Misc

Bigfoot Networks Killer 2100 Review

vortez
April 19, 2011 20 Min Read
2 0

A gaming network card that is said to improve transfer speeds. Does it really work, and should you buy one? We put it to the test.


Product on Review: Bigfoot Networks Killer 2100
Manufacturer and Sponsor: Bigfoot Networks
SRP: £69.99
Street Price: £56.65


Back in 2006, Bigfoot released the Killer NIC K1 and slightly later, the M1 models. These were a relatively new concept – high end network cards. Before the Killer NIC, network cards took somewhat of a back-seat in a high end gaming rig, with most opting to use the onboard network of the motherboard. This still really remains the case, though perhaps awareness that the Network Interface Card (NIC) can have an effect on the online game experience is increased. The problem with the K1 and M1 were that they were extremely expensive at £100-£150 and also that due to the benefits being difficult to test reliably and only small changes in performance observed, they were met with a very lukewarm reception. In fact, the general consensus amongst the gaming community was that there were no real gains to be seen and a normal NIC was good enough as makes no difference, and those who did splash out on the Killer NICs were often met with criticism for falling for marketing spiel and having too much cash to burn. This is a shame, as the theory behind the cards was solid, and these early models had support for other exciting technologies such as the inclusion of FNApps, which allowed users to run BitTorrent clients and more on the card itself, with strict controls over bandwidth usage and gaming traffic prioritised. BitTorent whilst you game? Sounds too good to be true, and unfortunately, often it didn’t work so well, perhaps cementing the cards reputation and contributing to their demise.

A re-think came in mid 2009, with the Killer Xeno Pro – a new, smaller, cheaper card with a more reasonable price tag of ~£75. This dropped the FNApps but included support for offloading VOIP traffic to the card. This, also, never really took off. Coupled with occasional problems with TCP throughput and still a relatively high price (compared to the ‘free’ network chip on motherboards) meant that, although this time the product had a much higher chance of becoming widely popular, it never saw mass appeal and was restricted to only the highest end PCs and a select number of OEM systems.



The latest model, the Killer 2100 is their most streamlined product yet, focussing directly on delivering the core feature, unrivalled gaming network performance. The wheat has been sorted from the chaff, leaving us with Bigfoot Network’s lowest priced Killer NIC thus far. An SRP of £69.99 and a market price of around £55 if you shop around; this card is now much more of an upgrade prospect for a large number of enthusiasts. How has it changed? The VOIP support is dropped, the card has been tweaked and updated under the hood, and the new software is more reliable and supposedly, delivers better overall performance than the Killer Xeno Pro.

The question now in your mind will be, ‘Should I buy one?’. This review aims to give you the answer.

Features and Specifications


Gaming network cards have proven to be a difficult sell, so as you may expect, Bigfoot Networks have a lot to say about their latest product. Here’s how they describe the features:

Game Networking DNA™ – Only Bigfoot Networks exclusive Game Networking DNA™ combines the benefits of a dedicated network processor, Advanced Game Detect™, Windows network bypass, Visual Bandwidth Control™ and other technologies designed to deliver the best online gaming experience possible.
Advanced Game Detect™ – Automatically classifies and accelerates traffic to and from your game faster than any other desktop, gigabit networking product on the planet.
Visual Bandwidth Control™ – See which applications are hogging bandwidth and tune performance for each, so they don’t interfere with your online gameplay.
Online Gaming PC Monitor™ – Monitor and manage the health and performance of your gaming rig with graphical displays, detailed logging and complete redesigned user interface.
Application Blocking™ – Provides per-application control to block programs that access the network for increased performance and safety.
Automatic Bandwidth Sensors™ – Provides per-application control to block programs that access the network for increased performance and safety.
Plug and Play – Simply plug in the card and install the software. Killer™ 2100 automatically works with all online games – no mods, patches or updates required


Why do Bigfoot Networks think you should buy one? Here are their ‘Benefits’:

Performance – Whether you’re playing MMOs or first-person shooters, the Bigfoot Networks Killer™ E2100 game networking platform delivers fast networking with a new level of online gaming performance and a competitive edge over standard network interfaces.
Intelligence – Killer™ E2100 hardware identifies and accelerates traffic to your game with Advanced Game Detect™ for unmatched online gaming performance.
Control – See which applications are hogging your bandwidth. Set limits using Visual Bandwidth Control™ or shut them down completely with Application Blocking, so they don’t interfere with your online games.
Visibility – Monitor and manage the health and performance of your gaming rig with at-a-glance Online Gaming PC Monitor™.
Optimized Voice Chat – Get clearer chat and lower-lag by prioritizing your favorite voice programs with Visual Bandwidth Control™.


For the more technical readers, here are the hardware specifications:

Specifications:

– Gigabit Ethernet
– 400 Mhz Network Processing Unit
– 128MB RAM
– PCIe
– Killer Network Manager Software Suite


Note that the Killer 2100 is, like its predecessors, basically a mini PC on a card!
If you’re already convinced (I wouldn’t be!), here’s what you’ll need to make the most of the card:

Requirements:

• PCIe motherboard with open PCIe slot
• Broadband internet connection (wired)
• CD/DVD ROM drive
Microsoft Windows Operating Systems:
– Windows 7 (32 & 64-bit)
– Windows Vista (32 & 64-bit)
– Windows XP (32-bit only)

Packaging and Bundle


As ever, the arrival of new hardware is always met by a certain, boy-like excitement in most enthusiasts, and when the Killer 2100 arrived on my desk, it was no exception.


The claims on the front of the box are simple, and to the point


As you can see, the box art is pleasing to the eye and sells the product in a very similar way to a GPU box would.


A feature list is on the back of the box


The rear of the box simply lists the features we saw on the previous page, as well as an image of the card. This box arrived slightly worse for wear thanks to our wonderful postal service, but it still protected the entire contents from damage.


The bundle is sparse


You’ll remember that I said this was a streamlined product, well the bundle reinforces this viewpoint. Bigfoot Networks have included no accessories, simply a quickstart guide and a driver CD, though I’d recommend you download the latest driver from the the Bigfoot Networks Website before you install. The card itself comes in an anti-static bag.

Killer 2100



The top of the card, showing the Bigfoot Networks logo on top of a mesh cover


The card isn’t as attractive as the original Killer K1, but there’s no doubt that it wouldn’t look out of place next to a GPU in a high end system. The black shroud covers the PCB and chips below, also providing some form of ventilation, not that the card ever gets that warm. You can also see the sole port – an RJ45 connection and the PCIe x1 connector, which will fit in any PCIe slot.


The back of the card is also ventilated


The attractive design is also extended to the rear of the card, where there is a ventilated section. No more connections here, in fact, the network connection is the only one available on this card, unlike the Xeno Pro, K1 and M1.


The bottom reveals a surprisingly complex PCB design


On the underside there are also two stickers for model information and serial number, but not much else of note.

Closer Look


Let’s take a closer look at the card.


The lone Ethernet port


The card, although not as well-finished some of the early models, has some nice touches such as the ‘K’ logo seen above next to the port. The quality is also very high throughout, there’s nothing flimsy about it.


Without the shroud, you can see the full PCB and the core


Beneath the cover, you’ll see the 400MHz Network Processing Unit (NPU) which shows the familiar ‘K’ logo. You can also see that this is quite a small card, so space considerations in most chassis shouldn’t be an issue, though Bigfoot Networks have made the decision to make a full-height card as opposed to a low profile compatible one, which may prevent some people from installing the Killer 2100 into their cases. However, it must be considered that the vast majority of people buying a gaming network card will have a high performance GPU, which are almost exclusively full-height. There are situations when purchasing this card may be useful, even if you don’t game, but that will be covered later.
Note the similarities (and more, importantly, differences) to the Killer Xeno Pro pictured below.


The Xeno Pro preceded the 2100, and supported VOIP features as well as a useless USB port


The 2100 clearly uses the same PCB design as the Xeno Pro, but it does not share all components. As mentioned in the introduction, the 2100 is streamlined (read – has features removed), but also sports higher quality components and improved network performance, according to Bigfoot Networks.


The 400MHz NPU


There is no cooling solution on the NPU pictured above, and the vented shroud is enough to keep the card barely warm whilst in use. That said, I wouldn’t put it somewhere with no airflow at all, but that is generally not recommended for any PC components.


64MB Samsung RAM chips for operation of OS


Here you can see the two Samsung memory chips that make up the 128 MB of onboard RAM. Now this memory doesn’t really have much of a direct bearing on the network performance, it’s just to allow the card to run its own proprietary operating system. If you think about it, you don’t want to put network packets into memory – it’s an unnecessary step that would increase latency.

Installation



The Killer 2100 installed into a spare PCIe slot


Installation of the Killer 2100 is a fairly standard affair. It’s as simple as installing any other PCIe card into your machine. There is no PCI alternative, so you must have an appropriate slot free on your mainboard. Also note, if you have to use a 16x slot due to space restrictions, or even a 4x slot on some boards, it may reduce the PCIe 16x lanes available for your graphics card from 16x to 8x. This totally depends on the motherboard you’re using, but it’s something that you should check with a tool such as GPU-Z if you’re unsure as to how your motherboard works. This would have the effect of possibly reducing FPS rather than the increases the Killer 2100 is touted to offer.


The Killer 2100 gives of a mean red glow


The LEDs have a pleasing appearance as the light goes through the mesh cover. Matches with some Lian Li and vented cases well, such as the one above. If this would ruin your lighting scheme (if you have one!), it can be turned off completely in the software that comes with the card.


Without drivers installed


Once you boot up your PC, before driver installation you’ll see two devices that need drivers in the device manager, as above. I recommend that you install the latest drivers from the Bigfoot Networks website, as there have been many revisions and improvements since release.


The main window for the Killer Network Manager Software


On installing the drivers and software, which come in a single package, the firmware of the card will be updated if it needs to be. This is relatively quick processes, but you may quickly stumble into it if you’re used to simply pressing ‘next’ at each screen. It’s very important that the process is not interrupted, as it may damage the card, so make sure you really do have all other open applications closed before you begin installing.

Above, you can see the first page presented to you by the Killer Network Manager software, which loads into the system tray at startup. This page provides basic information on your machine.

Installation cont'd



The PC Monitor tab


This screen allows you to monitor and log performance data relating to network gaming on your machine. The options include CPU usage, NPU usage, Internet Bandwidth monitor, Memory usage, FPS (if you also install FRAPS) and finally, and perhaps most importantly, ping.


The Applications tab


The Applications tab allows you to selectively control allowed bandwidth over the applications on your PC. As you can see, Battlefield: Bad Company 2 has been allocated the maximum for my line, 10 Mbps down, 0.5 Mbps up. The software is able to automatically detect your connection speed, or you can enter it manually, if you prefer. The control that this application gives the user is a very good feature. Although most Bittorrent clients allow bandwidth control, this tool will be useful for limiting them (whilst gaming) and other programs with no such controls. Coupled with the prioritisation of gaming packets, this theoretically makes the Killer 2100 a good option if you’re multi-tasking with your networking.


The Network tab


This tab allows you to set the connection speed for better control over applications that are using it. It is also allows you to choose LAN exceptions, which is an important option because without it enabled, you will throttle LAN traffic, which if you’re local file sharing or LAN gaming could cause some performance issues.


Advanced options


Finally, the advanced options page allows you to control the broader options of the application itself, and of the card. You can control which information is shown on the overview panel, default priority of the applications, the whole application control, you can reset application data to defaults, as well as change graph units and logging options and finally the LEDs on the card can be disabled here too.

Testing Methodology


In order to assess how well the Killer 2100 works, many different methods have been used, as testing overall network performance is surprisingly complex. The test system used is shown below:

CPU: AMD Phenom II X4 965BE @ 3.8GHz
CPU Cooler: Noctua NH-D14
Motherboard: ASUS Crosshair III Formula
RAM: 2x2GB G.Skill Ripjaw DDR3 1600MHz
Sound: Creative X-Fi Xtreme Gamer
GFX: AMD Radeon HD 5870
HDD: Mushkin Callisto Deluxe 40GB SSD, 2x1TB Samsung F1, 1x WD RE2 250GB
PSU: Corsair HX620
Case: Lian Li V1000 Plus II



The Gigabit switch used in the test setup with three connections. One to each network adapter and one to the router.


The network setup is important to this test as it can affect the results, especially since both of the network adapters on test can easily saturate a normal 100 Mbps Ethernet connection. Therefore, 1000 Mbps (Gigabit) connections were used between the cards on test. The setup used a Gigabit switch (D-Link DGS-1008D) and Cat5e cable to connect both network cards and the router together.

The first 2 tests make use of software and methods provided by Bigfoot Networks. These are the GaNE test and NetPerf tests and results should be observed with the caveat that they may be biased towards the Killer 2100 as they have been devised by its manufacturer. The second set of tests, including File Transfer testing and Real World testing are much more representative of everyday usage. The real world tests chosen to run are using two popular online gaming titles to test the ping or latency associated to online public servers.

Firstly, we must understand a little more about the nature of networking and what we’re actually testing:

Latency is a measure of the time it takes a network packet to reach a destination, i.e. from your PC to the game server. This is how fast the data are transmitted and is sometimes referred to as ‘lag’. Ping is a measure of latency that includes a round trip from client to server and back. This is like the speed limit on a motorway, how fast.

Throughput is a measure of bandwidth, i.e. how much bandwidth is available to transmit data every second to the destination. This is commonly referred to, perhaps incorrectly, as the speed of an internet connection and is measured in Megabits per second, such as a 10 Mbps or 50 Mbps service. This is like the number of lanes on a motorway, how much.

The amount of latency does not indicate the bandwidth, and we are in a situation where broadband internet services are sold on bandwidth and not latency, and online gaming only typically requires a very small amount of bandwidth (ca. 5 kbps to 25 kbps). The quality of online gaming is much more sensitive to latency or ‘lag’ yet we cannot purchase a service to optimise this once the signal is out of the house. Indeed, purchasing a higher bandwidth internet connection is absolutely no guarantee of getting better latency, and therefore online gaming experience. Modern networks and network adapters are optimised for highest throughput. The Killer 2100 however, aims to minimise latency to game servers as much as possible by bypassing the Windows network stack and fast tracking game data directly through its hardware. Specifically UDP (User Datagram Protocol – less reliable but better latency) packets, which are most commonly used by online games, whereas normal network adapters are optimised for TCP (Transmission Control Protocol – more reliable but worse latency) packets which are used for web pages and mail etc.

If you got through that, take a deep breath and we’ll dive into the testing.

Game Network Efficiency Testing


This test was developed by the manufacturer to simulate real world network performance of online games. The test simulates the load of a game on the PC (by running a game benchmark, in this case Resident Evil 5) and measures the transmission of gaming network traffic over a local network by utilising two machines, one to test and one to ‘listen’ for packets. The test is designed to examine the performance of two network cards simultaneously, so the onboard network card and Killer 2100 were tested. For the more technical of you, 100 byte packets are sent over the local network on a round trip every 50ms and the average ping and jitter (deviations in ping) are measured.

Each test was performed 3 times, and it was noted a gradual decrease in the performance was measured each time. This may have been a problem with windows networking or the ‘listen’ PCs internal network card. It could not be resolved, so it is probably safe to assume that these figures represent a more realistic real-world comparison of the adapters rather than the idealised system presented by the manufacturer.


The Killer wins out here, but not by the margin expected


Provided with the testing software, Bigfoot Networks include estimated values that are considered reasonable for this test. The results here show the Killer in a much less favourable light than what is expected. As above, the reason for this is unknown and despite much troubleshooting, no solution was found. The Killer 2100 still shows better all-round performance than onboard non the less.

NetPerf Testing


Netperf is an industry standard application to test throughput over networks. As with the GaNE test, the manufacturer has provided some instructions and files to test with. Both the UDP and TCP are tested for each adapter.


The Killer is clearly optimised for UDP packets


The test results show that the onboard adapter provides more bandwidth for TCP packets, but the Killer 2100 blows the onboard out of the water in terms of UDP performance. This is good to see, as if you remember back to the testing methodology page, UDP traffic is that which is most commonly used by online games.

File Transfer Testing


This is the first of what can be regarded as a ‘real world’ test. A simple 2.5 GB file transfer from one machine to another, using each adapter. Each run was repeated 3 times and averaged.


A win for the onboard network chip


The onboard performs slightly better here, which may be expected due to its TCP optimisation, but the times are actually quite similar. This is because the major bottleneck of performance was not the gigabit network, but in fact the write speed of the hard disk drive in the machine being transferred to. So, in the real world, it’s of little consequence which network adapter you use.

Battlefield: Bad Company 2


This game needs little introduction – an extremely popular FPS title with large maps and up to 32 players per server and a variety of game modes. Things can certainly get frantic, and lag is the last thing you’ll want whilst trying to snipe the advancing team or bring down a helicopter with a rocket launcher.

In order to test the performance, there are some qualitative notes, but also some hard numbers. A busy 32-slot server was chosen to test on. The ping number given on the in-game score board was recorded every 30 sec for 4 minutes two times with each adapter, and the scores averaged. The average values are seen below.


A 1% difference in the Killer 2100’s favour


These are averaged results, but they are within the margins of error, so it is difficult from these numbers alone to be convinced that the Killer 2100 offers a better experience.


The onboard shows the least variability


The standard deviation was measured for each card using the same data, and it is a surprise to see the onboard offering a more stable ping, but it also must be stressed that 0.61 ms is very far within error margins on these readings, and so, like the average ping, there’s not a lot to read into this other than that the results are essentially identical. The gameplay experience itself was felt to be the same on both adapters.

Day of Defeat: Source


Day of Defeat: Source is a popular Source engine based online FPS that is considerably faster paced than Battlefield: Bad Company 2 and so latency can be a game-changer in this shooter. Again, a busy server was chosen to best test the Killer 2100’s abilities.


The onboard was 10% better in this test


Curiously, the onboard has given the lowest pings, though once more, I believe this difference is within the error margins that can be expected with not running the tests simultaneously on the same server from the same broadband line and identical spec PC etc. There are a lot of variables in this equation.


Small difference is within expected error


Once again there is a slim victory for the onboard with stability, although it was noted several times that the connection to the server using the onboard was less reliable than when using the Killer 2100. Also, initially connecting to the servers and getting onto the game was quicker with the Killer 2100. This indicates that perhaps the Killer 2100, as expected, is the more stable gaming card, despite the hard numbers showing very little either way.

Conclusion


There’s a lot of information in this article, and that is testament to the complexity of networking and really trying to understand how well the Killer 2100 works. Well done if you’ve followed it all so far.



The Killer 2100 is a high end piece of kit, and it is presented very well. The software is also quite comprehensive and gives the user heaps of control over every aspect of its abilities. We’ve seen from the NetPerf testing that the Killer 2100 certainly has the raw performance that is claimed by the manufacturer.
However, back in the real world, the Killer 2100 is an expensive network card at £55+ compared to the £0 cost of an integrated solution. Even worse is the fact that when it comes down to it, the Killer 2100 is addressing only the smallest part of the latency incurred when online gaming, and the tests show you are much more at the mercy of the ISP and game servers than you ever will be your onboard network card and operating system.

Perhaps the Killer 2100 is better suited to LAN events, though since the infrastructure at such an event is so much better than the internet as a whole, the Killer 2100 is fighting over fractions of milliseconds that are most likely imperceptible to even the sharpest gamers. There is probably a niche somewhere where the Killer 2100 is ideal, but I haven’t found it and I’m struggling to imagine where it would be. For this reason, I can’t recommend shelling out for the Killer 2100 unless you know this is what you need.

Bigfoot Networks have crafted a beautiful and intricate supercar of a network card with the Killer 2100. It gets the job done fast, and in style. Unfortunately, the world of online gaming is somewhat akin to the daily commute to work. Using the Killer 2100 online is like driving the supercar to the train station. You’ll get there quick, but the chap also on the train with the suit and Mondeo will be at his desk at the same time as you.

Pros
+ Very high quality product
+ Lots of control over functions using software provided
+ Offers brilliant UDP latency


Cons
– Near useless in everyday gaming
– Much more expensive than integrated solution
– Very limited worthwhile applications


Share Article

vortez