Test Setup

Test System
CPU:	Intel E6300 Conroe 1.8 GHz, 2 MB Cache
Motherboard:	ASUS P5B Deluxe BIOS 0806 Intel P965 C1
Video Card:	ATI Radeon X850 Pro PCI-E
Harddisk:	Maxtor DiamondMax 160GB
Power Supply:	OCZ GameXStream 700W
Software:	Windows XP SP2, Catalyst 6.9

Performance

In the first test we evaluate how these modules do at their default settings of 5-4-4-12, to establish a baseline performance value.
Next we tried to determine how the memory handled CL5, CL4 and CL3 timings, at lower memory frequencies, but the same CPU clock speed. If you compare the first four results you can see that changing memory clock, at constant CPU clock, has little effect on performance. The only exception seems to be memory running at 1:1 which seems to starve the CPU, so it has to wait for memory in situations where it could perform calculations at a faster memory speed.
The next three tests show how well the memory responds to different voltages at the CL5 setting. The frequencies shown are the maximum stable clock at this setting. As you can see the overclocks scale very nicely, the more juice you give the memory. The same is happening for the CL4 and CL3 scenarios.
I find it very important to point out that the memory can run 266 MHz @ CL3 @ 1.8V and 400 MHz @ CL4 @ 1.95V. This gives you added flexibility in situations where you can to reduce the clocks a bit but tighten the timings.

After looking through the benchmarks you can see that maximizing CPU clock is the most important factor when it comes to performance, then you should optimize memory speed.

Mushkin XP2-8500 4 GB Kit
CPU Clock & Memory Ratio	Memory Speed	Memory Timings	Everest Read	Everest Write	Everest Latency	Quake 3 Timedemo	3DMark 2001SE	SuperPi Mod 1M
7 x 266 1:2	533 MHz	5-4-4-12 2.3V	7048 MB/s	4809 MB/s	67.9 ns	370.7 fps	26650	28.61 s
7 x 266 1:2	533 MHz	5-5-5-15 2.3V	7015 MB/s	4806 MB/s	68.1 ns	369.6 fps	26590	28.66 s
7 x 266 2:3	400 MHz	4-4-4-12 2.0V	6389 MB/s	4807 MB/s	72.6 ns	365.9 fps	26283	28.89 s
7 x 266 1:1	266 MHz	3-2-2-4 1.8V	5257 MB/s	4804 MB/s	95.6 ns	351.6 fps	24721	30.02 s
7 x 232 1:2	464 MHz	5-5-5-15 1.85V	6140 MB/s	4168 MB/s	78.0 ns	322.1 fps	23755	32.86 s
7 x 267 1:2	534 MHz	5-5-5-15 2.15V	7020 MB/s	4809 MB/s	67.8 ns	369.5 fps	26901	29.61 s
7 x 280 1:2	560 MHz	5-5-5-15 2.35V	7358 MB/s	5061 MB/s	65.4 ns	387.2 fps	27537	27.34 s
7 x 247 2:3	370 MHz	4-4-4-12 1.85V	5920 MB/s	4439 MB/s	78.5 ns	340.4 fps	24903	31.19 s
7 x 293 2:3	440 MHz	4-4-4-12 2.15V	5966 MB/s	3972 MB/s	77.9 ns	306.9 fps	23064	34.61 s
7 x 327 2:3	490 MHz	4-4-4-12 2.35V	6645 MB/s	4416 MB/s	69.9 ns	340.4 fps	25029	31.08 s
7 x 268 1:1	268 MHz	3-2-2-4 1.85V	5291 MB/s	4796 MB/s	96.0 ns	354.9 fps	24789	29.88 s
7 x 312 1:1	312 MHz	3-2-2-4 2.15V	6065 MB/s	5616 MB/s	82.2 ns	412.7 fps	27599	25.70 s
7 x 329 1:1	329 MHz	3-2-2-4 2.35V	6403 MB/s	5927 MB/s	77.6 ns	433.1 fps	28466	24.39 s

This graph shows you at a quick glance which clock frequencies you can expect at which voltage. Please note that each individual module overclocks different, the results here can only be seen as an indicator of performance.