Due to some complexities with some of our other testbeds at the time, for this DDR2-533 review we resorted to utilizing an ASRock motherboard. Although ASRock is generally dubbed as being a budget-oriented manufacturer, the 775Dual-880Pro that we'll be using today isn't actually all bad and we've even experienced relative success overclocking this motherboard previously using stock voltages. In addition, as the motherboard is based upon the VIA PT880 Pro + VIA 8237R Chipset, we've seen respectable results partially due to its superb Linux support. For the testing today we utilized the hardware and software components listed below.
|Processor:||Intel Pentium 4 530 (3.0GHz)|
|Graphics Card:||Leadtek PX6600GT 128MB|
|Hard Drives:||Western Digital 160GB SATA|
|Optical Drives:||Sony DVD-RW|
|Power Supply:||SilverStone Strider ST405 400W|
|Operating System:||Ubuntu 5.10 (Breezy Badger) Preview|
|GCC (GNU Compiler):||4.0.2|
|Graphics Driver:||NVIDIA 1.0-7667|
For our memory testing procedure, we proceeded to follow our traditional Phoronix practices for checking the stability, overclocking, and also benchmarking. During the overclocking process, Memtest86+ v1.60 was used as our main check for the stability and operation of the RAM. As the ASRock motherboard we used for our test system today lacks any adequate voltage adjustment options for the CPU and RAM, our main abilities were focused on the timings and FSB. As far as the benchmarking goes, we continued to use our latest Linux benchmarking packages and test settings, which consist of using Doom 3, LAME Compilation, LAME Encoding, FreeBench, and RAMspeed. For the specifics on our current memory benchmarking practices, we'd recommend you check out this page. With the Transcend 2 x 512MB DDR2-533 memory modules installed, we proceeded to first run our memory benchmarks at stock speeds and timings. With the stock benchmarks completed, we next went ahead to see just how low these timings could go with Elpida ICs, and to our surprise we were able to run these modules stable, as in stable through numerous loops of Memtest86+, at 3-4-3-8. We were able to boot at 3-3-3-8 but when doing so it resulted in countless memory errors. With the timings cleared, we went ahead to see at what speed we could maximize our Front Side Bus at while both the CPU and memory remained to their stock voltages, we maxed out the stable system with a mere 224MHz. Even while loosening the timings, we still hit a barrier at 224MHz due to conflicts with the onboard SATA controller. For comparison purposes on the ASRock motherboard, we also ran the Kingmax Mars DDR2-667 KLCC28F-A8EB5 modules but for the sake of a viable Elpida comparison, we ran this memory at DDR2-533 speeds with 4-4-4-11 timings (the same speed and timings for the stock Transcend modules).
Thus the various scenarios we'll be benchmarking with the Transcend memory is:
· 200MHz FSB (3.00GHz) - DDR2-533MHz - 4-4-4-11
- 1.80V (memory)
· 200MHz FSB (3.00GHz) - DDR2-533MHz - 3-4-3-8 - 1.80V (memory)
· 224MHz FSB (3.36GHz) - DDR2-595MHz - 4-4-4-11 - 1.80V (memory)
· Kingmax Mars DDR2-667: 200MHz FSB (3.00GHz) - DDR2-533MHz- 4-4-4-11 - 1.80V (memory)