Design a Simple 3-Address Code IR

Three-Address Code as IR

Three-address code (TAC) is the classic intermediate representation between AST and machine code. Each instruction has at most one operator and three operands: t1 = a + b. GCC lowers to GIMPLE; many textbooks use quadruples or triples. TAC flattens tree-shaped expressions into linear instruction sequences.

Instruction Forms

Binary ops: t = a op b. Copies: t = a. Jumps: goto L, if t goto L. Labels mark join points. Function calls use param and call instructions. Temporaries hold intermediate results.

cCopy
Loading…

• One operator per instruction simplifies optimization
• Temporaries map to registers or stack slots later
• Control flow uses explicit labels and branches
• Machine-independent: same TAC targets x86, ARM, LLVM

Connecting to real compilers

Production compilers embed this step inside a longer pipeline. GCC flows through cpp, cc1, assembly, and ld; Clang uses the driver, Sema, LLVM IR passes, and a target backend. LLVM bitcode, JVM bytecode, and WASM are other familiar IRs at the same layer. The exercise isolates one pass so you can test it alone before chaining it to the next stage.

Why for this exercise

You will design a simple three-address code IR with instruction and operand types. This exercise asks you to define the data structures representing TAC before you write the generator that emits them from the AST.