RT-AC68U Slow USB 3.0 speeds

[RMerlin]

For the people worried about this, here you go.

Test setup:

- RT-AC68U (got mine last week, kindly provided by Asus)
- Laptop running Win7
- Gigabit Ethernet interface
- USB-AC53
- NexStar3 USB3.0 enclosure, with a 2.5" Toshiba 750 GB HDD, formatted as NTFS
- USB 3.0 Interference reduction was Disabled (to ensure the HDD ran at USB 3.0 speed)

The first test was done with the laptop connected over gigabit to a LAN port on the router. I copied a 11 GB file from the router to the laptop (this is a compressed Acronis True Image backup). During the copy, I had my Nexus 7 tablet connected to the router on the 2.4 GHz band. A Speedtest was run at some point during the transfer, to generate some traffic. No disconnection, and the speedtest ran normally.

The speed never dropped below 49 MB/s, hitting around 52 MB/s at one point. This confirms that Asus did resolve the USB3 performance with a firmware update, and this router will match any of the other USB3-equipped routers.

The second test was to test the stability of the 2.4 GHz connection while there was activity on a USB 3.0 hard drive. Performance is lower for good reasons - this is a 300 Mbits USB-AC53. (I normally use the internal card on my laptop, but I wanted to also test the 802.11ac connectivity - I didn't screenshot it because it's irrelevant to this test). This time I copied a smaller 1.2 GB AVI (cause it would have taken far too with the 11 GB test file). There was no disconnection from the 2.4 GHz radio during the test, and speed remained constant:

So with these results, a few notes:

1) USB3 performance can be fixed with a firmware update (and Asus actually did).
2) It's possible to have a stable 2.4 GHz connection while accessing a USB3 drive on the RT-AC68U
3) HOWEVER: stability on the 2.4 GHz band with USB3 device isn't just dependant on the router itself. It will also depend on the HDD shielding (mine is an aluminium case, which was about one feet away from the router) and of the cable itself (the cable NexStar provides seems pretty well shielded - definitely thicker than many typical USB2 cables).
4) Regarding Tim's stability issues while copying the file with the disk running at USB3: this could be caused either by the firmware version he was using, or his test HDD is just more sensitive to radio interference than mine is. To reach any conclusion on this one, we'll have to wait for Tim to re-test once Asus releases the new firmware. For now, it's too early to tell if the AC68U has a shielding issue or not.


As for the inevitable question I'm half-expecting :

I still have no ETA as to when I'll have a firmware available for the RT-AC68U. First, I want to wait for Asus to release the new firmware which will include various fixes (starting with the USB one). I also still have a few issues left to resolve (major SMB performance issue after I merged the 979 GPL in, AiCloud from 979 doesn't compile properly due to the GPL archive not containing up-to-date build files for lighttpd), etc...

Otherwise, the rest is looking good. Stealth Mode is working (altho it's a bit redundant with Asus adding the LED button).

 

[ronan_zj]

just a quick question, could you tell me why the transfer speed is only 12.6 MB/s under 2.4G wifi? and the 49.7MB/s is the reading speed or writing speed?

 

[RMerlin]

just a quick question, could you tell me why the transfer speed is only 12.6 MB/s under 2.4G wifi? and the 49.7MB/s is the reading speed or writing speed?

Because a 300 Mbits connection results in roughly 150 Mbits throughput after you take into account encryption, packet retransmissions, etc... This is normal for wireless.

That speed was read speed. Write speed would probably be a bit lower (maybe around 30 MB/s - I didn't re-test it).

 

[BatJoe]

I assumed the issue could be resolved with a firmware update, but the fact it was broken out of the gate proves that ASUS is clearly not on their game with their router support.

 

[RMerlin]

I assumed the issue could be resolved with a firmware update, but the fact it was broken out of the gate proves that ASUS is clearly not on their game with their router support.

Ask Netgear then about shipping a dual core router with 256 MB of RAM with only one core enabled and 128 MB of RAM. They said they wouldn't fix that until this autumn - and this was about a router they released this spring (the R6250).

At least Asus will only take a few weeks to fix this, not 6 months...

 

[greggy101]

For the people worried about this, here you go.

USB 3.0 Interference reduction was Disabled (to ensure the HDD ran at USB 3.0 speed)
.

Please correct me if I am wrong - on .205 firmware disabling interference
reduction would turn this USB 3.0 port into USB 2.0
At least that was the case with my unit.
Are you on a different firmware?
Thanks

 

[nicholas]

Merlin do you believe that generally any router can ever reach speeds on usb 3 which can compare to a dedicated nas (not taking in to account better speed due to raid on the nas)...?

 

[PrivateJoker]

Merlin do you believe that generally any router can ever reach speeds on usb 3 which can compare to a dedicated nas (not taking in to account better speed due to raid on the nas)...?

A HD doesn't get better performance by being in a RAID array, if anything, it probably takes a hit. Only RAID 0, which is a non-redundant RAID (you'll see sometimes in high end video or digital photography workstations) are you using a RAID array to achieve better performance, not for file sharing.

The R&W speeds today on the Netgear R7000 reviews are probably equal to or better than the lowest entry level consumer NAS machines (like the 1 bay, sub $200 enclosures). So if you compare the most expensive consumer routers on the market with the least expensive consumer NASes, ya they're close.

I think most SOHO or "prosumer" type users that a standalone NAS appeals to will continue evolving to provide features & performance that a consumer router w/ a USB 3.0 port can't provide. Most people I know that have NASes have a largely wired backbone, whereas most consumers do not. Even with highly featured, big horsepower routers kind of closing the gap I think most people who choose to attack it with separate independent machines will continue to do so.

You could have asked the same question maybe 5ish years ago when routers started to get USB 2.0 ports that could share a drive, and probably get the same "it depends, kind of but not really" answer.

The Broadcom SoC family that the Netgear R7000 & Asus AC68U run (BCM4708x) has an extremely similar sister chip, the StrataGX (BCM5301x) was designed to run small enterprise routers and NASes.

 

[RMerlin]

Merlin do you believe that generally any router can ever reach speeds on usb 3 which can compare to a dedicated nas (not taking in to account better speed due to raid on the nas)...?

Depends on what NAS you are looking at. Some NAS come with CPUs that are barely any faster than the RT-AC68U, others come with actual Intel i3 which should be able to easily push full USB3 performance. So, performance will vary. But in general, no, a router will always be lagging behind a dedicated NAS, which usually features not only a quite faster CPU, but also a more optimized USB controller.

The way I see it, router-based storage is only good for sharing small amount of data which aren't bandwidth intensive (like music or photos), or backups (where you don't always care if the backup takes 6 hours instead of 3 hours as it happens during the night). Any serious file sharing should go with a dedicated NAS (and preferably not a 150$ entry level one - HDD included - either, cause these are probably not really better than an RT-AC68U).

 

[RMerlin]

Please correct me if I am wrong - on .205 firmware disabling interference
reduction would turn this USB 3.0 port into USB 2.0

This is the opposite. Enabling interference reduction will make the USB device downgrade to 2.0, so to avoid the radio interference introduced by the USB3 signalling.

 

[nicholas]

Thanks for the quick response guys So the limiting factor on routers is the Cpu and less optimised usb and a bit the slower and less ram.Maybe in 2-3 years routers with attached usb 3 hard drives working in raid 0 may hit the theoretical 125 MB ethernet can do.I was looking at he Synology 412+ and the Qnap ts-469l.The Synology is a bit faster but the Qnap has more inputs and gets updates faster.I really want to see a revised review for the QNAP with latest firmware to see how much better performance is.

 

[PrivateJoker]

Thanks for the quick response guys So the limiting factor on routers is the Cpu and less optimised usb and a bit the slower and less ram.Maybe in 2-3 years routers with attached usb 3 hard drives working in raid 0 may hit the theoretical 125 MB ethernet can do.I was looking at he Synology 412+ and the Qnap ts-469l.The Synology is a bit faster but the Qnap has more inputs and gets updates faster.I really want to see a revised review for the QNAP with latest firmware to see how much better performance is.

The biggest limiting factor I have in my Synology 212+ is the CPU & HD speed of the client I have connected to it. I can press 80Mb/s+ reads from my fastest machine w/ SSD drive, and 50Mb/s-ish on my slowest oldest machine w/ traditional HD.

PS, they do major updates on an annual or quicker basis, and have beta programs open to all, don't know where you read QNAP [which I still think is a very solid product] gets "updated faster". Also the last 3 revs of the NAS OS have extracted small but solidly better improvements in NAS throughput to my client devices (specifically my large Mac TimeMachine backups). So, if you want one, just get one - don't wait on the results of a particular firmware/OS results on one model.

FWIW - go look at the test machines they use to get the published #s, I recently posted the fine print on Synology's results, I guarantee that neither you or I have a machine config'ed to those specs! But to think you're going to get the published #s on what it can do is like thinking you can get 125MB/s over GbE. Yes the math works out to that but real world instantly slices out a chunk.

 

[tl2k2]

For the people worried about this, here you go.

Test setup:

- RT-AC68U (got mine last week, kindly provided by Asus)
- Laptop running Win7
- Gigabit Ethernet interface
- USB-AC53
- NexStar3 USB3.0 enclosure, with a 2.5" Toshiba 750 GB HDD, formatted as NTFS
- USB 3.0 Interference reduction was Disabled (to ensure the HDD ran at USB 3.0 speed)

The first test was done with the laptop connected over gigabit to a LAN port on the router. I copied a 11 GB file from the router to the laptop (this is a compressed Acronis True Image backup). During the copy, I had my Nexus 7 tablet connected to the router on the 2.4 GHz band. A Speedtest was run at some point during the transfer, to generate some traffic. No disconnection, and the speedtest ran normally.

The speed never dropped below 49 MB/s, hitting around 52 MB/s at one point. This confirms that Asus did resolve the USB3 performance with a firmware update, and this router will match any of the other USB3-equipped routers.

The second test was to test the stability of the 2.4 GHz connection while there was activity on a USB 3.0 hard drive. Performance is lower for good reasons - this is a 300 Mbits USB-AC53. (I normally use the internal card on my laptop, but I wanted to also test the 802.11ac connectivity - I didn't screenshot it because it's irrelevant to this test). This time I copied a smaller 1.2 GB AVI (cause it would have taken far too with the 11 GB test file). There was no disconnection from the 2.4 GHz radio during the test, and speed remained constant:

So with these results, a few notes:

1) USB3 performance can be fixed with a firmware update (and Asus actually did).
2) It's possible to have a stable 2.4 GHz connection while accessing a USB3 drive on the RT-AC68U
3) HOWEVER: stability on the 2.4 GHz band with USB3 device isn't just dependant on the router itself. It will also depend on the HDD shielding (mine is an aluminium case, which was about one feet away from the router) and of the cable itself (the cable NexStar provides seems pretty well shielded - definitely thicker than many typical USB2 cables).
4) Regarding Tim's stability issues while copying the file with the disk running at USB3: this could be caused either by the firmware version he was using, or his test HDD is just more sensitive to radio interference than mine is. To reach any conclusion on this one, we'll have to wait for Tim to re-test once Asus releases the new firmware. For now, it's too early to tell if the AC68U has a shielding issue or not.


As for the inevitable question I'm half-expecting :

I still have no ETA as to when I'll have a firmware available for the RT-AC68U. First, I want to wait for Asus to release the new firmware which will include various fixes (starting with the USB one). I also still have a few issues left to resolve (major SMB performance issue after I merged the 979 GPL in, AiCloud from 979 doesn't compile properly due to the GPL archive not containing up-to-date build files for lighttpd), etc...

Otherwise, the rest is looking good. Stealth Mode is working (altho it's a bit redundant with Asus adding the LED button).

hi,let say if asus correct the usb 3.0 speed issue in future firmware, do you think this speed reflect the interfering enable or have to be disable because if already fast right now with that enable.

 

[RMerlin]

hi,let say if asus correct the usb 3.0 speed issue in future firmware, do you think this speed reflect the interfering enable or have to be disable because if already fast right now with that enable.

The results I posted were with interference reduction disabled. Enabling the option will force the disk to run at USB2 speed, which means around 25 MB/s - what Tim posted in his review (where he had to enable that option for stability reasons).

 

[PrivateJoker]

I don't believe this has been posted yet, but here is Intel's 22 page white paper shared w/ USB trade group about a year and a half ago on 2.4ghz/USB 3.0 interference.

http://www.usb.org/developers/whitepapers/327216.pdf

Main points seem to be that the following things are important: shielding/grounding of USB 3.0 connectors inside devices (routers, computers, etc); cable's shield quality; and distance between USB 3.0 peripheral and the device it's plugged into (longer, non-physically touching distances are better).

Also, reading between the lines, seems like utilizing 5ghz as much as possible and maybe lowering your 2.4ghz TX power on AP to lowest # needed might also help.

And it might make sense to use a better/more shielded cable than the one that comes free w/ the drive. This is total speculation on my part, but couldn't hurt to check.

 

[mlai]

Yeah, a NAS will always be faster IN THE RIGHT CONDITIONS.....

Below is what I get from my QNAP TS559Pro2:




But then, I am running Netgear GS724T, QNAP NAS is on 5 disks RAID6, Both ethernet ports in LAGS to the switch, and client on a I7-4770K on all SATA3 SSDs, GB ethernet connectivity.

But if for home portable device use, I would say it is a bit over kill And the router USB3 on the AC68, if works correctly, oughta be sufficient for 80%+ of home use scenarios......

 

[nicholas]

The biggest limiting factor I have in my Synology 212+ is the CPU & HD speed of the client I have connected to it. I can press 80Mb/s+ reads from my fastest machine w/ SSD drive, and 50Mb/s-ish on my slowest oldest machine w/ traditional HD.

PS, they do major updates on an annual or quicker basis, and have beta programs open to all, don't know where you read QNAP [which I still think is a very solid product] gets "updated faster". Also the last 3 revs of the NAS OS have extracted small but solidly better improvements in NAS throughput to my client devices (specifically my large Mac TimeMachine backups). So, if you want one, just get one - don't wait on the results of a particular firmware/OS results on one model.

FWIW - go look at the test machines they use to get the published #s, I recently posted the fine print on Synology's results, I guarantee that neither you or I have a machine config'ed to those specs! But to think you're going to get the published #s on what it can do is like thinking you can get 125MB/s over GbE. Yes the math works out to that but real world instantly slices out a chunk.

Yea i know what you mean.I don't think i have such a bottleneck... I'm on an i7 920 at 4.2Ghz 6gb of 2ghz ram and 2x MZ-7TE1T0BW in raid 0.From the 2 nas which would you go for..? Ok over gigabit ethernet i'm hopeful i could hit around 100-110 ish MB/s Read-Write which i think is doable...

 

[Asusnet]

Please correct me if I am wrong - on .205 firmware disabling interference
reduction would turn this USB 3.0 port into USB 2.0
At least that was the case with my unit.
Are you on a different firmware?
Thanks

"Reducing USB 3.0 Interference" Enabled: Reduces USB 3.0 top speeds to USB 2.0 speeds, but decreases interference and attenuation over 2.4Ghz.

"Reducing USB 3.0 Interference" Disabled: Allows USB 3.0 top speeds, but can cause interference/attenuation over 2.4Ghz.

This should be the case for both FW .205 and .214.

 

[greggy101]

"Reducing USB 3.0 Interference" Enabled: Reduces USB 3.0 top speeds to USB 2.0 speeds, but decreases interference and attenuation over 2.4Ghz.

"Reducing USB 3.0 Interference" Disabled: Allows USB 3.0 top speeds, but can cause interference/attenuation over 2.4Ghz.

This should be the case for both FW .205 and .214.

On .205 in my case - when disabled I was able to mount USB 3.0 HD on the USB 3.0 port at the very slow speed (USB 2.0) and it would disappear at times.
It would not mount at all when Enabled.
Once installed .214 - it worked good. At least it got properly recognized as USB 3.0 HD on USB 3.0 port when Disabled.

 

[llc00ljoel]

On .205 in my case - when disabled I was able to mount USB 3.0 HD on the USB 3.0 port at the very slow speed (USB 2.0) and it would disappear at times.
It would not mount at all when Enabled.
Once installed .214 - it worked good. At least it got properly recognized as USB 3.0 HD on USB 3.0 port when Disabled.

I had the same exact issue - seemed like the drive disappeared with the older firmware. Thankfully the latest firmwares on the AC68U resolved that.