Page 1:Mainstream ready?
Page 2:Aeneon XTune AHX860UD20-16H-K-4G
Page 3:Buffalo FireStix FSI1600D3G-2G (FSI1600D3G-K4G)
Page 4:Corsair XMS3 TW3X4G1600C9DHXNV
Page 5:Kingston HyperX KHX12800D3K2/4G
Page 6:Mushkin HP3-12800 P/N 996629
Page 7:OCZ Reaper HPC OCZ3RPR16004GK
Page 8:Patriot Memory PVS34G1600LLK
Page 9:Wintec AMPX 3AXH1600C8WS4GK
Page 10:A-Data, Chaintech, and Super Talent
Page 11:Hardware Configuration
Page 12:Benchmark Settings
Page 13:Overclocking and Latency Results
Page 14:Benchmark Results: Games
Page 15:Benchmark Results: Applications
Page 16:Benchmark Results: Video and Synthetic
Overclocking and Latency Results
We tested each 4 GB (2 x 2 GB) dual-channel memory kit to find its highest stable speed in Memtest86+ v1.70, using a maximum 1.90 volts and CAS 9-9-9-24. One module set, Aeneon’s XTune AHX860UD20-16H-K-4G, didn’t respond to such high voltage levels and was instead tested at 1.54 volts. The results were somewhat surprising.
Buffalo FireStix uses a similar heat spreader design to Corsair, but without the internal rows of fins to cool the PCB. Buffalo’s “simpler” parts still took a clear overclocking lead, probably due to slightly higher-quality chips.
Aeneon XTune came in second place, even with the lower required test voltage. These appear to be true PC3-12800 components rather than the most overclockable of lesser-rated parts.
Kingston’s HyperX modules do without so much as a single fin on its heat spreaders, yet these edged out Corsair’s similarly-priced XMS3.
Corsair defends its parts by stating that these are optimized for Nvidia chipsets, and also claims an average overclocking capability of 1800 MHz at CAS 8 using an Asus Striker II Extreme. We didn’t use the Striker II Extreme for today’s comparison, but Corsair’s claims are worth investigating when searching for the best memory for Nvidia’s 790i Ultra SLI chipset.
We were hoping that at least some of the modules would overclock to DDR3-2000, and chose the appropriate 500 MHz FSB clock. While none of today’s competitors overclocked that far, we were able to get some interesting results at DDR3-1666 and DDR3-1333.
|DDR3-1600 at 1.90V (Max)||DDR3-1667||DDR3-1333|
|Kingston HyperX KHX12800D3G-2G||7-7-6-15||6-5-5-10|
|Buffalo FireStix FSI1600D3G-2G||7-7-6-15||6-6-5-12|
|OCZ Reaper HPC OCZ3RPR16004GK||7-7-7-15||6-6-5-12|
|Mushkin Enhanced HP3-12800 996629||7-7-7-20||6-6-5-12|
|Aeneon XTune AXH860UD20-16H||9-9-6-18||7-7-5-12|
|PDP Patriot PVS34G1600LLK||Crash||6-5-5-10|
|Wintec AMPX 3AXH1600C8WS4GK||Crash||6-6-7-13|
Two dual-channel kits achieved low 7-7-6-15 timings at DDR3-1667, and two achieved 6-5-5-10 timings at DDR3-1333, but only Kingston’s HyperX could do both. A second-place finish at both speeds sets OCZ Reaper HPC apart in value, since careful shoppers can find these around 25% cheaper than the Kingston parts.
- Mainstream ready?
- Aeneon XTune AHX860UD20-16H-K-4G
- Buffalo FireStix FSI1600D3G-2G (FSI1600D3G-K4G)
- Corsair XMS3 TW3X4G1600C9DHXNV
- Kingston HyperX KHX12800D3K2/4G
- Mushkin HP3-12800 P/N 996629
- OCZ Reaper HPC OCZ3RPR16004GK
- Patriot Memory PVS34G1600LLK
- Wintec AMPX 3AXH1600C8WS4GK
- A-Data, Chaintech, and Super Talent
- Hardware Configuration
- Benchmark Settings
- Overclocking and Latency Results
- Benchmark Results: Games
- Benchmark Results: Applications
- Benchmark Results: Video and Synthetic