by John Reynolds
When Windows XP was released, SimHQ did an extensive test series of popular sims comparing the new OS with tried-and-true Windows 98 SE, Windows 2000 Professional, and the Windows 98 SE hybrid, “98lite”. A new age of simulations, new hardware and new OSes bring us back to the testing lab.
Released on January 30th of this year, Windows Vista is the latest bailiwick in Microsoft’s desktop operating systems and is perhaps the Windows release most heavily touted by the company as a gaming OS. While not directly tied to Vista’s launch, MS intensified their “Games for Windows” marketing campaign in January in an effort to create the perception that their new OS is the gaming platform for the future. Beyond marketing efforts, though, the new OS does appear more oriented toward gamers; the new Start menu, for example, includes a shortcut to Games Explorer, which shows performance ratings and parental controls for newer titles supporting the aforementioned Games for Windows branding. Vista also includes some significant changes to past Windows GUIs, such as the new Aero interface, moving the graphics driver model out of kernel mode, and, as a direct requirement of this last change, making DirectX 10 exclusive to Vista. DX10 is a major sea change to Microsoft’s API that brings a number of new features and enhancements, such as a unified shader architecture, geometry shaders, and improved instancing. And the market is already seeing limited support for DX10 in older titles like Company of Heroes and new releases such as Lost Planet, Call of Juarez, and BioShock.
Yet as with any new OS, there are usually a few bumps along the road, hiccups largely known as 3rd-party driver maturity. Despite the fact that Vista saw repeated delays, even major IHVs have been struggling this year to get their hardware drivers as polished as their XP equivalents. We feel, however, that enough time has now passed that most performance testing on SimHQ’s part wouldn’t be entirely hindered by driver immaturity. While XP vs. Vista performance has been extensively written about by all the major hardware and corporate gaming sites, SimHQ’s benchmark suite includes titles that aren’t typically used by these other outlets, so we hope our readers find some added value in this feature.
Test System Details
All testing was performed with clean installs of both Windows XP Professional (32-bit) and Windows Vista Ultimate (64-bit) on separate hard drives. Features like System Restore and Automatic Updates were disabled in both OSes, numerous services were turned off, and new Vista features such as Windows Defender, UAC, and auto tuning on the network stack were likewise disabled. The test system used the following components:
- Intel 975XBX2 motherboard
- Core 2 Extreme X6800 CPU
- 2GB of Corsair DDR2/800 memory
- XFX ‘XXX’ edition GeForce 8800 GTX graphics board
- Creative X-Fi SoundBlaster sound card
- 2 Western Digital 150GB Raptor hard drives
- Plextor PX-712S DVD drive
- Dell 3007WFP 30” LCD
The latest BIOS and chipset and add-in drivers (NVIDIA’s 162.18s for XP and 162.22s for Vista) were installed. The NVIDIA control panel was set to 4x anti-aliasing and 8x anisotropic filtering for titles that lack in-game settings for those features, and texturing filtering was set to high quality. We felt that while these settings may shift the performance bottleneck more toward the graphics board, the GeForce 8800 GTX is a high-end, $500+ component and shouldn’t be too burdened by these settings, allowing our testing to better represent how an end user would actually play if using a similar gaming rig. And along that line of thought, the titles in our benchmark suite were generally configured with higher, more demanding settings enabled, though 1600×1200 was the highest resolution used. Last, FRAPS 2.9.1 was installed to record average frame rates for applications unable to output such data. The following list of gaming titles comprised SimHQ’s benchmark suite for this article:
- Microsoft’s FSX
- IL-2: 1946
- Falcon 4: Allied Force
- Silent Hunter: Wolves of the Pacific
- GTR 2
- Ghost Recon: Advanced Warfighter 2
- Call of Duty 2
- A Company of Heroes
For synthetic testing, we used various tests from the following applications, both by Futuremark:
All of the above were patched to their latest version except for one title, as noted.
For readers wondering why SimHQ decided to test the 32-bit version of XP against Vista x64, the decision was largely based on the fact that the 64-bit version of Vista will be OS of choice for SimHQ’s future hardware testing. We expect future games to start pushing the need for more system memory above and beyond the 2GB that’s been the standard with higher end gaming systems over the last few years, and as such 32-bit operating systems will not be ideal test beds. We also hoped to further differentiate our performance comparison testing from that already done by most other online sites.
Last, SimHQ would like to thank both XFX and NVIDIA for the GeForce 8800 GTX graphics board used in our test system, and Western Digital for supplying an additional 150GB Raptor hard drive for this feature.