NVIDIA Reflex – Better than AMD’s Anti-Lag?
|This week has been an amazing week to be a gamer, with both Sony and Microsoft announcing pricing for their next generation consoles, and NVIDIA launching the RTX 3080, to much fanfare. Hidden inside that launch though, is a new feature anyone with a GTX 900 series card or newer can make use of called NVIDIA Reflex. Much like AMD’s Radeon Anti-Lag, it’s meant to help reduce input lag – the time it takes from you doing something like clicking your mouse to it registering, and showing you the gun firing in game. So, is it any good? Is it better than AMD’s Anti-Lag? Lets find out. But first, if you haven’t already, consider subscribing for more videos every Monday, Wednesday and Friday!
Reducing the amount of input lag you have is really important for a responsive gaming experience. Playing on a projector with nearly 200ms of input lag makes it feel like you are moving your mouse through jelly, and makes it painfully difficult to hit any shots. Your current setup likely has something like 20-30ms of input lag, depending mostly on your monitor and for most people that’s a reasonable compromise, but NVIDIA aren’t satisfied.
They’ve introduced NVIDIA Reflex, a feature built into supported games, with a supported driver, to help reduce input lag. The games that are listed as getting support today are Valorant, Fortnite and Apex Legends, although as of my testing Apex didn’t have any options for it yet. NVIDIA’s approach here is pretty different from AMD’s, getting each game to support the feature rather than having a system-wide toggle and doing all the magic in the driver instead.
So, does it work? Well, if you have a 3080, or 2080ti, there isn’t much reason to enable it. In Fortnite I saw no appreciable difference between off, on and on + boost. Even the minimum and maximum’s are within margin of error of each other.
In Valorant it’s a little better, with a slight advantage of having On + Boost on, going from around 20ms to 18ms average, but that’s not game changing.
So it.. Doesn’t work? Well, there is a keyword NVIDIA uses, which is “in GPU bound scenarios”. It’s pretty rare a 2080ti is the main bottleneck, and even at 1440p and using a Ryzen 9 3900X can’t help that. So lets swap out the 2080ti for something like an RTX 2070 to see how it stacks up, especially since back when I tested Radeon Anti-Lag I used an RX 5700XT which is roughly equivalent.
In Fortnite, it makes a pretty massive difference. Going from 51ms average to 32ms with it on, that’s a pretty big deal. You’d see enemies 20ms faster, start reacting to them and be able to fire that much faster. It’s not visually noticeable, but it makes a difference in how well you play whether you know it or not.
In Valorant, it still doesn’t make any difference, which makes a lot of sense. In the training level I was using, it was running at something dumb like 600FPS with the 2080ti and not too much lower with the 2070 meaning it really isn’t a GPU bound scenario. It’ll make more of a difference in more demanding games, I imagine COD MW could see a decent benefit but sadly that’s not an option right now.
And that brings me nicely onto the downsides. Since NVIDIA is going with the “in game” approach, rather than AMD’s universal toggle, it relies on game developers implementing the feature, and doing it well. When I tested Anti-Lag, in CSGO, a similarly un-GPU bound game to Valorant, the RX 5700XT cut input lag by 30%, comparing that to 5% with the RTX 2070 in Valorant, it doesn’t seem quite as impressive.
Nevertheless, there doesn’t seem to be any downside to leaving this turned on anyway. There was no change in game performance, the average FPS was the same regardless, and the playing experience felt pretty similar across the board too so in the games that do get the feature I’d leave it on for the potential of lower input lag anyway. I personally prefer AMD’s approach here, but of course they aren’t directly comparable since you need a different GPU. Hopefully NVIDIA can improve this feature over time too, and might make for an interesting re-test video in 6 months to see what’s changed. If you want to see that, make sure you are subscribed and let me know in the comments too.