Far from being a beneficial function to balance performance and battery life, the framerate limiter inside the Steam Deck reportedly increases the input latency substantially. A Redditor accidentally stumbled upon the framerate limiter's malfunction while testing the latency between his wireless PlayStation 5 controller and docked Steam Deck. Other Redditors have pitched in on the testing and confirmed the issue.
The amount of input latency scales linearly according to the frame rate and refresh rate cap. For reference, the uncapped input latency is 31.8ms. Enforcing a 30 FPS cap increased to 145.9ms, more than 4X higher. According to the Redditor's results, the worst configuration was the 40 Hz, 20 FPS setting, causing the input latency to skyrocket to 232ms, over 7X more than the uncapped setting. The two settings with the minor input latency penalty are the 50 Hz or 40 Hz uncapped scenario, where the increase was less than 8%.
The Redditor conducted the tests in Rogue Legacy 2 at the settings menu with the Steam Deck undocked. His methodology included using an iOS application called "Is It Snappy?" that essentially captures video at 240 FPS and allows users to set a starting point and endpoint pin to calculate the difference in milliseconds. The Redditor established the starting point when he fully pressed down on the button, and the endpoint was when the screen changed. The user highlighted a margin of error of 4ms, so he averaged the five individual timings from his test runs.
In his experiment with Rocket League, the Redditor discovered a minimal to zero increase in input latency. Rocket League is one of those few titles with a built-in framerate limiter (not to be confused with V-Sync). For this experiment, the Redditor configured the target frame rate in Rocket League and the Steam Deck's refresh rate to 50 Hz.
Until Valve issues a fix for the Steam Deck's framerate limiter, the best course of action is to use the game's built-in framerate limiter to match your desired refresh rate on the Steam Deck or leave the latter uncapped. The problem is that not every title has an in-game framerate limiter. Furthermore, the Redditor highlighted he still experienced screen tearing and judder despite using Rocket League's internal framerate limiter.
The Redditor praised the Steam Deck's button-to-photon latency compared to other consoles in other news. The user recorded values between 70ms to 86ms for the Nintendo Switch a year ago, with latency performance similar to the PlayStation 5 or Xbox Series X. However, the Steam Deck (31.8ms) flaunts excellent input latency when uncapped. The only good news is that Valve has delayed Steam Deck pre-orders to after Q3, so it's more than enough time for Valve to fix the issue before the consoles get into gamers' hands. Unfortunately, existing Steam Deck owners will have to use the workaround until an official patch arrives.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Zhiye Liu is a news editor, memory reviewer, and SSD tester at Tom’s Hardware. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.
I get that as web "journalists" you get paid per word and click. But is this the best this article could offer? A blow by blow review of a Reddit post?
Did TH verify these findings themselves? Did you confirm them and then find that they were 'egregious' rendering the Steam Deck basically unplayable with these settings? Are they 'egregious' for all games available on the Steam Deck or just for highly competitive games? Will I notice if I'm emulating games or if I'm playing slower paced titles?
If the article isn't going to add context to what you are reporting, or demonstrate the expertise and perspective of staff at TH, then you might as well just link to the Reddit post and call it a day. No need to basically copy the post while adding nothing else of value.
Enforcing a 30 FPS cap increased to 145.9ms, more than 4X higher. According to the Redditor's results, the worst configuration was the 40 Hz, 20 FPS setting, causing the input latency to skyrocket to 232ms, over 7X more than the uncapped setting.
I somehow doubt that someone locking their frame rate to 20fps is going to care all that much about a 200ms increase in input latency. Most games would be uncomfortable to play at that frame rate either way.
As for the 60fps cap, the increased latency was still on par with their results for the Switch, so I don't see much problem there. And even latency at the 30fps cap is arguably not that bad, at least for the kinds of games one would play at 30fps.
This is a power-saving feature, so it's probably a bit of a stretch to call it a "malfunction", if it is in fact saving power when on battery. I suppose they could decouple the input sampling from the frame rate though, assuming that doesn't affect the power-saving capabilities much. Or add a reduced-latency power-saving option. I imagine the power saving feature is just inserting a delay between each frame, and the input may be getting averaged over the course of several frames.
