Futuremark has rolled out a handful of updates for a couple of its benchmarks this past week, including the just-released 3DMark. That update is an important one for Windows 8 users, as makes the benchmark “fully compatible”. Other enhancements or fixes include the program remembering the last-used resolution, a fixing of the GPU count detection and a block to stop Windows from entering sleep mode while the benchmark is running.
Here’s an official list of what’s new in 3DMark 1.0.4:
Improvements
- Keyboard focus is no longer checked during the demo or when running the benchmark in a loop. Note that if the display switches to the desktop due to a popup window during the benchmark, the run will still be aborted as the display mode is no longer the same.
- 3DMark 11 now remembers the last custom resolution used.
- GPU count detection for multi-GPU systems is now correct even when SystemInfo is explicitly disabled.
- 3DMark 11 will now prevent Windows from entering Sleep during the test.
- The license key is now reset when the benchmark is reinstalled, allowing removal of invalid keys.
Compatibility
- SystemInfo updated to version 4.15 for full Windows 8 compatibility.
- Updated DirectX redistributable.
- Now using static Visual C++ 2010 runtime libraries to fix rare installation issues on some systems.
- Removed dependency on EvaDotNet.dll and EvaDotNet.DirectX.dll for GPU enumeration to fix compatibility issues with some systems.
Bug Fixes
- Fixed display initialization code that caused “SetFullscreenState failed: DXGI_ERROR_NOT_CURRENTLY_AVAILABLE” errors on some systems.
- Fixed several memory and texture leaks, improving stability during very long looping runs.
- The HUD timer no longer rolls over if a single test takes more than ten minutes. (Yes, this could happen during the Combined Test on some low end DirectX 11 cards.)
Updates for both PCMark 7 (1.4.0) and PCMark Vantage (1.2.0) have also been released, and unlike most of Futuremark’s updates, this one does affect scores. Given that, we’re unlikely to upgrade on established test benches, but instead hold out for the next major release.
The reasons for the score changes boil down to fixed issues in the video transcoding tests, a fix for a data compression bug with storage workloads and finally, enhanced accuracy in video playback measurements. Instead of copy / pasting the updates list for these benchmarks, I recommend heading on over to the official site if you’re needing to know.
You can grab the updates from the usual places, all found through the official product pages.