We’ve spent the week testing some of the latest routers: Netgear’s Nighthawk X8, D-Link’s DIR-895L and Linksys EA9500. While our results resemble a group test this article focuses on the benefits of the latest MU-MIMO technology in an effort to see if it makes a big difference.
Our tests demonstrate that, with products on the current market, there isn’t a huge difference between using MU-MIMO compatible WiFi adapter to connect to a MU-MIMO router than there is a modern laptop/tablet. The main, visible benefits appear only when using a 5GHz WiFi band (rather than the more-common 2.4GHz band) and, even then, only at medium distance. That’s a simplification of the results, and what follows shows how we came by them. But first…
Old routers used to have one aerial and transmit slowly (it seemed fast at the time). Then MIMO routers appeared which had Multiple In and Multiple Out antennae. This has been the basis for routers based upon the 802.11n and subsequent 802.11ac WiFi standards for some time. The problem with both of these standards is that if there is a slow device on the network, in theory, everything on the network slows down. Enter MU-MIMO which caters for Multiple Users using different devices. As such different antennae can be focused at different users which, in theory, avoids any speed drop.
How we tested
Testing was performed in a three-storey, Sydney townhouse. On the ground floor was our wireless router which was connected via Ethernet cable to our Alienware Alpha test PC. We then performed three sets of speed tests: next to the router, in a bedroom one floor up and finally two floors up in a second bedroom. In addition to the three main routers we also used our existing router, an older AC3200 D-Link DIR-890L (which isn't MU-MIMO compatible and only has one 5GHz network) for comparison. Both D-Link routers offer a Smart Connect feature whereby only one network is displayed and the router chooses the fastest way to connect to it (the user doesn’t see if it’s 2.4GHz or 5GHz). As this is the default setting we left it as is for this test. The Netgear allows its own two 5GHz networks to Smart Connect as one but it's off by default so we left it. The Linksys does combine its 5GHz channels in a Smart Connect manner but it's always on and you can't separate them. Finally, note that the D-Link AC5300 DIR-895L is MU-MIMO compatible but the firmware that enables it is in beta and so neither of the D-Links use MU-MIMO here. Their results are used for comparison.
We ensured that each test was stable and ran it again if it was not (unless otherwise stated).
In order to test the MU-MIMO technology, we added other equipment to each network: we had an old iPad 2 playing V for Vendetta on Netflix (by the router) and Tinkerbell and The Lost Treasure playing on Netflix on an Xbox One upstairs in the second bedroom. We also performed tests on an empty network from the one-floor-up bedroom for comparison.
We then transferred a 2GB movie file across the networks using both 2.4GHz and 5GHz WiFi bands (the Netgear uses two 5GHz networks at once – both on different channels – so we tested them both). We did this using both the native WiFi adapter in a Surface Pro 3 – a Marvell AVSTAR Wireless-AC Network Controller – and then we disabled that and ran the tests using the Surface Pro 3 with a Linksys MU-MIMO Max-Stream AC600 Wi-Fi Micro USB Adapter. All computers were running Windows 10.
All results are posted in MegaBYTES per second. Results from the Surface Pro 3 are in orange while the results which used the MU-MIMO dongle are in Blue. As can be seen, the Surface Pro 3 speeds are almost always faster.
Up close and it looks as though the old D-Link destroyed the field on 2.4GHz. This is likely due to the proprietary Smart Connect feature which means it likely was actually working on a 5GHz band but due the automated nature of the network handling on D-Link’s devices we can’t know for sure. We do know it repeatedly did it, however, but for now this result should be taken with a grain of salt. Interestingly, the newer D-Link router didn’t seem as clever at picking the fastest channel as the older model and was noticeably slower than everything else here.
More importantly, the SP3 on its own scored faster times than those with the MU-MIMO adapter on each occasion. We suspect that the MU-MIMO USB dongle doesn’t like being too close to the broadcasting router.
One floor up
At 2.4GHz In the first bedroom, There wasn’t much change among most of the Surface Pro 3 scores (the older D-Link model looks to have dropped back to a decent 2.4GHz speed here). The routers seem comfortable beaming to this distance. Interestingly, while all the MU-MIMO dongle scores were lower than the SP3 (again) both the Netgear and D-Link scores stayed the same and the Linksys got faster. This suggests that the MU-MIMO adapter is a bit happier when slightly-further away from the router.
On the 5GHz band this was far more pronounced. Finally, the MU-MIMO dongle scores started to out-perform the SP3's. Except for with D-Link where the newer model looked like it chose the wrong channel again rather than missing out on any MU-MIMO boost - though we can't be completely sure.
As can be seen, we later ran these tests on an empty network (it had become evident that one-floor-up was the MU-MIMO dongle’s sweet spot) in order to see how much difference having ‘slow’ devices on the network actually made. On the SP3’s WiFi, at 2.4GHz, across the board, speeds slowed down less than 3MB/s when there was other traffic present. All speeds were still faster than with the MU-MIMO dongle although slowdowns only ranged between 0.3 and 1.1MB/s. That slightly-narrower range may represent a win for MU-MIMO, but when you’ve already lost the race, how much does it matter?
Here are the raw results in MB/s :-
|Netgear Nighthawk X8||4.9||9||6||12|
Things changed in the 5GHz test. Speed drops were minimal on the SP3’s WiFi (save for one of the Netgear channel's 3.1MB/s drop). However, when MU-MIMO kicked in, we finally saw some differences. Both the Linksys and Netgear results got faster when the MU-MIMO dongle was used. As for why this would be faster than an empty network is anyone’s guess. Perhaps, it was triggered into putting in extra effort? Meanwhile, D-Link’s low scores were likely caused by it’s “intelligent” Smart Connect technology using the 2.4GHz band rather than an absence of MU-MIMO but again we can't be completely certain.
Here are the raw results in MB/s :-
|Netgear Nighthawk X8 (1)||12.5||6.7||12||6.7|
|Netgear Nighthawk X8 (2)||17.6||4.1||9.2||7.2|
Two floors up
At two floors up, at 2.4GHz, the Surface Pro 3 won again. Speeds were noticeably faster than when using the MU-MIMO dongle. The D-Link couldn’t actually connect at all when the MU-MIMO dongle was used which is likely due to the small antennae within the dongle.
At 5GHz the Netgear couldn’t connect to the SP3’s WiFi or the dongle while the D-Link couldn’t connect to just the dongle. However, the Linksys did see the MU-MIMO beat the SP3’s WiFi.
It appears that having a decent WiFi adapter with a strong antenna – as the Surface Pro 3 does – is currently a better bet than a MU-MIMO dongle in most cases. It’s not surprising, the MU-MIMO dongle is very small whereas the SP3’s antennae array is integrated into the chassis.
The SP3 generally coped both better up close and far away than the MU-MIMO dongle but we suspect that’s more down to antennae integration than technology. When the MU-MIMO dongle was at a sweet spot (generally not too close and not too far from the router and on 5GHz) it’s turbo-charged prowess beat the Surface Pro 3 setup.
However, this doesn’t really show that MU-MIMO is much faster on a network containing other, slow devices. As we saw with the empty network test, the MU-MIMO dongle didn’t make much difference.
As such we are confident in saying the following:-
The latest AC5300 (and older AC3200 routers) routers are generally very fast but not because of MU-MIMO itself.
MU-MIMO dongles are not likely to unlock massive performance from your MU-MIMO router unless your mobile device’s WiFi adapter and/or antennae are bad.
We’re looking forward to laptops having MU-MIMO-compatible wireless adapters and decent antennae arrays as then we’d likely see better performance improvements across the board.