|
Kingston HyperX DDR2-1200 Dual Channel Kit |
|
|
|
Written by Fedja Drndarski
|
|
Wednesday, 24 September 2008 |
|
Page 1 of 4
 Kingston HyperX modules are made with selected series of memory chips which are usually soldered to quality 6-layer or 8-layer PCB circuit board. Multilayer PCB reduces noise in communication with memory controller, which is very important during overclocking. In this situation, things are very simple – if signals are pure and clear in communication between memory module and chipset, than it’s possible to attain higher working speeds.
Even though JEDEC memory standard considers DDR2-400, 533, 667, 800 and 1066MHz, Kingston like some other manufacturers of overclocked modules made up some “new” DDR2-1200 standard. Still, don’t assume that this product will work at 1200MHz. DDR2-1200 tells you that this type of modules in some combination with some motherboard can work at 1200MHz and maybe even higher. Yet we managed to pull out “almost 1200MHz” on AMD platform, which is great result! Overclock was accomplished at default power range from 2.35V. For 1066 CL4 frequencies, you will need only 2.05-2.1V.
HyperX DDR2-1200 has recognisable blue aluminum heatspreaders which have already become standard bundle even for some cheaper DDR2 modules. Having in mind that warm up of DDR2 modules is relatively low, we could say that these heatspreaders have actually protecting role because they prevent user to physically damage memory chips or PCB.
Chips on these modules are signed by Japanese manufacturer – Elpida, which is one of the biggest manufacturers in that area. Elpida’s chips comparing to Micron’s D9, need lower power to gain high overclock. Latencies for HyperX 1200 at 1066MHz are really solid CAS4 5-5-15 2T, while for frequencies higher than 1200MHz it’s necessary to “unloose” a bit timings. The question is: How big are benefits of higher frequencies and “unloosen” timings? Well that depends on what are you using your computer for. Generally speaking, memory latencies are calculated with timings but also with memory frequency parameters. For example, if you have memory that works at 400MHz on CL3 3-3-6 and some other modules that works at 1066MHz on CL4 5-5-15, the second one will have lower frequency i.e. access time, even though timings are more unloosen. Reason for that is quite simple – processor needs less cycles to get access to faster memory.
Also are important applications and algorithms they are used for memory access. In the case of processing huge amount of data in some predetermined order, memory flow rate will be crucial, while on the other hand in case of data sorting problems, work with databases, solving chess problems, using of lossless data compression etc. more important will be memory access time in nanoseconds than memory rate by itself. Of course, the best thing is to have both, low access time in nanoseconds and massive flow rate.
|
