Although only dual band, awesome looking gaming router with impressive specs and rave reviews to boot. Includes dual 2.5G ports. Asus AI Mesh 2.0 compatible. Outstanding features at this price point, IMO. Best price I've ever seen for this model. Getting this to supplement my 1.2 Gbps Xfinity service. Cannot wait for it to arrive and get it hooked up!
This collaborative space allows users to contribute additional information, tips, and insights to enhance the original deal post. Feel free to share your knowledge and help fellow shoppers make informed decisions.
Although only dual band, awesome looking gaming router with impressive specs and rave reviews to boot. Includes dual 2.5G ports. Asus AI Mesh 2.0 compatible. Outstanding features at this price point, IMO. Best price I've ever seen for this model. Getting this to supplement my 1.2 Gbps Xfinity service. Cannot wait for it to arrive and get it hooked up!
I'm actually eying this model due to the newer gen2 platform. I'd buy it @ $250..
jsz, i'm reading these router reviews, and the performance is all over the place, even on the ac2900 i just bought, nothing online matches up with anything on my end. even the ac86u by small netbuilder they only got 164 down, 263 up, that's just plain off, I'm finally getting ~630 down and 400 up using the unii 3 channels.
jsz, i'm reading these router reviews, and the performance is all over the place, even on the ac2900 i just bought, nothing online matches up with anything on my end. even the ac86u by small netbuilder they only got 164 down, 263 up, that's just plain off, I'm finally getting ~630 down and 400 up using the unii 3 channels.
What is the point of these reviews ?
Nothing. Reviews are constraint to the user reviewing it.
I guess this is the point I was trying to make when arguing for 4x4 MIMO. There is at least consistency in performance when increasing radio size.
Thats why I keep pushing you to buy a cheap AX3000 2x2 router to compare to your new GT-2900. I think it will be a good laugh.. 😂. (Especially considering that your intel 8260 would prob bottleneck harder).
I do have faith in this specific Gen 2 4912 + 6715 platform though. It's more of a legitimate upgrade from AC Wave 2 from everything I've seen and heard.
Gen 1 AX on Broadcom end is more like AC Wave 2.5 from my experiences...
The current/previous "high end" BCM 4906/4908 were released in 2016 for AC Wave 2.
4x4 BCM43684 (2018) follows the older 436XX AC naming scheme. Every other newer AX radio follows 67XX.
The newer 4x4 6715 replaces BCM43684 and solves throughput issues the previous radio had.
Edit:
Theres only 1 independent reviewer that I've seen with logical/reasonable testing. Every one else seems to think a large room constraints "performance" 🤦
Case and point: 2x2 MIMO with inflated numbers that won't ever see the light of day in real usage scenarios.
Nothing. Reviews are constraint to the user reviewing it.
I guess this is the point I was trying to make when arguing for 4x4 MIMO. There is at least consistency in performance when increasing radio size.
Thats why I keep pushing you to buy a cheap AX3000 2x2 router to compare to your new GT-2900. I think it will be a good laugh.. 😂. (Especially considering that your intel 8260 would prob bottleneck harder).
I do have faith in this specific Gen 2 4912 + 6715 platform though. It's more of a legitimate upgrade from AC Wave 2 from everything I've seen and heard.
Gen 1 AX on Broadcom end is more like AC Wave 2.5 from my experiences...
The current/previous "high end" BCM 4906/4908 were released in 2016 for AC Wave 2.
4x4 BCM43684 (2018) follows the older 436XX AC naming scheme. Every other newer AX radio follows 67XX.
The newer 4x4 6715 replaces BCM43684 and solves throughput issues the previous radio had.
Edit:
Theres only 1 independent reviewer that I've seen with logical/reasonable testing. Every one else seems to think a large room constraints "performance" 🤦
Case and point: 2x2 MIMO with inflated numbers that won't ever see the light of day in real usage scenarios.
The main diff between bcm4912 and bcm4906/08 is that 4912 has one internal 10GE and support one external PHY 10GE, but asus only give one 2.5GE. So I think it's more like a waste here though you can finally get fully pps speed of 2.5gpbs while the formal 4906/08 cannot perfectly handle the 2.5GE for small packages (like 64k package)
The main diff between bcm4912 and bcm4906/08 is that 4912 has one internal 10GE and support one external PHY 10GE, but asus only give one 2.5GE. So I think it's more like a waste here though you can finally get fully pps speed of 2.5gpbs while the formal 4906/08 cannot perfectly handle the 2.5GE for small packages (like 64k package)
4916 has 10G. (GT-AX11000 PRO is likely using this CPU variant unless 10G is fed via second controller IE: 2.5G + 10G setup.)
4912 doesn't have 10G, but it has 2.5G built in without external controller. (GT-AX6000 is running a 2nd external controller for second port)
And yeah 4906/4908 have no native support. Anything with 2.5G+ is external chip.
Improved packet processing and 28nm >16nm die shirk seem to be the biggest benefits.
I'm more interested in 6715 radios though. BCM43684 can be inconsistent. My GT-AC2900 trades off with my AXE11000 at a similar distance on certain clients.. its funny.
And yeah 4906/4908 have no native support. Anything with 2.5G+ is external chip.
question about these cpus jsz. how come I've never gotten my peak (service) throughput in any of my tests. it maxes out at 840mbps on my old router and this new one. I've tried multiple different speed test servers. Yet I'm able to get 890mbps on upload which is more than their 840 up advertised.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
Last edited by xtp July 23, 2022 at 05:29 AM.
Like
Helpful
Funny
Not helpful
Sign up for a Slickdeals account to remove this ad.
question about these cpus jsz. how come I've never gotten my peak (service) throughput in any of my tests. it maxes out at 840mbps on my old router and this new one. I've tried multiple different speed test servers. Yet I'm able to get 890mbps on upload which is more than their 840 up advertised.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable
Quote
from xtp
:
question about these cpus jsz. how come I've never gotten my peak (service) throughput in any of my tests. it maxes out at 840mbps on my old router and this new one. I've tried multiple different speed test servers. Yet I'm able to get 890mbps on upload which is more than their 840 up advertised.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
Broadcom router cpus have two levels nat acceleration
level 1: CTF/Flow Cache (this is software based, so it's compatible with some features like AQos, traffic analyzer, AIprotection, etc. It can handle 1gig throughput with big size packages, but you will see high cpu usage and the speed may drop down if you have those additional features turned on)
level 2: FA/Runner (this is hardware based, so the system will use special build-in ASIC to fully offload the package processing, but it doesn't compatible with most of the extra features. But with this offload turned on, you should see almost 0 cpu usage even on 1gbps throughput and for bcm4906/4908, its runner ASIC can handle fully 1gbps packages process no matter what sizes they are.)
Since you're using the 1gig connection, make sure you turn off the features I mentioned above (those features are meaningless in high bandwidth connection) and check if the level 2 acceleration is turned on. Also, if you did a firmware update, make sure you reset the configs after the upgrading since sometimes the old configs may cause some weird conflicts in the new firmware.
question about these cpus jsz. how come I've never gotten my peak (service) throughput in any of my tests. it maxes out at 840mbps on my old router and this new one. I've tried multiple different speed test servers. Yet I'm able to get 890mbps on upload which is more than their 840 up advertised.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
Nah, it should be able to do the full 900-940 DL via cable (unless ISP is limited innately, a lot of times, its "up to" speeds)
I would disable the other two settings like the other guy mentioned. Youre not going to need them with 1G.
Specific modem's sometimes hinder performance with certain hardware. I've found that Intel Puma based modems lower overall speed via wireless (when paired with BCM). The Broadcom based stuff usually does better with Broadcom hardware. (throughput wise, SNR aside). Not exactly the same issue (since wired is hindered), but I thought id mention it.
Nah, it should be able to do the full 900-940 DL via cable (unless ISP is limited innately, a lot of times, its "up to" speeds)
I would disable the other two settings like the other guy mentioned. Youre not going to need them with 1G.
Specific modem's sometimes hinder performance with certain hardware. I've found that Intel Puma based modems lower overall speed via wireless (when paired with BCM). The Broadcom based stuff usually does better with Broadcom hardware. (throughput wise, SNR aside). Not exactly the same issue (since wired is hindered), but I thought id mention it.
I don't have those options in my merlin firmware. I only see, Jumbo frame, Spanning tree, and Bonding/link aggregation.
BTW i'm talking about gt-ac2900, i don't have this ax6000.
I do see, runner and flow cache enabled under HW acceleration in Tools system information, but I don't see any button to disable them
don't sweat it jsz. i'm very much enjoying the 2900. this is a minor issues relative to the whole.
my torrents run alot faster and almost no router cpu load
The only downside of this older platform is that the wave 2 radios are clocked in at a lower speed, but the ASUS platform is A LOT faster than competitors using similar AC W2 radios. They might be overclocked past default 800mhz. I'm not sure lol
It's night and day vs something like a C4000 from TPLINK. (4906/4908 *version specific + 3x3 W2 4365E).
I'm assuming the newer BCM6715's have a bump over 1.5ghz internal radio speed. QCA 8x8 v2 high end is at 1.7ghz (1.5ghz v1) IIRC, but this spec isn't marketed and hard to find on internet.
I do know the gen 1 BCM43684 runs internally @ 1.5ghz because ARRIS marketing made a big mistake with wording on their boxes.. But it does make sense since the 6750/6755 SoCs are basically confirmed 1.5ghz Tri core + quad core A7 designs.
If it hasn't made sense yet, The lower end SoC is factoring the internal radio processor. Hence why the 6755 with 2 radios built in is a "quad core" and 6750 with 1 radio is a "tri core". It's kind of a 2+2 or 2+1 design.
Broadcom and manufacturer marketing is freaking weird.. GT-AC2900 should be considered quad core if were going by that logic (2+1+1), but I guess it doesn't matter.
to quote Rmerlin :
Quote
:
"The CPU has almost no impact on wireless performance because the wireless chip uses its own internal CPU. Since the charts primarily focus on wireless, the router's CPU won't really matter.
The extra CPU performance is mostly for VPN, USB file sharing, running complex firewall rules with NAT acceleration disabled, etc..."
This is only half true since the SoCs kinda use both . Depends on what you have.
Edit: This is also why I'm not against the AX86S. It's legit AX86U without 2.5Gbe port. The 4906/4908 swap wont matter for 99% of people. Prob why TP-link did it on their older C4000.
Btw, try control channel 161. For me, this is highest performance on unii-3.
what's so special about 161 ? tell us about the secret sauce.
I honestly don't know. 157 control has more spectrum available (for me) when using WIFI radar, but 161 just performs best for me with this specific router.
It's like the FCC Power output is false 😂. I doubt it, but simply weird anomaly for me.
Leave a Comment
18 Comments
Sign up for a Slickdeals account to remove this ad.
I'm actually eying this model due to the newer gen2 platform. I'd buy it @ $250..
I'm actually eying this model due to the newer gen2 platform. I'd buy it @ $250..
What is the point of these reviews ?
What is the point of these reviews ?
I guess this is the point I was trying to make when arguing for 4x4 MIMO. There is at least consistency in performance when increasing radio size.
Thats why I keep pushing you to buy a cheap AX3000 2x2 router to compare to your new GT-2900. I think it will be a good laugh.. 😂. (Especially considering that your intel 8260 would prob bottleneck harder).
I do have faith in this specific Gen 2 4912 + 6715 platform though. It's more of a legitimate upgrade from AC Wave 2 from everything I've seen and heard.
Gen 1 AX on Broadcom end is more like AC Wave 2.5 from my experiences...
Edit:
Theres only 1 independent reviewer that I've seen with logical/reasonable testing. Every one else seems to think a large room constraints "performance" 🤦
Case and point: 2x2 MIMO with inflated numbers that won't ever see the light of day in real usage scenarios.
I guess this is the point I was trying to make when arguing for 4x4 MIMO. There is at least consistency in performance when increasing radio size.
Thats why I keep pushing you to buy a cheap AX3000 2x2 router to compare to your new GT-2900. I think it will be a good laugh.. 😂. (Especially considering that your intel 8260 would prob bottleneck harder).
I do have faith in this specific Gen 2 4912 + 6715 platform though. It's more of a legitimate upgrade from AC Wave 2 from everything I've seen and heard.
Gen 1 AX on Broadcom end is more like AC Wave 2.5 from my experiences...
Edit:
Theres only 1 independent reviewer that I've seen with logical/reasonable testing. Every one else seems to think a large room constraints "performance" 🤦
Case and point: 2x2 MIMO with inflated numbers that won't ever see the light of day in real usage scenarios.
4912 doesn't have 10G, but it has 2.5G built in without external controller. (GT-AX6000 is running a 2nd external controller for second port)
And yeah 4906/4908 have no native support. Anything with 2.5G+ is external chip.
Improved packet processing and 28nm >16nm die shirk seem to be the biggest benefits.
I'm more interested in 6715 radios though. BCM43684 can be inconsistent. My GT-AC2900 trades off with my AXE11000 at a similar distance on certain clients.. its funny.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
Sign up for a Slickdeals account to remove this ad.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
level 1: CTF/Flow Cache (this is software based, so it's compatible with some features like AQos, traffic analyzer, AIprotection, etc. It can handle 1gig throughput with big size packages, but you will see high cpu usage and the speed may drop down if you have those additional features turned on)
level 2: FA/Runner (this is hardware based, so the system will use special build-in ASIC to fully offload the package processing, but it doesn't compatible with most of the extra features. But with this offload turned on, you should see almost 0 cpu usage even on 1gbps throughput and for bcm4906/4908, its runner ASIC can handle fully 1gbps packages process no matter what sizes they are.)
Since you're using the 1gig connection, make sure you turn off the features I mentioned above (those features are meaningless in high bandwidth connection) and check if the level 2 acceleration is turned on. Also, if you did a firmware update, make sure you reset the configs after the upgrading since sometimes the old configs may cause some weird conflicts in the new firmware.
Is fios just bullshitting me on the 940mbps downlink claim ?
Is the cpu still not fast enough with routing overheads ?
Some other issue ? I know I can transfer 95% of the full gigabit over lan. so that stuff all works. I have 23awg solid copper cable shielded with spine, so I know the cable is capable.
I would disable the other two settings like the other guy mentioned. Youre not going to need them with 1G.
Specific modem's sometimes hinder performance with certain hardware. I've found that Intel Puma based modems lower overall speed via wireless (when paired with BCM). The Broadcom based stuff usually does better with Broadcom hardware. (throughput wise, SNR aside). Not exactly the same issue (since wired is hindered), but I thought id mention it.
I would disable the other two settings like the other guy mentioned. Youre not going to need them with 1G.
Specific modem's sometimes hinder performance with certain hardware. I've found that Intel Puma based modems lower overall speed via wireless (when paired with BCM). The Broadcom based stuff usually does better with Broadcom hardware. (throughput wise, SNR aside). Not exactly the same issue (since wired is hindered), but I thought id mention it.
I don't have those options in my merlin firmware. I only see, Jumbo frame, Spanning tree, and Bonding/link aggregation.
BTW i'm talking about gt-ac2900, i don't have this ax6000.
I do see, runner and flow cache enabled under HW acceleration in Tools system information, but I don't see any button to disable them
BTW i'm talking about gt-ac2900, i don't have this ax6000.
I do see, runner and flow cache enabled under HW acceleration in Tools system information, but I don't see any button to disable them
And yeah I know. GT-AX6000 is kinda pricey still. Hoping it goes down to the $250~ area. That or the QCA 8x8 AX89X for the same price point..
Well I'm not on merlin ATM so I'm not sure
And yeah I know. GT-AX6000 is kinda pricey still. Hoping it goes down to the $250~ area. That or the QCA 8x8 AX89X for the same price point..
my torrents run alot faster and almost no router cpu load
my torrents run alot faster and almost no router cpu load
It's night and day vs something like a C4000 from TPLINK. (4906/4908 *version specific + 3x3 W2 4365E).
I'm assuming the newer BCM6715's have a bump over 1.5ghz internal radio speed. QCA 8x8 v2 high end is at 1.7ghz (1.5ghz v1) IIRC, but this spec isn't marketed and hard to find on internet.
I do know the gen 1 BCM43684 runs internally @ 1.5ghz because ARRIS marketing made a big mistake with wording on their boxes.. But it does make sense since the 6750/6755 SoCs are basically confirmed 1.5ghz Tri core + quad core A7 designs.
If it hasn't made sense yet, The lower end SoC is factoring the internal radio processor. Hence why the 6755 with 2 radios built in is a "quad core" and 6750 with 1 radio is a "tri core". It's kind of a 2+2 or 2+1 design.
Broadcom and manufacturer marketing is freaking weird.. GT-AC2900 should be considered quad core if were going by that logic (2+1+1), but I guess it doesn't matter.
to quote Rmerlin :
The extra CPU performance is mostly for VPN, USB file sharing, running complex firewall rules with NAT acceleration disabled, etc..."
Edit: This is also why I'm not against the AX86S. It's legit AX86U without 2.5Gbe port. The 4906/4908 swap wont matter for 99% of people. Prob why TP-link did it on their older C4000.
Btw, try control channel 161. For me, this is highest performance on unii-3.
Sign up for a Slickdeals account to remove this ad.
It's like the FCC Power output is false 😂. I doubt it, but simply weird anomaly for me.
Might be different for you.
Leave a Comment