Machine Benchmark Setup
Processor
|
AMD Athlon XP 2500+ (Barton)
|
Cooler
|
Swiftech Water Cooling Kit
|
Motherboard
|
ABIT NF7-S Rev. 2.0 motherboard
|
Memory
|
512MB Corsair TwinX512-3200LL modules
|
Graphics Card
|
Powercolor Radeon 9500 128MB
|
Hard Disk
|
80GB Maxtor D740X-6L (7200 RPM)
|
Overclockability Tests
Testing was done using Prime95 ver 23.7.1 using the 'Blend' test. It took up about 400MB of the 512MB we had for system memory. This test is actually more sensitive than Memtest86 running the standard tests. The duration for each test was approximately 1-2 hours. If any setting fails the test at any time, it is considered as 'Not stable'.
Default Settings
Parameters |
Values |
Unit |
Row Active Delay (tRAS) |
6 |
Clock Cycles |
RAS-to-CAS Delay (tRCD) |
3 |
Clock Cycles |
Row Precharge Delay (tRP) |
2 |
Clock Cycles |
Selected DDR Voltage |
2.60 |
V |
True DDR Voltage |
2.67 |
V |
We first tested the Corsair PC3200LL at its default timing of 2-3-2-6. Corsair did not state the rated voltages for these modules though. But since 2.60V is the lowest DDR voltage option in the ABIT NF7-S, we used that instead. We also tested the modules with a slightly less aggressive CAS latency timing of 2.5 cycles.
Here are the results:-
CAS 2.0
|
|
CAS 2.5
|
||
Memory Clock
|
Results / Comments
|
Memory Clock
|
Results / Comments
|
|
400MHz
|
Stable
|
400MHz
|
Stable
|
|
406MHz
|
Stable
|
406MHz
|
Stable
|
|
414MHz
|
Stable
|
414MHz
|
Stable
|
|
420MHz
|
Stable
|
420MHz
|
Stable
|
|
426MHz
|
Stable
|
426MHz
|
Stable
|
|
434MHz
|
Stable
|
434MHz
|
Stable
|
|
440MHz
|
Not stable
|
440MHz
|
Stable
|
|
446MHz
|
Not stable
|
446MHz
|
Not stable
|
|
454MHz
|
Cannot boot
|
454MHz
|
Cannot boot
|
|
460MHz
|
Cannot boot
|
460MHz
|
Cannot boot
|
|
466MHz
|
Cannot POST
|
|
466MHz
|
Cannot POST
|
At default timing of 2-3-2-6, the 400MHz modules managed to hit the speed of a PC3500 memory module, at 434MHz! That is 8.5% or 34MHz higher than the stock speed of 400MHz! When the Corsair PC3200LL is running at this speed, it will have a theoretical bandwidth of 3472MB/s or 6944MB/s when used in a dual-channel system.
When I increased the latency to CAS 2.5, disappointingly, it didn't improve
the overclockability much. It only managed to push a little higher to 440MHz.
Even though that is pretty good performance for PC3200 memory modules, I expected
it to do much better at CAS 2.5. Nevertheless, 440MHz is still remarkable performance,
isn't it?
Before you ask, there are no CAS 3.0 results because Winbond BH-5 memory chips do not support CAS 3.
<<< The Memory Chip & Chip Specifications : Previous Page | Next Page : Overclockability Tests (Continued) >>>