EVGA Killer Xeno Pro: The Impact of Network Offloading
by Derek Wilson on July 3, 2009 4:20 AM EST- Posted in
- Networking
Bigfoot Networks has, for the past few years, been trying very hard to bring high powered, intelligent network interface cards to the desktop. We previously looked at their Killer NIC with some interesting results, and today we've got the Killer Xeno Pro in our labs.
The major difference between the older Killer NIC and the newer Killer Xeno Pro is the inclusion of an audio path and audio processing for voice chat acceleration. They Killer Xeno Pro also has twice the RAM of the original. Despite the improvements, one of the major benefits is that the Killer Xeno Pro will be available at a lower retail price than the Killer NIC was. Oh, and it is sort of cool to see the new hardware dialog talking about a PowerPC Processor:
I sooo want to hack this thing now.
In our original investigation, we did see some situations where the Killer NIC could make some difference, but, for what you get, the cost was much too high. One of the ways that Bigfoot is trying to combat this is by selling chipsets and letting vendors like EVGA build and market boards. They've managed to get their costs down and the price of the Killer Xeno Pro, while very high for a network card, is much more reasonable than the original offering. The EVGA Killer Xeno Pro can be had for about $120 USD.
The EVGA Killer Xeno Pro in all its glory.
Let's start by saying that this isn't going to be a network card for someone hanging on to a 7 Series NVIDIA card or a Radeon 1k part from ATI in a single core CPU system. When upgrading, spending the $120 cost of the Killer Xeno Pro on a better graphics card will net you a great deal more performance. Even putting that money into the CPU is likely to get you more for your money in general. This is a card that should be targeted at the online gamer with a good system who wants to make sure every possible advantage is covered.
This hardware at this price is just not for everyone. It still needs to come down to more of a commodity price in order to see wider adoption. In our opinion, those who should even consider this card should already have a modern dual core system with single GPU graphics hardware capable of delivering a good, steady, high framerate at the preferred resolution in the majority of games. We don't expect that everyone who has such a system will want to invest in the Killer Xeno Pro either, but let's not get too far ahead of ourselves.
Up first we will look at the Killer Xeno Pro, its features, and why we should expect some level of increased performance at all from a typical network card.
Facebook
Twitter
RSS
121 Comments
View All Comments
Etern205 - Saturday, July 4, 2009 - link
How is this misleading? Most people already have onboard nic and it's already gigabit. Why would someone go out and spend extra money when they can spend it on somewhere else?Also some board sports 2 or even 4 (Asus P5Q Premium).
As for your graphic cards, there is a major performance difference between integrated and discrete. This is why tech sites will benchmark cards either all discrete or all onboard cause putting a discrete card against a onboard will be a disavantage.
mindless1 - Saturday, July 4, 2009 - link
It is misleading because the premise is that a potential buyer is someone who might buy a nic for increased performance, so they need to know how the difference add-on NICs compare. It is the most relevant testing possible and the article is sorely lacking this MOST relevant information.wicko - Monday, July 6, 2009 - link
Read the title of the article, you might get a different picture about what's relevant.wicko - Saturday, July 4, 2009 - link
That is a horribly misleading analogy.The title of the article is "The Impact of Network Offloading". Of course they are going to test it against integrated NICs, otherwise how would they demonstrate effectiveness or the lack thereof over integrated NICs, the most common NIC there is? This is nothing like integrated video, which was made very clear by the results.
kmmatney - Friday, July 3, 2009 - link
I would have liked to see testing with a lower power cpu. A lot of people don't want to spend the money on an i7 system.crimson117 - Friday, July 3, 2009 - link
You really think those same budget-minded people would consider dropping $120 on a NIC?Qi - Friday, July 3, 2009 - link
I agree. In addition to being compared to an integrated NIC, it would have been interesting if the Killer NIC was compared to one of the add-in Intel network adapters. Most, if not all, Intel adapters have CPU offloading too, and therefore, might improve latency/fps as well. I'm especially interested in a one on one comparison between the Killer NIC and the Intel Gigabit CT. The Intel Gigabit CT is primarily intended for desktop use.DerekWilson - Friday, July 3, 2009 - link
there aren't any (to my knowledge) desktop targeted add-in NICs that do network stack offloading. there is no add in card that is comparable to the Killer in terms of what it does.while some desktop add-in network cards do more than on-board cards, they don't do much more. and with no significant difference between the baseline option and the Killer NIC, certainly no other add-in board is going to be worse ... and they aren't capable of being better.
there really was no reason to test anything else ...
And if you need a good add-in board, buy the cheapest gigabit card you can from a reliable network hardware vendor and you pretty much won't go wrong.
rudy - Monday, July 6, 2009 - link
Well good then comparing to other add in NICs should show the Killer is better, It still needs to be compared to others half the point of a site like this is to tell us about value. Maybe the killer nic is not much better then a integrated solution but maybe an add on card can beat the killer nic and is worth 30$ to some people.mindless1 - Saturday, July 4, 2009 - link
You might be overlooking that some do more or less offloading, some do it faster or slower than others, some have more or less efficient drivers, and with aging systems or add-on cards some make the mistake of being PCI rather than PCI-e.There is always a reason to test something else, even if the result is showing there is no difference it is still significant to show that, especially when other NICs cost quite a bit less. In the end the point is a consideration of alternatives for someone who would add a nic, replacing the oboard networking adapter. Such a person could choose this or some other card so the other alternatives are exactly what should be tested.