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 |
|---|
Sidesteps that whiff online and connect offline? Opponents teleporting out of your punish?
Tekken 8 runs rollback netcode, and it shows every player's connection type before the match - including a marker for anyone playing on a wireless link. That indicator exists because the community learned, over many years and many arguments, that wireless connections break this genre in a specific way that ping figures do not capture. Tekken's three-dimensional movement then makes the consequences more visible than they are in a 2D fighter.
What this test sends: 60 packets per second of 50 bytes, matching Tekken 8 rather than a generic ping. verified
Rollback netcode: one input exchange per 60 fps frame between the two peers.
Run the test above for at least 60 seconds. Here is what each number means for this game specifically.
Wireless links retry silently and lose packets in bursts. Rollback absorbs latency gracefully and bursts of loss badly, producing corrections large enough to move a character across the screen. The indicator is a warning about loss, not about ping - which is why a wireless opponent with a good ping figure can still be unplayable.
Tekken's sidestep depends on your position relative to an attack's tracking, evaluated on frames that may be rolled back. When corrections vary in size, a step that cleanly avoided a move gets re-simulated into one that did not. This is the most Tekken-specific symptom of an unstable line.
One frame is about 17 ms at sixty frames per second. Most of the game tolerates a few frames of prediction without complaint. Where it bites is the punish game: a move that is minus thirteen on block is a specific, learnable window, and reacting to it through several frames of prediction is a genuinely harder task.
What you see in game, and which number on this page explains it.
| What you see in Tekken 8 | Look at | What it means |
|---|---|---|
| Opponent teleports out of a punish | loss | A large rollback correction after their inputs arrived late. |
| Sidestep clips an attack it should have avoided | jitter | The frames deciding tracking were re-simulated differently. |
| Reaction punishes fail only online | ping | Frames of prediction added to an already tight window. |
| Wi-Fi marker on your own name | loss | You are the wireless player in the match. A cable is the fix. |
Fighting games are the hardest genre to network well, because they are the only one where a single frame reliably decides the outcome and where both players are acting continuously rather than taking turns. Rollback is the accepted answer: predict, simulate forward, and rewind when the prediction was wrong. It works because predictions are usually right, and it degrades gracefully as long as the truth arrives roughly on schedule.
Tekken adds an axis that makes the failures easier to see. In a 2D fighter, a mispredicted position is a character standing a little further left or right than expected. In Tekken, movement includes sidestepping into and out of an attack's tracking, so a correction does not merely move the character - it can change whether an attack was going to connect at all. The same size of network error produces a more consequential visual result.
This is the real content of the Wi-Fi indicator. It is not a statement that wireless is slower, because often it is not: a modern wireless link can show a perfectly respectable ping. It is a statement that wireless links retry, and that retries mean packets arriving in clumps or not at all, and that clumps are exactly what turns small rollbacks into large ones. The community arrived at this conclusion empirically long before the games started labelling it.
The practical upshot is that the three numbers on this page should be read in reverse of the usual order. Loss first, because it is what rollback cannot fix. Jitter second, because consistent prediction depth is what makes a match feel offline-like. Ping last, because a few frames of it is precisely the problem rollback was invented to solve, and it does that job well.
If the chart above shows packet loss or high jitter, work through these in order.
The Symptom: Matches against wireless players feel broken.
The indicator tells you what to expect from an opponent, and it also tells everyone else what to expect from you. If you are carrying that marker, the fix is not a setting - it is a cable. This is the one piece of advice the fighting game community is entirely united on.
The Symptom: Large corrections during combos and punishes.
A cable barely changes ping and dramatically changes loss, and loss is the metric rollback cannot conceal. In a genre where a dropped punish is the whole round, that trade is not close.
The Symptom: Your combos are reliable offline and not online.
Rollback re-simulates frames when it guessed wrong, which changes what you were reacting to mid-input. If your execution holds up in practice mode and falls apart in ranked, measure the line before you rebuild your muscle memory around a problem that is not yours.
The Symptom: Bad sets that coincide with someone else uploading.
Fighting game traffic is a handful of tiny packets per frame, which makes it the first thing a saturated upload delays. Router QoS that prioritises interactive traffic removes a cause that is easy to mistake for the opponent's connection.
Because wireless connections lose packets in bursts, and bursts of loss are the one thing rollback netcode cannot absorb smoothly. The indicator warns other players about a problem that a ping number would not reveal.
It makes ping far less important than it used to be, not irrelevant. Prediction hides a few frames very well. Reaction punishes are where the remaining frames are felt, and packet loss is where rollback stops helping entirely.
Sidestepping depends on your position relative to an attack's tracking at specific frames, and those frames can be rolled back and re-simulated. An unstable connection changes the size of those corrections, which turns a clean step into a clipped one.
A wired one first, and a stable one second. The absolute ping matters less than whether it is steady and whether packets arrive. Run this test for a few minutes and look at loss and jitter before you look at the ping number.