Why don't I see better file transfer speeds over AC network?

[rlp2955]

Hi all, first time on the forum and looking for some help understanding why I am not seeing any difference in file transfer speeds between my old and new setup. I just upgraded to the Asus AC87R (from N66U)router and also bought the USB-AC56 adapter for my laptop downstairs. My hope was that I would see significant increase in file transfer speeds between that laptop downstairs and the USB 3.0 drive I have plugged into the router as well as a seperate laptop hardwired into the router upstairs.

I downloaded and ran the LAN Speed Test application and here are my results (using 5Ghz connection):

Laptop downstairs to USB drive on router: 200Mbps Write, 145Mbps Read
Laptop downstairs to wired laptop on Gig Ethernet: 160Mbps Write, 133 MBps Read

If I disconnect the USB AC adapter and use the internal wireless N adapter (Killer Networks), I get the following results:

Laptop downstairs to USB drive on router: 165Mbps Write, 97 Mbps Read
Laptop downstairs to wired laptop on Gig Ethernet: 36 Mbps Write, 111 Mbps Read

If I use the AC adapter and transfer to the USB drive or laptop hard drive, the max transfer speed that Windows typically reports is anywhere from 30-35MB/second. Is this right? Shouldn't I be getting almost double that with an AC adapter, AC router, and USB 3.0 drive plugged into router? Doesn't seem like I am seeing much of a performance increase from my previous network adapter. Help?

 

[JackMDS]

People do not realize that the numbers used to describe Wireless hardware are not functional result of Network transfer but rather a description of the inner working of the individual wireless components without taking into consideration of the whole Network Component and physical Environment (in case of Wireless).

That means that switching from an N900 Dual Band to AC2400 yields much less than x2.5 that "made believe" by the mathematical relations between the two numbers.

The USB USB-AC56 is a AC 1200. Actual transfer of 30-35 MB/sec. (B=Byte) is actually a good outcome for a system as yours.

 

[azazel1024]

Yes, it should be better...if you are in the same room with the router. Wireless signals attenuate, which means slower speeds the further away you are.

I can get 61MB/sec between my laptop with an Intel 7260ac in it (dual stream/AC866) if my laptop is about 6ft from you Archer C8. If I move about 15 feet away, it drops to maybe 55MB/sec at best. Put a wall in the way and 15ft and it is down to about 45MB/sec. Go a floor up, a couple of walls and about 25ft distance and is about 20MB/sec. Move 35ft away with another wall in the way and it is about 3.5MB/sec. Move just a tiny bit further and zero connection.

This is all 5GHz BTW. On 2.4GHz it is about 28.5MB/sec within about 15ft of the router. Move as far away as that last 5GHz location and it is about 8MB/sec. Move even further away and it is about 3MB/sec.

Wireless speeds you should get are roughly 50-75% of the listed numbers if you are within a few feet of the router and have no sources of interference occuring. There is overhead for error correction as well as other issues that result in this.

If you move further away, have walls or floors in the way, interference, etc. this all drops.

 

[inachu]

If you have USB 1.x and also a USB 2.0 slot then try both. If your laptop is pretty modern with USB 3.0 then always plug the AC usb dongle in the usb 3.0 slot

 

[rlp2955]

I tried copying an 8GB file from the laptop with the AC adapter this morning to the USB 3.0 drive (Western Digital Elements 1TB) this morning and the max I saw it spike at was 35MB/s when I was one floor down. If I moved it right next to the laptop it spiked at 46MB but mostly hovered between 40 and 46MB.

 

[Gryz]

I don't know about the speeds of USB3.0. But from experience, I know that when people try to measure network performance by copying disk-to-disk, they often measure the performance of the disk(s). Not the performance of the network.

You need a benchmark-tool to just measure network performance. Or TCP performance. Anything that doesn't involve disks. I haven't used any of those tools in ages, so I can't tell you what would be the best. Maybe others here on AT have a suggestion ?

Here's a page I found with google:
https://www.raymond.cc/blog/network-benchmark-test-your-network-speed/
The 2nd one (LANBench) seems to do the right type of testing.

After you have ensured that network-performance is good, then you can start doing tests which involve harddisks.

 

[RadiclDreamer]

To save you a long drawn out post:

AC is faster

1300mbps is in theory and only on 5ghz and only when using 4xchannels (80mhz)

1300mbps includes any retransmits, header data such as frame checks etc and is NOT simply the "payload" data.

5ghz doesnt go as far as 2.4 so you need to be closet to get those awesome speeds.

 

[drebo]

The realistic possibility of being able to use 80mhz or 160mhz channels is, honestly, slim to none.

Hell, most residential areas are so congested that 40mhz channels are unlikely.

You'll probably get better real performance sticking with 20mhz channels.

 

[RadiclDreamer]

The realistic possibility of being able to use 80mhz or 160mhz channels is, honestly, slim to none.

Hell, most residential areas are so congested that 40mhz channels are unlikely.

You'll probably get better real performance sticking with 20mhz channels.

Depends entirely on your environment, my personal testing with cisco 2702i AP and 7260AC intel wireless adapters I see better performance with 80mhz, not crazy better, but enough to be outside the margin of error and enough to notice on large file transfers.

 

[PliotronX]

Depends entirely on your environment, my personal testing with cisco 2702i AP and 7260AC intel wireless adapters I see better performance with 80mhz, not crazy better, but enough to be outside the margin of error and enough to notice on large file transfers.

That's what he was saying; every situation is different and we share the airspace with too many random devices (looking at you Bluetooth and console controllers) and neighbors. Because of the amount of APs I've usually encountered where I have lived, my routers rarely get to spread their wings with even 40MHz. I am glad to be out of this one neighborhood where some idiots had a router that would always automatically hop channels to whatever mine was on. I had tried hiding the SSID in case humans were controlling it but sure enough, I don't know if it was a bug or they really had no lives but yeah. All in all I will do my damndest to make sure stationary devices are hard wired.

 

[azazel1024]

That's what he was saying; every situation is different and we share the airspace with too many random devices (looking at you Bluetooth and console controllers) and neighbors. Because of the amount of APs I've usually encountered where I have lived, my routers rarely get to spread their wings with even 40MHz. I am glad to be out of this one neighborhood where some idiots had a router that would always automatically hop channels to whatever mine was on. I had tried hiding the SSID in case humans were controlling it but sure enough, I don't know if it was a bug or they really had no lives but yeah. All in all I will do my damndest to make sure stationary devices are hard wired.

Also being 5GHz, unless you are in a dense apartment building, you absolutely would be able to use 80MHz or 160MHz channels. In my fairly rural house, completely across the house through 3 walls and 45ft my 5GHz signal from my 11ac router on my LAPTOP (which is a lot stronger than the signal on my phone) is about -80dBm. If I go outside, even in the room next to the access point, my 5GHz signal is about -78dBm. If I walk 30ft from my house I can't pick up the 5GHz signal at all. Let alone trying to pick up a 5GHz signal from another house. Heck, in my old townhouse neighborhood I've gone over to my rental property with my tablet before. "war walking" I can pick up about a dozen 5GHz networks walking down the row about about 20 townhouses that my rental TH is the second from the end. Of course they drop in and out as I walk along. Inside the townhouse itself, I can pick up 2 5GHz networks other than my renter's network, I assume it is the immediate neighbors SSIDs. They come in at around -75 to -80dBm, about 30-40dB lower than what my renter's 5GHz network is at.

That is compared to the 9 different 2.4GHz SSIDs I can pick up standing in the living room, ranging in signal strength from -48dBm to -90dBm, which is nearly as strong as my renter's 2.4GHz network for the stronger couple.

Yes, interference on 2.4GHz is a big thing, but even in most fairly dense living, interference on 5GHz just isn't a big deal, because the signal has much worse penetration. 80MHz is absolutely viable (I have no issues running two 80MHz 5GHz access points, one in the 36-48 range and one in the 147-161 range). 160MHz is a bit more problematic, both from a price of components stand point, but also that it causes interference for ANY OTHER 5GHz network that might be within range, unless of course DFS channels become more common (which I doubt, unless the FCC changes them away from DFS).

So it can become a problem with your own network setup if you have more than a single router/AP. With 5GHz 80MHz, you have two non-overlapping channel selections you can use (without DFS, with DFS you have 4 non-overlapping selection ranges), in 2.4GHz you only have a single 40MHz non-overlapping selection, plus 2.4GHz penetrates a lot further through obstacles.

 

[Emulex]

Nowadays with all the neighbors having 802.11N on 5ghz - it is quite crowded and the distance of being 1 floor is enough to lost signal integrity.

802.11AC doesn't necessarily give better performance, except in perfect conditions. And a crappy 802.11ac router could very well be outperformed by an 802.11N router for sure! Antennas,chipset,power output, and bios of the router counts a lot.

Best to stick to one known chipset across the board. Mixing Broadcom to Atheros tends to muck up some of the fancy tech used in newer routers!

Intel to Atheros seems to be pretty stable.

 

[RadiclDreamer]

AC will typically give better performance than N even with all environmental factors being the same.

A good, and bad thing about it is the 5ghz, its less crowded but a the same time the penetration sucks compared to 2.4, which I guess depending on your viewpoint could be a good thing.

I do agree that a good wireless N setup can outperform a crappy AC setup.

I recommend nothing but the intel wireless cards, all others pale in comparison when it comes to performance and reliability.