The node server

The node server does a fairly simple job: It receives a chunk of code, and the data this code needs for its computations, executes the code, and returns the result.

In the node server, the focus is on the virtual machine. The virtual machine must be efficient. Even though the language provides instructions that can do fairly complex (i.e. computationally expensive) operations, like matrix multiplication, there will be cases where we need to execute a larger number of instructions to complete some simple task. There is a constant overhead in the virtual machine for executing an instruction, no matter which instruction. This overhead is mainly

• Calling the correct instruction executing code (all instructions have a corresponding C++ implementation in the virtual machine)
• Setting up pointers to register values (needed every time we change stack frames)
• Moving the instruction pointer, and various other virtual machine state maintenance

The Fibonacci number calculation program which will be introduced in Section 5.2 is an extreme example that almost does no real work but has it's execution speed limited only by the number of instructions that are executed (not the work they do).

<