Soldato
- Joined
- 17 Oct 2002
- Posts
- 10,321
- Location
- Stoke
Killer NIC
- Thread starter Slackworth
- Start date
I can't really help but be sceptical tbh, most lag is beyond the control of the local network & with the processing power of new PCs the network load will make no difference.
for the biggest boost in FPS the money would be better spent on more RAM or a better graphics card
EDIT:
And using a naff router / switch will be far worse for packet loss and latency
for the biggest boost in FPS the money would be better spent on more RAM or a better graphics card
EDIT:
And using a naff router / switch will be far worse for packet loss and latency
Biggest load of carp I've ever seen. First, it's a PCI card so it is limited by the PCI bus (which, if it is a gigabit LAN card) means it is useless for wire transfers anyway as it will be limited by the PCI bus - especially if you have any HDU's or similar running on the same bus. Also, the price - surely my £5 Realtek chipset card can cope just fine. Oh, I might lose that ms but really....
So methinks gimmick
.
+ Who will spend on that thing and use up the already increasingly limited number on PCI slots.
So methinks gimmick
.+ Who will spend on that thing and use up the already increasingly limited number on PCI slots.
Someone reviewed it here: http://www.gdhardware.com/hardware/networking/bigfoot/killer/review/001.htm
To summarise:
No effects on pings (as to be expected really)
Slight increase in FPS through taking network overheads away from the CPU
However, that review must be taken with a slight pinch of salt as if you note the test system CPU is way underpowered (deliberately?) compared to the other components. This might have been to artificially exaggerate the FPS difference. I suspect the network overhead would be a lot less of an issue on a dual core chip and the FPS difference a lot less as a result.
To summarise:
No effects on pings (as to be expected really)
Slight increase in FPS through taking network overheads away from the CPU
However, that review must be taken with a slight pinch of salt as if you note the test system CPU is way underpowered (deliberately?) compared to the other components. This might have been to artificially exaggerate the FPS difference. I suspect the network overhead would be a lot less of an issue on a dual core chip and the FPS difference a lot less as a result.
Last edited:
seems a bit weird to me, all other network cards which exist have to adhere to the standard IEEE 802.3, and so does everything else which uses Ethernet.. Things like lag and ping spikes are completley beyond the control of your NIC unless its totally broken. Its easy to test, do a tracert to your ISPs default gateway, the latency will be really low... then as it goes through the internet to a destination it gets higher.. no NIC can change this..
Also, take into consideration that there are far more important apps and machines which use NICs, servers, router/switch interfaces, aggregation devices which are always striving to work as fast as possible with minimum delay and latency... i've never seen anything in the real world using anything like this.
It might take some of the load off your CPU, but the entire PC still has to process the networking side of the game, as the game runs in the CPU <obviously>
I'd have to see it working side by side in proper testing to be convinced by any of it.
Also, take into consideration that there are far more important apps and machines which use NICs, servers, router/switch interfaces, aggregation devices which are always striving to work as fast as possible with minimum delay and latency... i've never seen anything in the real world using anything like this.
It might take some of the load off your CPU, but the entire PC still has to process the networking side of the game, as the game runs in the CPU <obviously>
I'd have to see it working side by side in proper testing to be convinced by any of it.
Satellite Orbit Height = 36,000,000m
Speed of Light = 299,792,458 m/s
Time = Distance / Speed
Time = 36,000,000 / 299,792,458
Time = 0.12 seconds
Time = 120 milliseconds
That's just one way from the end user to sattellite, now double it for the round trip
Time = 240 milliseconds
So on sattellite internet the minimum ping is 240ms no matter what spec your pc, and to get 25ms ping on sattellite internet would have to defy Einsteins universal speed limit, so if the reviewer would like to expalin his findings i'd be very interested in the explanation.
Speed of Light = 299,792,458 m/s
Time = Distance / Speed
Time = 36,000,000 / 299,792,458
Time = 0.12 seconds
Time = 120 milliseconds
That's just one way from the end user to sattellite, now double it for the round trip
Time = 240 milliseconds
So on sattellite internet the minimum ping is 240ms no matter what spec your pc, and to get 25ms ping on sattellite internet would have to defy Einsteins universal speed limit, so if the reviewer would like to expalin his findings i'd be very interested in the explanation.
Last edited:
V-Spec said:
Well, they do, they use NIC's which can offload the TCP/IP processing bit to their own processor, saving on CPU cycles. Been established in server based NIC's for a while now. Which is all this seems to be....
<pirate-themed nitpicking>
Yarrrr! That be Einstein's speed limit now... Yarrr!
</pirate>
V-Spec said:
Fair enough. I was thinking that in the case where you've got a game which is hammering the CPU, and you're chucking a hell of a lot of packets down the line, then you could improve latency by using a TOE card, but I imagine the amount of difference isn't going to be huge. You'd have to be sending a *lot* of data to have a noticable effect....
Associate
- Joined
- 8 Mar 2003
- Posts
- 1,297
- Location
- London
The only thing this card is likely to kill is your wallet.
