higher.
How much does Anti-Lag 2 reduce input lag in CS2?
In PCGH's tests, 1.1 ms on an RX 6700 XT at 1080p (12.7 to 11.6 ms) and 2.5 ms on an RX 7900 XTX at 4K with a 300 fps cap (20.5 to 18.0 ms). AMD's own figure is up to 40% on an RX 7900 GRE at 4K, from AMD's lab.
What is the difference between Anti-Lag and Anti-Lag 2?
Anti-Lag is a driver setting. It adds a delay inside the driver so the CPU and GPU line up. Anti-Lag 2 does the same job from inside the game, just before your mouse and keyboard are read. It has to be built into each game. In PCGH's CS2 tests, driver Anti-Lag saved 0.5 and 1.8 ms where Anti-Lag 2 saved 1.1 and 2.5 ms.
Which graphics cards support Anti-Lag 2?
AMD Radeon RX 5000 series and newer, meaning RDNA cards, on Windows 10 or 11, per AMD's GPUOpen page. Older Radeons from the RX 400 series on get only driver Anti-Lag. On NVIDIA, Reflex works on GTX 900 series cards and newer, per NVIDIA.
Should I use Reflex On or On + Boost in CS2?
PCGH measured On + Boost, which saved 0.9 ms at a 300 fps cap. It did not publish a figure for On alone. Boost keeps GPU clocks higher, per PCGH. So On + Boost is the setting with a measured result behind it.