Mushkin HP2-5300 DDR2 1GB Kit

Test Setup

Processor: Intel P4 630 3.0ghz Prescott 2mb Cache 800mhz FSB
Motherboard: Abit AW8D (Intel 975X Chipset)
Graphics Card: XFX 6600GT PCI-E
Hard Disk: Hitachi Deskstar 80gb SATA-II 8mb Cache
Power Supply: Silverstone Strider 600w
Software: Windows XP SP2

Benchmarks

When benchmarking the HP2-5300 I wanted to be sure to cover all angles. Some people like to run their memory at low latencies (mostly gamers), some like to run their memory at high frequencies, whilst others prefer a mixture of both settings. Below is some information regarding the testing procedures and their results:

Stock Settings
I’m pleased to say that the HP-5300 had no problems running at its stock speed of DDR667, 4-4-4-10 with voltage set to 1.8v. This is a great start, as some memory (including some that my colleague is currently reviewing) can have trouble running at stock timings.

Lowest Timings at Stock Frequency
For this test I increased the voltage to a maximum of 2.1v which should be more than safe for this memory, and proceeded to reduce the timings. After many BIOS reset’s, the lowest stable timings I was able to achieve was 3-4-3-8 at DDR667. These timings are very respectable for DDR2 memory and should offer a reasonable performance increase over the stock timings.

Highest Frequency at Stock Timings
In this test I aimed to find the maximum MHZ overclock achievable while leaving the timings at stock (4-4-4-10). By increasing the voltage to 2.1v, I was able to overclock the HP2-5300 to DDR820 which is a massive increase over the memory’s stock speed of DDR667

Highest Frequency Overall
It is well known that by relaxing timings the memory speed can normally be further increased. For this test, I relaxed the memory timings to 5-5-5-15, increased the voltage to 2.1v and overclocked the memory as far as it would go while remaining stable. The maximum overclock achieved was DDR850, which was yet again a very respectable result.
What was even more interesting is that using Windows based overclocking tools, I was actually able to take this memory all the way to DDR900 and perform 1m SuperPI tests – however, it is worth noting that the system was not stable at this speed.

Everest Readeverest Write

Everest LatencySisoft Sandra

CSS FPSSuperPI 1m