| Kingston HyperX Genesis 16GB DDR3-1600 |  |
| Reviews - Featured Reviews: Memory | ||||||||
| Written by David Ramsey | ||||||||
| Thursday, 29 December 2011 | ||||||||
Page 4 of 6
AIDA64 Memory TestBenchmark Reviews has used AIDA64 and its predecessor Lavalys Everest for years, mainly for its processor benchmarks and CPU stress-testing features. But it also includes a "Cache and Memory Benchmark" that performs read, write, and copy bandwidth tests on a system's installed memory. 
The Kingston HyperX Genesis 2133MHz memory wins on the Read and Copy tests, but loses out very slightly to the 1600MHz memory in the Write test. If you're impressed by the 7.35% better Read score the DDR3-2133 memory (red bar) turns in relative to the DDR3-1600 memory (blue bar), bear in mind that you'll pay over three and a half times as much for the former as you will for the latter ($318 vs. $90). The generic DDR3-1333 memory trails in all three tests. SiSoft Sandra Memory TestSiSoft's Sandra Lite is a free version of SiSoftware's "Sandra" benchmarking utility. Its comprehensive memory benchmark tests report a number of items, but we're interested memory throughput tests. Unlike some other benchmarks that merely perform straight sequential reads and writes, Sandra reads and writes different areas of memory, using integer as well as floating point data, all in SMP mode. 
The HyperX 2133 memory wins decisively in these tests. Interestingly the looser-timed Kingston 1600 memory turns in fractionally better scores than the tighter Corsair 1600 memory, which retails for $60 more. Euler 3DEuler 3D is a Computational Fluid Dynamics (CFD) program that is multi-threaded and computationally very intensive. The benchmark version by Case Labs is built with the Intel FORTRAN compiler and uses 8-byte double precision floating-point math. The test case simulates Mach 0.5 airflow over a NACA 65A004 airfoil section. The benchmark score is the CFD cycle frequency, with higher scores being better. 
The Kingston 2133 "wins" again, albeit by only 4% over the much cheaper Kingston 1600 memory. Join me in the next section as I run the application benchmarks.
|
||||||||
Comments
If it cannot be OC'd then the memory is of very low quality, thus, that is how I will think of Kingston Memory from now on. Not overclockable according to Benchmark Reviews.com.
Memory overclockability was more important back in the days when raising the FSB was the only way to overclock the CPU. These days, Intel processors either have unlocked multipliers or locked-down BCLKs (like Sandy Bridge) and almost all AMD processors allow multiplier overclocking. Granted, X79 does bring back a limited BCLK adjustability, but again, it's just not going to make any real difference. Granted, some people just like to see how high their memory benchmark scores can be...
I also disagree that "if it cannot be OC'd then the memory is of very low quality." Overclocking is never guaranteed and is dependent on the motherboard as well as the memory; just because I could take (for example) this memory to 1800MHz at 9-8-9-22 is no guarantee at all that you could. High quality memory runs at its specs (XMP if so equipped) reliably; there's really no other criterion that makes sense.
If you're interested in real world performance improvements (as opposed to benchmark scores), concentrate on overclocking your CPU and video card.
Note, though, that they were using Corsair Dominator memory, which will typically have more overclocking headroom than this stuff.