That's an interesting start to handling traffic that a Killer NIC sends. First hop latency on an internal network won't be a problem and maybe there's some actual benefits an end user can realize for traffic leaving the LAN though I'm skeptical still that this is going to be a magic bullet solution for online gaming given the potential lack of prioritization of packets outside the reach of the local router and ethernet adapter. It's worth a look, of course, but latency probably won't magically go away.
No way, this thing talks directly to the BGP routers and tells them to freaking carry your sh*t faster 'cause you have a damn collection of Killers in your house and they be wild things.
Also 1 year from now they'll stop releasing new firmware because they'll have already launched the new WRT41Z that will have even more optimization for your gaming traffic. Then maybe you can put an open FW on it that does the exact same thing but without commanding a $100 premium just to get a long list of techno-babble* features that are 99.9% present in any other higher end router.
The first hop is many times the problem. Bufferbloat means that any time someone attempts to use even a small amount of average bandwidth, a short burst of high bandwidth causes jitter and loss.
Oh crap, it's a GAMING™ router. That really appeals to the GAMER™ inside me who craves GAMER™ quality products to give me every amount of placebo edge over the competition using well-priced normal computer equipment in GAMING™ events.
I don't think my $200 ASUS AC3200 router cuts it anymore. It's not GAMER™ certified and it's not the quality that fellow GAMERS™, such as myself, need during intense online GAMING™ sessions. Linksys gets GAMERS™ though. They know GAMERS™ need GAMING™ prioritization during GAMES™, so that frames don't get dropped and headshots always land their mark.
Definitely in for 2.
-----This post was sponsored in part by Linksys and viewers like you. Thank you-----
IMO it doesn't look much worse than the average N external antenna model does. Give or take the angular faceplate - which is a crime against good taste that's gone well beyond just gamer products - I don't see anything spectacularly horrid as I'd expect from the branding.
The sooner people realise that Rivet Networks have nothing to contribute to the PC industry but a bunch of jargon and advertising slogans, the sooner they go bust and stop messing with everything. It's bad enough that so many machines in certain market segments have their shitty, bugged wireless adapters, at least they tend to be socketed.
One should check if their existing router has QOS settings before buying this. Or one can get a router with QOS settings and just set them (see Ubiquiti). Much cheaper than this monstrosity.
having tried them, I can say that either my router isn't very good at it (it's not an ubiquiti) or I didn't set it properly, or it doesn't really stop a max bandwidth download on another computer from making your ping jump all over.
I find the 3 160MHz channel claim amusing since at present there are only at most 2 160MHz channels licensed in the 5ghz ISM band. The US is considering some spectrum reallocation's that would open two more (at the expense of a dedicated band for semi/autonomous car to car communications); but AFAIK it hasn't gone beyond the talking stage. And I haven't seen any rumbles about non-US regulators even thinking about doing the same.
Dan, Maybe my wording was not clear. Can you point out why you think there is a claim of '3 160MHz channels'? We are not talking about a tri-radio solution (like Broadcom's XStream which requires two 5 GHz channels and 1 2.4 GHz channel), but a single radio solution.
In this particular case, the configuration is a 160 MHz channel with three spatial streams. When discussing N spatial streams, we have N data streams that are transmitted at the same time in the same channel, but, by different antenna. They are combined at the receiver end by MIMO processing. So, Marvell's solution will not face the limitations imposed by the number of available 160 MHz channels, as long as at least one is available.
Please do provide more clarification on your interpretation so that we can ensure we are all on the same correct page.
The QoS problem has been virtually solved with fq_Codel and upcoming Cake. All you need to do is set the bandwidth, nothing else. Some people have unknown bandwidth because of massive over-subscription. In these cases, some have made scripts that monitor latency and reduce the assigned bandwidth when latency goes up.
It's a hard problem, but it has been solved for well over 5 years now.
Considering that Killer NICs work better with vanilla Qualcomm drivers than the Killer Bloated Krap I wold not touch this router if it was free. I avoid motherboards now that have Killer NICs.
My wife's Skylake mobo came with a Killer NIC. I had to iperf it just to see how "fast" it is. Turned out it used about 4x more CPU and was about 10% slower than the integrated Intel NIC in her 8 year old Nehalem mobo.
Needless to say, I disabled the Killer NIC in the bios and installed an Intel i210
So Marvell hardware with Rivet's Killer software? Why not ditch the Killer software or use Rivet's own chipset in this product, if they want to use that Killer branding? I would probably pay money so I don't have to use a product with any "Killer" tech or branding though.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
21 Comments
Back to Article
BrokenCrayons - Monday, January 9, 2017 - link
That's an interesting start to handling traffic that a Killer NIC sends. First hop latency on an internal network won't be a problem and maybe there's some actual benefits an end user can realize for traffic leaving the LAN though I'm skeptical still that this is going to be a magic bullet solution for online gaming given the potential lack of prioritization of packets outside the reach of the local router and ethernet adapter. It's worth a look, of course, but latency probably won't magically go away.close - Monday, January 9, 2017 - link
No way, this thing talks directly to the BGP routers and tells them to freaking carry your sh*t faster 'cause you have a damn collection of Killers in your house and they be wild things.Also 1 year from now they'll stop releasing new firmware because they'll have already launched the new WRT41Z that will have even more optimization for your gaming traffic. Then maybe you can put an open FW on it that does the exact same thing but without commanding a $100 premium just to get a long list of techno-babble* features that are 99.9% present in any other higher end router.
The other 0.1% is the Killer branding.
*[airtime efficiency Exclusive optimization clear DFS channel airspace SOHO-grade TCAM NAS-like network storage Custom-built firmware tuned]
bcronce - Monday, January 9, 2017 - link
The first hop is many times the problem. Bufferbloat means that any time someone attempts to use even a small amount of average bandwidth, a short burst of high bandwidth causes jitter and loss.JoeyJoJo123 - Monday, January 9, 2017 - link
Oh crap, it's a GAMING™ router. That really appeals to the GAMER™ inside me who craves GAMER™ quality products to give me every amount of placebo edge over the competition using well-priced normal computer equipment in GAMING™ events.I don't think my $200 ASUS AC3200 router cuts it anymore. It's not GAMER™ certified and it's not the quality that fellow GAMERS™, such as myself, need during intense online GAMING™ sessions. Linksys gets GAMERS™ though. They know GAMERS™ need GAMING™ prioritization during GAMES™, so that frames don't get dropped and headshots always land their mark.
Definitely in for 2.
-----This post was sponsored in part by Linksys and viewers like you. Thank you-----
DanNeely - Monday, January 9, 2017 - link
It's not a Real Gaming(WFTM) router yet. There's no indication that all the status lights are programmable RGB for maximum eye cancer yet.Michael Bay - Monday, January 9, 2017 - link
At least it`s properly ugly.DanNeely - Monday, January 9, 2017 - link
IMO it doesn't look much worse than the average N external antenna model does. Give or take the angular faceplate - which is a crime against good taste that's gone well beyond just gamer products - I don't see anything spectacularly horrid as I'd expect from the branding.fanofanand - Monday, January 9, 2017 - link
Thanks for the laugh joeyjo, your satire is sadly spot on!Colin1497 - Monday, January 9, 2017 - link
WUT?Azurael - Monday, January 9, 2017 - link
The sooner people realise that Rivet Networks have nothing to contribute to the PC industry but a bunch of jargon and advertising slogans, the sooner they go bust and stop messing with everything. It's bad enough that so many machines in certain market segments have their shitty, bugged wireless adapters, at least they tend to be socketed.ChefJeff789 - Monday, January 9, 2017 - link
This.^ "Killer" branded crap is a reason for me to avoid something, not to buy it. Stop putting them in systems, PLEASE.IKeelU - Monday, January 9, 2017 - link
One should check if their existing router has QOS settings before buying this. Or one can get a router with QOS settings and just set them (see Ubiquiti). Much cheaper than this monstrosity.Murloc - Tuesday, January 10, 2017 - link
having tried them, I can say that either my router isn't very good at it (it's not an ubiquiti) or I didn't set it properly, or it doesn't really stop a max bandwidth download on another computer from making your ping jump all over.DanNeely - Monday, January 9, 2017 - link
I find the 3 160MHz channel claim amusing since at present there are only at most 2 160MHz channels licensed in the 5ghz ISM band. The US is considering some spectrum reallocation's that would open two more (at the expense of a dedicated band for semi/autonomous car to car communications); but AFAIK it hasn't gone beyond the talking stage. And I haven't seen any rumbles about non-US regulators even thinking about doing the same.https://en.wikipedia.org/wiki/List_of_WLAN_channel...
http://specmap.sequence-omega.net/blog/2013/02/how...
fanofanand - Monday, January 9, 2017 - link
But.....this is KILLER! You think KILLER cares about the availability of a 3rd 160 MHz channel? Brah you got pwned by KILLER!Sorry if the above teenspeak was inaccurate or incorrect. I never did learn the proper jargon.
ganeshts - Monday, January 9, 2017 - link
Dan, Maybe my wording was not clear. Can you point out why you think there is a claim of '3 160MHz channels'? We are not talking about a tri-radio solution (like Broadcom's XStream which requires two 5 GHz channels and 1 2.4 GHz channel), but a single radio solution.In this particular case, the configuration is a 160 MHz channel with three spatial streams. When discussing N spatial streams, we have N data streams that are transmitted at the same time in the same channel, but, by different antenna. They are combined at the receiver end by MIMO processing. So, Marvell's solution will not face the limitations imposed by the number of available 160 MHz channels, as long as at least one is available.
Please do provide more clarification on your interpretation so that we can ensure we are all on the same correct page.
DanNeely - Monday, January 9, 2017 - link
I think we're good now, and I've been misunderstanding the feature since first reading about it years ago...bcronce - Monday, January 9, 2017 - link
The QoS problem has been virtually solved with fq_Codel and upcoming Cake. All you need to do is set the bandwidth, nothing else. Some people have unknown bandwidth because of massive over-subscription. In these cases, some have made scripts that monitor latency and reduce the assigned bandwidth when latency goes up.It's a hard problem, but it has been solved for well over 5 years now.
Zak - Monday, January 9, 2017 - link
Considering that Killer NICs work better with vanilla Qualcomm drivers than the Killer Bloated Krap I wold not touch this router if it was free. I avoid motherboards now that have Killer NICs.bcronce - Tuesday, January 10, 2017 - link
My wife's Skylake mobo came with a Killer NIC. I had to iperf it just to see how "fast" it is. Turned out it used about 4x more CPU and was about 10% slower than the integrated Intel NIC in her 8 year old Nehalem mobo.Needless to say, I disabled the Killer NIC in the bios and installed an Intel i210
Penti - Tuesday, January 10, 2017 - link
So Marvell hardware with Rivet's Killer software? Why not ditch the Killer software or use Rivet's own chipset in this product, if they want to use that Killer branding? I would probably pay money so I don't have to use a product with any "Killer" tech or branding though.