Tests
Counter-Strike 1.6

Our map of choice in Counter-Strike 1.6 was de_dust2. We chose this map because everyone and every server has this map, and because it’s really the only map shoesy is any good on. The spawn we stood in for forever for the ping test was the defensive spawn, chosen for no real good reason at all. Here are the results of our Counter-Stirke 1.6 FPS tests:


As you can see, there was a seemingly slight performance increase with the KillerNIC installed. However, these results highlight a distinct disadvantage of using a very high-end system with an older game like Counter-Strike 1.6. We should preface that by saying that in CS 1.6, we cannot do ANYTHING to make our system’s FPS performance drop below 100. A graph of these results would result in collinear results at the 100 mark - boring. Since 100 FPS is the standard (and max) frame rate in CS 1.6, we had to enable the developer console to get a better idea of the system’s performance. The average frame rate without the KillerNIC was 763.5 FPS, as measured by Fraps benchmarking. The average frame rate with the KillerNIC was 813.6 FPS, an increase of just over 6.5%. First of all, there is no way that anybody in their right mind could tell the difference between 760 and 810 FPS. However, while the performance benefits in a game like Counter-Strike 1.6 are not really practical, they are a farily good representation of the general performance increase you can expect. This is due to the fact that CS 1.6 is extremely CPU limited. Taking away the networking load from the CPU would then be expected to increase the performance of the game, and indeed, this was the case. While 6.5% is not very much, it is still pretty significant, and certainly an increase that provides the KillerNIC with some merit.

More interesting than these results are the values we obtained in our ping tests.



Here you can see that our ping was reduced by 7.5% (.8ms) when playing in the Los Angeles server, and by 3.5% (2.2ms) when playing in the Chicago server. These are pretty significant improvements when you consider that the onboard LAN chipset on this motherboard is the Intel Pro 1000 - no slouch of a NIC itself. We really don’t know why these two numbers are as far apart from each other as they are, but repeat testing offered similar results. Unfortunately none of us here are uber networking experts that can draw up some sort of explanation for this, but we digress.