Like every other website on the planet, SmallNetBuilder uses cookies. Our cookies track login status, but we only allow admins to log in anyway, so those don't apply to you. Any other cookies you pick up during your visit come from advertisers, which we don't control.
If you continue to use the site, you agree to tolerate our use of cookies. Thank you!

Wi-Fi Router Charts

Click for Wi-Fi Router Charts

Mesh System Charts

Click for Wi-Fi Mesh System Charts

Multiband Throughput

The Multiband benchmark simultaneously loads all radios in a router/AP/mesh node and measures throughput and latency under load. It's a good way to see whether (or how much) Ethernet port speed limits what you can get out of an AP. And since it's run on each node in a mesh system, it's also a good way to gauge backhaul bandwidth. Keep in mind, however, that you'd be hard-pressed to push an AP to the limits tested by this benchmark. The traffic generator for this test is connected to the root mesh node's WAN port. Channel bandwidth settings are 40 MHz for 2.4 GHz and 80 MHz for 5 GHz.

The first plot shows throughput for the 2.4 and 5 GHz radios at the root mesh node for the three ASUS systems. I also threw Linksys' Atlas Max 6E so you could compare to a product with really poor backhaul. The AX92U's legacy radios and its 1 GbE WAN port put it at a disadvantage vs. the two Zens that have 2.5 GbE WAN ports. Adding up throughput for both bands yields 1396 Mbps for the XT8 and 1414 Mbps for the ET8, which is essentially the same. But the XT8 seems to allocate more throughput to 5 GHz than the ET8. For comparison, the Linksys' total bandwidth is 1309 Mbps.

Multiband Throughput per radio - Root node

Multiband Throughput per radio - Root node

Moving to the Hop1 node brings backhaul performance into play. The ET8 has a clear advantage in this test, thanks to its 6 GHz backhaul. But even though the Linksys also has a 6E radio for backhaul, it appears to not have used it for backhaul. At least that's my explanation for why only 99 Mbps is available for 5 GHz STAs at the Hop 1 node.

Multiband Throughput per radio - Hop 1 node

Multiband Throughput per radio - Hop 1 node

Tables 2 and 3 show the throughput change for each band, referenced to Root throughput, for 2.4 and 5 GHz, respectively. The ET8 is the clear winner for 2.4 GHz, with basically no loss of throughput through the backhaul. Worst of the bunch is the Linksys, losing 38% of 2.4 GHz throughput backhaul.

  ASUS ET8 ASUS XT8 ASUS RT-AX92U Linksys Atlas Max 6E
  Throughput % change from Root Throughput % change from Root Throughput % change from Root Throughput % change from Root
Root 432 - 399 - 230 - 376 -
Hop 1 429 -0.7 340 -14.8 219 -4.8 232 -38
Table 2: Multiband throughput change, Root to Hop 1 node - 2.4 GHz

For 5 GHz, the ET8 is again the winner, even though backhaul throughput loss is a bit over 20%. Both the XT8 and AX92U lose around half the throughput available at the root node. The Linksys again surpasses the other three products, but not in a good way for 5 GHz, losing almost 90% of root node throughput through the backhaul.

  ASUS ET8 ASUS XT8 ASUS RT-AX92U Linksys Atlas Max 6E
  Throughput % change from Root Throughput % change from Root Throughput % change from Root Throughput % change from Root
Root 982 - 997 - 697 - 933 -
Hop 1 763 -22.3 470 -52.9 373 -46.5 99 -89
Table 3: Multiband throughput change, Root to Hop 1 node - 5 GHz

Multiband Latency

For latency, CDF plots tell the story better than bar plots because they show latency and latency spread (jitter). Unfortunately, the SNB Chart system can't produce them; I generated these using Excel.

The Root node 2.4 GHz latency CDF plot below shows the AX92U's 90th percentile latency around 2X of the two Zens (362 ms vs. 163 ms and 190 ms). The AX92U's has pretty long tails at the start and end of its plot. We'll see why in the time plots below.

Multiband Latency CDF plot - 2.4 GHz comparison - Root node

Multiband Latency CDF plot - 2.4 GHz comparison - Root node

It took awhile to understand what was happening to make the Hop 1 node 2.4 GHz Latencies lower than the Root latencies. After all, you would think that packets running through three radios (root client connect, root to hop1 backhaul and hop1 client connect) would have more latency than those using just the root node radio. But with help from an octoScope colleague, the cause was found to be the difference in airtime use.

Since packets on the root node come from a 2.5 Gbps Ethernet port, the 2.4 and 5 GHz radios can be fully loaded, which produces higher latency. Hop 1's client-facing radios source their packets from the backhaul connection, which has limited throughput shown in the throughput measurements above. So airtime isn't as congested, resulting in lower latency.

Multiband Latency CDF plot - 2.4 GHz comparison - Hop 1 node

Multiband Latency CDF plot - 2.4 GHz comparison - Hop 1 node

The Root 5 GHz Latency CDF again shows the AX92U with a wide latency spread, indicating significant jitter. But its 90th percentile value isn't too much more than the two Zens, coming in at 87 ms vs. 82 for the ET8 and 80 for the XT8.

Multiband Latency CDF plot - 5 GHz comparison - Root node

Multiband Latency CDF plot - 5 GHz comparison - Root node

Moving along to the Hop 1 node again shows lower latencies for all three products, 31 ms for the ET8 and around 38 ms for the XT8 and AX92U.

Multiband Latency CDF plot - 5 GHz comparison - Hop 1 node

Multiband Latency CDF plot - 5 GHz comparison - Hop 1 node

Time plots of Multiband throughput, latency and loss provide a quick feel for link stability. The Root node plots below show very steady throughput, even though the radios are both running full blast. Latencies are also surprisingly stable (L2E_png = 2.4 GHz, L3E_png = 5 GHz). The AX92U latency plot shows a ramp from ~ 125 Mbps to ~ 350 Mbps, which explains the long tail at the start of its CDF plot above. The spikes up to 500 Mbps are the cause of the long tail at the end of its CDF.

Multiband throughput comparison, latency, loss vs. time, Root node

Multiband throughput comparison, latency, loss vs. time, Root node

The Hop 1 plots are also remarkably steady and show the reduced latency. The XT8's latency plot had me checking and rechecking my setup and results many times to make sure I was actually measuring both radios (I was). Since most of us don't load our Wi-Fi radios as heavily as this test, it's likely the latencies you experience will be more like those in the Capacity test, which is up next.

Multiband throughput comparison, latency, loss vs. time, Hop 1 node

Multiband throughput comparison, latency, loss vs. time, Hop 1 node

Mesh System Capacity

The last benchmark associates one STA per mesh node, runs TCP/IP traffic to all STAs simultaneously, limiting each STA's throughput to 250 Mbps. Since a two-stream AX STA can produce around 900 Mbps of throughput with a strong signal, this test should not fully load a 5 GHz channel. The ideal throughput result for this test should be 500 Mbps for two nodes and 750 Mbps for three nodes.

The Charts produce bar charts of the results. But the time plots below, which the Charts engine can't produce, are more informative. All three systems achieve the expected 500 Mbps total throughput, which should not by any means be a stretch for mesh systems with dedicated four-stream backhaul radios. Latencies are all in the same ballpark, generally below 20 ms.

Mesh capacity time plot comparison

Mesh capacity time plot comparison

The bar plots do a better job of clearly ranking the produce latencies and show the ET8 with best Root and Hop 1 scores (lowest latencies), followed by the AX92U and the XT8.

Mesh capacity latency score comparison

Mesh capacity latency score comparison

Closing Thoughts

Our Wi-Fi Mesh System Ranker places the ET8 and XT8 in the #1 and #2 slots, respectively, with the RT-AX92U in the current bottom, #4 slot. But these three ASUS mesh systems provide a good / better/ best scenario for Wi-Fi shoppers. With essentially the same hardware design, the price/performance gap between the XT8 and ET8 is narrower than either of them vs. the RT-AX92U. But if you don't have any Wi-Fi 6 devices, don't plan on adding a bunch real soon and have a sub-Gigabit internet service, the RT-AX92U might be a good fit.

If you want the top-of-line ET8, however, you may want to wait for better supply. Amazon third-party sellers are really sticking it to prospective buyers as I write this, asking $1100, which is waaay above its $530 MSRP.

Support Us!

If you like what we do and want to thank us, just buy something on Amazon. We'll get a small commission on anything you buy. Thanks!

Don't Miss These

  • 1
  • 2