User panel stuff on forum
  9 posts on 1 page  1
Client Talk
2008-05-09, 12:01
Member
569 posts

Registered:
Feb 2006
After reading article: http://www.tomsguide.com/us/killer-m1-nic,review-1083.html
i am a bit confused. If I understand it correctly, it states that "real world" round trip time will be shorter using one NIC over the other.

Looking in quakeworld, the client will display 13ms regardless what NIC i use.

Is the difference between NICs (~avg 20ms) in that test due to how those games handles frames, network packets,etc.. and that it offloads the CPU or is it due to some other thing?

(like qw client only display ping between client NIC and server NIC rather than between server application and client application.
2008-05-09, 12:51
Member
705 posts

Registered:
Feb 2006
well it offloads the cpu, but you don't want that card. get an intel high-end card instead.
and yes, response times are usually faster in 'real-world' compared to cheapo-cards
2008-05-09, 13:44
Member
129 posts

Registered:
Mar 2007
no, it's a heap of crap

that is by far the worst review of anything i have ever read.
2008-05-09, 14:12
Member
569 posts

Registered:
Feb 2006
driz wrote:
no, it's a heap of crap

that is by far the worst review of anything i have ever read.

Guess i can agree that the review isn't very good, but usually it is possible to give a good explanation of what in their test setup causes the observed difference.
2008-05-09, 14:30
Administrator
1265 posts

Registered:
Jan 2006
They're suggesting a ethernet card can be responsible for ~30ms delta?

get real.

I remember isdn card's quality DID matter, but ethernet? I don't believe it
never argue with an idiot. they'll bring you back to their level and then beat you with experience.
2008-05-10, 01:32
Moderator
1329 posts

Registered:
Apr 2006
mushi wrote:
They're suggesting a ethernet card can be responsible for ~30ms delta?

get real.

I remember isdn card's quality DID matter, but ethernet? I don't believe it

Totally agreed.

Killer-nic is a bullshit product with little real-life performance improvement. It's true that QoS:ing (=prioritizing) protocol can cause lower ping times when UDP and TCP are being used. However who would keep downloads enabled while playing an online game that requires low pingtimes?

Any network card younger than 15 years has less than one millisecond latency in a small LAN. (I consider the connection between NIC and ADSL modem a small LAN)
Servers: Troopers
2008-05-10, 07:27
Member
569 posts

Registered:
Feb 2006
The question was if the messured delta is something not included in normal ping times.
2008-05-10, 10:16
Member
401 posts

Registered:
Mar 2006
Ok heres a trace route to my favourite qw server.

Can this NIC really reduce 1st hop ping? I don't think it can.

Tracing route to quakeworld.ipgn.com.au [210.50.4.11]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 8 ms 8 ms 8 ms 10.20.20.224
3 8 ms 8 ms 8 ms nme-nxg-ibo-zeu-1-port-channel-1.tpgi.com.au [22
0.245.179.193]
4 9 ms 8 ms 8 ms nme-nxg-ibo-zeu-1-port-channel-1.tpgi.com.au [20
2.7.162.65]
5 8 ms 8 ms 8 ms nme-pow-ibo-zeu-1-pos-1-0.tpgi.com.au [202.7.162
.209]
6 8 ms 9 ms 8 ms AS9443.melbourne.pipenetworks.com [218.100.13.15
]
7 8 ms 8 ms 9 ms ge5-2.sw01.mel.idc.iprimus.net.au [210.50.0.33]

8 9 ms 9 ms 8 ms quakeworld.ipgn.com.au [210.50.4.11]

Trace complete.
2008-05-11, 11:55
Moderator
1329 posts

Registered:
Apr 2006
Running routed mode it seems. But even if you had direct connect to the internet (bridged mode if you have adsl) it still wouldn't matter. Only thing that matters is that NAT can actually increase the ping by a millisecond or two to the first hop (that is not local) but the cause is the performance of the ADSL modem. (in case of ADSL connection)

Again, the answer to "Can this NIC really reduce 1st hop ping? I don't think it can." is quite simple: there's just no way it can.

* with a disclaimer of course: if your network card has 10000000 mips processor then it's obvious that the pingtime can decrease.
Servers: Troopers
  9 posts on 1 page  1