|
US Modular Cold Fusion DDR2-800 2x1GB Memory Kit
Testing - Games Our game benchmarks use the same custom timedemos that we use to benchmark video cards, but to minimize the chance of the video card affecting results, the resolution and special effects are set to absolute minimums. Half-Life 2: Episode 1 Our Half-Life 2 Episode 1 benchmark consists of Gordon emerging into the streets to find ant lions and Combine soldiers fighting. Gordon proceeds to kill a mother ant lion and lots of baby ant lions while generally blowing up anything in the way. The gravity gun is used to push a couple of cars over ant lion dens, then some more Combine soldiers are killed. The timedemo ends after Gordon enters a small room and sets off a mine. Overall, the timedemo is pretty representative of gameplay.
Half-Life 2 Episode 1
137
136
136
136
133
0
FPS
200
Prey While not as taxing as some of our DX9 titles, Prey is an important benchmark as it is one of a few OpenGL games out there that can put pressure on a system. It uses a very heavily modified Doom 3 engine with a lot of open spaces and inventive portal-centric level designs, not to mention visceral, moving surfaces and "Spirit Walking" effects. We've recorded our own timedemo that starts by flying an exo-suit through a huge series of tubes, a handful of portals, an aerial battle and finally some emergency sniping on foot.
Prey
183
180
179
179
177
0
FPS
250
Testing - Synthetic Applications SiSoft Sandra Sisoft's Sandra is a comprehensive benchmarking utility with very nice memory tools. We use it to measure system bandwidth and memory latency, both of which have direct impact on a computer's overall performance.
SiSoft Sandra Bandwidth
(Show All Graphs)
8264
8257
8256
8247
7917
0
MB/s
12000
8257
8246
8235
8204
7985
0
MB/s
12000
SiSoft Sandra Latency
19
19
20
20
21
0
ns
50
SiSoft Sandra Latency
95
95
91
88
88
0
ns
150
SuperPi SuperPi is used to relate the real-world benefits of lower latencies and greater system bandwidth. We use SuperPi Mod 1.5 for finer definition; milliseconds instead of seconds.
SuperPi
28.91
29.01
29.03
29.11
29.125
0
sec
50
SuperPi
66.78
67.03
67.06
67.25
67.34
0
sec
100
SuperPi
146.94
147.28
147.34
147.64
147.75
0
sec
200
Overclocking Pushing the limits is a part of the fun of building a computer, so it's important to check how well memory overclocks. Of course, as with all overclocking, your mileage may vary. We check three things: how high the memory overclocks with tight timings at 2.4 volts, how high the memory overclocks with loose timings at 2.4 volts, and, lastly, how high the memory overclocks at stock timings and voltages. To make sure that the processor is not affecting these results, we use a low multiplier; we underclock it.
Overclocking
(Show All Graphs)
260
234
222
218
216
0
MHz
500
212
208
208
204
0
0
MHz
500
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
