多任务环境下，进程执行速度的不确定性会导致测控任务响应时间的不可预测性．针对此问题，文中首先基于虚拟仪器系统的结构及实时工作条件，建立了系统的响应时间模型，采用微秒级计时方法，准确计算虚拟仪器实时性能计量指标；然后，通过数据处理超时公式推导，找出实时性能优化的主要途径，并得到实验证实；最后，基于实验数据分析，提出最大数据采集频率估算公式．利用该公式，可在运行效率高达80％的负载条件下，保持稳定的响应时间，实现实时性能的约束优化．

In multitasking environment, the response time is unpredictable because the execution speed of measurement and control tasks is uncertain. In order to solve this problem, first, a response time model is put forward based on the virtual instrument architecture and the real-time working conditions. Next, real-time measuring indicators are exactly computed via the timing measurement at a microsecond level. Then, the main approaches to real- time performance optimization are proposed and experimentally verified based on the derivation of data-processing delay formula. Finally, an estimation formula for the highest data acquisition frequency is presented based on the analysis of experimental results, which enables virtual instruments to maintain stable response time under the system load as high as that with 80% CPU usage, and thus realizes the constrained optimization of real-time performance.