Single page Print

Ethernet throughput
Despite its gaming focus, we want to see how the Killer NIC performs as a network card, so we'll kick things off with a simple throughput test. We evaluated Ethernet performance using the NTttcp tool from Microsoft's Windows DDK. The docs say this program "provides the customer with a multi-threaded, asynchronous performance benchmark for measuring achievable data transfer rate."

We used the following command line options on the server machine:

ntttcps -m 4,0,192.168.1.25 -a
..and the same basic thing on each of our test systems acting as clients:
ntttcpr -m 4,0,192.168.1.25 -a
Our server was a Windows XP Pro system based on Asus' P5WD2 Premium motherboard with a Pentium 4 3.4GHz Extreme Edition (800MHz front-side bus, Hyper-Threading enabled) and PCI Express-attached Gigabit Ethernet. A crossover CAT6 cable was used to connect the server to each system.

The nForce and Yukon GigE controllers were tested with jumbo frames disabled. The Killer NIC doesn't actually support larger frame sizes.

That's not a good start. Regardless of whether it's running in app or game modes, the Killer NIC delivers dismal throughput in NTttcp. At least the Killer's CPU utilization is lower than that of the Yukon and nForce controllers. Interestingly enough, Marvell's GigE controller actually consumes fewer CPU cycles than Nvidia's nForce 680i SLI, despite the fact that the latter boasts a TCP/IP checksum offload engine.

Network file transfer performance
So the Killer NIC doesn't fare well in the synthetic NTttcp throughput test, but does that affect file transfer performance? To find out, we moved two sets of files between our NTttcp server and our test system using a crossover cable and standard Windows file sharing. The "small" file batch was made up of 1.8GB of high-bitrate MP3s, while the "big" batch contained 4.1GB of movie files. This should give us a better idea of how the Killer's throughput stacks up in the real world.

With these file transfers, the Killer NIC doesn't fare nearly as poorly as it did in NTttcp. Interestingly, app mode appears to be faster for larger files, while game mode seems to work better with smaller ones. Bigfoot doesn't take top honors with either file set, but it's right in the mix with the nForce and Yukon alternatives.

During our file transfer tests, the Yukon GigE chip managed the lowest CPU utilization of the lot. That puts the Killer NIC in second place, using a little less CPU time than the nForce 680i SLI. App mode appears to use ever-so-slightly fewer CPU resources than game mode here.