Killer Xeno Pro: Do You Really Need A Gaming Network Card?

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

royalcrown

Distinguished
Why do you even bother to test this, it is obvious to anyone that knows the internet isn't "tubes", that the net itself is the biggest source of latency. Why not mention that part ? Even a nic with NO latency at all won't make any noticable difference. After the packet leaves your super duper killer nic, it has to get on the 4 lane traffic jam of an internet no matter what. As far as lan parties, well the ping using onboard nic to the router is 1ms or less on cheap realteks even. Seriously, it makes me sad that some ppl are this clueless.
 

anamaniac

Distinguished
Jan 7, 2009
2,447
0
19,790
Sure, it might help with handling network information assuming you have a solo core and a absolutely pathetic integrated NIC.
Come on, I got a bargain bin computer and it still came with an NIC expansion card.

If I were to go for a cheap build, there is no way this would go in.
If I were to go for an extremebuild, there'd be no space after a tri crossfire setup.

=D
 
Yes indeed, this is not needed at all nor it gives you better ping. Total nonsense just like the previous cards. For the price/performance ratio this is a joke. Steer away and save yourself from banging your head on the wall when you see the exact same results w or w/o the card.
 

haplo602

Distinguished
Dec 4, 2007
202
0
18,680
[citation][nom]Neggers[/nom]I have already reduced my latency in WoW, from 500-600 to 250-350 with TCP tweaks in windows. This card claims to bypass and optimise TCP use, which means it can still lower ping, by using TCP more effectively.When I ping the WoW servers i can get a reply as low as 150ms, however I dont see this kind of latency in game. This is because more data is transfered during actual gameplay, compared to just a ping.If this card can optimise or prioritise that data better it may be able to give me in latency closer to my ping times.[/citation]

basicaly this. bigfoot networks is harvesting on the clueless people that don't know how to tweak the mess called MS Windows to full potential.
 

royalcrown

Distinguished
[citation][nom]bellbillsnow[/nom]One other issue:I am going to have to come clean here: I own a K1. The review I read indicated that it made a difference and avoiding the network stack made sense to me.Here is the issue: not only must the card be in and installed, but any other NIC's must be disabled and then the Killer must be put into Game Mode. A support tech told me that programs launched before the Killer is put into game mode would likely operate in App Mode. Were you sure that you launched the program in Game Mode.I have similar concerns about your secondary review.My take as an owner of a K1:It seems to help - especially in big battles. The problem is that any time I lose my connection I blame it on the Killer. For awhile it did seem like it was crashing. Your review missed it's biggest weakness: it introduces a whole other computer that can crash in your computer.That said, I have lost my connection a couple times (I play WAR) and then switched to the onboard NIC only to lose my connection again. As with all things in online gaming, there are always sooooo many variables outside of my system, so I am left wondering is it: my vid card drivers, the router, the NIC, my ISP or the game servers!? Drives me crazy. (BTW, I am looking for any network diagnostic tools that might help.)The latest round I have been investigating routers. Seems like there is very little research on router quality for gaming purposes. Seems like sessions are key because games like to have lots of small UDP sessions open. And we have about 10 network devices now and multiple browser tabs seem to be overlaoding the NAT tables... so I reboot my router before gaming sessions and it seems to have helped.[/citation]

Are you really that clueless ?! just open a cmd window and ping your router. My ping time to the router and back is less than 1 millisecond, it actually says ZERO milliseconds with my onboard gigabit lan. so now type "ping google.com".

Microsoft Windows [Version 6.1.7100]
Copyright (c) 2009 Microsoft Corporation. All rights reserved.

C:\Users\Thomas>ping 192.168.0.2

Pinging 192.168.0.2 with 32 bytes of data:
Reply from 192.168.0.100: Destination host unreachable.
Reply from 192.168.0.100: Destination host unreachable.
Reply from 192.168.0.100: Destination host unreachable.
Reply from 192.168.0.100: Destination host unreachable.

Ping statistics for 192.168.0.2:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),

C:\Users\Thomas>
C:\Users\Thomas>ping 192.168.0.1

Pinging 192.168.0.1 with 32 bytes of data:
Reply from 192.168.0.1: bytes=32 time
 

royalcrown

Distinguished
stupid ass forum...ping google.com and tell me how a killer nic lowers the ping once the packet leaves you router. Do you really think 1 MS is going to matter when it takes 40 times that or more to get to the internet router and back ? quit being a sucker for junk science. Look up "Rube Goldberg"
 

Marcus52

Distinguished
Jun 11, 2008
619
0
19,010
First thing I'd ask, did you set the TcpAckFrequency to 1? You should see a significant improvement in latency in TCP games like WoW if you do - depending on your server. On Draka, my lag went from 150-200ms down to 40-70, as measured in Dalaran at peak times. On another server, my lag went from 40ish to around 10ish (!).

That's the biggest difference the Killer cards will make for you; however, it can be done with a hack, so the card isn't needed for WoW.

It's hard for me to believe that if you use the Killer firewall, voice chat, and optimize your network settings, that it won't make at least a small amount of diffeence at least in a mediocre computer. Indeed, it was my experience in such a computer that I did have a slight benefit in feel of WoW and using the hardware firewall slightly improved it again, even though I did the TcpAckFrequency = 1 hack independent of the card.

Now that I've said that, I really think you did bigfootnetworks a slight disservice by using WoW to measure their card; it is primarily a UDP enhancing card. While in the past they made claims about improvements in WoW, when people started pointing out in their forums (me for one) that close to the same results could be achieved by 'hacking' the registry, they backed off thier claims for WoW. They have said many times the Killer is primarily for UDP protocol games.

Bigfootnetworks realy though in my opinion has done themselves the biggest disservice; they released the Xeno Pro too soon. They built the previous cards (at least the M1) with the idea in mind that users could write apps for it, but when that didn't happen they didn't do much about it.

I'm no network guru - far from it. However, it seems to me that network connections are a tricky business, and it is easy for a driver or hardware to go haywire if things aren't perfect. Bigfootnetworks didn't, from what I've read, do everything they could to make sure that as little trouble would happen as possible. People are having trouble that shouldn't be - as your attempts on multiple suystems showed.

About your comment that 'we are tired of spending needless money' - what does that even mean? 'Needless? lol none of it is 'needed', it's about gaming. however, I love hardware and building systems, so none of it is 'needless' to me; I'm NOT tired of spending money on it, lol.

My recommendations for WoW - hard to justify this Xeno for it; if your guild uses Teamspeak, it might be more attractive. Use ALL it's features for max effect. Instead though, I'd say do the TcpAckFrequency hack (there are multiple guides on Youtube and other places), and if you have 4 cores or more on your processor, optimize WoW to run on all 4 cores - or whatever other config suits it on your machine. There's a guide for that on the WoW forums.

Prioritizing your traffic might help, too, but I don't run much when I game so it's not been much help for me.

If you are running UDP based games, I suggest reading more on the Killer cards before you decide against it. However, I'd say either buy an older card (K1 or M1), which are more expensive but less buggy, or wait a bit for the bugs in the Xeno line-up to be worked out.

;)
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
[citation][nom]royalcrown[/nom]stupid ass forum...ping google.com and tell me how a killer nic lowers the ping once the packet leaves you router. Do you really think 1 MS is going to matter when it takes 40 times that or more to get to the internet router and back ? quit being a sucker for junk science. Look up "Rube Goldberg"[/citation]

I guess I am that clueless. Sorry to annoy you.
Ping times have little to do with what the KillerNic is trying to accomplish. The killerNic is about UDP sessions... not tcp pings. Lowering pings is nice, but games loose their "smooth" from the game waiting for the windows network stack to process packets. The Killer is supposed to offload this process and pass packets directly to the game.
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
One more thing I have noticed...
When I am using my KillerNic in Warhammer Online, I find that I spawn in dungeons several seconds sooner than others who entered at the exact same time... and this does not happen with the Onboard NIC. That said, I have not adjusted the TCPAckFreq in my registry for the onboard NIC.

I think it is really interesting how so many people seem to decide out of nowhere that this is snake oil... and then get really angry about it. I am not sure why people want to call me an idiot for trying the KillerNIC. I think smart people (most of us are way above average) have too much pressure on them to be the smartest... probably because we end up deriving so much of our self worth from being the "smartest".

Back to the Killer... I end up feeling like the guys at Killer would not have made it to making a second version of the card if there was not something to it. As has been mentioned, the something may just be suckers like me. Still, I feel like testing FPS's and ping times are the wrong paradigm.

This is from the 2008 M1 review:
"Our own testers (which included a couple of diehard full-on teenaged gamers from our neighborhood, as well as author Toby Digby and his cousin, Steven Tran) also reported a definite improvement in the fluidity and smoothness of their gaming experience when using the Killer NIC as opposed to the built-in Realtek GbE interface on the Gigabyte motherboard. Such comments are hard to quantify or pin down precisely, but gamers reported quicker reaction times in their virtual worlds, a better ability to anticipate and track opponents and general improvements in their scoring as well. Tough though these qualities may be to measure objectively, they should (and apparently do) constitute a powerful lure for serious gamers."
http://www.tomsguide.com/us/killer-m1-nic,review-1083-5.html

Cheers from clueless a$$hole that bases his decisions on numerous reviews indicating a performance benefit of the Killer NIC!
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
I think the best way to settle this is with a blind experiment... give five gamers each identical machines except that one is using a killer nic and the other is using an Intel network card (which most say is the best alternative). Then see which machine they would rather play on. Then interview them.

Somethings things just cannot be measured and I think this is one of them.
 

haplo602

Distinguished
Dec 4, 2007
202
0
18,680
[citation][nom]bellbillsnow[/nom]I think the best way to settle this is with a blind experiment... give five gamers each identical machines except that one is using a killer nic and the other is using an Intel network card (which most say is the best alternative). Then see which machine they would rather play on. Then interview them. Somethings things just cannot be measured and I think this is one of them.[/citation]

paying 130$ for something that can be tuned in the registry or other places is nonsense. optimising a system for latency instead of bulk download speed is the key here. once you have that done, no killer nic will improve anything. they are just picking on people that are clueless and MS helps them with their mess of an OS.

btw your system is way faster than the internet connection. so IRQ processing and protocol routing on your local machine will always beat the rest in speed. however if it is messed up, it also creates a large gap in the processing. desktop network is usualy tuned for throughput and not latency, because latency tuning can influence user experience when something goes wrong. f.e. a DoS attack on a latency tuned system is way more effective than on a throughput tuned system.
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
[citation][nom]haplo602[/nom]paying 130$ for something that can be tuned in the registry or other places is nonsense. optimising a system for latency instead of bulk download speed is the key here. once you have that done, no killer nic will improve anything. they are just picking on people that are clueless and MS helps them with their mess of an OS.btw your system is way faster than the internet connection. so IRQ processing and protocol routing on your local machine will always beat the rest in speed. however if it is messed up, it also creates a large gap in the processing. desktop network is usualy tuned for throughput and not latency, because latency tuning can influence user experience when something goes wrong. f.e. a DoS attack on a latency tuned system is way more effective than on a throughput tuned system.[/citation]

There is nothing compelling about your post. Thanks for furthering insulting me though.

In order for that to mean anything to me, I would have to believe that you understand the issues involved better than the pros at Bigfoot with established careers. I know nothing about you. Fail.
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
I decided my last post was not very clear...

The problem is that you are not really providing a satisfying explanation, haplo602. The question is whether it provides benefit beyond optimizing one's system for latency.

So yes, IRQ processing is very fast... but the windows stack has to process the packet, which must take more time that a dedicated processor processing it and passing it directly to the game. There is no way that the Killer is NOT faster than the windows network stack. The question is: is the difference noticeable.

I have yet to read anything in forums or on tech review sites where the author offers up an explanation that answers the claims of the Bigfoot Networks. There are at least three reviews on Tom's Hardware and thus far none of them inspire any confidence. There is a great deal of complexity to installing and optimizing the card... and there is no explanation of what was done to setup the card in any of the reviews.

So far, from what I have read, the people that have taken the time to install the card correctly and optimize it have experienced an improvement. These examples are very rare compared to the volumes of "I put it in and it didn't seem to do anything" reviews.

Perhaps this generation is spoiled. When installed the first soundblaster card it took hours online with tech support to get the IRQ's figured out. Now most hardware just works, but it takes time to optimize.

Now haplo602, you seem to know what you are doing. How about reading Bigfoot's LLR white paper and answering their claims?

http://www.killernic.co.uk/news/llr.pdf

 

cleeve

Illustrious
[citation][nom]bellbillsnow[/nom]There is a great deal of complexity to installing and optimizing the card... and there is no explanation of what was done to setup the card in any of the reviews.So far, from what I have read, the people that have taken the time to install the card correctly and optimize it have experienced an improvement. [/citation]

Whoa there. 2 things:

1- I'm the author, and I spent a lot of time on the phone with Bigfoot engineers making sure everything was working just fine. As well, I posted every setting of the card's control panel in images.

2- I experienced a big improvement with the bigfoot card.
The problem is that so far, I'm the only person who bothered to benchmark the card against a software solution (as far as I know).
And as it happens, the software solution ALSO caused a big improvement.

If I hadn't bothered testing vs. a software solution, I'd be singing the Killer Xeno Pro's praises like everyone else...
 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
Cleeve,

Okay, I see that you did have the card in game mode. Still not sure if you made the TCPAckFreq changes to the onboard NIC. And why did you leave the delay on 2?
And why no explanation of why previous reviews seem to indicate there is a benefit to the M1?
I still have not seen an explanation of why ping times are an effective measure of game responsiveness. This seems to be assumed. Or FPS's? Who cares if it is throwing up frames, but not showing what other players are doing accurately?

Finally, I do not think the simultaneous download issue is worth exploring until we get a handle on the no-load situation. I am just talking about optimization for game play.

 

bellbillsnow

Distinguished
Jul 2, 2009
25
0
18,530
Also, were you running the KillerNIC machine in the DMZ with the FNA Firewall?
Did you disable the onboard NIC when running the killer?

I don't get how you switched between the cards in your test? If I go from the onboard to the killer, I disable one, enable the other and reboot. Killer to onboard rarely needs a reboot.
 

razor512

Distinguished
Jun 16, 2007
2,130
68
19,890
the problem with these companies is they never offer public beta test to people outside of the company. if they did, they would have been able to quit before mass producing this crap


if you want to invest in networking hardware, buy a good router that you can install ddwrt or tomato firmware to.

a crappy router can cause a lot of networking lag especially when dealing with a lot of connections

another benefit

a router offers QOS which adds the benefit of better gaming to all computers on the network instead of just 1

I have had a 30 computer lan party go over a single linksys wrt54gl, some wireless and some on a switch connected to it (going over a 50mbit connection) to battle other other groups of people (fios connection)

kinda a get together after a convention that happened

no network problems at all (especially considering that it was a open connection and other people there may have even downloaded other crap while there, but tomato handles it all with no problem (wrt54gl running at 250MHz tomato firmware)

if you ping your router, the ping is often like 0MS and with modern motherboards, the onboard network card has a lot of dedicated hardware and the cpu usage is almost nothing
 
G

Guest

Guest
A much better solution is to implement a gateway/router using a spare older PC lying around and putting OpenBSD (http://www.openbsd.org/) on it. With a quick read through the FAQ and some time spent on Google, you can enable the Packet Filter engine and not only get a very solid and secure firewall, but by enabling queuing, get lag free game play no matter what else you're downloading. Since doing this myself, I can have literally hundreds of downloads going and still have the lowest ping in any game I play. Its just a matter of figuring out which ports the game communicates on and configuring your firewall to queue its traffic ahead of other things.
 

cleeve

Illustrious
[citation][nom]bellbillsnow[/nom]Still not sure if you made the TCPAckFreq changes to the onboard NIC.[/citation]

I did not, the onboard NIC was comepltely left at the defaults it was installed with originally.

[citation][nom]bellbillsnow[/nom]And why no explanation of why previous reviews seem to indicate there is a benefit to the M1?[/citation]

I never indicated there wasn't a benefit; I indicated that the benefit can be replicated with a freeware firefox plugin.

If you're specifically talking about no-lag results, I've seen other reviews showing that the no-lag results are also insignificant. As far as reviews that show a no-lag improvement - well, I can't explain another reviewer's experience, I know as much about them as anyone else reading an article on the internet.
But if you're asking me to make a blind guess, maybe the machines they were testing on had an existing problem with the way the network was working, and bypassing the stack bypassed the problem.

[citation][nom]bellbillsnow[/nom]I still have not seen an explanation of why ping times are an effective measure of game responsiveness. This seems to be assumed. Or FPS's? Who cares if it is throwing up frames, but not showing what other players are doing accurately?[/citation]

Well, higher pings=lag, and lag=poor game responsivemness at best, and being kicked from the game server at worst. As for FPS, it's only a problem if it's dropping very low. Even the amount of FPS improvement claimed by Bigfoot isn't going to make a real-world difference, and frankly if they hadn't made any claims I wouldn't have bothered testing the FPS.


[citation][nom]bellbillsnow[/nom]Finally, I do not think the simultaneous download issue is worth exploring until we get a handle on the no-load situation. [/citation]

IMHO, we already have a handle on the no-load situation - the card doesn't do anything better than an integrated NIC. I've tested it, and the results prove that pretty conclusively unless you're seeing something I'm not in those results.

I had a lot of back and forth with Bigfoot and they told me that everything was working fine. These are the results I got. I don't see what more there is to get a handle of.
 
Status
Not open for further replies.

Latest posts