Your router has two bands. Use both.
Your laptop uses only one. We gave ours a Wi-Fi adapter for each band and sent live video over both with NanoPing. On its own, each band made the video wait now and then. With both, not one frame was late.
Five rounds, three ways to send the video
Each round sent the same minute of live video three times, one right after the other: over 2.4 GHz alone, over 5 GHz alone, then over both bands at once. All three went through NanoPing, so the only difference was the number of bands. On every line, lower is better.
All five rounds together; a check marks the best on the line. Every figure is what the receiving program saw: how long each video packet took from the sending program to the receiving one. The video has to wait in any tenth of a second in which a packet sent then took more than 100 ms. A frame is late if any of its packets is. No packet was lost.
Neither band was the safe one
2.4 GHz alone made the video wait in rounds 1, 3 and 4, and 5 GHz alone in rounds 2 and 5. Each band failed in rounds where the other was fine, so there was no good band to pick in advance. Over both bands, no round came near the limit.
Each strip is one minute of video. A bar is the slowest packet sent in each tenth of a second, from the sending program to the receiving one, red where it passes the dashed 100 ms line and cut off at 200 ms. Right: the slowest packet of that minute, and the frames it made late.
The bands kept stalling, just never at the same time
The other traffic did not stop when NanoPing used both bands. Each band still had moments when its packets took longer than 100 ms: eight in all, up to 254 ms. But in none of the 3,000 tenths of a second were both bands too slow at once, so the video always had a good band to go over.
Each band: every packet NanoPing sent over it, video and repair packets alike, and how long it took to cross that band. The video: how long its packets took from the sending program to the receiving one, as above.
Every time a band was too slow
| Band | Band's slowest | Video's slowest | |
|---|---|---|---|
| Round 1, 43.6 s in | 2.4 GHz | 114 ms | 4 ms |
| Round 2, 26.0 s in | 2.4 GHz | 144 ms | 11 ms |
| Round 2, 47.1 s in | 2.4 GHz | 123 ms | 5 ms |
| Round 3, 0.4 s in | 2.4 GHz | 133 ms | 15 ms |
| Round 3, 40.0 s in | 5 GHz | 254 ms | 35 ms |
| Round 5, 3.9 s in | 2.4 GHz | 107 ms | 14 ms |
| Round 5, 5.1 s in | 5 GHz | a packet lost | 22 ms |
| Round 5, 49.2 s in | 2.4 GHz | 107 ms | 24 ms |
A band counts as too slow in any tenth of a second in which a packet NanoPing sent over it took more than 100 ms or never arrived; moments less than half a second apart count once. Video's slowest: the slowest video packet sent during that moment.
When a band froze, the video moved to the other one
Round 3, 40 seconds in: 5 GHz froze, and packets sent over it took up to 254 ms to arrive. 31 ms later NanoPing had sent the 11 stuck packets again over 2.4 GHz and stopped putting new video on 5 GHz. The slowest video packet in that moment took 35 ms.
Swipe sideways to see the whole moment.
Top: every packet NanoPing sent over each band, by when it was sent and how long it took to cross that band. Middle: the slowest packet of each video frame, from the sending program to the receiving one. Bottom: which band carried each frame. Hover or tap the chart to read any frame.
Far faster at the slow end, a touch slower in the middle
The slowest packets decide whether live video stutters, and there both bands won by far: 999 of 1,000 packets arrived within 39 ms, against 98 ms over 2.4 GHz alone and 176 ms over 5 GHz alone. Typical packets were a little slower than over 5 GHz alone, 2.7 ms against 2.4 ms for half of them, because the video moves at the pace of the slower band.
All five rounds together, 219,900 video packets for each way of sending. Each bar runs from zero to the time within which that share of packets had arrived, from the sending program to the receiving one, red where it passes the limit.
Why the middle is slower
- NanoPing hands the video on in order, so a packet that crossed 5 GHz waits for the one before it on 2.4 GHz. Over both bands, half the video packets sent over 2.4 GHz arrived within 2.8 ms and half of those over 5 GHz within 2.7 ms, although 5 GHz itself was the faster band (0.9 ms to cross it for half the packets, against 2.0 ms).
- NanoPing sent 47% of the video over 2.4 GHz, so the slower band set the pace.
- In the three rounds where 5 GHz alone never made the video wait, 99 of 100 packets arrived within 8 to 16 ms over 5 GHz alone and 17 to 22 ms over both bands, well inside the limit either way.
All latency figures
| NanoPing, both bands | 2.4 GHz alone | 5 GHz alone | |
|---|---|---|---|
| Half of the packets within | 2.7 ms | 4.2 ms | 2.4 ms |
| 9 of 10 packets within | 7 ms | 12 ms | 8 ms |
| 99 of 100 packets within | 18 ms | 33 ms | 37 ms |
| 999 of 1,000 packets within | 39 ms | 98 ms | 176 ms |
| Slowest packet | 56 ms | 198 ms | 374 ms |
| In time at 100 ms | 100.00% | 99.90% | 99.68% |
One-way times of every video packet over the five rounds, from the sending program to the receiving one.
Round by round
| NanoPing, both bands | 2.4 GHz alone | 5 GHz alone | |
|---|---|---|---|
| Round 1 | 2.5 / 17 / 56 | 4.3 / 40 / 165 | 2.1 / 16 / 52 |
| Round 2 | 2.5 / 15 / 56 | 4.0 / 23 / 79 | 2.2 / 14 / 173 |
| Round 3 | 2.7 / 20 / 51 | 4.0 / 36 / 158 | 2.2 / 8 / 16 |
| Round 4 | 3.2 / 22 / 49 | 4.1 / 36 / 198 | 2.1 / 8 / 12 |
| Round 5 | 2.5 / 18 / 39 | 5.4 / 33 / 81 | 6.4 / 126 / 374 |
Milliseconds within which half and 99 of 100 video packets arrived, and the slowest; red where the slowest passed the limit.
Two bands, used together
A laptop's Wi-Fi joins one band of the router and stays on it, however busy it gets. This laptop had two adapters, one on each band, and NanoPing used both at the same time, leaning on whichever was good at that moment.
One adapter per band
A second USB Wi-Fi adapter let the laptop join the router's 2.4 GHz and 5 GHz networks at the same time. Both stayed connected the whole time, so there was never anything to switch.
It moved the video off a stalled band
When a band stalled, NanoPing stopped sending new video over it within 24 to 31 ms and carried on over the other one. Over the five rounds it split the video about evenly: 47% over 2.4 GHz and 53% over 5 GHz.
It sent what got stuck again
Packets caught in a stalled band were sent again over the other one: 704 packets sent again and 3,660 repair packets over the five rounds, so the video never had to wait for them.
How the test was run
One laptop was both ends of the stream. It sent the video over Wi-Fi to the router and got it back over its Ethernet port, through the router's own network. Two other computers on the same router streamed YouTube the whole time, one on each band.
The video
A live stream at 7 Mbit/s and 30 frames a second for 60 seconds, 43,980 packets each time. Each packet carries the time it was sent, and both ends run on the same laptop clock, so every packet's travel time is exact.
The router
One router with two radios, each its own network: 2.4 GHz on channel 6 and 5 GHz on channel 36. The laptop had one USB Wi-Fi adapter for each. Two other computers streamed YouTube on it the whole time, one on each band.
The rounds
Five rounds, 6 to 8 minutes apart, on 8 October 2026. Each round sent one minute of video over 2.4 GHz alone, then 5 GHz alone, then both, about 90 seconds apart and always in that order. All three went through NanoPing.
What is measured
Whether each packet arrived within 100 ms of being sent. A frame is late if any of its packets is. While NanoPing used both bands, every packet it sent over a band also shows what that band did at that moment.
Reading these numbers fairly
- One after the other, not side by side. The three tests of a round ran back to back, so the other traffic was not the same in each. That is why the tests over both bands count on their own: each band still stalled during them, and the video did not notice.
- The YouTube traffic was not recorded, so how hard it loaded each band, and when, is not known.
- One router behind both bands. They share its processor, its cable and its power, so a fault in the router itself would take both down at once. None happened here. Two separate access points would not share that.
- Five minutes for each way of sending, from one laptop standing still in one place.
The whole test as a PDF
Every number on this page, with the setup, every round and every stall in detail.
Measured 8 October 2026, five rounds of three 60-second tests, one after the other. Results reflect the router, adapters, other traffic and settings described on this page.
Reliable real-time connectivity in just a few steps
Every signup includes 30 days of full Pro access, no credit card. Spin up NanoPing on your device in just two commands and watch your traffic become reliable and stable right away.