System setup and notes
Specifications| Memory | Kingston HyperX T1 2,000 | Corsair TR3X6G2000C7GTF | Crucial Ballistix Tracer LED 6GB |
|---|---|---|---|
| Memory capacity | 6GB (3x 2GB) | 6GB (3x 2GB) | 6GB (3x 2GB) |
| Memory speed and timings | 2,000MHz, 8-8-8-25 2T | 2,000MHz, 7-8-7-20 2T | 1,430MHz, 8-8-8-24 2T |
| Rated voltage | 1.65V | 1.65V | 1.65V |
| CPU | Intel Core i7 965 EE ES (3.20GHz) | ||
| Core speed | 3.289GHz | 3.29GHz | 3.289GHz |
| Uncore speed | 4.0GHz | 4.0GHz | 3.2GHz |
| Motherboard | Gigabyte EX58-UD5 | ||
| BIOS revision | F5d (08/02/2009) | ||
| Graphics Card | Force3D Radeon HD 4870 512MB | ||
| Disk drive(s) | Samsung Spinpoint F1 1TB | ||
| Mainboard software | Intel Inf 9.1.01007 | ||
| Graphics driver | Catalyst 9.1 | ||
| PSU | Corsair HX1000W | ||
| Operating System | Windows Vista Business SP1 64-bit | ||
| Approx. price at time of writing | £311 (pre-order) | £475 (posted to UK) | £254.14 |
Tests
| 2D benchmarks | SiSoft Sandra 2009 SP2 (15.72) Win64 memory bandwidth
(float) Lavalys Everest Ultimate Edition v5.00.1650 ScienceMark 2.0 memory latency HEXUS.PiFast wPrime (1024M calculation) DivX 6.8.3 encode of 1.22GB file Far Cry 2 benchmarking loading time |
|---|
| 3D Benchmarks | Far Cry 2 v1.01, 1,680x1,050 4x AA, vhq,
ranch long demo. Enemy Territory: Quake Wars, OpenGL, 1,024x768 0xAA, 0xAF low Enemy Territory: Quake Wars, OpenGL, 1,680x1,050 4xAA, 16xAF vhq |
|---|
Setup notes
Three high-speed modules, all toting 6GB capacity, and run on an Intel Core i7 965 Extreme Edition with a 143MHz BCLK, enabling the 2GHz kits to hit, well, 2GHz, by using a 14x DRAM multiplier. The HyperX modules completed a three-hour Memtest86+ burn-in test without failure.Comparison Crucial Ballistix Tracer was run with a 10x multiplier, leading to a clock-speed of 1,430MHz, or 170MHz below its rated frequency. The need to run a 143MHz BCLK clock means that this is the closest speed we can achieve without going over 1,600MHz. The next step up, using a 12x multiplier, would lead to DRAM operating at 1,716MHz.
We're looking to see what effect, if any, having slightly higher latencies has on performance. Going by historical results it will be negligible, but let's head to the benchmark section and find out.
