Killer Xeno Pro: Do You Really Need A Gaming Network Card?

[bellbillsnow]

That explains why your responses have not made sense for awhile and have just misrepresented the conversation. You know, you are right that I am hypocrite and likely rude. I got frustrated when instead of you addressing my points, you defaulted to saying that you were just verifying the MFGs claims. To me that showed that you were being insincere which spurned an insult. I apologize. But then you started to to mischaraterize mt statements.
You said that I "lashed out" at you for testing their claims. Really, I was "lashing out" at you for avoiding the question of UDP packets and their relevance to ping times and retreating to "just verifying the MFG claims. I think it is disingenuous to say "I just did what the mfg" says when your conclusions go beyond the mfg claims. Instead, again, you make claims about routers and hd4850.
You still cannot respond to the difference between UDP/TCP/ICMP packets in their relative function with respect to pings.

In fact, you still have not disproven the mfg's claim that ping times are not a full measure of the cards performance or established in your review that it was bigfoot's measure of performance. It just occurred to me that they only said that ping times could drop, but never say that is the only reason to believe that UDP packets are being optimized. And if ping times show a small drop, then UDP packets could show an even larger drop due to their random sequencing. So MFG claims still stands.

 

[bellbillsnow]

That explains why your responses have not made sense for awhile and have just misrepresented the conversation. You know, you are right that I am hypocrite and likely rude. I got frustrated when instead of you addressing my points, you defaulted to saying that you were just verifying the MFGs claims. To me that showed that you were being insincere which spurned an insult. I apologize. But then you started to to mischaraterize mt statements.
You said that I "lashed out" at you for testing their claims. Really, I was "lashing out" at you for avoiding the question of UDP packets and their relevance to ping times and retreating to "just verifying the MFG claims. I think it is disingenuous to say "I just did what the mfg" says when your conclusions go beyond the mfg claims. Instead, again, you make claims about routers and hd4850.
You still cannot respond to the difference between UDP/TCP/ICMP packets in their relative function with respect to pings.

In fact, you still have not disproven the mfg's claim that ping times are not a full measure of the cards performance or established in your review that it was bigfoot's measure of performance. It just occurred to me that they only said that ping times could drop, but never say that is the only reason to believe that UDP packets are being optimized. And if ping times show a small drop, then UDP packets could show an even larger drop due to their random sequencing. So MFG claims still stands.

 

[cleeve]

[citation][nom]bellbillsnow[/nom]So MFG claims still stands.[/citation]

The MFG's claims are based on the same testing method they suggested I use, though.

You can't have it both ways; they measure latency the same way I did. It's all laid out in the reviewer's guide they provided, along with the results they claim to have recorded using the same tests we performed.

Either the card has demonstrated lower latencies, or it hasn't. Which is it? Did they use a useless method and base their performance claims off of it? Or did they use a solid method which demonstrates no advantage in independant testing?

 

[bellbillsnow]

You are describing a subset of the MFG's claims. The MFG also claims in the link I provided that the PING times do not represent UDP performance, which requires game play to "feel" the difference. The only feel test that you used is WoW, which from what I can tell is TCP dominant and therefore minimally affected by the Killer.

Now you say this means there is no data to back up their claim. I will agree that at this point there is no direct method for measuring changes in UDP latency. I agree that directly measurable data is superior, but sometimes the only data available is based in human perception. Still empirical, just not measured in International Units. I work as an economic analyst and for many things there are no dollar amounts. We have to do opinion surveys.

I agree that I cannot have it both ways. My understanding, at this point, is that they used a useless method (ping times) for one of their claims (that it reduces latency, which it still does for most, but not by much). Also, I still think it would be more interesting to look at minimum ping times on two machines; if ping times to indicate something, this would be the more important stat to me. The increase in FRAPS's for CPU bound applications still stands. As someone on the EVGA site said, a system using double 295's or an older processor with a newer card could benefit from the killer due to it being CPU bound. We do know that the average system will see a minimal benefit in FRAPS, but still have no data on the potential effect on CPU bound systems.

Now as for why Bigfoot suggests those tests and invalidates them; why John Peddie/Daily Boot/Hot Hardware see improvements in ping times under no load; it's a mystery to me. My point has been that I just want to know what's really going on and don't much care what the MFG uses as a benchmark... though it was nice to establish that there is little change in ping times. It adds another data point.

I also wonder if the FRAPS Benchmarks show lower max FRAPS because the killer is constantly feeding them network data, whereas the windows stack is slowing packets down, so the game has time to generate frames without making calculations about other character movements. Just a thought.

 

[bellbillsnow]

I would add that every review that I have read shows lower ping times with the Xeno... they just complain that it is not enough for the $$$.

 

[cleeve]

Honestly, for me it comes down to value. Slight ping decrease or no ping decrease, I can't sense any difference whatsoever between the Killer Xeno and an integrated NIC when using a freeware plugin when loaded, and under no load I can't sense a difference whatsoever.

I do, however, have extensive experience with graphics cards, and I can say with absolute certainty that a Radeon 4870 (now the same price as the Xeno) will provide massive, measurable and subjective performance increases over an older card, such as a GeForce 8800 GT.

I can also say that I've experienced a massive benefit from using a gigabit router, with wireless N speeds and gigabit wired transfer.

To me, that's what it boils down to. It's simply overpriced for a nebulous benefit that's difficult to even prove the existance of.

 

[Guest]

What about the cpu load reduction? About how much cpu load IS taken off by the NIC, vs how much workload your CPU will bear with its onboard motherboard based network ports?

Especially if your computer processes are primarily used in something other than graphic rendering that could be assisted with a Nvidia / Radeon product.

 

[cyberkuberiah]

What about the cpu load reduction? About how much cpu load IS taken off by the NIC, vs how much workload your CPU will bear with its onboard motherboard based network ports?

Especially if your computer processes are primarily used in something other than graphic rendering that could be assisted with a Nvidia / Radeon product.

you'll get better performance by upgrading to a faster processor for that money .