X-bit labs

DDR500 Memory Roundup: Corsair, GeiL, Kingston, SimpleTech and Transcend (page 2)

Category: Memory

by Kirill "ALT-F13" Balalin

[ 04/09/2004 | 01:10 PM ]

Regrettably, our decision about carrying out tests in the dual-channel mode expelled A-Data Vitesta DDR500 memory from the list of our testing participants since we had only one module like that.

Once again, overclockability depends on the particular properties of the given memory stick. However, the dispersion is smaller with PC memory than with graphics cards and processors, so our picture of the current DDR500 market is quite true to life.

Besides the main parameter, the performance, we evaluated two subsidiary ones: the looks of the modules and their accessories shipped with the modules.

A remark for the curious: by sending a good voltage (like 3.5V and more) to modules on Winbond BH+5 chips, you can make them work in the dual-channel mode at 260-270MHz and higher frequencies and with 2-5-2-2 timings. That’s the only way to combine really low timings with high clock rates. The same experiment with DDR500+ memory produces another result: 300MHz and higher clock rates in the synchronous mode, but the timings remain “bad”. Extreme memory overclocking is somewhat off topic for this review, so we only used easy methods (like a simple modding of the Vdimm for mainboards) to increase the overclockability of the tested modules.

Testbeds

We used two testbeds for our testing session. The first of them was working in practically standard conditions. We set the Vdimm to 2.85V and the real value was fluctuating from 2.88 to 2.89V. For testing purposes (to be able to swap memory modules often and quickly), we used a short-length graphics card, which didn’t block the DIMM slot clips:

• Intel Pentium 4 3.20GHz CPU;
• ASUS P4P800 mainboard;
• GeXCube RADEON 9600 XT Extreme graphics card, 128MB;
• Western Digital 800BB HDD;
• Sirtec HighPower 420W PSU.

The second testbed was used for checking the ultimate performance of the modules. The Vdimm was volt-modded to the maximum possible 3.2V, while the real voltages were 3.25-3.31V (thanks to the PSU that offers control over the voltages). This testing demonstrated the maximum achievable frequencies at “moderately extreme” overclocking.

• Intel Pentium 4 Extreme Edition 3.20GHz CPU;
• ASUS P4C800-E Deluxe mainboard;
• Sapphire Atlantis RADEON 9800 XT 256MB graphics card;
• Western Digital 800BB HDD;
• Antec TrueControl 550 PSU;
• nVENTIV Mach II cooling system.

Other system components didn’t affect the performance values.