Skip to content

Wi-Fi performance

All wireless adapters were tested under consistent conditions - each positioned in close proximity (1-2m) and connected to the same wireless access point (AP). The adapters utilized various interface types, including USB, SDIO, and PCI, to evaluate performance across different hardware configurations.

Results

Measured 2026-07-01 11:47 UTC

48 wireless link(s) · ✅ 48 pass · 📶 peak ↑ 681 / ↓ 868 Mbps

Wi-Fi 7 · 802.11be

Chip Attach Channel ↑ Up (Mbps) ↓ Down (Mbps)
MediaTek MT7925 PCIe 6215 MHz @ 160 681 868
Qualcomm WCN7851 PCIe 5240 MHz @ 160 232 784

Wi-Fi 6 · 802.11ax

Chip Attach Channel ↑ Up (Mbps) ↓ Down (Mbps)
MediaTek MT7921 USB 5240 MHz @ 80 646 350
Realtek RTL8852BE PCIe 5240 MHz @ 80 199 609
Realtek RTL8852BE PCIe 5240 MHz @ 80 348 450
Realtek RTL8852AE PCIe 5240 MHz @ 80 332 416
Realtek RTL8852BS SDIO 5240 MHz @ 80 352 394

Wi-Fi 5 · 802.11ac

Chip Attach Channel ↑ Up (Mbps) ↓ Down (Mbps)
Realtek RTL8812AU USB 5240 MHz @ 80 200 271
Realtek RTL8822BU USB 5240 MHz @ 80 265 132
Realtek RTL8821CU USB 5240 MHz @ 80 259 241
Broadcom (brcmfmac) SDIO 5240 MHz @ 80 247 182
MediaTek MT7610 USB 5240 MHz @ 80 131 245
Broadcom BCM4356 SDIO 5240 MHz @ 40 122 209
Unisoc UWE5622 — — 132 93
Realtek RTL8822CS SDIO 5240 MHz @ 80 119 103
Broadcom (brcmfmac) SDIO 5240 MHz @ 80 19 108
Broadcom (brcmfmac) SDIO 5240 MHz @ 40 105 97
Realtek RTL8822CS SDIO 5240 MHz @ 80 44 88
Realtek RTL8821CU USB 2437 MHz @ 20 45 36
Unisoc (unisoc_wifi) — — 44 22
Realtek RTL8821CU USB 2437 MHz @ 20 28 43
Broadcom (brcmfmac) SDIO 5240 MHz @ 80 39 42
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 29 37
Realtek RTL8822CS SDIO 2437 MHz @ 20 15 28
Broadcom (brcmfmac) SDIO 5240 MHz @ 40 25 22
Broadcom (brcmfmac) SDIO 5240 MHz @ 40 6 22

Wi-Fi 4 · 802.11n

Chip Attach Channel ↑ Up (Mbps) ↓ Down (Mbps)
Ralink RT5572 USB 5240 MHz @ 40 56 93
Ralink RT5572 USB 5240 MHz @ 40 53 73
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 33 44
Ralink RT5370 USB 2437 MHz @ 20 40 25
Realtek RTL8192CU USB 2437 MHz @ 20 37 27
Realtek RTL8189FS SDIO — 35 26
Ralink RT5370 USB 2437 MHz @ 20 34 18
Realtek RTL8189FS SDIO — 33 27
Realtek RTL8189ES SDIO — 29 29
Ralink RT2870/RT3070 USB 2437 MHz @ 20 28 10
Realtek RTL8189ES SDIO — 26 25
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 17 24
XRadio XR819 SDIO 2437 MHz @ 20 12 23
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 21 22
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 15 22
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 14 21
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 16 19
XRadio XR819 SDIO 2437 MHz @ 20 5 17
Realtek RTL8723BU 802.11b/g/n USB 2437 MHz @ 20 13 17
Broadcom (brcmfmac) SDIO 2437 MHz @ 20 13 15
XRadio XR819 SDIO 2437 MHz @ 20 1 14
MediaTek MT7601 USB 2437 MHz @ 20 12 13

Test Equipment

All adapters connect to a single common access point — a Zyxel NWA130BE (Wi-Fi 7) — so throughput is comparable across every board. The boards under test and their radios are listed in the results above.

Software and Infrastructure

NetBox is the lab’s source of truth — every board, its capabilities and its wiring. The Armbian autotests framework reads that inventory and drives each board over self-hosted GitHub Actions, power included via armbian/infra.

Methodology

Wireless throughput is one step of the per-board Armbian autotests pipeline. For each board:

  1. Power on & flash — the board is powered up and, where the hardware allows, flashed with a clean Armbian image; boards that can’t take a clean flash are tested in place.
  2. Provision & connect — first login, a stable lab IP, and the Wi-Fi association to the common access point (SSID) are configured, then connectivity is verified.
  3. Measure with iperf3 — the board runs iperf3 as the client, source-bound to each interface, against the lab’s central server in both directions: upload (board → server) and download (-R, server → board). Goodput is recorded in Mbit/s.
  4. Enrich from NetBox — each result is annotated with the interface type, Wi-Fi chip, and channel frequency/width, so every number carries the hardware that produced it.
  5. Restore & power down — the original wired configuration is restored, results are uploaded, and the board is powered off.

Results are published as a time series, so a regression shows up the moment throughput drops on a given board.

Contribute

  • Assist us in developing and maintaining our testing system: Your expertise can help us enhance and optimize our test infrastructure. By contributing your skills, you can play a key role in ensuring the accuracy and reliability of our test results.
  • Donate hardware: Your contribution of new hardware, whether it’s a wireless adapter or any other equipment, helps us expand our testing capabilities. We’re specifically looking for new wireless adapters that haven’t yet been added to our system. Your donation can directly impact the scope and depth of our tests.

Other resources