|
|
|
Noob
Group: Forum Members
Last Login: 9/20/2006 9:23 AM
Posts: 7,
Visits: 47
|
|
Hi, I'd like to find out some more details on the tests reported in the Killer NIC white paper.
1. What exactly were the command lines used for PCATTCP. Why were those options/values selected?
2. Were the same command lines used for the calls/sec and throughput graphs? If not, what were the differences and why?
3. By "UDP - Throughput in MB", do you mean "UDP - Throughput in MB/s"? If not, what exactly do you mean?
4. Does the killer NIC support jumbo frames? What sizes?
5. Were jumbo frames used in the tests? What sizes?
6. Were any non-default NIC options selected for the tests? Which ones? Why?
7. Were any non-default OS/networking options selected for the tests? Which ones? Why?
8. Which OSs were used for the tests?
9. Were any other applications putting significant load on the system(s) run during the tests?
I understand that not all of my questions / perspective will be the best ones from a point of view of gaming, esp. on-line gaming.
However, I'm sure that anyone considering buying such a NIC would also want good performance in things like high-speed internal file transfers, and that some such questions (e.g. regarding jumbo frames) could apply in those cases.
Thanks.
|
|
|
|
|
ELN Board Member, Bigfoot Networks CEO
Group: Administrators
Last Login: 9/8/2008 7:13 PM
Posts: 266,
Visits: 636
|
|
Madwand, Thanks for taking the time to ask these questions, I'm sure there are others who want to know:
I'll respond from memory now, and update the rest later.
1. What exactly were the command lines used for PCATTCP. Why were those options/values selected?
I'll get you this by eoday today.
2. Were the same command lines used for the calls/sec and throughput graphs? If not, what were the differences and why?
Yes, same commands. The tool gives both results simultaneously.
3. By "UDP - Throughput in MB", do you mean "UDP - Throughput in MB/s"? If not, what exactly do you mean?
Yes MB/s.
4. Does the killer NIC support jumbo frames? What sizes?
No. Games don't use Jumbo frames. Jumbo frames are a LAN only concept.
5. Were jumbo frames used in the tests? What sizes?
Nope. To simulate real internet performance (as opposed to LAN only)
6. Were any non-default NIC options selected for the tests? Which ones? Why?
Everything default.
7. Were any non-default OS/networking options selected for the tests? Which ones? Why?
Everything default: a fresh install of XP Pro.
8. Which OSs were used for the tests?
XP Pro.
9. Were any other applications putting significant load on the system(s) run during the tests?
None. No other applications were running in the background.
And in general: these benefits are a clear showing of the Different kind of stack we've developed (one that is more hardware interrupt driven vs. one that does a lot of cacheing and copying (ala windows). This is only possible with a hardware accelerated stack of course.
I don't believe the CPU was the bottleneck in either of these tests: it was purely stack overhead (that's why we published the Calls-per-second figure!)
Hope this helps,
Tytus
-------------------------
[ELN]Tytus - EndLagNow.ORG
Member of the Board of Directors of ELN
CEO + Mad Scientist of Bigfoot Networks, Inc.
http://www.bigfootnetworks.com
|
|
|
|
|
Noob
Group: Forum Members
Last Login: 9/20/2006 9:23 AM
Posts: 7,
Visits: 47
|
|
| Thanks for the quick initial reply dealing with most of the questions. I look forward to the answer for the first question.
|
|
|
|
|
ELN Board Member, Bigfoot Networks CEO
Group: Administrators
Last Login: 9/8/2008 7:13 PM
Posts: 266,
Visits: 636
|
|
jeepdave was involved in the testing, so I'm hoping he'll answer #1! 
Tytus
-------------------------
[ELN]Tytus - EndLagNow.ORG
Member of the Board of Directors of ELN
CEO + Mad Scientist of Bigfoot Networks, Inc.
http://www.bigfootnetworks.com
|
|
|
|
|
Noob
Group: Forum Members
Last Login: 9/20/2006 9:23 AM
Posts: 7,
Visits: 47
|
|
Tytus (8/10/2006)
jeepdave was involved in the testing, so I'm hoping he'll answer #1!
In my experience, when CEOs mention something like this, smart people in the company usually take it as more than a mild suggestion.
|
|
|
|
|
Member
Group: Forum Members
Last Login: 9/5/2006 4:37 AM
Posts: 18,
Visits: 78
|
|
Bah, Tytus is a *****cat.  I'll do one ya one better, Here are the actual screen shots from the PCATCP test I ran a couple weeks back. I ran each test 3 times, the network card tested was typed in the command line, then screen shot taken. 
|
|
|
|
|
Member
Group: Forum Members
Last Login: 9/13/2006 11:18 PM
Posts: 39,
Visits: 98
|
|
thats great lets see what ping times and tracerts look like over the net now
ok so it can sustain high transfor speeds thats nice
thats worth maybe 50 bucks
but can can it do if your gateway is weakpoint?
as you can see here its 1ms hop to my gateway and my gateway is 60ms to 70ms to google
now tell me how this card makes my gateway send packets and get packets faster?
|
|
|
|
|
Member
Group: Forum Members
Last Login: 9/5/2006 4:37 AM
Posts: 18,
Visits: 78
|
|
OK...not sure where that fit's into the initial question...but your a spunky little guy, that's cool. I posted those screen shots just to share how powerful this thing really is...figured y'all would like a little insight and transparency to the testing methods / results. Maybe you should read the FAQ's that Bigfoot published recently: http://www.bigfootnetworks.com/FrequentlyAskedQuestions.aspx In particular, this one may be appropriate to your question / point: Q: | How do you expect to lower ping, given that most latency is caused by the internet? You can't control that! | A: | True, our engineers are the best in the business...but they can't break the laws of physics, nor get ISPs to fix their problems. However, latency (Ping) as measured in-game (UDP Ping) also includes the OVERHEAD of the Operating System and the Network stack, which can add many milliseconds of ping during game-play. Killer shaves off these extra milliseconds while also prioritizing gaming network packets. ICMP Ping that can be run from a DOS prompt (ICMP Ping) does not actually measure the overhead of the operating system's network stack. |
|
|
|
|
|
|
Member
Group: Forum Members
Last Login: 9/13/2006 11:18 PM
Posts: 39,
Visits: 98
|
|
guess what im trying to say is my PC more the fast enough for the games i play so
i dont see how this will help
and any one that has a pc that is gettin lower framerates i think there are things that 300 bucks can fix better then a nic
so i dont see how this fits in
yes the XP TCP/IP stack is crap Vistas is MUCH better
you talk about other things that can add to 'lag' some how i dont see a nic that uses maybe 5% less CPU doing much in a high end PC thats GPU limited to start with
and any one that thinks a 300 buck nic can fix that i got a bridge to sell them cash only please in small bills
|
|
| |
