The front end of this compiler consists of following phases:

• Lexical Analysis
• Syntax Analysis(Parsing)
• Abstract Syntax
• Semantic Analysis
• Activation Records
• Translation to Intermediate Code

The back end of this compiler consists of following phases:

• Basic Blocks and Traces
• Instruction Selection
• Liveness Analysis
• Register Allocation

The following picture in Tiger Book illustrates why there is a need to divide a compiler into two ends (IR stands for Intermediate Code):

We can see that, without IR to separate compilers into two ends, If we want to compile five different languages into four different machines, we have to write 5*4 (namely 20) compilers. But if we do use IR to divide one compiler into two parts, then we have to merely implement 5 front ends and 4 back ends which greatly reduces the whole work! (Programmers are lazy-_-)

Other benefits includes:

• Better modularity
• Better portability
• Better code reuse