ASUS ROG Rapture GT-AC5300 Wireless-AC5300 Tri-Band Gaming Router Reviewed

[thiggins]

Heck, as a former Router/AP systems design guy - this is a good place to be, honestly

For you, maybe. Regular consumers should not be made to be unwitting beta testers.

 

[pege63]

FWIW - I spent $350 for a dedicated pfSense box without wireless - and that's money well spent, considering needs at the time...

Not everybody is Router PROs and money does not grow on trees yeat

 

[thiggins]

Not everybody is Router PROs and money does not grow on trees yeat

That was the point I was trying to make.

 

[sfx2000]

Most of the candidate customers for this device - they've probably followed Asus across multiple devices. These folks know that any time a new device ships, there's going to be some opportunities for improvement.

[opinion] == The device performed well in Tim's testing overall - good enough across all parts... Asus knows this is a premium/flagship device, and fixes will be forthcoming. I do take exception that people take a couple of small issues, and blow them way out of proportion, and this is what this thread has turned into.

Could Asus have done better? Maybe, we won't likely know, as those decisions are part of the daily Project Meetings that all companies have when a product is close to shipping.

 

[Xentrk]

That CPU also has the potential to reach pretty high performance levels with VPN, as it contains an AES crypto engine, so it's not just about the increase in clock speed or IPC. I don't know however if Asus leverages it in their firmware.

As someone who relies on access to streaming media from USA from the other side of the globe, VPN is very important to me. The first thing I always zero in on when these reviews come out is the CPU capabilities and the number of cores. Adding a VPN speed test to the router reviews at a few different encryption levels would be a nice addition.

Thank you @thiggins for the excellent review of the router. I was curious about it since it was announced.

 

[CaptainSTX]

As someone who relies on access to streaming media from USA from the other side of the globe, VPN is very important to me. The first thing I always zero in on when these reviews come out is the CPU capabilities and the number of cores. Adding a VPN speed test to the router reviews at a few different encryption levels would be a nice addition.

Thank you @thiggins for the excellent review of the router. I was curious about it since it was announced.

I agree that it would be interesting to see "lab" tests of what a router can do with VPN with various levels of encryption you also need to test at various WAN speed levels. A router that can give you 95% of your ISP through put on a 75/10 connection may not handle 95% at 300 Mbps, 500 Mbps or 1000 Mbps.

Also some people want to run a VPN client and others a server. Just as WAN- LAN speeds can very from LAN-WAN speeds s0 possibly can VPN client and server speeds.

While adding VPN testing to the all ready involved router tests would be time consuming for the testers it certainly would be interesting to see the comparisons for some of the top of the line newer routers with fast processors.

 

[Xentrk]

I agree that it would be interesting to see "lab" tests of what a router can do with VPN with various levels of encryption you also need to test at various WAN speed levels. A router that can give you 95% of your ISP through put on a 75/10 connection may not handle 95% at 300 Mbps, 500 Mbps or 1000 Mbps.

Also some people want to run a VPN client and others a server. Just as WAN- LAN speeds can very from LAN-WAN speeds s0 possibly can VPN client and server speeds.

While adding VPN testing to the all ready involved router tests would be time consuming for the testers it certainly would be interesting to see the comparisons for some of the top of the line newer routers with fast processors.

Excellent point about VPN speeds correlating with native WAN speeds. My VPN speeds improved when my fiber connection was upgraded from 100Mbps to 200Mbps. I then was able to go from no encryption to AES-128-CBC with no impact on streaming media performance.

 

[CaptainSTX]

Excellent point about VPN speeds correlating with native WAN speeds. My VPN speeds improved when my fiber connection was upgraded from 100Mbps to 200Mbps. I then was able to go from no encryption to AES-128-CBC with no impact on streaming media performance.

My experience was the opposite of yours. When my speed package went from 85/11 - to 180/23 my VPN speed decreased from 75Mbps down to 65 Mbps down. The Atom processor running at 1.86 Ghz just didn't have enough processing power. I changed the device out for one with an I7 processor with Turbo Boost to 3.0 Ghz. I can now down load at speeds up to 150 Mbps. Ciper on both devices was AES-256-CBC.

 

[avtella]

That's interesting, the recent Atoms should be fairly powerful for VPN, did your Atom not support AES-NI? Just out of curiosity could you mention the model.

 

[sfx2000]

Ciper on both devices was AES-256-CBC.

You should try AES-128-GCM on the i7 - pretty fast

 

[sfx2000]

That's interesting, the recent Atoms should be fairly powerful for VPN, did your Atom not support AES-NI? Just out of curiosity could you mention the model.

Have to be careful with the Atoms - some do, some don't - marketing segmentation.

The 14nm stuff (Braswell) all do support the AES-NI instructions (along with RDRAND which also helps with the OpenSSL side)

 

[CaptainSTX]

You should try AES-128-GCM on the i7 - pretty fast

If had a faster connection I might. As it is now even with the AES-256 I think I am maxing out the bandwidth from my VPN provider. Before replacing my unit with the Atom processor I did make some adjustments to the encryption but it didn't make much of a difference and I never could not get the speed even up to 100 Mbps.

The I7 seems to have plenty of processing power and I have never been able to get the CPU load over 26% and that was running three HD video streams, a 4K stream from Netflix plus three audio streams and simultaneous speedtests on two PCs. Download during this period was averaging 50 - 60 Mbps all routed over the VPN. During the duration of the test I downloaded five GB of data.

The processor is over kill but if the mini PC I bought for the experiment hadn't worked out I wanted to be able to use it as a normal computer.

Power consumption isn't terrible on the unit. When it isn't in TurboBoost mode it is pulling just five Watts according to my Kill A Watt meter.

 

[Skyler LaVey]

My vpn settings for IPVanish do not stick. I can get through two speed test at full speed (100Mbps+)and the third one my download speed drops to 2Mbps or less. The internet becomes unusable until I turn the VPN off. I upload the openvpn and cert and type my username and password. It simply doesn’t work on the newest firmware or the previous firmware.


Sent from my iPhone using Tapatalk

 

[CaptainSTX]

That's interesting, the recent Atoms should be fairly powerful for VPN, did your Atom not support AES-NI? Just out of curiosity could you mention the model.

Sorry I don't have the unit anymore. It was about five years old so it probably wasn't Intel's latest and greatest Atom processor.

 

[primitivo]

For me VPN achievements or speedtests are really irrelevant unless someone post the latency to the VPN server as well. From my fiber at home of 200/45 I can get 111/45 via VPN if the latency to the VPN server is 6ms (usually meaning same country vpn) & lower further away, on RT-AC68U using Astrill RouterPro AES-256-CBC-SHA256. However if the latency is is 130ms I can get half of it. That means 1.7Ghz CPU should do at least 2x better.

R9000 is quite VPN efficient but DD-WRT is a mess in terms of stability for the time being.

 

[RMerlin]

For me VPN achievements or speedtests are really irrelevant unless someone post the latency to the VPN server as well. From my fiber at home of 200/45 I can get 111/45 via VPN if the latency to the VPN server is 6ms (usually meaning same country vpn) & lower further away, on RT-AC68U using Astrill RouterPro AES-256-CBC-SHA256. However if the latency is is 130ms I can get half of it. That means 1.7Ghz CPU should do at least 2x better.

R9000 is quite VPN efficient but DD-WRT is a mess in terms of stability for the time being.

VPN latency has virtually nothing to do with the router hardware. It depends on the distance between both endpoints of the tunnel.

 

[thiggins]

Any tests we do are always done on a local private network.

 

[primitivo]

VPN latency has virtually nothing to do with the router hardware. It depends on the distance between both endpoints of the tunnel.

Latency has nothing to do with hardware but the speed to remote VPN server is affected by both the latency & hardware (& encryption). Example: RT-AC68U on 130ms can get 50Mbps but R9000 can easily max 100Mbps, while the same AC68U on 6ms latency can reach 110Mbps. I mention this because saw people here posting VPN providers speed.

 

[RogerSC]

Most of the candidate customers for this device - they've probably followed Asus across multiple devices. These folks know that any time a new device ships, there's going to be some opportunities for improvement.

[opinion] == The device performed well in Tim's testing overall - good enough across all parts... Asus knows this is a premium/flagship device, and fixes will be forthcoming. I do take exception that people take a couple of small issues, and blow them way out of proportion, and this is what this thread has turned into.

Could Asus have done better? Maybe, we won't likely know, as those decisions are part of the daily Project Meetings that all companies have when a product is close to shipping.

I'm with Tim...I've learned not to jump and buy hardware with either buggy firmware or firmware with missing features until the bugs that affect me are fixed, or the features that I need have been added and are working. You can also assume that there will be third-party firmware that will fill in the gaps, but I've learned not to assume that as well. It pays to do your homework, watch the reviews, and see what's what before jumping.

My own opinions from experience with several routers, take it or leave it *smile*.

 

[sfx2000]

Here's the deal - you guys have kinda pinned me into a corner, but here's some facts...

This device is built on a new ARM architecture - the previous RT-AC*** were all on Cortex-A9 (and some were even on MIPS, but that's another story).

The older RT-AC's that supported ARM, they were Cortex-A9 - it's a dual core SMP CPU, and as an A9, it has the functional stuff only, as Cortex-A9 has optional features that Broadcom didn't include - that being said, Broadcom and Asus have done a nice job with that chip on the RT-AC legacy devices.

The new SoC arch is Cortex-A53, with some Broadcom special sauce left over from their Vulcan ARMv8 development which was sold off to AMCC, and then to MaCOM... this gets complicated as there was the whole Avago merger and subsequent Cypress divestiture...

This chip is different enough that it's classified as a ARM Cortex-A53 variant - and it's better than most would guess...

That being said - it's actually pretty quick, and depending on Broadcom's investment in GCC (time, not money, but time is money) - it will improve over time - kernel and complier updates here...

On another thread - we know that the SoC is running generally in Aarch64 mode - some older code is still 32-bit, and much of that is ARMv7-a optimized for the castrated Cortex-A9 that was part of the Broadcom SDK for those devices.

Going back to some personal experience in Q4-2016 thru Q1-2017 - we developed a code base (cafeole) around Cortex-A53, and had to backport against Cortex-A9 - and our experience here was there were two profiles, as the A53 was ARMv8, and the A9 is obviously ARMv7-A... some things were better, some had opportunities to improve.

I'm still confident that Asus, and Broadcom, will find additional performance in the future - esp. as the third party objects/software are ported over, and the obvious bugs are fixed.

From a HW perspect - the SoC is superior here to the other RT-AC's - the interfaces are faster/newer, and lower power consumption and heat - as they're on a small geometry - 28nm vs 40nm for the older chips, and we have a couple of years of dev behind the switching fabric - e.g. the GT-AC is NBase-T on all ports internally vs gigabit...

Be patient - there's plenty over performance on this device - as it is, it's shipping, and performs well enough, but as it has a lot of new stuff inside - new cores, new SW/Kernel, there are going to be bugs...

And I'm probably the last guy that would defend Asus or Broadcom as many would know across the forum threads.

I can also appreciate the effort going into this device - HW, it's stout on both the wired and wireless side.