⚙️ GearBench v1.5-beta
This update introduces improvements to reliability and latency measurement.
🔍Changes
-Offline fallback added – the benchmark can now run without internet connection. If the leaderboard cannot be reached, results are still shown locally and Best Score defaults to 0.
-Latency measurement updated for better consistency across modern CPUs.
-Added thread affinity, warmup phase, and median sampling to reduce scheduler noise and outliers.
-Latency timing now uses RDTSCP cycle timing combined with pointer-chasing methodology.
📊Result
Improved latency stability and more consistent results across different CPU architectures.
📝Developer Note
Huge thanks to the Overclock.net community for the direct testing and feedback.
I’m actively following your posts, results, and suggestions.
Your input helps shape GearBench and many of the recent improvements came directly from community testing.
We’ll keep refining the benchmark together for a more accurate and fair memory benchmark.
Thanks for being part of the process.