Back to Top

Nvidia GeForce Now’s Competitive mode cuts input latency 30% lower than Stadia

Nvidia logo

It looks like ‘Netflix for games’ services, like Google Stadia and Nvidia GeForce Now, are here to stay. But there is one major problem that every game streaming provider must contend with: input lag. With every step taken from away from a local machine there’s the risk of increased latency. The good news is that some heroes from GameStar have tested and confirmed that GeForce Now’s ‘competitive’ option provides the lowest latency – even getting pretty close to the latency of a local connection.

GameStar tested the latency of the two game streaming services via a novel method involving a Makey-Makey board ($50) and Olympus TG-6 camera ($379). The results show not only that GeForce Now achieves less latency than Stadia, but also that it comes pretty close to achieving the low latency of an entirely local machine. Who would have guessed that Makey-Makey boards would be the tool to solidify Nvidia GeForce’s Now’s latency superiority?

GameStar used the Makey-Makey board for input and then set up the TG-6 to record at a high-speed 480 FPS. This allowed for latency testing by comparing the number of frames between input registry to the corresponding action’s on-screen occurrence. The Makey-Makey board makes for the most meticulous input measurement machine because it lights up red on input, making it easier to see exactly when input is registered.

The streaming services were tested on the following stream conditions:

GeForce Now Google Stadia
Balanced (1080p, 60 FPS, V-Sync on) Balanced: (1080p, 60 FPS, V-Sync not specified)
Competitive: (720p, 120 FPS, V-Sync off) Limited data usage: (720p, 60 FPS, V-Sync not specified)

GeForce Now’s ‘competitive’ option set seems to be geared more towards those gamers wanting a competitive edge in games (thus the name). On the face of it this seems odd, because game streaming services like GeForce Now aren’t really targeted at competitive gamers – I’m not entirely sure the market is currently there for this streaming feature. However, once people see the competitive option set’s latency results, maybe this will change.

Latency was tested in both Destiny 2 and Metro: Exodus using each of the aforementioned option sets for GeForce Now and Google Stadia, as well as on a local PC. The results, averaged across 10 measurements, are as follows:

Destiny 2 latency Metro: Exodus latency
Local PC 39.5ms 41.3ms
Google Stadia ‘balanced’ 95.4ms 100.5ms
Google Stadia ‘restricted data usage’ 93.4ms 97.1ms
GeForce Now ‘balanced’ 97.8ms 81.1ms
GeForce Now ‘competitive’ 69ms 68.6ms

From this, we see that GeForce Now can get you much closer to the latency of a local PC than either Google Stadia’s ‘balanced’ or ‘restricted data usage’ option sets. It’s not quite 40ms, but it’s certainly not 100ms, and that’s quite an achievement. Also of note is that there doesn’t appear to be much difference at all in latency between Google Stadia’s two option sets. It might restrict data usage, but Stadia’s reduced settings stream conditions don’t quite hit the latency mark like GeForce Now’s do.

This isn’t good news for Stadia, a streaming service that performs technically worse than GeForce Now. It’s also not quite the ‘negative latency’ Google’s been targeting. But it’s not all that straightforward. GameStar points out that, during testing, there were “quite high fluctuations” in latency, suggesting that even GeForce Now’s stellar results might not have been consistently stellar.

Game streaming services like Nvidia GeForce Now and Google Stadia haven’t been quite the flop that many of you cynical readers (and I myself) may have been expecting. In fact, GeForce Now has had over 1 million sign-ups since launch – this despite both Blizzard and Bethesda pulling many of their games from the game streaming platform. There’s definitely promise for this level of support continuing, providing these services continue to work on input latency issues.

These latency issues are a massive hill for any game streaming service to climb. And frankly, it’s a mountain that might be unscalable, at least until data transfer and encoding speeds improve. It can’t help but feel like we’re reaching for next-gen services on current-gen technology. We can’t feel too bad for Nvidia and Google, though – it was a hill they knew they’d have to climb from the start.

Back to Navigation