Wireless AC vs. Wireless N, my little test.

[SVTmaniac]

I setup my new Airport Time Capsule today and decided that I would do a little test to see just how much faster wireless AC was. I know this test isn't scientific in any way and it's just a gee whiz kind of thing, but I did find it interesting.

I used Black Magic Disk Speed Test to read and write to the SSD drive of my late 2012 quad core i7 Mac Mini that was plugged into the Time Capsule via Gigabit Ethernet.

On the left is my late 2013 rMBP connected via 5ghz Wireless AC. On the right is my late 2012 rMBP connected via 5ghz Wireless N. Obviously I didn't run the tests at the exact same time, but they were within about 30 seconds of each other. I did the tests a few times and got roughly the same exact numbers each time. It's interesting to see that the wireless AC didn't double the results like I had expected it to.

 

[dirtnnasty]

I think the limiter is the harddisk?

My non-raid NAS is fast like that...

 

[SVTmaniac]

I think the limiter is the harddisk?

My non-raid NAS is fast like that...

It was an SSD drive in my Mac Mini. When I run Black Magic on the mini I get read speed of like 450 and write speed over 500 so I don't think the SSD is the bottleneck here.

 

[Quu]

The real bottleneck here isn't the wireless of your notebooks but the Airport Extreme.

If you check it out online it suffers from bad 802.11ac performance, in some tests it is 1/3rd the performance of the best 802.11ac routers which cost less than the Airport Extreme does.

Just look at this chart from Small Net Builder:

Notice where the Airport Extreme is? ... Right at the bottom. By the way if you look your high speed in your benchmark is 41MB/s and in the above benchmark it is only reaching 40MB/s.

I myself have an Asus 802.11ac router, the AC66U. And I get 90MB/s sustained over 802.11ac between my rMBP 2013 and my server which is connected to the router over 1Gb Ethernet.

 

[SVTmaniac]

Well that's discouraging to know. I figured the Apple would be pretty fast. Oh well, it was a Christmas gift from the wife so I can't complain too much.

 

[Quu]

Well that's discouraging to know. I figured the Apple would be pretty fast. Oh well, it was a Christmas gift from the wife so I can't complain too much.

It is discouraging I agree. I used to have the older 802.11n Airport Extreme which was a perfect product but when it came to buy an 802.11ac router Apples one wasn't out yet so I grabbed the Asus and couldn't be happier with it, amazing router and Apples one is just shockingly rubbish in comparison. The benchmarks speak for themselves.

 

[SVTmaniac]

It is discouraging I agree. I used to have the older 802.11n Airport Extreme which was a perfect product but when it came to buy an 802.11ac router Apples one wasn't out yet so I grabbed the Asus and couldn't be happier with it, amazing router and Apples one is just shockingly rubbish in comparison. The benchmarks speak for themselves.

I wonder if a firmware update can fix some of the speed issues or if it will always just be slow?

 

[jav6454]

The real bottleneck here isn't the wireless of your notebooks but the Airport Extreme.

[...]

Notice where the Airport Extreme is? ... Right at the bottom. By the way if you look your high speed in your benchmark is 41MB/s and in the above benchmark it is only reaching 40MB/s.

I myself have an Asus 802.11ac router, the AC66U. And I get 90MB/s sustained over 802.11ac between my rMBP 2013 and my server which is connected to the router over 1Gb Ethernet.

Before acting high and mighty, would you like to clarify your results? Are your personal results in MB/s or Mb/s? The table you posted is in Mb/s, which can make some serious difference.

 

[Quu]

Before acting high and mighty, would you like to clarify your results? Are your personal results in MB/s or Mb/s? The table you posted is in Mb/s, which can make some serious difference.

I take offence with you claiming I'm acting high and mighty. But I am happy to clarify my results.

My own personal results are in MB not Mb. The table I posted is not mine but from netbuilder a very very reputable community. They list their results in Mb which I converted to MB when I quoted the Airport Extreme transfer speed as 40MB (325.5Mb = 40.6MB). My own tests show 90MB (720Mb/s) transfer speeds between my 2013 Retina MacBook Pro Haswell (2.6GHz / 1TB SSD) and my Asus AC66U Router.

My server was connected to the router via 1Gb Ethernet (125MB/s not including overhead). The server was serving files to the notebook from a RAID6 array capable of sustained 600MB/s reads (4800Mb/s). When transferring from my Server to my Desktop when both systems are connected to the router over 1Gb Ethernet I see transfer speeds at 117MB/s (936Mb/s) that is with jumbo frames disabled and a significant cable length.

I would really love to show you pictures of a benchmark but I already wiped my 2013 rMBP and boxed it up ready to be returned back to Apple due to a screen fault but you can trust the netbuilder benchmark I posted above and I promise I'm not blowing smoke about my own routers performance over 802.11ac - My own router is also featured in the netbuilder benchmark.

----------

I wonder if a firmware update can fix some of the speed issues or if it will always just be slow?

I wouldn't discount the possibility but I didn't design the product so I cannot say definitively. It could be antenna placement, chipset or software related.

 

[muyaad]

I take offence with you claiming I'm acting high and mighty. But I am happy to clarify my results.

My own personal results are in MB not Mb. The table I posted is not mine but from netbuilder a very very reputable community. They list their results in Mb which I converted to MB when I quoted the Airport Extreme transfer speed as 40MB (325.5Mb = 40.6MB). My own tests show 90MB (720Mb/s) transfer speeds between my 2013 Retina MacBook Pro Haswell (2.6GHz / 1TB SSD) and my Asus AC66U Router.

Thanks for an insightful response and politely reply.

 

[SVTmaniac]

Thanks for an insightful response and politely reply.

Yeah he replied much nicer than I would have. Thanks for offering your experience with the Wireless AC. Still kind of bummed that this Apple is so slow.

 

[Hitch08]

Interesting thread with good info.

This is making me reconsider my recent purchase of an A/C AirPort Extreme. I read a number of reviews, but didn't go to Small Net Builder. For example, I read the CNET review, which included a chart on throughput.

CNET's throughput chart showed the following:
ASUS RT-AC66U: 339.2 Mbps
AirPort Extreme: 287.2 Mbps

http://reviews.cnet.com/wireless-ac...port-extreme-base/4505-3265_7-35781432-2.html

Although the ASUS was faster, it was not enough for me to decide to get it over the AirPort.

Do these two charts show similar tests? Or am I not comparing "apples to apples"? Maybe Small Net Builder just got more out of the AirPort and a lot more out of the ASUS?

I haven't opened the AirPort yet. I may return it and get the ASUS....

 

[Bear]

If you have 802.11N and 802.11AC devices connected the same wireless Network, they're all probably running at 802.11N speeds.

 

[Quu]

Interesting thread with good info.

This is making me reconsider my recent purchase of an A/C AirPort Extreme. I read a number of reviews, but didn't go to Small Net Builder. For example, I read the CNET review, which included a chart on throughput.

CNET's throughput chart showed the following:
ASUS RT-AC66U: 339.2 Mbps
AirPort Extreme: 287.2 Mbps

http://reviews.cnet.com/wireless-ac...port-extreme-base/4505-3265_7-35781432-2.html

Although the ASUS was faster, it was not enough for me to decide to get it over the AirPort.

Do these two charts show similar tests? Or am I not comparing "apples to apples"? Maybe Small Net Builder just got more out of the AirPort and a lot more out of the ASUS?

I haven't opened the AirPort yet. I may return it and get the ASUS....

In my opinion CNets data is questionable. I'm much more likely to trust SmallNetBuilder where reviewing network equipment is their bread and butter. CNet is a jack of all trades.

To put it another way, if you had a brain tumor you wouldn't have your local general practitioner give you an opinion on it would you? You'd want the best neurosurgeon in the world.

I can tell you this though, the AC66U I have works fantastic. Uptimes of 100+ days, no dropped clients, no lag issues, no jitter in its latency, serves 17 different WiFi enabled devices in my home perfectly, streams 1080p content to our HTPC (connected via 802.11ac) perfectly. It worked briliant with my MacBook Pro 15" Haswell.

In-fact I did a time machine restore like this: rMBP 15" 802.11ac -> Router -> Gigabit Ethernet Server. I transferred 300GB or so of personal data in just over an hour and 10 minutes. That is an average data rate of 70MB/s (560Mb/s) which is insanely good for WiFi in my opinion especially when you consider it was having to unpack the disk image and do whatever else processing it does when restoring a Time Machine backup.

 

[Hitch08]

Quu - Great points and thanks for the quick response!

 

[PatientWaiter]

This is intriguing. Does anyone know of an AC router on this list that has the audio streaming functionality of the Apple routers? I'd like to hook up my entertainment center to the router to stream music.

 

[Quu]

This is intriguing. Does anyone know of an AC router on this list that has the audio streaming functionality of the Apple routers? I'd like to hook up my entertainment center to the router to stream music.

Only the Airport Express and Apple TV can stream music. Not the Airport Extreme.

In my home we use an Airport Express for audio streaming, it works great and isn't too pricey. We don't use its router features.

 

[tubbymac]

Notice where the Airport Extreme is? ... Right at the bottom. By the way if you look your high speed in your benchmark is 41MB/s and in the above benchmark it is only reaching 40MB/s.

I wonder if my Apple Time Capsule Wifi AC edition is as slow as the Airport Extreme. Too lazy to test it right now, but file transfers to my NAS have not subjectively gotten that much faster versus when I had the Wifi N version. At most things seem to copy about twice as fast, but that's ideal. Most times I don't seem to notice any difference. I wonder if my iPhone 5 connected only at Wifi N speed slows everything down.

 

[Quu]

I wonder if my Apple Time Capsule Wifi AC edition is as slow as the Airport Extreme. Too lazy to test it right now, but file transfers to my NAS have not subjectively gotten that much faster versus when I had the Wifi N version. At most things seem to copy about twice as fast, but that's ideal. Most times I don't seem to notice any difference. I wonder if my iPhone 5 connected only at Wifi N speed slows everything down.

The 802.11ac Airport Extreme and Time Capsule are identical. The only physical difference is the Time Capsule has a notch on its motherboard for a SATA connector to attach so that a hard drive can be connected inside. The Airport Extreme is actually just a Time Capsule without a hard drive in its central cavity.

 

[ElectronGuru]

If you have 802.11N and 802.11AC devices connected the same wireless Network, they're all probably running at 802.11N speeds.

Could the published tests have missed this, like brought competing stations into and against a 'working' ap station?

 

[y0zza]

The real bottleneck here isn't the wireless of your notebooks but the Airport Extreme.

If you check it out online it suffers from bad 802.11ac performance, in some tests it is 1/3rd the performance of the best 802.11ac routers which cost less than the Airport Extreme does.

Just look at this chart from Small Net Builder:

Image

Notice where the Airport Extreme is? ... Right at the bottom. By the way if you look your high speed in your benchmark is 41MB/s and in the above benchmark it is only reaching 40MB/s.

I myself have an Asus 802.11ac router, the AC66U. And I get 90MB/s sustained over 802.11ac between my rMBP 2013 and my server which is connected to the router over 1Gb Ethernet.

Be careful of mixing routing performance (wired WAN to wired LAN) and wireless throughout. The graph above is for wired routing performance only.

If you compare say 5GHz wireless downlink performance, SmallNetBuilder reports a *best case* throughout of 352Mb/s for the AC66U vs 269.6Mb/s for the AirPort Extreme. Asus still substantially faster, but I've never seen speeds of 90MB/s (720Mb/s) over 3x3:3 802.11ac - not sure how you measured/what you did to get those figures...

 

[Quu]

Be careful of mixing routing performance (wired WAN to wired LAN) and wireless throughout. The graph above is for wired routing performance only.

If you compare say 5GHz wireless downlink performance, SmallNetBuilder reports a *best case* throughout of 352Mb/s for the AC66U vs 269.6Mb/s for the AirPort Extreme. Asus still substantially faster, but I've never seen speeds of 90MB/s (720Mb/s) over 3x3:3 802.11ac - not sure how you measured/what you did to get those figures...

I do have my server connected to the WAN port. I have the internet go in to the server and then the server goes in to the router. The router itself is used more like an access point with the server handling DHCP.

I'm sorry I didn't specify this earlier. I didn't think there would be any performance difference but that is due to my own ignorance.

 

[y0zza]

I do have my server connected to the WAN port. I have the internet go in to the server and then the server goes in to the router. The router itself is used more like an access point with the server handling DHCP.

I'm sorry I didn't specify this earlier. I didn't think there would be any performance difference but that is due to my own ignorance.

Ah yes - your numbers would make sense for purely wired throughput if that is what you tested. The OP posted quite decent 5GHz wireless down/uplink throughout at approx 332Mb/s down and 246Mb/s up. I would expect some overhead from the Time Capsule HDD as well, so the numbers compare well vs AC66U figures from SmallNetBuilder.

In your case, if you are not using the routing function of the AC66U (ie a pure access point with NAT/routing done by your server) it would probably make more sense to put the server on the LAN ports or bridge the WAN port. An AirPort Extreme in this configuration should also deliver largely similar results to your AC66U (ie wired throughput to server will approach Gigabit line rate for simple layer 2 switched LAN-LAN traffic).

The AirPort Extreme does seem relatively 'slow' at routing (300-500Mb/s) for some reason (could be firmware related since it uses a similar dual-core A9 ARM CPU to the AC66U). If this is important in your application, eg Gigabit internet or a WAN-side local sever, there are much better options for routing performance. Most people with less-than 200-300Mb/s internet speeds will not see any difference, although one could argue it is better to future-proof.

As evidenced above, most BCM4360-based 802.11ac solutions will show largely similar radio performance in the ideal case (assuming same 3x3:3 configuration). Where differences become much more apparent is after RF attenuation or noise is factored in (real-world longer distance connections) - power amplifier, antenna design/gain, etc, then become significant differentiating factors between models. However, very few review sites test weak-signal performance consistently or thoroughly...

 

[glenthompson]

This reinforces my decision to use wired connections whenever possible. I get about double this throughput to my Synollgy NAS over gigabit Ethernet. That's with spinning drives, not an SSD.

 

[tubbymac]

The AirPort Extreme does seem relatively 'slow' at routing (300-500Mb/s) for some reason (could be firmware related since it uses a similar dual-core A9 ARM CPU to the AC66U). If this is important in your application, eg Gigabit internet or a WAN-side local sever, there are much better options for routing performance. Most people with less-than 200-300Mb/s internet speeds will not see any difference, although one could argue it is better to future-proof.

Oh he was talking about wired speeds? No wonder those numbers seemed so much faster than my wifi ac speeds.

My experience lines up with the OP. In general the fastest I see over wireless on my wifi ac to the time capsule is about 40 MB/sec. This is just glancing at activity monitor, never bothered doing strict benchmark tests. For comparison I think the fastest I got on my wifi n older generation time capsule was about 16 MB/sec.

Over a wired connection though even my old wifi n time capsule had no problems hitting 100 MB/sec to my ubuntu based NAS (all connected via gigabit ethernet). Here's the only strange thing. I could only get this speed via a samba connection via Windows (whether through bootcamp or parallels it didn't matter). If I tried a samba connection via Mac Mountain Lion, max speed even via gigabit ethernet was around 45 MB/sec. I never figured out why it was so much slower through OS X.

Mavericks doesn't even work properly with samba, and is broken all over for me so I have to wait for Apple to fix all the bugs before I can test on Mavericks.

So the TLDR of all this is that my time capsule doesn't seem to have any of the "slow routing" issues the Airport Extreme had over wired gigabit ethernet only if using Microsoft Windows. It only had those issues when using OS X.