System setup and notes
| System name | Chillblast Fusion Gemini | PC Specialist Apollo Q260GTX | HEXUS E6700 GTS |
|---|---|---|---|
| Processor | Intel Core 2 Quad Q9550 @ 3.485GHz, 12MB L2 cache, 1,600MHz, quad-core | Intel Core 2 Quad Q9450 (2.67GHz, 12MB L2 cache, 1,333MHz, quad-core) | Intel Core 2 Duo E6700 (2.67GHz, 4MB L2 cache, 1,066MHz, dual-core) |
| Motherboard | ASUS Rampage Formula (X48) | ASUS P5N-D (nForce 750i SLI) | Intel Bad Axe 2 (i975X) |
| Memory | 4GB (2x 2GB) Corsair XMS2-8500 | 4GB (2x 2GB) Corsair XMS2-6400 | |
| Memory timings and speed | 5-5-5-15-2T @ 1,092MHz | 5-5-5-15-2T @ 800MHz | |
| Graphics card(s) | ATI Radeon HD 4870 X2 2,048MB | ASUS GeForce GTX 260 896MB | NVIDIA GeForce 8800 GTS 512MB |
| Disk drive(s) | Samsung 1TB Spinpoint F1, 7,200rpm, 32MB cache | Western Digital 750GB Caviar SE, 7,200rpm 16MB cache | 2x Seagate 500GB Barracuda 7200.10, 7,200rpm, 8MB cache |
| Optical drive(s) | LG H20L combo drive | LG H20L combo drive | Pioneer 110D DVD ReWriter |
| Graphics driver | ForceWare 177.41 | ||
| Operating System | Windows Vista Home Premium 64-bit, SP1 | Windows Vista Home Premium 64-bit, SP1 | Windows XP Pro, SP3 |
| Base unit price | £1,289 | ~£1,050 | ~£700 |
| Benchmarks | HEXUS.boot - cold-boot time from power-on to Gadwin PrintScreen pop-up HEXUS.power - idle and 3D load measurements from the mains HEXUS.transcode - 200MB, 1080p clip to iPhone (480x272, medium quality) HEXUS.optical - DVD Decrypter of Batman Begins DVD HEXUS.photofix - 152MB photos auto-fixed - 25 images HEXUS.squeeze - 205MB photos compressed into one file using WinRAR HEXUS.pro - CINEBENCH R10 multi-CPU render HEXUS.usb - 4GB (180 files) transfer from USB drive to primary drive HEXUS.gaming - Enemy Territory: Quake Wars - 1,920x1,200 4xAA 16xAF, VHQ |
|---|
Notes
We encountered no stability-related problems during a weekend of testing, and the errant on-board CPU sensor reckoned that the processor was running at just 44C under load.
Further, we've completely re-jigged our testing suite to reflect task-based tests rather than isolated benchmarks such as ScienceMark 2.0 and Pov-ray.
The benchmarks, we hope, reflect real-world usage of what a consumer might want to do with their shiny, new machine.
For example, we've hand-timed the boot sequence, from switch-on to usable Windows, because system integrators may forget to optimise the BIOS.
Another example is auto-fixing photographs in the free-to-use Adobe Photoshop Starter Edition 3.2 - correcting, say, 25 photographs is something most users would look towards doing.
Stacking on the hurt, a 1080p clip is compressed to a 480x272 resolution using a codec, H.264, suitable for the Apple iPhone. We reckon it's another activity that's quite plausible for many folk.
File compression and USB transfer rates are evaluated with real-world apps rather than synthetic HD Tach, and we've tipped our hat towards the professional community by keeping CINEBENCH.
Long story short, a system is more than the sum of its parts and benchmarks should reflect real-world usage.
Reckon we're off our collective rockers or is this a step in the right direction? Feel free to commentate in the HEXUS.community.
The control (comparison) machines are a PC Specialist rig that we reviewed recently and a cheaper HEXUS machine that's seen a few days. The specification of the latter is indicative of a mid-range PC from early last year, so it will be interesting to see how it compares against something a little newer and vastly more powerful.
All benchmarks are run three times and the scores averaged. If the benchmarks show a larger-than three per cent difference between slowest and fastest runs we keep running until the standard deviation is reduced.
