ASUS RT-AC87U Dual-band Wireless-AC2400 Gigabit Router Reviewed

[dsring]

Hey everyone, hopefully this is the right place for this question.

By the way, love the site and have been reading for a while.

Anyway, my issue is with my Xbox One. I just got this router and an Xbox One recently and I can't get the NAT to be "Open" on the Xbox One. It is Strict currently and won't budge. UPnP is enabled as well.

Does anyone know what the issue here might be? I can't figure it out.

Thanks!

EDIT: Well, it appears to be Open now. Didn't do anything other than a power cycle of the Xbox One. Hopefully it stays open!

EDIT 2: Now I'm having NAT problems with my Xbox 360 and PS4, ugh.

I also have NAT problems with gaming consoles that I did not have with the RT-N66U. My theory is that UPnP is not working properly on this router. I did submit a ticket with Asus for this issue about a month ago, and have yet to hear anything back from them. It should 'open' the NAT on one of the consoles if you put its IP in the DMZ, but you should not have to do that.

 

[yooper2001]

5 ghz still stinks

Due to placement of any router requiring to be in my basement, I was exited to buy this router as it would finally be strong enough to fill my 2700 sq ft house. Most of my devices were 2.4 so everything went smooth there. I decided to try my Asus 66U as a media bridge, and it connected with a strong signal....I then tried to stream an .iso file.....that's when I realized something was up.....when connected its hella fast.....unfortunately it drops and reconnects every 5 to 15 minutes....causing freezes in streaming...I then tried my main media bridge a trendnet tew-800. Again...connects with 70% + signal strength. insane speeds until random drops. I have a total of 8 different 5 GHz devices and it does it on all of them. I have upgraded the bios and tried every setting and speed and channel combination with the same results. The 2.4 channel works fine, but I would have bought a 50 dollar router if that is what I wanted....I am praying that another bios fix will solve this piece of junks issues or I will be forced to go back to my 66u. If anyone else finds a fix or has advice it would be greatly appreciated.

 

[TTT]

Due to placement of any router requiring to be in my basement, I was exited to buy this router as it would finally be strong enough to fill my 2700 sq ft house. Most of my devices were 2.4 so everything went smooth there. I decided to try my Asus 66U as a media bridge, and it connected with a strong signal....I then tried to stream an .iso file.....that's when I realized something was up.....when connected its hella fast.....unfortunately it drops and reconnects every 5 to 15 minutes....causing freezes in streaming...I then tried my main media bridge a trendnet tew-800. Again...connects with 70% + signal strength. insane speeds until random drops. I have a total of 8 different 5 GHz devices and it does it on all of them. I have upgraded the bios and tried every setting and speed and channel combination with the same results. The 2.4 channel works fine, but I would have bought a 50 dollar router if that is what I wanted....I am praying that another bios fix will solve this piece of junks issues or I will be forced to go back to my 66u. If anyone else finds a fix or has advice it would be greatly appreciated.

The 5GHz is still not stable on this router.

 

[sfx2000]

Been busy with other stuff - so this slipped under my radar a bit - but it's worth discussion...

DYIPC posted a basic block diagram of the RT-AC87U, and it's interesting what design choices they made.

DISCLAIMER - I'm looking at things from the outside, and have signed no NDA's and have no inside knowledge of Broadcom or Quantenna, much less Asus

Let's start with the router core - this is a BCM4709 - this is a dual-core ARM Cortex-A9 dual CPU platform with an embedded 1Gb ethernet switch - it's got a BCM4360 on the 2.4GHz side (3 stream 11n/11ac) with 256MB of RAM - the BCM4360 is sitting on the PCIe 2.0 bus, and I think this is generally a very solid design here - fast, stable, and now mature...

Where it gets interesting, and where I question the design is on the right hand of the diagram - the QSR1000 complex, the 1Gb ethernet port, and how it interfaces back into the BCM hardware...

From an RF perspective, it's pretty solid perhaps, however, what's not clear in the diagram is the 2.4GHz vs. 5GHz antenna's on the radiated side - as we don't see 7 antenna's on the final design, so those resources are shared, and not likely optimal.

Going over to the right side of the diagram - e.g. the 11ac 5GHz portion..

From a HW perspective - the 5Ghz complex is an AP with a 1Gb interface going into it, with it's own RAM/Flash.

From an integration view - this must be a tough problem - and might be why the 87U seems to have stability problems only on the 5Ghz side.

The QSR is on BCM RGMII - this is a 1Gb connection - with reduced signal paths - not a big deal, but one that does require some attention on the PCB layout and layers - this is interesting as more layers on the PCB, more cost here - but over a production run... This is basically a QSR AP hanging off a Broadcom reference design - read on... the QSR runs it's own SW, independently of the BCM 2.4Ghz platform

The BCM platform, on RGMII treats the QSR complex as just another 1Gb connection - it's up to the ASUS gui to configure both here.

Why did ASUS include the RT8211E 1Gb interface and expose it to the LAN? If you have a wired connection to that LAN port, one runs the risk of saturating the RGMII interface back to the BCM core switch, and starving the QRS 5GHz radio - I would not have done this.

Again, at the end of things - the AC87U is really two devices in 1, IMHO - a strong broadcom 2.4GHz design along with a Quantenna AP that is loosely coupled.

It seems rather rushed to market, and not as tightly integrated as some of the other solutions I've looked at.

rmerlin had an interesting post from the SW side - can't find it at the moment.

 

[RMerlin]

Why did ASUS include the RT8211E 1Gb interface and expose it to the LAN? If you have a wired connection to that LAN port, one runs the risk of saturating the RGMII interface back to the BCM core switch, and starving the QRS 5GHz radio - I would not have done this.

That's an interesting point, unless the block diagram doesn't tell the accurate story there.

If someone wanted to test that theory - the Realtek port is LAN1. Someone could try having a large data transfer running between LAN1 and LAN2, and see what happens to the 5 GHz performance then. If it does drop like your theory says (and again, if that block diagram is accurate, then your theory should also be accurate), then it definitely shows a pretty serious flaw in that design.

If that's really the case, then it means people should try to avoid putting anything likely to generate a large amount of traffic between LAN1 and either LAN2-4, WAN, or the 2.4 GHz band.

Even in the case of traffic between LAN1 and the 5 GHz band I have to wonder if that traffic would stay local to the QTN machine, or if that traffic would go from the 5 Ghz radio, to the Broadcom side, then back to QTN (saturating that RGMII channel in both directions), then finally out through the Realtek port.

So many scenarios someone with enough free time could test...

 

[T-Nation]

Funny this came up... the past several days I've been digging through the reviews I could find and see what these "team ports" are all about. Like Tim said in his review there is... nothing... nada! But based on everything I've read I would have to assume port 1 was tied to the 5Ghz side. This like sfx2000 noted would make me think port 1 would become saturated with heavy traffic loads.

Now with people having issues with streaming might want to switch to ports 2-4 because it makes me wonder if its the saturation between port 1 and 5Ghz transfer causing the speed drops or so called 5Ghz instability.

So, I decided to ordered one of these AC87's to test it against the r7500 ( I personally hate the genie interface btw ). I was going to plug everything into ports 2-4 and stress test, then switch to port 1 to see if it bombs. Personally I see the RGMII connection a huge design flaw but as noted, the block diagram maybe incorrect on how everything is at play on this design.

 

[sfx2000]

Even in the case of traffic between LAN1 and the 5 GHz band I have to wonder if that traffic would stay local to the QTN machine, or if that traffic would go from the 5 Ghz radio, to the Broadcom side, then back to QTN (saturating that RGMII channel in both directions), then finally out through the Realtek port.

So many scenarios someone with enough free time could test...

Going from the Realtek over to the 5GHz - this is layer 2/3, so I suspect it would stay local on the Quantenna - but with Asus' traffic shaping/prioritization, maybe it does have to go offboard over to the BCM switch/router SOC.

Hmm... I'm not as savvy on the Asus SW here as others might be.

 

[RMerlin]

Going from the Realtek over to the 5GHz - this is layer 2/3, so I suspect it would stay local on the Quantenna - but with Asus' traffic shaping/prioritization, maybe it does have to go offboard over to the BCM switch/router SOC.

Hmm... I'm not as savvy on the Asus SW here as others might be.

I suspect there must be some of that traffic hitting the BCM side of things because of the numerous network functionality that are implemented on that side: the bridge, IGMP/EMF proxying, and so on.

This whole design and its implications aren't really clear for me either.

 

[LoneWolf]

I suspect there must be some of that traffic hitting the BCM side of things because of the numerous network functionality that are implemented on that side: the bridge, IGMP/EMF proxying, and so on.

This whole design and its implications aren't really clear for me either.

Assuming the diagram is correct, I don't get why there would be one LAN port (LAN 1, Realtek) separated from the other three LAN ports in the first place. What would be the point for this? You would think there would be four LAN ports tied to a single switch, in a logical arrangement.

I might be able to understand (under some circumstances) having the WAN port separate, but not the LAN.

 

[sfx2000]

Assuming the diagram is correct, I don't get why there would be one LAN port (LAN 1, Realtek) separated from the other three LAN ports in the first place. What would be the point for this? You would think there would be four LAN ports tied to a single switch, in a logical arrangement.

I might be able to understand (under some circumstances) having the WAN port separate, but not the LAN.

The Broadcom embedded switch on the router SOC only has 5 ports, and they're all used - 1 for WAN, 1 for the RGMII over to the Quantenna solution for 5GHz, and three LAN ports... two hit a product spec of 4 LAN ports, hanging one off the Quantenna SOC would meet that goal.

In any event, from the outside looking in, I wouldn't hang the WAN port off that interface, IMHO, as this would make Asus job even more of a challenge - esp. when they have a stable SW based for the BCM solution that pretty much does what they need - it's risk reduction.

Which kind of makes sense from a product perspective, and given the short run to market, probably a pragmatic decision.

 

[sfx2000]

I also want to go on record here - I have no agenda, and as such, it's an objective view.

I have no direct knowledge of the AC87U, my analysis is based on the block diagram only. There's no way to verify if the block diagram is accurate or not, but my comments are based on design experience - it looks good, and it's a valid approach if I were tasked with getting something to market on a short timeline and minimal budget.

I think this is an ambitious product, and Asus has pretty much nailed their goal - being first to market with an early 11ac Wave 2 solution.

Kudos to them!

 

[sfx2000]

I suspect there must be some of that traffic hitting the BCM side of things because of the numerous network functionality that are implemented on that side: the bridge, IGMP/EMF proxying, and so on.

This whole design and its implications aren't really clear for me either.

I agree... the integration challenges on the SW side are fairly significant, as the Quantenna complex pretty much runs on it's own...

I finally did track down your excellent SW discussion on the 87U - confirms what I suspected once having gone thru a high-level block review.

nicely done there.

http://forums.smallnetbuilder.com/showthread.php?t=18415

sfx

 

[sfx2000]

I have no direct knowledge of the AC87U, my analysis is based on the block diagram only. There's no way to verify if the block diagram is accurate or not, but my comments are based on design experience - it looks good, and it's a valid approach if I were tasked with getting something to market on a short timeline and minimal budget.

BTW - the AC87U block diagram was hosted on an ASUS.com site, so this lends come credibility there.

Click the link and see for yourself.

sfx

 

[sfx2000]

I suspect there must be some of that traffic hitting the BCM side of things because of the numerous network functionality that are implemented on that side: the bridge, IGMP/EMF proxying, and so on.

This whole design and its implications aren't really clear for me either.

Did some thinking here... if the ASUS special sauce is in the BCM chipset, then treat the Quantenna 5Ghz complex as an distinct AP associated with it over ethernet, so everything still applies....

Just don't hang a wired connection off the LAN1 port - if what your suggesting that LAN1 is on the Quantenna Realtek 1GB interface is true...

Most of the ASUS special sauce is in the Router side, not the wifi side - It's an odd design, no doubt...

 

[scavenger]

the Realtek port is LAN1.
[...]
Even in the case of traffic between LAN1 and the 5 GHz band I have to wonder if that traffic would stay local to the QTN machine, or if that traffic would go from the 5 Ghz radio, to the Broadcom side, then back to QTN (saturating that RGMII channel in both directions), then finally out through the Realtek port.

Wow, is LAN1 really the Realtek port? On my 87U it's teamed with LAN2, or at least that's what the case label tells me.

It would be an even bigger sin, I would think, to have such an asymmetrical path for a (potential) bonded pair, especially when one of the pair is squirting through two RGMII interfaces, one of them thick with 5GHz traffic.

 

[RMerlin]

Wow, is LAN1 really the Realtek port? On my 87U it's teamed with LAN2, or at least that's what the case label tells me.

It would be an even bigger sin, I would think, to have such an asymmetrical path for a (potential) bonded pair, especially when one of the pair is squirting through two RGMII interfaces, one of them thick with 5GHz traffic.

I will have to physically check to confirm it. From the software side of it, it's LAN1. Port order could possibly be reversed however (it's the case with a few router products).

 

[sfx2000]

I will have to physically check to confirm it. From the software side of it, it's LAN1. Port order could possibly be reversed however (it's the case with a few router products).

I think you're correct...

Looking at the Board Pix - http://www.lostrealm.ca/temp/rtac87u.jpg

The BCM - can follow the traces over from the 4709 on the right hand side over to the magnetics for the WAN port and the LAN ports - so LAN2-4 is on the BCM, and Quantenna is furthest away from the WAN port as LAN1

Would have been cool to get a pic with the cans/heat sinks removed to confirm things.

Dug thru the AC87U User Guide, and there is only mention about Teaming for Link Aggro on LAN1 and LAN2 on the fuctional port description page, there is no mention of it anywhere else - so I wonder if this got turned off and they couldn't update the artwork in time for the user guide (believe it or not, the User documentation has a fairly long lead time as it goes thru many layers of review).

I think at the end of the day, if one can avoid using LAN1, that would be a very good thing.

sfx

 

[paraplu]

Quoted from Asus' http://pcdiy.asus.com/2014/08/august-2014-the-best-802-11ac-router-rt-ac87u-rt-ac87r/

"Users looking to maximize media bridge mode performance can enable teaming functionality allowing for greater than gigabit throughput."​

Has someone already tried bridge mode? I would be interested to see if there is an option for port aggregation in the web interface.

 

[RMerlin]

Quoted from Asus' http://pcdiy.asus.com/2014/08/august-2014-the-best-802-11ac-router-rt-ac87u-rt-ac87r/

"Users looking to maximize media bridge mode performance can enable teaming functionality allowing for greater than gigabit throughput."​

Has someone already tried bridge mode? I would be interested to see if there is an option for port aggregation in the web interface.

There's no option for link aggregation in the webui's code.

In fact, I still don't know if this feature was ever fully implemented or not. I saw some code added to the GPL source code a few months ago, but it never got user or enabled.

 

[scavenger]

There's no option for link aggregation in the webui's code.

In fact, I still don't know if this feature was ever fully implemented or not. I saw some code added to the GPL source code a few months ago, but it never got user or enabled.

I can imagine the software guys, well after functional spec and the case was printed, saying of the hardware guys...

"They did what?"

Though it's not too hard to patch a case mold. Free stickers for everyone.