Page 1 of 1

Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 3:11 pm
by Apoptosis
Killer 2100 Blows Away Standard NICs in Online Gaming Benchmarks

New Analyst Report Shows KillerTM 2100 Reduces Latency up to 33X and Jitter up to 85X on High-End Gaming PCs
REDWOOD CITY, CA--(Marketwire - July 8, 2010) - The newly released Killer™ 2100 gaming network card from Bigfoot Networks sharply improves network performance for online games, even on high-end PCs, according to a new analyst study from Jon Peddie Research.

The study tested network performance on enthusiast and mainstream gaming PCs. The high-end enthusiast test systems were equipped with an Intel® Core™ i7 980-X six-core CPU and an ATI Radeon™ HD 5870 graphics card, and the mainstream systems with an Intel Core i5 CPU and an ATI Radeon HD 5550 graphics card.

The results revealed the strengths of the Killer 2100's Game Networking DNA™ technology, which maintains extremely low latency even while a PC is running resource-intensive applications such as online games. On the enthusiast platform, Killer 2100 reduced latency as much as 33 times and jitter as much as 85 times -- differences that translate to clear improvements in users' online gaming experience. By contrast, standard networking interfaces delivered comparatively poor performance, even inside the enthusiast PC, operating with Microsoft Windows® 7 and premium components.

"Most surprising was how rock-solid the Killer 2100 remained when faced with the gaming-type load while the standard NIC was negatively impacted and exhibited much higher average latency results," Peddie wrote. "Additionally, Bigfoot's new user interface allows the user to control the network stack using advanced features such as Visual Bandwidth Control to make gameplay the priority for the NIC. We believe this capability will be a true benefit to real-world online gamers."

Test details:

Enthusiast test: GaNE test, Resident Evil 1920x1200 8X AA
Mainstream test: GaNE test, Resident Evil 1650x1080 2X AA

Enthusiast Gaming PC Testing:
Iteration 1: Killer 2100 delivered 33.0 times faster latency and 63.0 times better jitter performance.
Iteration 2: Killer 2100 delivered 30.3 times faster latency and 79.2 times better jitter performance.
Iteration 3: Killer 2100 delivered 27.7 times faster latency and 85.2 times better jitter performance.

Mainstream Gaming PC Testing:
Iteration 1: Killer 2100 delivered 6.2 times faster latency and 6.3 times better jitter performance.
Iteration 2: Killer 2100 delivered 6.2 times faster latency and 7.7 times better jitter performance.
Iteration 3: Killer 2100 delivered 14.9 times faster latency and 15.4 times better jitter performance.

To download the full report, visit http://jonpeddie.com/publications/bigfoot-killer-2010/

About Bigfoot Networks
Bigfoot Networks is a leading networking technology company that develops hardware and software solutions that significantly improve the online gaming experience. Its products are designed to provide maximum network performance for online gaming and other real time applications. To learn more, visit Bigfoot Networks at http://www.bigfootnetworks.com.

BIGFOOT NETWORKS, the BIGFOOT NETWORKS logo, KILLER and all other Bigfoot Networks product names are trademarks or registered trademarks of Bigfoot Networks Inc. KILLER Reg. U.S. Pat. & Tm. Off. All other trademarks and registered trademarks are the property of their respective owners.

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 3:17 pm
by Apoptosis
Yeah... just read their 'review'...

They didn't test it online and only used one game from 2009 that not too many people play online.

i'm laughing at this report. sorry

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 3:57 pm
by Sttm
I just looked through that report. So they hook up 2 pcs to another pc thru a gigabit switch, and measure performance across it. And they use Resident Evil 5; who the hell cares about multiplayer performance in RE5. I did not even know it had a competitive feature. And they get the result that the Killer 2100 gives better numbers in their Metric, which seems odd as it should just be ping in ms and packet loss percentages, but it doesnt say on charts.

I think it stands to reason they designed this test to give this result, and they did so knowing it has 0 impact on real world gaming performance; but would give the company something to point to in their marketing.

I wonder how much money it costs to get Jon Peddie Research to make up worthless tests that show I have the largest cock in the world.

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 4:41 pm
by Apoptosis
Sttm,

You are spot on... Glad you have the same exact feelings that I do and you are right they got paid to do that 'review' if you want to call it that. I'm really happy to see my thoughts are not alone.

We've asked for one of these cards for months and they won't send us one and keep telling us we are on the waiting list... I know why they aren't sending one.

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 5:05 pm
by Digital Puppy
We saw this card up close and personal last month at E3. For those that didn't see, here is the complete "live" demo of the Killer 2100 card. Looks like the set up at E3 was very similar to the setup Jon Peddie Research had.




Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 5:14 pm
by Digital Puppy
I'm also noticing the various media sites who got this card in the past few months. Interesting to see who got the Killer 2100 and who didn't. I talked directly to a couple of people at Bigfoot and they told me (multiple times) that we were going to be among the first to have the card for testing. :roll:

(I might even have some of these encounters on film...gotta dig out my raw footage...)

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 8:29 pm
by Apoptosis
A response from Killer on my forum post here:

Hi Nate,

I saw your post and I want to clarify how we test latency and why we think this matters to online gamers. We use a tool called GaNE (Gaming Network Efficiency) that we developed which is a software tool that measures latency impact on a PC under a gaming load. For anyone who is familiar with latency testing on a network, this is usually a time consuming, complex and expensive undertaking. GaNE is something we use in the development of our performance based networking products and it is a tool that we shared with Jon Peddie and other people who wish to review the product. (And yes, I welcome Legit Reviews doing their own testing to verify the results!) The advantage to GaNE is that it is relatively simple to use and it is much less expensive than a network analyzer.

GaNE uses a simple loopback test under load to record latency for each packet in a log file. It allows a tester to measure how latency is impacted when the PC is running a game. The test is run on a clean LAN to eliminate as many external variables as possible. The point of the test is to isolate and measure the performance of two different network interfaces on the same PC under the identical gaming load. The network traffic generated by the tool emulates online gaming traffic. The Resident Evil 5 Benchmark is used simply to load the PC in a way in the same way that any online PC game would load a system. The point is simply to get the CPU, memory, graphics card, etc. busy. Jon Peddie’s results are consistent with what we see in our labs – the on-board NIC latency performance degrades while the Killer 2100 latency remains consistent. Legit Reviews recorded this testing in person recently at E3 http://www.youtube.com/watch?v=dnQBNiYY ... re=related

The reason we issued an announcement about Jon Peddie’s report today is that we think these results are important to many gamers. Most people are unaware that their PC and on-board NIC may be the source of significant lag. There is a common belief that all Gigabit Ethernet interfaces are fast and that latency is usually around 1 ms. The reality is that this is untrue, especially when a PC is running an application such as a game. The Peddie results show average latency for an on-board NIC as high as 6 or 7ms with lag spikes over 20ms. Under the identical load conditions, the Killer 2100 stayed below 0.5ms average latency. For online players engaged in an MMO raid or a heated shootout in their favorite FPS, a 20ms spike over their baseline ping could ruin their day.

I’m happy to set aside some time to discuss this further if you have other questions.

Regards,

John Drewry
VP Marketing
Bigfoot Networks

Re: Killer 2100 Blows Away Standard NICs in Gaming Benchmarks

Posted: Thu Jul 08, 2010 11:23 pm
by Sttm
The Peddie results show average latency for an on-board NIC as high as 6 or 7ms with lag spikes over 20ms. Under the identical load conditions, the Killer 2100 stayed below 0.5ms average latency. For online players engaged in an MMO raid or a heated shootout in their favorite FPS, a 20ms spike over their baseline ping could ruin their day.
So the test shows that the Killer NIC has less latency over an ideal connection by on average 6-7ms, but sometimes peaking at 20ms. Or that integrated NIC's have more fluctuation in their ping by a small amount. I can believe that, and would admit that at the peak of 20ms it might have a very slight positive effect.

Here is the problem though:
"Table 2: Result: Killer 2100 delivered 6.2 times faster latency and 6.3 times better jitter performance."
I can not help but think they are going to use a quote like this one to say that it offers 6.2 times faster performance than integrated NIC in bold letters on the product; when that claim is created using a scenario no gamer is ever going to use.

I think they would be better off showing real gaming situations; but then getting results showing a 6-7ms increase every time seems very unlikely to me with how variable server connections can be.