We measured a ~70% improvement when taking both throughput and user CPU utilization into consideration (normalizedthroughput in the chart in Figure 1 below).
考虑到吞吐量和用户 CPU 利用率后，我们通过衡量得出性能大概提高了 70% （以下图表中的正常吞吐量）。
2
The charts in Figure 1 and Figure 2 show the throughput in calls per second, normalized to the 1-thread synchronized case, of the various implementations.