Routing throughput was measured using our standard router test process with the router using 126.96.36.199 firmware. Table 4 summarizes the results and once again includes the Linksys WRT1900AC, NETGEAR R7000 and ASUS RT-AC68U for comparison.
All these routers have plenty of routing throughput and you would be hard pressed to find a difference among them in real-world use.
|Test Description||NETGEAR R8000||NETGEAR R7000||Linksys WRT1900AC||ASUS RT-AC68U|
|WAN - LAN||806.4 Mbps||931.4 Mbps||631 Mbps||754.5 Mbps|
|LAN - WAN||782.1 Mbps||941.7 Mbps||926 Mbps||824.6 Mbps|
|Total Simultaneous||1391.8 Mbps||1378.3 Mbps||975 Mbps||1226.8 Mbps|
|Maximum Simultaneous Connections||34, 083||38,793||30,557||33, 263|
Table 4: Routing throughput
The IxChariot unidirectional composite plot for the R8000 shows a cyclical variation between low 700s Mbps and peak speeds near 950 Mbps for downlink. Uplink also peaked up to the same 950 Mbps region, but only with brief spikes.
NETGEAR R8000 routing throughput unidirectional summary
The R8000 simultaneous up/downlink benchmark plot shows some battling at the beginning, that is probably due to IxChariot's Nagle's algorithm implementation. Once that settles down, throughput is smoothly shared with higher uplink vs. downlink throughput.
NETGEAR R8000 routing throughput bidirectional summary
I mentioned earlier that I'm still finalizing the new SmallNetBuilder Wireless Testbed, so won't be presenting complete wireless performance data in this part of the review. Instead, I chose to explore NETGEAR's SmartConnect feature.
To recap, SmartConnect is NETGEAR's implementation of Broadcom's XStream architecture. SmartConnect uses two 5 GHz radios and assigns client to each one when they connect. The assignment is static until the connection between router and client is broken.
Dynamic client assigment is part of the XStream architecture, but vendors can choose which XStream features they use. NETGEAR told me they disabled dynamic client assignment because some devices "do not make a graceful switch".
Note that SmartConnect does not support 2.4 / 5 GHz band steering. Clients are only steered between the two 5 GHz radios.
I assembled the following devices to use for the test:
- Moto X smartphone (1x1 AC)
- NETGEAR R7000 in client bridge mode (3x3 AC)
- Laptop with NETGEAR A6200 USB adapter (2x2 AC)
- iPad 2nd gen (1x1 N)
- iPod Touch 5th gen (1x1 N)
The R8000 was located in the wireless testbed upper test chamber with the door open. The bridge mode R7000 was in the lower chamber with the door closed. This allowed me to use the testbed programmable attenuators to control the signal and therefore the link rate of the R7000. So that it didn't totally dominate the other AC devices, I programmed 20 dB of attenuation for the testing. All the other devices were located within 6 feet of the R8000, outside the test chamber, and all received a nice, strong signal.
I started by disabling SmartConnect and directed devices to connect to the upper band radio. The Attached Devices screen below confirms that everything except the laptop with A6200 adapter was connected. I used that in a second set of tests.
Device connection - SmartConnect Off
I then set up an IxChariot script that simultaneously ran a stream to each client. The composite plot below shows a just under 63 Mbps of total throughput delivered to all devices. Not very impressive and somewhat surprising given previous experiments with multiple AC clients. Note the highest-class client, the 3x3 R7000 ended up with the lowest throughput in the group--only 13 Mbps.