Abstract: To address the energy efficiency and data throughput limitations of the Von Neumann architecture, computing-in-memory (CIM) systems based on spiking neural networks (SNNs) impose rigorous demands on the performance of non-volatile memory technologies. Among the various emerging memory solutions, spin-transfer torque magnetic random access memory (STT-MRAM) is regarded as a highly promising candidate for low-power CIM applications. This is attributed to its nanosecond-scale read/write speeds, high endurance, excellent data retention, low power consumption, and compatibility with complementary metal-oxide-semiconductor (CMOS) back-end-of-line processes. The core component of STT-MRAM is the magnetic tunnel junction (MTJ), which operates based on the physical principle of current-driven magnetization switching, enabling both fast and energy-efficient functionality. This review summarize recent advancements in the fast switching dynamics and performance optimization of STT-MRAM, focus on analyzing the influence law of key structural parameters of MTJ (such as free layer materials, perpendicular magnetic anisotropy energy, etc.) on its switch performance and reliability. We also introduce the advanced electrical characterization techniques in details, offering a valuable guidance and technical path for designing high-performance STT-MRAM tailored to computing-in-memory applications.