What's new

AC1900 First Look: NETGEAR R7000 & ASUS RT-AC68U

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Status
Not open for further replies.
Guys I was reading the HW WiKi page for R7000: http://wikidevi.com/wiki/Netgear_R7000

and I found what kind of memory Netgear is using: RAM1: 256 MB (Samsung K4B2G1646E-BCK0)

and if I am reading it right, those memory chips can go up to 2133MHz with relaxed timings:
http://www.samsung.com/global/business/semiconductor/file/product/ds_k4b2g1646e_rev121-0.pdf

so in theory, someone can overclock R7000 from 1000/800 to 1200/1066 and that will make Netgear one insanely fast router!

ASUS, in contrast, chose [omitted] memory for AC68U - it maxes out at 1600MHz: http://www.esmt.com.tw/DB/manager/upload/M15F2G16128A.pdf - same as on AC56U

Now I am just dying to get R7000 and put DD-WRT on it :)
 
Question ...

Hi Tim,

Slightly off topic but with privacy at a premium these days I think there'd be a strong market and interest in VPN based home security routers that would, ideally, offload the encryption overhead to a hardware chip so as to offer the best performance. I'm not sure how common this kind of product really is or whether they're even priced for the home consumer market. But it's definitely something that could be in high demand if produced and marketed properly.

Your security router section is rarely updated with products so my guess is that the ideal solution I'm wanting is probably not very common. So I wanted to solicit your opinion on the performance hit of running some of these consumer routers as a dedicated VPN gateway or whether overheating is something I really even need to worry about anymore. It used to be that having a dedicated onboard encryption chip would work wonders but with the increased processing power you can find in home consumer routers these days it might not really be necessary. You didn't mention any VPN performance here, I'm assuming maybe because the out of the box firmware didn't offer it. I'm not worried about that so much as many of us either roll our own or install 3rd party firmware to get the most out of the features the hardware offers, but it is something I've been wondering for awhile as an investment I'd like to make.

How do you figure VPN would preform on routers such as the Netgear R7000 which seems to have the best processing performance thus far? Having a VPN Gateway that also acts as my home wireless gateway would rock.
 
I am running VPN client on my overclocked AC56U... and to my knowledge, no stock firmware offers you an option to run VPN client - only third party like OpenRouter (in case of R7000) or Merlin allow you to do that.

Yes, there is a quite measurable performance drop and it does depend on CPU/memory clocks.

First, you need to establish a baseline:
- Naked (no VPN) Internet speed on wired client - measure using testmy.net
- Internet speed on a reasonably fast system with VPN client installed and running
- same, but this time the router runs the VPN

if your internet speed is <25Mbps, I expect R7000 with stock clocks (1000,800) should be able to handle it with tolerable drop - let's say 22? The router may get warm - I do not know how good is Netgear's cooling)

If you are on much higher speed - like 100Mbps - the impact is hard to predict, but I do expect it to be very noticeable...
 
I am running VPN client
First, you need to establish a baseline:
- Naked (no VPN) Internet speed on wired client - measure using testmy.net
- Internet speed on a reasonably fast system with VPN client installed and running
- same, but this time the router runs the VPN

if your internet speed is <25Mbps, I expect R7000 with stock clocks (1000,800) should be able to handle it with tolerable drop - let's say 22? The router may get warm - I do not know how good is Netgear's cooling)

If you are on much higher speed - like 100Mbps - the impact is hard to predict, but I do expect it to be very noticeable...

Thanks for your insight. You say 22, so are you saying that a router at 25Mbps or just under would only drop 10-12%? That's really not bad at all. What sort of hit do you take on your AC56?

My speeds on average push anywhere between 75-100Mbps measured at speedtest.net so I was thinking I'd keep a portion of my traffic unencrypted that comes in through my feed and VPN everything else if the hit is too considerable. But that's not as secure or preferable as being able to op completely from VPN. 10-12% I'd take gladly. 50% ... probably not.
 
Thanks for your insight. You say 22, so are you saying that a router at 25Mbps or just under would only drop 10-12%? That's really not bad at all. What sort of hit do you take on your AC56?

My speeds on average push anywhere between 75-100Mbps measured at speedtest.net so I was thinking I'd keep a portion of my traffic unencrypted that comes in through my feed and VPN everything else if the hit is too considerable. But that's not as secure or preferable as being able to op completely from VPN. 10-12% I'd take gladly. 50% ... probably not.

Honestly I do not know what the hit is going to be on 100Mb connections... I am running overclocked AC56U at 1200 CPU/800 memory in order to achieve low impacts (this thread has the details http://forums.smallnetbuilder.com/showthread.php?t=12578, one the pages has my results with VPN speeds ), I do expect 1000/800 to take a significant hit; how bad - hard to predict, need to try. ASUS stock 800/533 yields abysmal performance running VPN client even on 25Mb connection.

Also the firmware has big impact on performance too - if you are considering R7000, that limits your FW choice to OpenRouter - and I have no idea how fast is that firmware. Big part of the problem is that OpenVPN client (used on routers) is only using single core and that limits performance on dual core CPUs big time...
 
Also the firmware has big impact on performance too - if you are considering R7000, that limits your FW choice to OpenRouter - and I have no idea how fast is that firmware. Big part of the problem is that OpenVPN client (used on routers) is only using single core and that limits performance on dual core CPUs big time...

Why wouldn't dd-wrt be developed for the R7000? Of course right now it's too new, but I would assume that porting it would be in the works after it comes out and the source is released.
 
Why wouldn't dd-wrt be developed for the R7000? Of course right now it's too new, but I would assume that porting it would be in the works after it comes out and the source is released.

It is already available as of Oct 1: http://www.myopenrouter.com/download/51635/DD-WRT-Firmware-for-NETGEAR-R7000-Wi-Fi-Router/

My point: it is unknown at this point what is the performance of DD-WRT vs. stock. Simply not enough R7000 are sold yet and not enough people are playing with them... and yes, VPN client support and performance is a very important factor in my buying decision.
 
I guess I'm not crazy then.

No, you are not :)


This screech, static, whatever sound is sort of annoying.

Here's a link to the sound file that describes the sound I hear:
http://www.dslreports.com/r0/downloa...b/IMG_0073.MOV

Indeed it is, and the reason why that EA6700 went back so fast. I must admit the R7000 has it, but not nearly as an issue as it can be with others. I feel that it would be more annoying to me if I had user using wireless while I was near it.

I do feel that there needs to be a caveat emptor about this. I hear it, Froach hears it, and you recorded it. BTW, that is the sound too; like a hard drive screechy-electric-buzzy sound.

Oh to note, when I placed the EA6700 exclusively to 802.11n it still made this sound.


Seems like this is plaguing Broadcom setups. Netgear said that they tested this when I reported it, but they didn't hear any sound. I just wonder if they weren't listening close enough or not.

Possibly a "problem" exclusive to Broadcom. That is why I did asked the curious question if other 802.11ac chips from other vendors do this too. And this is one of the issues of sound tests. They are very, very relative.
 
My point: it is unknown at this point what is the performance of DD-WRT vs. stock. Simply not enough R7000 are sold yet and not enough people are playing with them... and yes, VPN client support and performance is a very important factor in my buying decision.

If it is affordable, and VPN is an considering factor, I would make a pfSense system/router with hardware offloading for VPN. I think you would get more mileage out of it. Personally, that is what I would have done long time ago but is still unnecessary for my needs.

http://www.pfsense.org/
 
Just did some testing with the VPN server on the R7000,

My speeds are 39-41mbit/s for both uploads and downloads.

If I telnet into the router and run top, the VPN server is only using a single core

when idle or very little traffic, the CPU usage is 1%

and at full saturation, (transferring data over the LAN between 2 computers) it maxes out 1 core.

If the VPN service was allowed to use both cores and probably just run at a lower priority to avoid lagging the router, it could potentially double the VPN throughput

Hopefully smallnetbuilder can do a proper test of the VPN, and possibly also retest the 256 QAM since it is now confirmed that the router can report the link rate.
 
Power consumption

My WNDR3700v1 just died so I'm in the market for a replacement - possibly one of these two routers. I've been reading a lot of customer reviews of higher-end products from these manufacturers, and with some users saying they can run a bit hot, combined with dual core CPU implementation, I wonder about longetivity and my utility bill.

During your review, did you get a sense for which router produced less heat, and did you plug either unit into e.g. a kill-a-watt at idle? I haven't found any information about power consumption for the R7000 so far and I'd love to compare before making a purchase decision in the next few days.
 
I don't bother measuring power consumption in routers. The few spot checks I have done showed 7 W or so. Your power bill is safe.
 
During your review, did you get a sense for which router produced less heat, and did you plug either unit into e.g. a kill-a-watt at idle? I haven't found any information about power consumption for the R7000 so far and I'd love to compare before making a purchase decision in the next few days.

Not directly answering your question, but FWIW I attached a killawatt directly to my N66, an old Linksys WRT-54gl, and my Synology DS-212+ the other day. My Asus averages about 12w.

 
Thanks. My old WNDR3700v1 ran about 7W so that suggests newer models are using about 70% more power. While still low, it explains why they're getting hotter. The fact that the R7000 power brick is so large also made me curious.
 
Just did some testing with the VPN server on the R7000,

My speeds are 39-41mbit/s for both uploads and downloads.

If I telnet into the router and run top, the VPN server is only using a single core

when idle or very little traffic, the CPU usage is 1%

and at full saturation, (transferring data over the LAN between 2 computers) it maxes out 1 core.

If the VPN service was allowed to use both cores and probably just run at a lower priority to avoid lagging the router, it could potentially double the VPN throughput

Hopefully smallnetbuilder can do a proper test of the VPN, and possibly also retest the 256 QAM since it is now confirmed that the router can report the link rate.

Hi Razor, thanks for sharing.
there is a quite difference between running VPN Server (i.e. connecting from outside) and running VPN Client (your entire network traffic goes thru the VPN tunnel)
Will you consider trying OpenRouter DD-WRT + VPN Client? Like connecting to Private Internet Access VPN or another similar VPN service?
 
weird how the DD-WRT page doesn't list QoS?
Yeah I saw this CFW out of box support...I'm curious of Merlin or Shibby's Tomato will eventually make it to the R7k.
Do you have any experience with the cloud-based QoS yourself Tim?

merlin...what are the chance of you bringing merlin to the R7k given that you will be working on the ac68u? Would same chip but different radios/firmware make it very difficult to code over?

None on the R7K, definitely on the ac68u tho no ETA.
 
I tried the dd-wrt build at myopenrouter and it did not work, sometimes it will try to flash, other times it will complain, overall it is not a well made dd-wrt release

From my experience, (at least with the r6300V2), the openvpn is not very reliable in dd-wrt (for me, it kept failing to connect and when it did, it was extremely slow and would eventually stop working all together.)

For general routing and wifi performance, both tomato and dd-wrt have been slower than the offerings on the stock netgear firmware. (the only areas where they showed improvement was in USB storage performance)

I have mostly given up on 3rd party firmware, the drop in general routing performance, and the large drop in wifi performance, have made them less appealing.
 
Last edited:
Just did some testing with the VPN server on the R7000,

My speeds are 39-41mbit/s for both uploads and downloads.

Interesting. What are your speeds at when you run VPN server on your main box? And what kind of non-VPN speeds do you get?

If the VPN service was allowed to use both cores and probably just run at a lower priority to avoid lagging the router, it could potentially double the VPN throughput

Ideally yes. Or even just switching to using the other processor exclusively when the VPN dedicated processes gets close to to maxing its load could be a quick and dirty, albeit less efficient solution.

I have mostly given up on 3rd party firmware, the drop in general routing performance, and the large drop in wifi performance, have made them less appealing.

Really? That seems counterintuitive to me. Early development of any 3rd party firmware is going to necessarily be buggy. But the benefits after some time of active development should far exceed what you've come to expect from stock. That's assuming there's enough interest and critical mass to spur active dev of course.
 
Prior to this modem, I had a medialink http://www.amazon.com/dp/B0044YU60M/?tag=snbforums-20

I also had and asus usb-n53 thank linked the router to my HTPC

Hulu worked ok, some mild lagging but nothing too painful
NHL CenterIce was horrible
No wifi reception outside the house (ie smartphone in backyard)

Waited for r7000 reviews, SNB finally reviewed it vs the ASUS ac68 and the R7000 came out on top. I went to a locak B&M store and got it.

Hooked up and linked up in about 5-10 min, setup was a snap! Devices connected without a problem

Using USB-N53 at full potential (maxed at 300 mbps)

Hulu now works flawlessly, NHL Gamecenter is much better.

Can get reception well outside the house

I have a PCE-ac68 coming in the mail today, will update!

******* Got the PCE-AC68 installed
I purchased the R7000 Netgear AC router

I used it with a USB Asus N53 wireless adapter. At 5Ghz, I got 300Mbps per the asus utility (pretty much maxed out the wireless adapter)

I installed the PCE-ac68 just now. Install was easy, just slides into the PCI slot, restarted. Deleted the old N53 driver and installed the new driver from the disc (same version as the one online).

Per the ASUS utility, I am getting 433 Mbps .

A lot lower than the 1300Mbps I thought I'd get.

I played around with the antenna settings and I have full 5 bars (Router is next room over)

Unless someone can advise me otherwise, this does not seem to get the 1300Mbps that I thought 802.11ac would get me. Any advice?
 
Status
Not open for further replies.

Support SNBForums w/ Amazon

If you'd like to support SNBForums, just use this link and buy anything on Amazon. Thanks!

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top