Test System And Benchmarks
Testing network performance is tricky enough, but we're going to be testing network gaming performance and latency in real-world situations. That means dealing with the Internet, dropping players, laggy servers, changing imagery, and all that it implies.
We will be testing two multiplayer games. Team Fortress 2 will represent multiplayer first-person shooters (FPS) and World of Warcraft will represent the massively multiplayer online role-playing game (MMORPG) category.
Bigfoot Networks suggested that we run 30-minute test runs at each setting and sample latency every minute. Because so much can change on a game server in 30 minutes, we decided to take three 10-minute alternating samples to minimize the impact of changes on the server. For example, we'd test an integrated network card for 10 minutes, then the Killer Xeno Pro for 10 minutes, then the integrated network card for 10 minutes again, until our test samples were complete.
Even though the large 30-minute sample size should deliver some solid trends in the data, we have to point out that no two benchmarks are identical since there are so many changing variables in each test.
Note that there will be two main categories of benchmarks: "No Load" with the tested game being the only application using network resources, and "Load" where two large Ubuntu ISO image file downloads will be running in Firefox during testing.While testing the Load scenario, we will set the Killer Xeno Pro's bandwidth control to limit Firefox downloads to 256 Kb/s.
Tests will be taken with the motherboard's integrated network card as well as with the Killer Xeno Pro. In the Load benchmarks, we will test a third option: using the integrated network card in conjunction with freeware bandwidth-limiting software called Firefox Throttle, which will limit bandwidth to the Firefox downloads similar to the way the Killer Xeno Pro's software accomplishes the task. We will also set Firefox Throttle to allow 256 kbps of bandwidth to the Firefox downloads.
Here is our test system:
| Killer Xeno Pro Test System | |
|---|---|
| CPU | AMD Phenom II X4 955 Black Edition (3.2 GHz, FSB-200) 6 MB Cache.Factory OC at 3.6 GHz, FSB-225, 2,022 MHz HyperTransport link |
| Motherboard | Gigabyte GA-MA790XT-UD4P, BIOS F4a, AMD790X |
| Networking | Onboard Gigabit LAN controller |
| Memory | Kingston HyperX 9905403-048.A00LF PC3-107002x 2,048 MB, 1,498 MHz, CAS 9-9-9-24 2T |
| Graphics | 2 x HIS Radeon HD 4890 in CrossFire850 MHz GPU, 2200 MHz RAM, 1 GB Per Card |
| Hard Drive | Seagate Barracuda ST31500341AS1.5 TB, 7200.1RPM, 16 MB Cache, SATA 3.0 Gb/s |
| Power | Corsair CMPSU-650TX650 W, ATX 12V 2.2 |
| Software and Drivers | |
| Operating System | Microsoft Windows Vista 64-bit 6.0.6001, SP1 |
| DirectX version | DirectX 10 |
| Graphics Drivers | Catalyst 9.5 |
| Benchmarks and Settings | |
| World of Warcraft | Version: 3.1.3, 1920x1200, Visual Effects at Maximum, Shadows at Maximum -1, Vsync off |
| Team Fortress 2 | Build: 3862, 1920x1200, video settings at maximum, 4xAA, 16xAF |
