Measuring…
The verdict for this game appears once the first full second of packets is in.
Comparing with other tests to this server…
Measuring…
The verdict for this game appears once the first full second of packets is in.
Comparing with other tests to this server…
Are you sure you want to delete all results?
| Date & Time | Duration | Avg Ping | Avg Jitter | Packet Loss | Actions |
|---|
Modules that never cycled? Locked a target and nothing happened for a full second?
EVE Online is the slowest-ticking game in this catalogue and one of the least forgiving of a lossy line, and both facts have the same cause. The server resolves the universe once per second. Your client does not predict anything - it sends a command, waits for the tick, and displays what the server decided. That means a hundred milliseconds of extra latency costs you almost nothing, and a single dropped command costs you a whole tick of a fight that may only last twenty.
What this test sends: 1 packets per second of 150 bytes, matching EVE Online rather than a generic ping. verified
CCP's server runs a 1 Hz simulation tick; time dilation stretches that tick under load rather than dropping it.
Run the test above for at least 60 seconds. Here is what each number means for this game specifically.
There is no prediction layer to smooth over a lost packet. Your module activation, your target lock, your warp command - if the packet carrying it does not arrive, the server resolves the tick without it and you find out a second later that nothing happened. In a fight where a hardener cycle decides whether you survive an alpha strike, that is the whole engagement.
With a one second tick, the difference between 40 ms and 200 ms is the difference between using 4% and 20% of the window you have. Both fit. This is why EVE genuinely does support one worldwide server: a player in Australia and a player in Germany are resolving against the same tick, and neither of them is losing a tick to distance alone.
Jitter only hurts here when it pushes a command across a tick boundary. A command that would have made this tick lands in the next one instead, and the effect is identical to having clicked a full second later. If your ping line is spiky rather than high, that is what is happening to your commands.
What you see in game, and which number on this page explains it.
| What you see in EVE Online | Look at | What it means |
|---|---|---|
| Module clicked, nothing cycled | loss | The activation packet missed the tick and the server resolved without it. |
| Everything slow, TiDi percentage visible | server | Node load, not your line. Every pilot in the system sees it. |
| Dropped to login as 'socket closed' | loss | The connection ended. Sustained loss is the usual cause. |
| Commands land a beat late, ping looks fine | jitter | A late packet crossed a tick boundary and cost you a full second. |
EVE's server resolves the state of the universe once per second. Every module activation, every target lock, every warp command is collected during that second and applied at the boundary. There is no client-side prediction to hide the wait, which is why the game feels deliberate even on a perfect connection - you are genuinely waiting for a server decision rather than watching a guess that gets corrected.
That design is what makes a single worldwide shard possible. In a 128 Hz shooter, a tick is under eight milliseconds and a transatlantic route eats several ticks of it, so the game has to be split into regional servers. Here a tick is a thousand milliseconds and the whole planet fits inside one. The trade is that when something does go wrong, it goes wrong at the granularity of a whole second.
This inverts the usual advice about what to measure. In most games, latency and loss both degrade the experience gradually. In EVE, latency is absorbed almost entirely by the tick, and loss is not absorbed at all: a command that does not arrive is not applied late, it is simply not applied, and the next thing you can do about it is a second later. A connection with 200 ms of ping and no loss is genuinely better here than one with 30 ms and occasional drops.
Time dilation is worth understanding because it is frequently mistaken for a personal connection problem. When a system's node cannot process everything queued for a tick, CCP slows the clock rather than discarding commands - the simulation can run down to a small fraction of real time, and the UI shows the percentage. This is the server protecting fairness under load. If you can see a TiDi figure, your own line is not the story; if you cannot, and things are still going missing, this chart is where to look.
If the chart above shows packet loss or high jitter, work through these in order.
The Symptom: Everything crawls during a large fleet fight.
Time dilation is CCP slowing the simulation down on purpose so that commands queue rather than drop, and it is displayed in the UI as a percentage. If you can see a TiDi figure, the slowness is the server node and every other pilot in that system is experiencing the same thing. This test cannot fix that, and nothing on your end can - but it can tell you whether you also have a local problem stacked on top.
The Symptom: Commands go missing under fleet warp.
Wireless links drop packets in bursts, and a burst during a fleet warp is a ship left behind on grid. EVE sends very little data, so bandwidth is never the issue - what matters is that every small packet arrives, and a cable is the cheapest way to make that true.
The Symptom: You are dropped to the login screen mid-fight.
A socket closure means the connection between your client and the server ended, not that the server crashed. Sustained loss on this chart is the most common reason, and it usually shows up here minutes before it shows up as a disconnect.
The Symptom: Spikes appear when someone else uploads.
EVE's packets are tiny and time-critical, which is the worst combination behind a large upload: they wait their turn in the queue and miss the tick. Router QoS that prioritises small, latency-sensitive traffic removes that entirely and costs nothing on the download.
Much less than in almost any other online game. The server resolves once per second, so even 250 ms of round trip fits comfortably inside a tick. Packet loss is the metric that decides fights, because a lost command is not late - it never happened.
Time dilation is CCP deliberately slowing the simulation when a system's node is overloaded, so that thousands of queued commands are processed slowly instead of being dropped. It is server-side and affects everyone in the system equally. Nothing on your connection changes it.
Two different causes look identical from the client. A node under extreme load can end sessions, and so can sustained packet loss on your own line. Run this test during a quiet period: if it is clean, the fights are the server; if it is not, it is your connection and you can act on it.
No - that is the tick, not the geography. One shard is possible precisely because a one second tick absorbs intercontinental latency. The pace of the game is a design choice, and it is the same choice that lets everyone play in the same universe.