This release introduces several improvements across the OBUX benchmark suite, focused on scoring accuracy, application consistency, and usability.
Key updates include the introduction of Formula v5, replacing the previous v4 comparison model while maintaining backward compatibility for score comparisons. The updated formula reduces the weight of ultra-fast Category A transactions based on MIT blink-of-an-eye research, improves score calculation accuracy, and introduces the new Optimal rating threshold for high-performing environments.
System score calculations were further refined by excluding fixed-duration video playback activities from performance calculations. Since the playback duration is static and does not reflect actual system responsiveness, removing it prevents bias in mean, MAD, and total score calculations while improving overall benchmark accuracy.
OBUX PixelGen received multiple improvements, including full-screen maximized startup behavior, dynamic resolution-aware pixel placement, and updated window handling to ensure consistent behavior across all screen sizes and resolutions.
Application startup behavior was standardized across all OBUX applications by setting fixed startup positions and predefined window sizes using manual positioning instead of centered screen placement. This creates more consistent and repeatable benchmark execution scenarios.
Additional fixes and improvements include:
- Corrected assembly information handling
- Updated PixelGen assembly metadata
- Renamed the score label from Excellent to Optimal for consistency with Formula v5 terminology
- Updated transaction naming from "S: Video plays 30s" to "S: Video plays 90s" to better reflect actual playback duration
These changes collectively improve benchmark consistency, scoring reliability, and cross-environment comparability within the OBUX framework.