Test Setup
Test System |
---|
CPU: | AMD Athlon64 3000+ Venice |
---|
Motherboard: | DFI LanParty NF4 UT |
---|
Memory: | 2x 512 MB G.SKILL F1-4400DSU2-1GBFC |
---|
Video Card: | ATI Radeon X850 Pro PCI-E |
---|
Harddisk: | Maxtor Diamondmax 160GB |
---|
Power Supply: | HEC PurePower 475 |
---|
Software: | Windows XP SP2, Catalyst 5.10 |
---|
We will be testing this memory at 1T timing only, because that's what enthusiasts are using to get maximum performance out of their memory.
The first test we did, was test how the memory performs at a stock frequency of 200 FSB. We tested both 2-2-2-5 and 2.5-4-4-8 timings. Here you can clearly see how big the performance difference between both timings is.
After this, we tested how far we could overclock the memory at 2.6V. Even at 0.1V below the recommended voltage (GSkill recommends 2.7 - 2.9V) we saw massive overclocking at 295 MHz.
Now we increased voltage to 2.8V and 3.1V and 3.3V. More voltage does not help the overclocking much, but we did break the magic 300 MHz at 2.8V. Going beyond 3.1V did not net any performance or clock increase, so we left these tests out. Also running TCCx memory at such high voltages is not recommended.
The last test "JEDEC DDR-400A" is for comparison with a generic DDR module running at JEDEC standard timings.
GSkill F1-4400DSU2-1GBFC |
---|
CPU Clock & Memory Ratio | Memory Speed | Memory Timings | Everest Read | Everest Write | Everest Latency | Quake 3 Timedemo | 3DMark 2001SE | SuperPi Mod 1M |
---|
9 x 200 1:1 | 200 MHz | 2.5-4-4-8 2.6V | 5515 MB/s | 1845 MB/s | 53.5 ns | 247.6 fps | 19896 | 46.97 s |
---|
9 x 295 1:1 | 295 MHz | 2.5-4-4-8 2.6V | 7635 MB/s | 3212 MB/s | 37.0 ns | 367.0 fps | 27219 | 31.97 s |
---|
9 x 300 1:1 | 300 MHz | 2.5-4-4-8 2.8V | 7667 MB/s | 3269 MB/s | 35.4 ns | 371.6 fps | 27267 | 31.30 s |
---|
9 x 300 1:1 | 300 MHz | 2.5-4-4-8 3.1V | 7667 MB/s | 3269 MB/s | 35.4 ns | 371.6 fps | 27267 | 31.30 s |
---|
9 x 200 1:1 | 200 MHz | 2-2-2-2 2.6V | 5723 MB/s | 2036 MB/s | 48.5 ns | 254.6 fps | 20615 | 46.09 s |
---|
9 x 222 1:1 | 222 MHz | 2-2-2-2 2.9V | 6272 MB/s | 2174 MB/s | 44.0 ns | 281.0 fps | 22438 | 41.61 s |
---|
JEDEC DDR-400A | 200 MHz | 2.5-3-3-8 2.6V | 5642 MB/s | 2273 MB/s | 49.8 ns | 247.0 fps | 20269 | 46.58 s |
---|
Running this memory at CL2 timings does not make much sense in my opinion, since your overclocks are rather limited there. You specifically buy TCCD memory for high clocks at less aggressive timings. If you want tight timings you should look at memory which uses Winbond's BH-5 chips.
For easier comparison with other modules, we set a maximum voltage of 3.1V and 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 DDR-400 memory running at JEDEC standard timings (2.5-3-3-8). The average percentage of the three benchmarks is listed in following table: