Testing:

   One method I like to use when reviewing memory or computer hardware is researching known or common issues. This way I am better prepared and know what to do in the event something comes up. I searched some forums to see what kinds of experiences are being had. The results that I came across had nothing to do with the memory, but instead user error in timing settings and incompatible motherboards. You have to be able to discern between trivial and serious information in the forums or net. It's a good sign when researching a product, turns up good comments and easily reconcilable questions.

Note: For my testing and user purposes, I find it makes more sense to find the system's best performance using four main settings: Front Side Bus (HTT), Memory Voltages, CPU Multiplier, and the HT Multiplier. The benefits of lower memory latencies far out weigh the the hours spent finding of a few extra overclocked hertz. Those extra extreme tweaked gains can also cost you the life of your components over time, due to wear and out of spec abuse.
 

Test System:
Motherboard:	FOXCONN C51XEMAA2
Processor:	AMD AM2 3500+ (2.2Ghz)
Memory:	G.Skill F2-PC6400CL4D-2GBPK
Video:	EVGA 7600GT PCI-E
Power Supply:	Thermaltake Toughpower 700 Watt

 

Installed

Dual Channel DIMMs

Test System

   After installing the memory, the first test was to check for stock timings in the bios and set them if needed. One trivial issue that occurs between motherboards and memory is the memory may not "Auto" post the specified timings in the bios upon first boot. It's always a good idea to double check this and manually adjust the timings if needed. The system booted and the bios reported timings of 5-5-5-12 due to AUTO settings. The BIOS Temperature Shutdown value was also set to keep the system from exploring temperatures over 70 Celsius. This is a good preventative measure in case a CPU fan fails or something goes wrong.
 

Shutdown Temp Set

Default in Auto

Manually Adjusted

   Since Prime95 is great for ensuring your memory is performing at one hundred percent, especially when overclocked, each setting was tested and confirmed to ensure maximum stability.
 

CPU Info

Motherboard Info

Memory Slot 3

Memory Slot 4

    The memory was overclocked using 11x, 10x multipliers and 5x, 4x HT Multipliers. The memory overclocked to 850Mhz before becoming unstable. At this point the CPU voltage was raised to 1.5 volts and the memory voltage to 2.1 volts. The system then posted up to 870Mhz before becoming unstable again. Another quick memory voltage adjustment to 2.2 volts brought the system to 900Mhz without any issues. Pushing the memory voltage past 2.3 volts achieved 906Mhz with complete stability for 12 hours of Prime 95. Anything more than this only added more heat to the system.

I was able to boot in to 922Mhz and 938Mhz, but was unable to complete and serious testing. Adjusting the timings and voltages seemed to kick up a bit more heat than the chipset heat sink could handle. What this does is show the memory's ability to be more than just a sticker.
 

800 @ 4-4-4-12 @ 2.0v

830 @ 4-4-4-12 @ 2.1v

868 @ 4-4-4-12 @ 2.2v

906 @ 4-4-4-12 @ 2.3v

800 @ 4-4-4-12 @ 2.0v

830 @ 4-4-4-12 @ 2.1v

868 @ 4-4-4-12 @ 2.2v

906 @ 4-4-4-12 @ 2.3v

   Again, these results were obtained simply using the four main BIOS options to overclock: CPU Multiplier, HT Multiplier, Memory and CPU Voltages. Interestingly, the latest revision of Nvidia NTune Control Panel brought nearly the same exact overclocking results adjusting the same four settings repeatedly.

Some 4-4-4-12 memory has been known to run 3-4-4-12, which was not the case with this G. Skill memory. However, a timing of 4-4-3-12 would operate normally and perfectly stable, but only afforded a very, very small benchmark difference if any. The memory would even overclock to 850Mhz, but couldn't achieve anything higher. At this point, it simply doesn't pay off to use these timings unless you plan to keep your system stock. Your motherboard may also prevent you from using these settings as well.


Summary:

   What do we think of this 2GB PC 6400 Kit that can be found for $199.00? There is really nothing bad to say about this memory. The kit is only rated to 800Mhz at 4-4-4-12 timings, but held up to 900Mhz with typical voltage changes. A few forums indicate others have had success a little above 900Mhz with much more in depth voltage adjustments as well as having a better chipset heat sink and CPU cooler to handle the heat.

There are PC 6400 kits that cost more but have 5-5-5-12 timings and can't overclock near 850Mhz even with higher voltages. This memory performs more like PC7200 (900mhz) memory except you gain the added bonus of lower memory latencies. Naturally, the lower the latencies are the better the system's abilities will be. The choice is pretty simple.

The G. Skill F2-6400CL4D-2GPK in this test officially passed it's rated specifications. It also passed its way to 900Mhz with its rated memory latency timings. Couple this with the cost plus the customer service and warranty provided by G.Skill, you will definitely be happy. Only those needing a greater maximum memory frequency need consider something faster. If you do not need higher frequencies, then this memory will definitely make you happy.

Conclusion:

   The G. Skill F2-6400CL4D-2GPK performs beyond its official specifications and affords a much better range of results especially on a reliable motherboard. The 6400CL4D modules performed up to 900Mhz at 4-4-4-12 timings and remained stable converting my AM2 3500 processor in to a whole new beast. At its current price around $200, coupled with G.Skill's Customer Service and Warranty, it really is an exemplary choice. Looking for a more budget oriented PC 6400 Kit that gives you something extra?  This 800Mhz memory is definitely a good recommendation when you consider its stability, performance, and value.