Test SetupHere you can see the system we tested on.
All parts are selected with high performance and good overclockability in mind to reach high test results. We used BIOS 510-FIX2 (updated version of Mushkin's recommended 510-2) for the DFI LanParty nForce4 SLI-D.
|CPU:||AMD Athlon64 4000+ (S939; 1024KB; SanDiego)|
|Motherboard:||DFI Lanparty nForce4 SLI-D|
|Memory:||2x 512 MB Mushkin Redline XP4000|
|Video Card:||2x Pine XFX 6800 GT SLI @ 450/1200|
|Harddisk:||2x 74GB WD Raptor RAID 0|
|Power Supply:||OCZ ADJ 520W|
|Software:||Windows XP SP2, Forceware 77.60|
All AlphaTimings in the Bios were set to auto, except for:
- CPC - Enabled
- DRAM bank Interleaving - Enabled
- Idle Cycle Limit - 256
- Dynamic Counter - Enabled
- R/W Queue Bypass - 16x
- Bypass Max - 07x.
- TRAS - 5
- TRC - 7
- TRFC - 14
Very good results, but a motherboard with enough VDIMM is required.
|Mushkin Redline XP4000|
|CPU Clock & |
|11 x 200 1:1||200 MHz||2-2-2-5 2.9V||5390 MB/s||2496 MB/s||47.1 ns||365.2 FPS||28129||37.8s|
|11 x 250 1:1||250 MHz||2-2-2-5 3.1V||6716 MB/s||3113 MB/s||38.5 ns||453.1 FPS||34126||30.0s|
|11 x 251 1:1||251 MHz||2-2-2-5 3.1V||6719 MB/s||3121 MB/s||38.4 ns||455.4 FPS||34180||29.9s|
|11 x 257 1:1||257 MHz||2-2-2-5 3.2V||6902 MB/s||3206 MB/s||36.9 ns||464.1 FPS||35438||28.9s|
|11 x 262 1:1||262 MHz||2-2-2-5 3.4V||7028 MB/s||3271 MB/s||36.1 ns||471.2 FPS||35809||28.4s|
|11 x 265 1:1||265 MHz||2-2-2-5 3.5V||7101 MB/s||3305 MB/s||35.7 ns||477.8 FPS||36512||28.2s|
|JEDEC DDR400A||200 MHz||2,5-3-3-8 2.6V||5130 MB/s||1710 MB/s||54.3 ns||354.9 FPS||27615||39.2s|
Reduced latencies did not give much overclocking headroom. This shows, that this memory is designed for low latencies. Faster than DDR500 with tight timings is very fast, compared to 2.5-3-3-7.
In the factory, the SPD Timings were not set to the correct values of 2-2-2-5 for DDR500. They were set to 2-3-2-6 for DDR400. Those are the best possible timings for DDR400 using 2.6V. So it's not completely wrong.
265 MHz @ 2-2-2-5 (3.5V)
For easier comparison with other modules, we set a maximum voltage of 3.1V und tested until we found the highest clock frequency and fastest timings for this memory. The Benchmarks Everest Read, Everest Write and Quake 3 were run. We then calculated the performance increase in percent compared to some standard DDR400 memory running at JEDEC DDR400A (2.5-3-3-8). The average percentage of the three benchmarks is listed in following table: