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 |
|---|
Skills firing seconds after you pressed them? Dropped with a long colon-separated error code?
Guild Wars 2 behaves differently from most action games on a imperfect connection, and the reason is that its gameplay traffic runs over TCP rather than UDP. TCP does not tolerate a lost packet - it stops and retransmits it, and everything behind it in the stream waits. So where a UDP game degrades into a bit of rubber-banding, Guild Wars 2 tends to freeze for a moment and then catch up in a rush. That distinctive stutter is the shape of loss, not of a slow computer.
What this test sends: 25 packets per second of 100 bytes, matching Guild Wars 2 rather than a generic ping. estimated
ArenaNet publishes no tick rate; 25 Hz is a community estimate from packet captures. Note that the game's own gameplay traffic is TCP, so this is a rate, not a protocol match.
A note on what this test sends: this test uses UDP, and Guild Wars 2's gameplay traffic uses TCP. That is deliberate and it does not weaken the result - loss and jitter are properties of the path, and this measures the path directly instead of through TCP's retransmission, which hides loss by paying for it in time. If anything, loss you can see here costs you more in this game than the number suggests.
Run the test above for at least 60 seconds. Here is what each number means for this game specifically.
One percent loss here is not one percent of your gameplay. Every dropped packet stalls the stream until it has been retransmitted and acknowledged, so a single loss event costs you at least a round trip of frozen game state. This is why Guild Wars 2 players describe hitches rather than warping.
Your skill presses are queued and sent; the server confirms them. Uneven delivery makes the confirmation arrive in clumps, which is what produces the pattern of pressing three skills and watching all three resolve at once a beat later. That is not skill lag in the World versus World sense - it is your own line.
Under 80 ms is comfortable for open world and instanced content, and dodge timing in raids and fractals stays honest well past that. High ping alone rarely breaks Guild Wars 2 - it is the combination of high ping and loss, where every retransmission costs a full round trip, that turns unpleasant quickly.
What you see in game, and which number on this page explains it.
| What you see in Guild Wars 2 | Look at | What it means |
|---|---|---|
| Brief freeze, then everything catches up at once | loss | TCP retransmitted a lost packet and the stream waited for it. |
| Three skills resolve together a beat late | jitter | Confirmations arrived in a clump instead of evenly. |
| Seconds of skill delay in a WvW zerg only | server | Map-server congestion. Everyone in that fight has it. |
| Numeric error code and back to character select | loss | The session broke. Check whether loss was present at that moment. |
Almost every game in this catalogue sends its gameplay state over UDP, because UDP lets a late packet be thrown away. In a shooter that is the right trade: a position update from 200 ms ago is worthless, so there is no point retransmitting it. Guild Wars 2 made the other choice and runs its gameplay traffic over TCP, which guarantees that everything arrives, in order, no matter what it costs in time.
The consequence is that loss converts into latency rather than into missing information. When a packet is dropped, TCP detects the gap, requests it again, and holds everything that arrived afterwards until the gap is filled - a behaviour usually called head-of-line blocking. Your client has newer data sitting in its buffer that it is not allowed to use yet. That is the freeze, and the rush that follows is the buffer being released at once.
This is why the same one percent loss produces such different complaints across games. A Counter-Strike player at one percent describes shots that did not register; a Guild Wars 2 player at one percent describes the game hitching every few seconds. Both are correct, and both are describing the same line. It also means the usual reassurance that low single-digit loss is tolerable does not hold here.
Skill lag in World versus World is a genuinely separate problem and worth keeping apart from the above. It scales with the number of players fighting in one place, it affects everyone on that map identically, and it happens on connections with no loss at all. The practical rule is that anything you can reproduce alone in the open world belongs to your connection, and anything that only happens in a three-way zerg belongs to ArenaNet.
If the chart above shows packet loss or high jitter, work through these in order.
The Symptom: Skills take seconds to fire in a large WvW fight.
In World versus World, hundreds of players in one map queue more skill activations than the map server can resolve promptly, and everyone in the fight experiences the same delay. That is server-side and no local change touches it. The test for it is simple: if the same delay happens in the open world with nobody around, it is your line, and this chart will show it.
The Symptom: Regular brief freezes followed by a catch-up.
Wireless links lose packets in short bursts. Over UDP that is a moment of imprecision; over TCP it is a stall while the burst is retransmitted, which is exactly the freeze-then-rush pattern. A cable removes the most common source of it.
The Symptom: Disconnected with a long colon-separated number.
Guild Wars 2's network errors are reported as a chain of numbers rather than a message, and the chain identifies where the failure happened. Note it down together with the time, then check whether this chart showed loss at that moment. The two together tell you whether to look at your line or at ArenaNet's status.
The Symptom: Stalls line up with someone else's upload.
A TCP stream that has to retransmit is much more sensitive to a full upload queue than a UDP game is, because each retransmission has to survive the same queue again. Router QoS that keeps small interactive packets ahead of bulk traffic is worth more here than in most games.
Because its gameplay traffic runs over TCP. A lost packet is retransmitted rather than skipped, and the rest of the stream waits for it. The result is a short stall followed by a rush of catch-up, instead of the sliding and snapping you see in UDP games.
It is server-side congestion, not latency. When a World versus World map holds hundreds of players in one fight, the map server cannot resolve every skill activation promptly and everyone present sees the same delay. If you get the same delay alone in the open world, that is your connection instead.
It is the most common cause you can act on. Sustained loss eventually breaks the TCP session, which is what produces the numeric error and the return to character select. Run this test for a few minutes and see whether the red lines are there.
Rarely, and it can easily make things worse. A VPN adds a hop and cannot recover packets your own line dropped. It only helps in the narrow case of a genuinely bad route from your ISP, which is not what this test measures.