📌 引言
在剑桥A-Level计算机科学（9608）的考试中，浮点数表示法（Floating-Point Representation）是Paper 3高级理论部分的核心考点。本文基于2018年10月/11月真题（9608/32），深入解析浮点数在计算机系统中的存储方式、二进制补码（Two’s Complement）格式下的转换技巧，以及规范化（Normalised）浮点数的计算方法，帮助考生快速掌握这一高频难点。
📌 Introduction
In Cambridge A-Level Computer Science (9608), floating-point representation is a core topic in Paper 3 Advanced Theory. Based on the October/November 2018 exam (9608/32), this article dives deep into how floating-point numbers are stored in computer systems, conversion techniques under Two’s Complement format, and how to compute normalised floating-point numbers — helping students master this frequently tested topic.
🔑 核心知识点一：浮点数由尾数和阶码组成
浮点数在计算机中以 尾数（Mantissa）× 2^阶码（Exponent） 的形式存储。在9608/32真题中，浮点数采用8位尾数 + 8位阶码的配置，两者均为二进制补码形式。尾数决定数值的精度，阶码决定小数点的位置（即数值的量级）。理解这一结构是解题的第一步。
🔑 Core Concept 1: Mantissa and Exponent Structure
Floating-point numbers are stored as Mantissa × 2^Exponent. In the 9608/32 exam, the format uses 8-bit mantissa + 8-bit exponent, both in Two’s Complement. The mantissa determines precision, while the exponent determines scale (where the decimal point sits). Understanding this structure is the first step to solving any floating-point problem.
🔑 核心知识点二：二进制补码（Two’s Complement）的识别与转换
二进制补码是理解浮点数的关键。最高位（MSB）为符号位：0表示正数，1表示负数。对于正数，直接按二进制权重转换为十进制；对于负数，需要先取反再加1求绝对值，然后加负号。在真题中，尾数 00101010 最高位为0，表示正尾数；阶码 00000101 同样为正。因此该浮点数 = 正尾数 × 2^正阶码，结果为正。
🔑 Core Concept 2: Two’s Complement Recognition and Conversion
Two’s Complement is fundamental to floating-point understanding. The Most Significant Bit (MSB) is the sign bit: 0 = positive, 1 = negative. For positive numbers, convert using binary weighting. For negative numbers, flip all bits and add 1 to find the absolute value, then add the negative sign. In the exam, mantissa 00101010 has MSB=0 (positive), and exponent 00000101 is also positive. So the value = positive mantissa × 2^positive exponent, yielding a positive result.
🔑 核心知识点三：非规范化浮点数转十进制（Denary）
真题要求将非规范化浮点数 0 0 1 0 1 0 1 0 | 0 0 0 0 0 1 0 1 转为十进制。步骤：① 尾数 00101010 的二进制小数 = 0×2⁻¹ + 0×2⁻² + 1×2⁻³ + 0×2⁻⁴ + 1×2⁻⁵ + 0×2⁻⁶ + 1×2⁻⁷ + 0×2⁻⁸ = 0.125 + 0.03125 + 0.0078125 = 0.1640625；② 阶码 00000101 = 5；③ 最终结果 = 0.1640625 × 2⁵ = 0.1640625 × 32 = 5.25。要点：二进制小数点默认在尾数最高位之后。
🔑 Core Concept 3: Converting Unnormalised Floating-Point to Denary
The exam asks to convert 0 0 1 0 1 0 1 0 | 0 0 0 0 0 1 0 1 to denary. Steps: ① Mantissa 00101010 in binary fraction = 0×2⁻¹ + 0×2⁻² + 1×2⁻³ + 0×2⁻⁴ + 1×2⁻⁵ + 0×2⁻⁶ + 1×2⁻⁷ + 0×2⁻⁸ = 0.125 + 0.03125 + 0.0078125 = 0.1640625; ② Exponent 00000101 = 5; ③ Final = 0.1640625 × 2⁵ = 5.25. Key insight: the binary point is placed immediately after the MSB of the mantissa.
🔑 核心知识点四：十进制数转规范化浮点数
真题要求将 +7.5 转为规范化浮点数。步骤：① 7.5转二进制 = 111.1₂；② 规范化要求尾数以 0.1（正数）或 1.0（负数）开头，将小数点左移3位：0.1111 × 2³；③ 尾数 01111000（8位，正数补0），阶码 00000011（3的8位二进制补码）。规范化结果 = 0 1 1 1 1 0 0 0 | 0 0 0 0 0 0 1 1。注意：规范化确保用最少的尾数位获得最大的精度。
🔑 Core Concept 4: Converting Denary to Normalised Floating-Point
The exam asks to convert +7.5 to normalised floating-point. Steps: ① 7.5 in binary = 111.1₂; ② Normalisation requires mantissa starting with 0.1 (positive) or 1.0 (negative), so shift binary point left 3 places: 0.1111 × 2³; ③ Mantissa 01111000 (8-bit, pad zeros), Exponent 00000011 (3 in 8-bit Two’s Complement). Result: 0 1 1 1 1 0 0 0 | 0 0 0 0 0 0 1 1. Note: normalisation maximises precision using the fewest mantissa bits.
🔑 核心知识点五：考试策略与常见失分点
在9608/32考试中（满分75分，90分钟），浮点数题目通常占3-6分。常见失分包括：混淆补码的正负数判定、未将尾数的小数点置于正确位置、规范化后忘记补零、阶码计算方向错误。建议考生：① 先在草稿纸上写出完整的二进制展开，避免跳步；② 验证结果：逆向计算确认；③ 注意题目要求的格式（规范化/非规范化）。
🔑 Core Concept 5: Exam Strategy and Common Pitfalls
In the 9608/32 exam (75 marks, 90 minutes), floating-point questions typically account for 3-6 marks. Common mistakes: confusing positive/negative in Two’s Complement, misaligning the binary point, forgetting to pad zeros after normalisation, and getting exponent direction wrong. Tips: ① Write out full binary expansion on scratch paper — don’t skip steps; ② Verify by reverse calculation; ③ Pay attention to the required format (normalised vs unnormalised).
🎯 学习建议
浮点数表示法是A-Level计算机科学的基础但易错内容。建议每周练习2-3道真题，从简单的正数转换开始，逐步过渡到负数补码的复杂情形。可以使用在线浮点数转换器验证答案，但务必先独立完成再对照。推荐结合9608历年真题（2017-2021）系统训练。
🎯 Study Tips
Floating-point representation is foundational yet error-prone in A-Level CS. Practice 2-3 past paper questions per week, starting from simple positive conversions and progressing to complex negative Two’s Complement cases. Use online converters to verify but always attempt independently first. Systematic training with 9608 past papers (2017-2021) is highly recommended.
📧 联系方式：16621398022（同微信）
📧 Contact: 16621398022 (WeChat) for learning resources