Cloud Gaming Latency Against a Local Machine
Streaming a game means the button and the picture are a round trip apart. We measured that trip with a camera and a photodiode on five different connections.
Tech Daily Review Desk
The shift
Cloud gaming latency fell far enough that the services stopped being a curiosity and started being sold as a replacement for hardware. The marketing now compares picture quality, which is the easy half; the half that decides whether a game feels right is the delay between the press and the pixel.
What matters is the whole chain, not the network portion of it: the controller's own polling, the trip out, the frame rendered somewhere else, the encode, the trip back, the decode, and the display. Most published figures measure one link in that chain, usually the network one, which is why they look so much better than the experience does. The honest measurement is click to photon.
What is at stake
Because latency is the one thing a faster connection cannot fix beyond a point — distance to the data centre is a physical floor — and because it is the difference between a game that plays and a game that merely runs.
Our method
A modified controller closing a circuit that also triggers an LED, filmed alongside the screen with a 1,000 fps camera, so every figure is the true interval between the button bottoming out and the first pixel changing. Two hundred presses per condition in the same game at the same point. The local machine is the same laptop running the same title natively, as the floor.
The measurements
| Connection | Median | 95th percentile | Jitter | Visible artefacts |
|---|---|---|---|---|
| Local machine, same laptop | 38 ms | 44 ms | 3 ms | none |
| Cloud, wired ethernet | 71 ms | 86 ms | 7 ms | rare |
| Cloud, Wi-Fi 6, same room | 84 ms | 109 ms | 12 ms | occasional |
| Cloud, Wi-Fi, evening peak | 112 ms | 188 ms | 31 ms | frequent |
| Cloud, 5G | 138 ms | 294 ms | 58 ms | frequent |
| Cloud, wired, different country | 164 ms | 201 ms | 9 ms | rare |
What the data shows
Even at its best, wired and nearby, the cloud path roughly doubled the local latency — 71 ms against 38 ms, which is perfectly playable and unmistakably different.
Jitter separates the conditions better than the median does: the evening figure and the 5G figure have similar medians and completely different feel, because the 95th percentile is where the game stutters.
Distance sets a floor nothing else can lift — the different-country test had excellent jitter and the worst median in the table.
Wi-Fi at peak hours was worse than 5G at its median and better at its tail, which is the opposite of what most people assume.
Where this could mislead you
Three things worth holding on to when you meet these figures somewhere else:
- A published latency figure that does not say click to photon is measuring one link of the chain, and the chain is what you feel
- Bandwidth is not the constraint above about 35 Mbps — past that, more speed buys picture quality and nothing else
- Controller connection is part of the budget: Bluetooth added 9 ms over a wired pad in our own measurements, on every condition equally
What we would buy
Use a cable and check where the nearest data centre is before you subscribe to anything. Those two facts decided more than everything else combined in our testing, and the second is the one you cannot change afterwards. If the nearest region is in another country, the floor is around 160 ms and no amount of money will move it.
Match the service to what you play. At 71 ms a slower game is indistinguishable from local and this is a genuinely good way to avoid buying hardware. Anything that depends on precise timing — a competitive shooter, a rhythm game, a hard platformer — is a different proposition, and the tail latency rather than the median is what will make it feel wrong.
Our verdict
Wired and nearby, the cloud costs about 33 ms over local and plays well. On Wi-Fi at peak, or over 5G, the median is survivable and the tail is what ruins it. Distance to the data centre is a floor you cannot buy your way past.
Behind this piece
Every product covered here was bought at retail and lived with for at least a fortnight before a word was written. Battery figures come from a scripted day rather than a manufacturer estimate, and sustained performance is measured warm — after half an hour of load — because that is the state a device spends most of its life in.
Where a firmware update changes a result, the piece is re-tested and the change noted rather than quietly corrected. No review unit arrives with conditions attached, and no score is agreed in advance with anybody.
Where to go next
If you are weighing this against the rest of the laptops we have covered, the category archive lists every review in date order. The closest comparisons we have published are Dell XPS 14 Review: Beautiful, Compromised and Dell XPS 14 Review: Good Idea, Uneven Execution.
For the reference material behind these figures, the specification pages carry the measured numbers, and the common questions collect what people ask once something is in daily use.
