Because GWifi continues to operate without an internet connection, I was able to run our Version 4 router performance tests on it with 8872.40.9 firmware. Table 2 summarizes routing test results; this Excel test summary contains all functional and performance test results.
The app let me forward the two static ports required for the test suite, but didn't support setting up triggered ports or setting the DHCP server range. So the lower functional score reflects this.
Table 2 summarizes the performance test results and includes Luma. eero isn't in this table because it drops its connections five minutes after losing internet connection. So it can't be tested with the Version 4 process, which requires devices be disconnected from the internet.
GWifi's lower Simultaneous TCP throughput appears to be due to a larger number of total retries vs. Luma (36,726 vs. 12,625). Similarly, GWifi's lower simultaneous UDP throughput is due to higher packet loss of about 10% in each direction vs. Luma 0.1%.
|Test Description||Google Wifi||Luma|
|WAN - LAN TCP (Mbps)||941||941|
|LAN - WAN TCP (Mbps)||941||941|
|Total Simultaneous TCP (Mbps)||1539||1764|
|WAN - LAN UDP (Mbps)||950||960|
|LAN - WAN UDP (Mbps)||950||950|
|Total Simultaneous UDP (Mbps)||1710||1898|
Table 2: Routing performance comparison
Luma has no functional score because none of the settings required for the test could be configured.
Google Wifi is not Wi-Fi Certified and does not support Wi-Fi Protected Setup (WPS). So any client you connect to it must be able to have its Wi-Fi network password entered. It supports WPA2/AES PSK wireless encryption only. Google also provided the following information about GWifi:
- Band steering is supported
- Devices are not load balanced among Wifi points
- 802.11v, k and r are currently not supported. RSSI based roaming assistance is used
- DFS channels are not supported
Google was very cooperative in accommodating our V9 test process requirements. In addition to a standard three-pack, they provided one Wifi point with 8872.40.9 firmware and the 2.4 GHz radio set to 20 MHz bandwidth for our throughput vs. attenuation test and one set to 40 MHz for maximum throughput testing. Both had the 2.4 GHz channel set to 6 and 5 GHz set to 40. Throughput vs. range testing was performed on the 20 MHz bandwidth sample.
GWifi was centered on the test chamber turntable in normal operating position, as shown in the photo below. The 0° position for had the connectors facing away from the chamber antennas as shown.
Google Wifi in test chamber
Although eero and Luma have been retested using the V9 wireless process, they are still marked with a "Special" test method because they could not run the latest Router test process. This is why you won't find either as AC1200 class products in either the Router Ranker or Charts. So I had to create the comparison plots below of the three plus NETGEAR's Orbi using Excel.
The 2.4 GHz downlink profile shows a stark difference in both throughput and range between eero and Luma and Orbi and GWifi. Given the design similarity, this is puzzling.
2.4 GHz Downlink Throughput vs. Attenuation
A similar pattern holds for 2.4 GHz uplink.
2.4 GHz Uplink Throughput vs. Attenuation
The performance stratification continues for 5 GHz downlink. Orbi continues its pattern of remaining slightly above GWifi throughout much of the tested attenuation range.
5 GHz Downlink Throughput vs. Attenuation
5 GHz uplink is the one test that mixed things up. eero suddenly jumps ahead of everything else for throughput, but still has shorter range, disconnecting after the 30 dB attenuation test. GWifi starts out stronger than Orbi until the 15 dB test, then the two track very closely for the rest of the run.