[H] Killer NIC Evaluation

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
[H] Killer NIC Evaluation - You have seen the benchmarks and read the ping data. Now find out what the Killer NIC can do for you in real world gaming situations. No canned benchmarks, just feedback from gamers that have used the card.

Some gamers will see a benefit while others do not. Hardcore deathmatchers are likely to feel the Killer NIC advantages while the middle-of-the road player will not be fine tuned enough to benefit from the experience. Certainly though, the hardcore online gamer is exactly who this product is targeted at. The World of Warcarft advantages we witnessed were impressive to say the least as the Killer NIC imparted real gaming benefits and advantages.
 
I would be all over one of these IF:

-they were less than $100
-they get rid of that stupid heatsink

Nice article BTW.
 
Excellent review, thank you so much. Based on your findings, I'm going to be passing on this product for now.

Thanks again.
 
Good review. I'm glad you guys are giving it a fair shake. A lot of people are ragging on it badly. It's just too friggin' expensive right now, but it seems like a really cool piece of hardware. I think if they stick with it and put out a 2nd version that's cheaper, we'll definitely see people picking it up.
 
I really think that this is one of those things that someone doesn't need but if you already have everything, why not?


In my personal opinion this is one level above PPU's and hardly worth the current cost and its only one level above because it supports all online games and activity. Also a +1 for Linux.

World of Warcraft has the biggest boost and if your hardcore I guess I can see someone investing in this so they don't fuck up in raids or something due to lag spike. I dout it would help during a primetime server spike which many times we would be at Nef and lag out but I would help during the overall experience.
 
Very nice article... It outlined exactly what I wanted to know and i'll pass as well unless the price hit under the 100$ mark.
 
Malk-a-mite said:
It's your site disclaimer, you can do what you what, etc....

But... is putting the ad for the Killer NIC on the same page as the review... ummm, damn. I don't even know how to phrase this question.

EDIT: refreshing a few times I see it's part of a rotation of ads at least.

They have advertised with us for months and months now. They hold a contract on that ad space on all content we produce.

So what exactly is your problem with this? We are a publication that exists ONLY due to advertising dollars. I don't think you expect Road and Track to not have car advertisements...

If you think we are not capable of seperating editorial and business, well, just say it.
 
Bigfoot is stuck.

If the Killer NIC was cheaper (read: <=$100USD range), I think more people would go for it.

But...Bigfoot will recoup little if none of their development costs selling at that price point. But, the price point is rather high right now...until enough people buy it at the higher price point, I think it's just kinda going to stay there.

I don't think they expect high-volume, low-price sales to carry 'em sadly. :(
 
I still have not seen the tests for this product I would like to.

I want to see results with gaming while your full bandwidth is being used to download pr0n or whatever...
 
J-Mag said:
I still have not seen the tests for this product I would like to.

I want to see results with gaming while your full bandwidth is being used to download pr0n or whatever...
That's more QoS on your router's end, though would be nice to see the Killer NIC do some QoS work on the PC end too....FNA APP!
 
Interesting--it seems like it makes a bigger impact once your ping hits a certain level, like around 100ms.

I don't think I'll trust any benchmarks of the technology until I see somebody build two identical machines from scratch and do a double-blind test on them in their own controlled circumstances. I'm not saying that anything was underhanded--I just had a red flag pop up when I saw that you went to their offices and played on their machines that they set up. I don't think they did anything wrong, I've just been jaded by "reviews" where a certain CPU vendor said "Here, test these machines that we built, with only the benchmarks we put on there, with only the settings we say, and don't touch anything else".
 
I have some questions:

You tested at the Bigfoot offices... but were they your computers or theirs?

Can you tell us the specs/components in the computers you were using?

Were the KillerNics set to "app mode" or "game mode"?

I play WoW a lot and I have a KillerNic, and it often seems like WoW runs better in App mode, because sometimes in game mode, the server and my client seem out of sync, but it doesn't happen in App Mode.

Based on other reviews on the web, KillerNic seems best suited for low-end computers with CPUs that will benefit from the off-loading of tasks. High end computers have seen little benefit in the other reviews since they have cycles to spare for network activities. I'd be interested to find out what sort of systems you were using.

The value of the KillerNic seems pretty small (even though I have one... I don't always act logically... :p )... if you have a high end system... you might not see a difference. If you have a low end system... you're better off spending dollars elsewhere because a better CPU/Video card and you'll see a bigger difference.

That is unless we find out the [H] reviewers used some high end hardware for their tests.
 
J-Mag said:
I still have not seen the tests for this product I would like to.

I want to see results with gaming while your full bandwidth is being used to download pr0n or whatever...


The Killer card has the ability to prioritize applications that you want to have access to your bandwidth first. You just set it in the control panel. It is very simple and works perfectly. I have seen it do it first hand.
 
Mohonri said:
Interesting--it seems like it makes a bigger impact once your ping hits a certain level, like around 100ms.

I don't think I'll trust any benchmarks of the technology until I see somebody build two identical machines from scratch and do a double-blind test on them in their own controlled circumstances. I'm not saying that anything was underhanded--I just had a red flag pop up when I saw that you went to their offices and played on their machines that they set up. I don't think they did anything wrong, I've just been jaded by "reviews" where a certain CPU vendor said "Here, test these machines that we built, with only the benchmarks we put on there, with only the settings we say, and don't touch anything else".

Actually there is already a ton of data out there on this from indepent sites that have done the testing. There is NO DOUBT that ping rates are impacted. Again, this was not the focus of of our article, but the data is there with other sites should you require it. Actually pulling ping data is a lot easier than gameplay testing.....
 
Thanks for the article... but everything you say is in the realm of subjectivity. I know you hate canned benchmarks but until you (or someone else) can provide hard data that shows the benefits in actual framerates and pings I'm going to leave this one alone.
 
EODetroit said:
I have some questions:

You tested at the Bigfoot offices... but were they your computers or theirs?

Can you tell us the specs/components in the computers you were using?

Were the KillerNics set to "app mode" or "game mode"?

I play WoW a lot and I have a KillerNic, and it often seems like WoW runs better in App mode, because sometimes in game mode, the server and my client seem out of sync, but it doesn't happen in App Mode.

Based on other reviews on the web, KillerNic seems best suited for low-end computers with CPUs that will benefit from the off-loading of tasks. High end computers have seen little benefit in the other reviews since they have cycles to spare for network activities. I'd be interested to find out what sort of systems you were using.

The value of the KillerNic seems pretty small (even though I have one... I don't always act logically... :p )... if you have a high end system... you might not see a difference. If you have a low end system... you're better off spending dollars elsewhere because a better CPU/Video card and you'll see a bigger difference.

That is unless we find out the [H] reviewers used some high end hardware for their tests.

We tested on a SINGLE COMPUTER owned by Bigfoot. We simply changed between the onboard NIC and the Killer NIC. Switching modes required pulling the Cat 5 and plugging it into the proper port. And turning on the Killer software. We did restarts between rounds.

Again, one computer, so there were no differences in test setups. FX-55, 2GB Ram, 7900GT.
 
Cobalt33 said:
Thanks for the article... but everything you say is in the realm of subjectivity. I know you hate canned benchmarks but until you (or someone else) can provide hard data that shows the benefits in actual framerates and pings I'm going to leave this one alone.

There are already several sites with that data should you wish to use Google. :)
 
I'm glad [H] did a review, but I was wondering how Killer compared to a NIC. Where onboard chips have to rely on the CPU, would an add in card off load some of the cycles? I have spent much time on the forums, but I'm sad to say that networking is my weakest subject.

I guess I'd like to see a 3-way shoot out between onboard, high end NIC, and the Killer. Unless of course, the NIC has no benefits over onboard.
 
Thanks.
I think you were very objective.
This hardware does what I expected it to. Improve things a little under certain circumstances, in certain games, for certain people.
I'm just not one of those people.
Good article. :D
 
What I am wondering is would this be more a benefit to all if you use the Killer NIC on the server. If the server could send out UDP packet accellerated and the application had more CPU.....

Does this mean the server could load more players without lag?

Would this be a killer application for the killer NIC.

In other words, could you get more benefit from installing it in a server then on the end user.

I understand the benefits would be different, I just want to know if someone has tried it yet?

Kyle... You have to had thought of this...
 
J-Mag said:
So, there is no perceivable difference in ping rate when downloading? I have a GL-4300 and it prioritizes (only outbound I think and I can't recall if I can set specific bandwidth rates), but I can still tell the difference when running a gaming and downloading at full speed vs. just running the game.

Also, how would I know what kind of bandwidth is needed for any given game?

I have talked to some people that use this and it is not an issue. I would guess though it would have a lot to do with your available bandwidth as well. As noted in the article, I would do some research into to specifics if this is a big point with you. Our article did not focus on this. I think that most serious/hardcore online gamers are going to turn off any other bandwidth hogs while gaming. I know that is what I would do, perceptible after the prioritization or not.
 
Mackintire said:
What I am wondering is would this be more a benefit to all if you use the Killer NIC on the server. If the server could send out UDP packet accellerated and the application had more CPU.....

Does this mean the server could load more players without lag?

Would this be a killer application for the killer NIC.

In other words, could you get more benefit from installing it in a server then on the end user.

I understand the benefits would be different, I just want to know if someone has tried it yet?

Kyle... You have to had thought of this...

We talked to Bigfoot about this several months ago and at that time they seemed to think there could be some big benefits by using a Killer NIC on the server. The did not seem interested in pushing into that market at that time. For anything more specific you might drop them a mail.
 
I was wondering when the test results would be posted! Yeah!

Time to go read it now.
 
Great article. I plan on spending many hours with Crysis, UT2007 and my favorite Source games
after I build my new PC in late January or early February. If I can get this NIC for $200 or less
and there are people reporting some gains in the games I play I will get one. Of course I will need
an accessable PCI slot as well...
 
I really wonder if it's possible to buy a much cheaper PCI and simply do some of the Windows TCP/IP registry mods to get similar results. I know Windows doesn't handle high bandwidth traffic as good as is possible with default settings. Add in a NIC processor, and I'm gambling some net tweaks would yield similar results.
 
Mohonri said:
Incidentally, I very definitely fall into the camp that would benefit from upgrading everything in my computer before the network card....

I am with you on that. Maybe if I was playing WoW and was not happy with my experience, but I am not a WoW player.
 
Thanks for the article.

I wonder if you could actually host a linux dedicated game server off it at the same time you were playing on it?

Has anyone tried this?

I guess I will go try and find some more info on their forums. :D
 
I think that the concept of the Killer card is something to be interested about, however the testing methods need some improvements made i think.

This was posted on [H], so where is the [H]ard review? I have to say that after reading the article, i know exactly as much as i did after reading the marketing hype from Bigfoot themselves. Hopefully Hardocp has a more extensive, or comprehensive review in the works and are just posting this to tide everyone over till its been completed.

Lets compare apples to apples, or atleast get a little bit closer.. My suggestion: How about comparing a regular 20 dollar NIC to the killer card instead of using the onboard network, or even all three for that matter. Maybe im really outdated, but the last time i checked basically every onboard device used more CPU and resources than dedicated NIC, Sound, Video, USB, and other controllers, and generally caused the system to be slower than with their add-in counterparts.

Compare it against an add-in NIC solution too, on machines that you built. If you want a fair test, dont test their product in their labs, under their conditions, on their machines and net connection.

Take a handful of 20 dollar network cards and test Killer against them. One at a time. If a 20 dollar solution beats out your onboard solution too, or by nearly as much, lets get realistic. $180-230 buys major hardware upgrades for the average gamer, a few extra gigs of ram, faster drives, and in quite a few cases, faster processors. All lead to likely some sort of advantages in the gameplay, and thus a better experience. Its a major expense, lets have an in-depth review.

Granted thats from the point of view of looking at the Killer card as just a card, and not for what else it could bring to the table once everyone starts programming apps to run on it, or the fact that its better looking than most NIC's.

Just as onboard video, and onboard sound never 'cut it' comparing the Killer card to a single onboard networking solution dosnt seem like a fair testing proceedure to base anything off of. This is particularly true not knowing the specifications of the test machines, the speed of the connection, its usual latency to what servers were tested.. what the other computers (if any) on the same internet connection were doing at the time of the testing, the list goes on and on.

When performing tests there usually needs to be a constant, and in an area as inconsistant as networking, extra care should be taken in the testing process. How do the pings respond when other computers on the network are browsing or downloading, were those computers even connected to a network or were they connected directly, were they each using a seperate net connection if testing occured at the same time, was there a router or a modem in the way, what models, what speed was the processor, how much ram, what videocard, what soundcard, drivers?

The Killer NIC get an Editors Choice [H] Enthusiast Silver Award, but we, the consumers still know pretty much nothing about the thing. We didnt get any test configurations from the guys doing the review or any testing proceedures. [H] is supposed to be a symbol of quality assurance, yet nothing in the article at all made me believe that there was any control over the tests performed. While [H] dosnt owe me anything, i feel as though ive been let down as a reader.

This product should get the buyer beware award until further testing can be done in a more (trustable) controlled environment, testing in different configurations that the end user might experience in the real world, as well as a best case scenario.
 
I've been very skeptical of this card from day one and I honestly expected it to fail. Having read the [H] reviews and other sites benchmarks I have to admit I was wrong. It actually does do what it says it does, to one extent or another. Unfortunately it's out of my price range or I'd probably get one... something I would never have considered doing previously. Thanks for pursuing this and telling it like it is guys.
 
Sorry some of you guys did not like it. If you have noticed over the years we are moving away from presenting a mountin of data with a bunch of charts and graphs and then making conclusions. We are trying to focus more on the experience provided by the hardware. Understandably, this may not be your type of coverage. I am sure you will have no trouble finding a myriad of other sites that can give you exactly what you are looking for. Your readership will be missed. We appreciate the support you have given in the past.
 
Hi everybody, just thought I'd jump in to answer any questions folks have on the technical side. I'm an engineer and I work at Bigfoot (though I wasn't present for the testing performed).

What I can say is that we did give 1 (maybe 2) cards to HardOCP to take with them, and they've had them for a while, and did more testing than was posted (though I don't know to what extent).

Double Blind testing, performance testing, etc. etc. all show the same results.

We've done testing against other PCI NIC cards, and saw the same benefits from Killer vs. other NICs (even one our own CEO helped design, the Intel Pro 1000).

Now, on to some of your interesting tech questions:
1.) Yes, one COULD run a small game server in Killer (fairly easily with our latest FNA features), and it would ROCK. (And Bigfoot would likely award that person with hard cash for the effort). [See Bounty Program]
2.) Killer's Packet Prioritization testing: Other reviews have looked at this... and seen that we perform well (as good or better than NVIDIA's solution). Our Prioritization does both Inbound and Outboud priorities (NVIDIA is just outbound). Downloading and gaming is a breeze. (happy to explain more about how it works if there are questions).
2.b.) But OUR Soon to be released FNApp (a BitTorrent Client) Will be much Better, because not only are packets prioritized: no CPU power or even Hard-drive usage is needed (Killer has a USB Port and can attach a USB Flash or HD to it).... I'm excited about that (It's a project I'm involved with).
3.) Killer's in Servers: Yep. great idea. And It's no secret that we've been working on Server Technology For some time... (announcements about that will be forthcoming, and I'm sure HardOCP will be among the first to know).

Finally, I just want to say thanks to HardOCP for saying it how it is (good and bad).... and I think the product has just gotten better since the review (COD2 and BF2 for example are starting to show improvement now, since our last patch).

Hope this helps, I'm here to answer any other questions as well.
 
We are not going to stop reading just because of a (in my opinion) bad review, however i dont feel it was up to [H] quality standards. Just recently you guys reviewed a GF8800, http://www.hardocp.com/article.html?art=MTIxOCwxLCxoZW50aHVzaWFzdA== and had a 18 page article on it.

Yet this is the first time an enthusiast NIC is really in the spotlight for reviews, and our favorite place which usually does 10+ pages of review on a new piece of hardware, does 3 pages instead. This is a follow up to the 2 page review done earlier, and now a 3 page review in which the only reviewing to be had is being shown Bigfoot's card, in their machines, under their test conditions. Many of which conditions are either not known, or not explained in the article. I mean, the article itself dosnt even say what hardware the tests were performed on, we had to read the forums to find that out.

Like i say, as an [H] reader im just a little disappointed in the Killer NIC article and didnt feel it was up to standard, or contained the sort of information that the enthusiast, or even the average consumer needs to know in order to make an informed buying decision. Basically it says 'Of our four test candidates, two people couldnt tell a difference, two people could, may only help in certain games, go do more research.' Its an expensive card, that makes big claims, only works in some cases, and gets an award without any data to back up the claims it makes.

That said, im aware it wasnt tested in the Hardocp test labs, which is half the reason why i find it to be a very disappointing review, IE: how can ya reliably review a product that you dont get your hands dirty on. But the other half is filled in with all the missing information.

I think all of us who read the article would love to get more information on the testing proceedure, or have it re-tested in a non-engineered environment, on middle-grade and high end hardware, on ideal, and not so ideal connections to see what its really made of.

It would be wrong for me to tell you guys how to do your jobs, but reviews which clearly state the test machines, test proceedures, and look closely at what the product can, and cannot do, are clear winners in todays market. Sell your readers on your opinions, and the facts, both the good points and the bad. But back it up with data too so that we can see for ourselves.

It seems like the Killer review was more of a first look than a review, except that we already had a first look. And maybe im totally off in the wrong direction thinking that you guys 'reviewed' the Killer card. Obviously what you had set out to do was test its real world performance, and who is to say that you didnt get exactly what you set out for. But if the article dosnt tell us what any of that real world testing included, the hardware or the configuration, or the test conditions, or show us any of the numbers.. then it dosnt read as a complete review. (Maybe it wasnt supposed to be?)

Not to single you out or anything Kyle, but did you guys confirm what hardware was in the systems, or look at the network configuration, or ask all those questions which both myself and the other forum-goers have asked? If so then why not include them in the article? It just seems that a lot of very important questions were left unasked. And in the world of today, you have to ask questions, but more importantly you have to ask the right questions.

Ive gone to countless dealer meetings for the products we sell in our shop, and every single company has some sort of marketing scheme, rigged test, or misleading information which leads you to think that they are better than everyone else. And since you are going to them, to test their product, in their configuration, they can skew the results in any way that they want, and they often do.

We arnt saying that we dont appreciate the work you guys do, ask pretty much anyone who reads and they will tell you that Hardocp is a great resource. But now and then you guys stumble along the way. Its our job as your readers to catch you when you fall, or atleast point out what we want so that when you go into the field, you can better report to our needs, questions, and concerns related to a product.
 
Ratha,
I think you articulate a lot of good points in your last post, but this particular arena of hardware testing runs up against a real problem with the normal testing methodology.

One of the reasons that testing this kind of equipmet is so difficult revolves around one of the basic aspects of the Internet that is both an advantage and a disadvantage.

One of the strengths of the Internet is the "many paths up the mountain" approach to traffic routing.
Packets may not all travel the same route, but they all get there.
That's a good thing if your only concern is simply that the data arrives.

That same dynamic, however, insures that in many circumstances there is zero chance of repeatability when it comes to IP packet travel.

This makes it virtually impossible to have the repeatable structured testing that we all crave in reviews.

Given that limitation, a different test location and different computers in an environment that [H] controls isn't going to give results that are any more repeatable.

I think that Kyle stated the limitations and benefits quite clearly. The fact that there were no numbers attached to those statement *in this particular siuation* doesn't relegate it to being an opinion.

The Silver Award was given because the Killer NIC does what it says it will do.

Whether or not someone feels that what the Killer NIC does is worth the price, however, is going to be a completely individual decision.
 
First of Thx Kyle for good review.

Killer NIC was/is something that was just to exprencive to just get it for a test it and see that with game I play online it doesn't make any difference.

NIC's and routers and switches realy are so many times overseen by people but can make much difference. Mostly because some onboard solutions are working good and cost nothing. Some onboard NIC's and their implementation and drivers are pure punishment for Online gamer.

From moment I've moved in a appartment where I live at this moment I have had many problems with Interet connections and online gaming.
Finaly I have it my way. To get there I went true a Hell and back.

Funy thing about it all is that I live from Computers and networking and it took few months to have it sorted out.

Well on some things I haven't had much influence either (Provider and their Helpdesk., like : "Sir but Your connection is up, there is nothing wrong with it and We can't do anything for You....").

Well I've changed provider.

Some thing's I've changed and many things I've tested till I've got maximum out of my new connection. Like few switches, few routers and more then few NIC's. Needless to say there are worlds of difference between two network devices designed to do same things. Maybe not for a normal Internet user but surely for a Online gamer.

So Killer NIC is more then interesting thing that could potentialy help a lot.

Well many people talk about Ping and FPS and those are important values that define quality of online gaming. Though not die only one related to networking.

Like ability to make more connections per minute is also important. Also upload speed is more important then download speed.

Like it's better to have constant 75 FPS then to have 200 FPS if You play on LCD monitor. Having 200 FPS constantly You are missing so many action. Also frame rates peaking up and down aren't giving You smooth gameplay.

Server with low ping doesn't asure You good server registration. Without Registration You can't kill some one. Also some games show in server list one ping and when You connect ping is higher. Then You look to a netgraph and server list connected nad see difference od 10+ mS. As bigger a difference gets server will be worse for gaming.

After testing routers I've ended up with small Linux box with a two network cards. There is a simply no Router within normal price range that can outperform Linux box.
Specialy new lines of Routers with switches and wireless AP's are prone to degrading game expiriance limmiting transfer of packeges and easly getting flooded and maxed out.

Like broadband connection (even 8 Mb and more) and two players, one playing Wow and one CS with third person just chating will max out many home devices and make games unplayable.

On other hand using Linux box with separate switch and separate AP only will do the job with ease.

NIC's are mostly of two types. Active and passive. Active do have some processing power and bigger cache and do processor offloading. Pasive are mostly just an interface and CPU with drivers do all the work. Killer NIC is hyperactive NIC :)

There are server NIC's work great even for gaming. You can use many 64 bit PCI NIC's in normal 32 bit PCI bus with great results and they cost less then Killer NIC but do not have Killer NIC abilities.

Also Windows PC will benifit a lot from active NIC. Linux will not benifit that much. (talking about server linux apps without Xwin) Somehow under a Linux networking works great on almost any NIC :)

I have had friend using one of my old systems playing with me on some server and He have had better server registry and lower Ping then my main gaming PC :(

Well that old PC used old 3Com 3c905 parallel tasking NIC and gaming PC just a onboard NIC.

I've even seen that some network cards do not like some switches and work better in 100 Mb mode then in 1Gb mode. I mean they copy faster on 1G but gaming was worse.

Also some cards with same chipset do not perform same.

Also installing some other drivers for some other hardware can limmit ability to make lots of connections per minute.

ADSL 2+ makes so much lag if You live far from a first telecomunication switch. Over a 1,8 miles and it can addas much as additional 30 mS on every ping and limmit a connection to ~700 per minute.
Same location cable Internet and I get max 2300 connections per minute.
Well just nummbers but those nummers for me and game I play means better palying expiriance.


My advice would to any gamer playing on Windows machine and has problems with online gaming try some A brand active Nic it can make so much difference.
I have good expiriance with Intel NIC's, active Broadcom's, active 3Coms (with broadcom, Marvell and 3C chips on it)

Also some mainboards use just cheep Marvell chips or have bad impementation or drivers (some mobos with nvidia chipsets) and are useless for online gaming.

Then some other with same chip on it works great with other mainboard chipset :(

My expiriance is that A brand name active NIC between 35 and 75$ almost always makes difference in online gaming.

Now how much more difference makes 150+ $ more exprencive Killer NIC?

I guess it is a time for NIC Shootout on [H]ard|Ocp





MD
 
I suppose one of my biggest complaints about the article was the lack of hardware information, and setup information. It was said that in some cases the card helped, and in some it did not. But lets go into marketing real quick.

http://www.killernic.com/KillerNic/KillerAboutGameplay.aspx

"41% improvement in FPS in World of Warcraft using Killer"
and
"23% improvement in FPS for F.E.A.R"

Sounds like Killer is going to be my next videocard. :cool:

My last videocard didnt give me that much of an increase in frame rates, if i buy that into that marketing hype expecting to even get a 10% increase in framerates id be quite disappointed. Theres no link to reviews/articles to back up these claims. 41% increase from how many frames per second? If your game runs at 5fps and with killer it runs at 7fps, thats a 40% improvement. Dosnt mean much in the real world yet it verifies its claims. Without test configuration the numbers are meaningless.

http://www.killernic.com/KillerNic/KillerAboutLLR.aspx

They say here that the Max for world of warcraft is 12.6%, EDIT: Some test configuration was displayed for this number, i overlooked it.

12.6% is more believable, but this is world of warcraft, a very dynamic game, whats to say one of the models didnt run off on a quest somewhere. Or a sound/music/effect *not* playing caused the framerate to jump fro 30 to 34? (a random number times 1.126 rounded up)

In the end, consistancy matters, when you hear about a car having a top speed of 166 miles per hour, is that at sea level, what was the temperature and barometric temperature that day, was that up hill, or down hill. Was that top speed from point A to point B, or the top speed they could obtain on a straight and perfectly level track with the gas floored, what was the quality of the gas, was it free from contamination, how much gas was in the tank/what was the weight of the vehicle at the time of testing, was the engine cold or warm before the test began?

Im not trying to say that Killer dosnt do what they claim it does, but as an enthusiast who has to pick very carefully where every dollar goes, i require more information than the deep-pocket believe what you read sort of person. Id love to have a network card that was magic like this thing claims to be.

The bottom line is this: Numbers are meaningless if they dont have any context to be put into, just as a review of a product is if theres no specifications to which the tests had to adhere to. After reading the review i was left with many questions:

1: What system was Killer tested on?
2: What connection tests were performed to see how Killer handles different latencies?
3: What will Killer behave like on a non-state of the art machine?
4: Will the networking benefits of Killer be nullified if you are behind a router? One which has other traffic?
5: Did the mice or keyboards or other combinations of hardware that the test users employed cause different test results between different people? "Our testers were given time to setup their own keyboards, mice, and needed add-on software so that they had a system close to what they were used to using at home"
6: Was Killer tested on a network of one, or of many, behind a router?
7: Did the server load differ between tests?
8: Was the Killer card disabled so that no drivers were loading that could interfere with the onboard NIC, was the onboard disabled so that it didnt interefere with Killer?
9: Whys there no mention of testing Killer against an add-in NIC. Is that an intentional omission to obscure the test results?
10: Graphics cards get frame rate tests, Killer says it increases frame rates, why were no numbers provided for us to look at? A 50% improvement over 2 fps is 3fps. Now if its a 50% improvement over 50fps, thats worth looking at.

I mean i could keep going on and on and getting more technical with each question. Obviously you cant expect a single person to do a full review of every software and hardware configuration out there, but without knowing what tests were performed, and what measures were taken to avoid inconsistancies you cant judge anything for yourself to see how it it fits into your unique application.

If this was *not* a review of the Killer NIC, then i take it all back. However if this was, then im pointing out a lot of things that this review did not answer for me, and these are probably similar things that the other gamers/enthusiasts here are thinking to themselves.

Like i say, id love to see Killer be the next best thing, its expensive, but if its what they say it is then it'd be great. However, the review was not thorough in explaining the testing proceedure, and gave absolutly none of the important details. It gave 4 people's opinions and experiences using this hardware in a phantom environment with who knows what controls in place if any to insure accurate testing, and didnt provide us any of the test data to analyze.

I dont mean to be hard on the reviewer, but after i look at the detail and work that went into the GF8800 review for example, complete with framerate graphs, average framerates, game settings, AA/AF, test setup and configuration data, patch numbers, driver revisions, and then real world thoughts and 'feel' tests.. this one just seems very.. uninforming. We dont know anything about the settings the games used, or even the network configuration. Both of which could obscure or inflate the results of the tests in either direction.

Anyway, without writing any more pages (dont want to kill this thread) ill just say that i feel a more comprehensive review with more controlled tests would be more beneficial to both the review of the Killer NIC and to the end user in their search to make a buying decision. Highlight where the product shines, show where it fails, and what configurations are needed for optimal results and what ones could benefit from an upgrade to a 20 dollar card and a 200 dollar processor, instead of a 225 dollar network card.

Anyway, goodnight and good luck with your questions. :)


Ratha
 
Wow, great article packed with tuns of info and incite
Gives a little better idea on just what this thing does
though i still think its not worth that price tag....
 
Good article. The article conveys to me exactly the information that I was interested in getting. Like many others, I'd be better off upgrading just about everything else before going with the Killer NIC.

I would also like to see a few less expensive add-in NIC versus the Killer NIC, or even an add-in NIC versus a couple of on-board solutions. Personally, I think that could really highlight whether onboard solutions actually take up a noticeable amount of CPU cycles, or not.

If a sub-$50 add-in NIC helped my gaming experience a bit, I probably would buy it, since it's just easier to spend that amount than it is to spend $200+ (new video card, new CPU, etc.).
 
Thanks for the series of interesting articles on this Killer NIC, Kyle.

I was an avid CS: Source player for a while and I must say, the volunteer who tested the Killer NIC for CS:S probably did not have enough time to adjust to the increase in ping rates. This is quite obvious when he joined a server with higher pings that became lower due to the Killer NIC. I assumed the higher pings were adjusted to the usual ping rates he experiences at home.

Many an FPS gamer who started back in the dial-up days knows about the amount of adjustment switching to broadband took since you no longer had to compensate and lead due to the lag (well, not as much, anyways).

Right now, the consensus is the price of Killer NIC is discouraging gamers to buy it. If Bigfoot can release a lower tier model of the Killer NIC, where the lowest end card offers the Windows TCP/IP Stack bypass for ping enhancement, but does not include the extra memory or ability to run FNA apps and price this at $75, I'd definitely consider a purchase.

I think they'd do better to market the high-end card for power users and server applications. Especially when FNA apps library grows to accommodate their needs.
 
QuakerOatz said:
Sure it's nice that [H] is different than the "myriad of other sites with pages of #'s", but if "different" is something that doesn't add value for the reader, why do it? It's like buying a Zune because you hate Ipods.

I am sorry our differences do not add value for you. We have a lot of feedback that shows that our approach is adding a depth of value for many of readers that they cannot find elsewhere. Should you like a bunch of graphs and data, there are plenty of other sites that cater to this. In regards to the Killer, I see no reason to regurgitate and spend cycles on things that have already been established by hardware reviewers and folks in the community.

We take each evaluation we do on its own merits and decide the avenue of attack we are going to take.
 
..WoW and linux...great for some, not for all. Seems this product has 'Striker' disease - the costly product you buy, when you want to blow some dough...good arti, despite what some may say...
 
Back
Top