Testing Methodology
This is our first DDR2 review, but just like the past DDR reviews we have done, we are only concentrating on the overclockability of the memory, since there is no reliable and consistent benchmark of memory available.
We tested the modules at two sets of timings, which were 3-3-3-8, and 4-4-4-10 (stock latency).
For those who are not familiar with memory timings, the first number refers to the CAS Latency, the second number to the RAS-to-CAS Delay (tRCD), the third to the Row Precharge Delay (tRP), and the final number refers to the Row Active Delay (tRAS). For more information on these timings, please consult our BIOS Optimization Guide.
It wasn't quite so easy for use to test the limits of these modules; we needed to choose the components of the testbed with some care.
We had to use the 'slower' 3.4GHz Intel Pentium 4EE ES with a lower multiplier limit, a good Intel i925X motherboard (DFI's LANParty 925X-T2), an ATI Radeon X600 XT-based PCI Express card, and a PATA hard disk to get up to a maximum FSB of 286MHz.
We tested the memory modules at two different voltages of 1.8V and 2.0V.
For stability at overclocked speeds, we tested the ram modules with Prime95's Torture Test 3.
The memory modules had to survive at least 1 hour of stress tests before we considered them to be stable. Testing was done with a CPU multiplier of 12X, to reduce the CPU bottleneck.
Testbed
Processor
|
3.4GHz Intel Pentium 4EE ES
|
Motherboard
|
DFI LANParty 925X-T2 motherboard
|
Memory
|
1GB Kingston HyperX KHX3200ULK2/1G modules
|
Graphics Card
|
ASUS AX600XT/TD
|
Hard Disk
|
Maxtor 541DX 20GB
|
Power Supply Unit
|
460W HiPro
|
Test Results
Results - CAS 3
Parameters |
Values |
Unit |
Row Active Delay (tRAS) |
8 |
Clock Cycles |
RAS-to-CAS Delay (tRCD) |
3 |
Clock Cycles |
Row Precharge Delay (tRP) |
3 |
Clock Cycles |
Stock Voltage |
1.8 |
V |
Boosted Voltage |
2.0 |
V |
1.8V |
2.0V |
|
DDR2 Speed |
Results |
Results |
533MHz |
Stable |
Stable |
539MHz |
Stable |
Stable |
544MHz |
Stable |
Stable |
549MHz |
Not Stable |
Stable |
555MHz |
 Not Stable |
Stable |
560MHz |
Cannot Boot |
Stable |
565MHz |
Cannot Boot |
Stable |
570MHz |
Cannot Boot |
Not Stable |
576MHz |
 Cannot Boot |
Not Stable |
580MHz |
 Cannot
POST |
Not Stable |
587MHz |
- |
Not Stable |
592MHz |
- |
Cannot
Boot |
At the tight latency, there isn't much room for overclocking. At stock voltage, the 544MHz was the best it would go. That's just a boost of 11MHz or 2% over stock speed.
Increasing the DDR2 voltage to 2.0V helped boost the maximum stable speed to 565MHz, 32MHz (5.9%) higher than the standard frequency of 533MHz. At 565MHz, the maximum memory bandwidth is roughly 4520MB/s.
Despite running at a tight (for DDR2) latency of 3-3-3-8, those timings are still much looser than those of DDR modules, which are capable of running at 2-2-2. The tightest timings that DDR2 support are 3-2-2. Running at 533MHz merely compensates for the higher latency.
<<< The Memory Chips, Chip Specifications : Previous Page | Next Page : Test Results (Continued) >>>