How much can a network card improve game performance? Can an onboard
card really hinder your gaming performance? What's the difference
between a $20 card and a $130 card?
Earlier
this year we were introduced to Bigfoot Networks' new Killer
2100 Gaming Network Card. As we didn't have a chance to test out the
card for ourselves at that time, some of you said: "Hey, what's up with
that?" Well, the good news is that Bigfoot has provided us
with a
testing unit so we could check out the power of the card ourselves.
The card itself is considered a gamer's network card. What this means
is that Bigfoot Networks claims the Killer 2100 Gaming Network Card is
the fastest networking product available for online games. It
"maximizes online game performance, improves game play and reduces
problems associated with lag."
How does it make such a claim? First, the card analyzes the packets
going in and out of your PC and is able to identify if it is network
traffic from gaming. If it is, it gets priority over other tasks your
computer may be trying to use your network connection for. It
completely bypasses the Windows Network Stack and gives users full
control over which applications get bandwidth and offers the ability
to customize how much bandwidth, if any, each application will be
alotted.
In addition, since the NIC has its own processor, it takes the
processing load off the CPU. As you'll see from our tests coming up,
this can mean a finite boost in frames rendered per second.
Before we get into the results, here are the specs of the PC with which
I outfitted the Killer 2100.
style="display: table-cell; vertical-align: middle; text-align: center;">
style="color: rgb(255, 204, 51);">Mac's Test PC Rig Specs
Motherboard:
Asus P5E3
Premium/Wifi-AP@n (slot 775)
Processor:
Intel Core 2 Quad
Q9450 @ 2.66GHZ (per core) Yorkfield
1333FSB
RAM: 4x 2GB
OCZ Platinum
DDR3-1333 Dual Channel (8GB total)
Video: 2x
Sapphire Radeon 4890
HD in Crossfire
Sound: Sound
Blaster X-fi
Xtreme Gamer FATAL1TY Professional
Series
OS: Windows
7 Home Premium
Onboard NIC: Marvell Gigabit
LAN
Installed NIC: Killer
2100 Gaming Network Card
I also connected an HP Pavilion laptop with a gigabit ethernet
connection to the network from which I ran part of my tests.
The installation was simple. The network card is a PCIe card, so you'll
need to have a PCIe slot free. The card can be installed in any PCIe x1
or faster slot. The only issue I had with the installation process was
trying to fit the card into the only PCIe slot I had free. I ended up
having to remove one of my video cards, installing the NIC card, and
then reinstalling the second video card. It's important to note that
this was solely due to the layout of my motherboard, and had nothing to
do with the Killer Network Card itself.
Once installed, the software installed quickly and easily and I was
good to go. I booted up some of my favorite games and started running
some tests. The following are the games I tested.
Age
of Conan: All settings set to max. I ran the DX10 client
at 1920x1200 and ran around Conall's Valley during the tests.
APB:
The newest MMOG to hit the gaming space, I felt this test would be a
good one as it relies more on pinpoint accuracy and response time than
the other games tested. Max settings. 1920x1200 racing around the
Financial District.
style="font-weight: bold; color: rgb(255, 204, 51);">Dungeons
& Dragons Online: All settings to max with DX10
enabled. Resolution was 1920x1200. I ran around Stormreach Harbor.
EverQuest
2:
All settings to max except foliage density. Shader 3.0 enabled.
1920x1200. Qeynos.
style="font-weight: bold; color: rgb(255, 204, 51);">World
of
Warcraft: Dalaran. Full effects. 1920x1200
Onto the test results!
style="font-weight: bold;">Test Chart
For the first test, I took a simple average framerate. The test in each
game was run 3 times for one minute each. The result in the FPS column
is the average FPS from all 3 tests. The purpose of this test is to
show how much smoother a game can run when the processing is taken off
the CPU and put onto the Killer 2100.
The second test was a result of a series of 100 32-bit packets sent
from the laptop to the test machine as it was running the game. Why is
this test valid? Again, it shows how quickly the Killer 2100 can
process traffic. This test is a more accurate test of the cards
capabilities as it eliminates other variables over the internet that we
have no control over (ex. a routing ISP down in San Diego as our
traffic tries to continue to the game's servers in Redwood.)
title="Game"
src="http://www.tentonhammer.com/image/view/76987/preview">
src="http://www.tentonhammer.com/image/view/86882/preview">
src="http://www.tentonhammer.com/image/view/86887/preview">
src="http://www.tentonhammer.com/image/view/86886/preview">
src="http://www.tentonhammer.com/image/view/86884/preview">
src="http://www.tentonhammer.com/image/view/86885/preview">
src="http://www.tentonhammer.com/image/view/86888/preview">
src="http://www.tentonhammer.com/image/view/86885/preview">
src="http://www.tentonhammer.com/image/view/86888/preview">
src="http://www.tentonhammer.com/image/view/86885/preview">
src="http://www.tentonhammer.com/image/view/86888/preview">
src="http://www.tentonhammer.com/image/view/86885/preview">
src="http://www.tentonhammer.com/image/view/86888/preview">
alt="Age of Conan" title="Age of Conan"
src="http://www.tentonhammer.com/image/view/77007/preview">
FPS
FPS
title="APB"
src="http://www.tentonhammer.com/image/view/86883/preview">
FPS
FPS
alt="Dungeons and Dragons Online"
title="Dungeons and Dragons Online"
src="http://www.tentonhammer.com/image/view/77008/preview">
FPS
FPS
alt="EverQuest 2" title="EverQuest 2"
src="http://www.tentonhammer.com/image/view/77005/preview">
FPS
FPS
alt="World of Warcraft" title="World of Warcraft"
src="http://www.tentonhammer.com/image/view/77011/preview">
FPS
FPS
style="font-weight: bold;">Results
It's possible that the results are even more obvious if we start
measuring the fractions of milliseconds, but when you're talking about
those
kinds of speeds, it's doubtful they will even be noticed. So any result
that was a fraction of a millisecond received the appropriate
"<1ms".
Even so, the results are pretty clear that the Killer 2100 NIC does do
exactly what it sets out to do. It's faster than the onboard card, and
it takes enough load off the processor that we even see a framerate
increase in all games. This is the biggest advantage to the card for
most MMO gamers as millisecond response time difference that we see in
the ping tests aren't usually a concern if you're playing an RPG. In
shooters though, it becomes much more important and noticeable.
style="font-weight: bold;">Software
We would be amiss if we didn't mention the network card's software
suite. The power and versatility of the included software is one of the
card's most valuable assets. With the ability to throttle your network
activity on any given application, and view every application that is
using bandwidth at a quick glance, the advantage of the Killer 2100
becomes clear.
As gamers we often are patching our games, downloading music, and
streaming video out to the television in the entertainment room. All of
this network activity takes resources and no other card that we've seen
is able to manage those resources as well as the Killer 2100.
The ease of use of the suite is what makes it favorable. The interface
is intuitive and streamlined, so it shouldn't take even the very
unsavvy long to figure out how to get the card to work to their
advantage.
style="font-weight: bold;">Conclusion
Bigfoot Networks' Killer 2100 Gaming Card provides the
fastest, most responsive, most controllable card we've seen in a
desktop computer. Its performance is simply unmatched. However, at a
price tag of $130 it's hard to recommend it to anyone but the
most hardcore of gamers. The performance increase in framerates is
respectable, but in most cases tested, it was only an increase of a
couple of frames. The actual network response times were the same.
While there was a definite improvement, those numbers probably won't be
noticed by anyone but the most extreme performance-seekers.
On the other hand, if you are the type of gamer who just needs to
optimize their gaming experience in any way possible the Killer NIC
offers a performance boost in an area that you'll be
unlikely to improve otherwise without it.
So, in that way, this card is a must-have for elite
performance tweakers. For everyone else the most valuable aspect of the
card is in its software suite. The ability to micro-manage all your
applications and bandwidth is something that's hard to put a price tag
on as it will vary according to your needs. If you only play one game
and patch while at work, you probably won't need the card. If you're
like me, though, and are constantly flipping through multiple tasks
online and through your network, there is definite value to the Killer
2100.
