220 likes | 330 Views
ITEC 352. Lecture 11 ISA - CPU. Review. Questions? HW 2 due on Friday ISA Machine language Buses Memory. Outline. P1 is posted CPU. Architecture. • Contains: Data Section and Control Unit Data Section: Registers and ALU (they do the processing)
E N D
ITEC 352 Lecture 11 ISA - CPU
Review • Questions? • HW 2 due on Friday • ISA • Machine language • Buses • Memory
Outline • P1 is posted • CPU
Architecture • • Contains: Data Section and Control Unit • Data Section: Registers and ALU (they do the processing) • Control Unit: controls the processing.`
CPU • Registers • One type of memory. • Each register can usually store a word (there are some that store a byte). • They are few in number (usually range from 16 onwards to something higher). • Hence, they require fewer bits to represent their addresses. • E.g., if we had 16 registers, we would only need 4 bits to address them (contrast this with 32 bit memory). • Hence, it is very fast memory. • ALU: Arithmetic and Logic Unit: This implements a variety of binary and unary operations. Examples: ADD, MULTIPLY, AND, NOT, OR etc…
ISA definition • Every control unit is pre-programmed with a set of basic instructions. • E.g., • IA-32 (Intel 32 bit processor) instruction set: • http://siyobik.info/main/reference • ARM processors (used in DROID and other mobile devices) instruction set: • http://infocenter.arm.com/help/topic/com.arm.doc.qrc0001m/QRC0001_UAL.pdf • SUN SPARC instruction set (32 bit) (version 7) • http://www.cs.utexas.edu/users/novak/sparcv7.pdf • These basic instructions are called Instruction sets and the architecture supporting the instruction set is refereed as Instruction Set Architecture (ISA). • E.g., SUN SPARC Version 7 32 bit ISA • ARM 11 ISA • NEXT: A second in the life of a program (from program to a process).
Program to Process: The process of compilation, linking and loading (overview) Program in high level language (e.g., C/C++/Java/ADA) A second (or much less) in the life of a high-level program Libraries (e.g., stdlib) COMPILATION Assembly code Assembly code (Java is a little different) ASSEMBLY Waiting (and happy) to be executed Machine Code Machine Code LINKING LOADING Combined machine code
Compiler/OS COMPILER Program in high level language (e.g., C/C++/Java/ADA) Libraries (e.g., stdlib) COMPILATION Assembly code Assembly code (Java is a little different) ASSEMBLY Waiting (and happy) to be executed By the CPU Machine Code Machine Code LINKING LOADING Operating System Combined machine code
Portability • Java programs are compiled into “Java bytecode” • Java bytecode is analogous to an ISA. • Depending on the specific CPU (Intel or ARM etc.), • Java’s bytecode is mapped to specific ISA.
Program to Process A specific ISA can only execute the instructions it supports. • E.g., to execute a Java program: • It has to be finally translated into the specific ISA.
Compilation • Consider the following harmless statement in a high level language: A = B + 4 To compile this program in a high level language (say Java), it must first reduce the statement to various symbols in the language. Symbols in the example statement: A = B + 4 constant Identifiers Operators (delimiters) This phase in compilation is called Lexical Analysis.
Lexical Analysis (2) • How does compiler store information about the symbols in a statement ? • Answer: Symbol table. • Symbol table contains several records. Each record in the table usually has the following information: • Name of the symbol if it is an identifier (Stored as a String) or a key word like “constant” if the symbol happens to be a contant value. • E.., A, B • Type of the identifier or constant. E.g., (integer , …) • This value cannot be filled up during the lexical analysis. We will see how this value is derived later. • Scope of the identifier (if the identifier is a variable – what is its scope). Remember scope from a previous slide.
Lexical Analysis (3) • Example • For the statement: A = B + 4 The compiler builds the symbol table: Name Type Scope Name: A UNKNOWN UNKNOWN Name: B UNKNOWN UNKNOWN CONST:4 UNKNOWN UNKNOWN
Static Analysis • Lexical analysis is done in parallel with syntactic analysis. Here, the compiler checks to see if the syntax of the statement is correct. • For instance, in Java, the statement A = B + 4 is incorrect, because at the end of the statement Java expects a “semicolon”. • We will not go into details of how it check the syntax. Lets just assume it does. • As part of these analysis, a brief step is the building of a syntax tree.
Syntax tree • A syntax tree captures the syntax of a statement. The nodes of the tree are operators (or delimiters). The leaves are the symbols. = + Symbol table entry of B Symbol table entry of 4 Symbol table entry of A
Semantic Analysis • After determining the syntax is correct, the compiler does a semantic analysis. • Usually, in many programming languages this requires that the compiler make a second pass across the program. • THIS ALLOWS A PROGRAM TO USE VARIABLES OR FUNCTIONS DEFINED LATER. This is called FORWARD REFERENCING • In this phase, the compiler determines and check the types of variables and functions. E.g., In A = B + 4, it determines the type of A and B (perhaps int A and int B) and then checks to see if the statement is correct when A and B are ints). E.g., if B is a String, the above statement will be incorrect – a compilation error is thrown here. • Also in this phase the compiler adds the remaining entries of the symbol table
Semantic Analysis (2) • After the type checking phase during semantic analysis Name Type Scope Name: A int scope of A Name: B intscope of B CONST :4int
Code Generation • In this phase, the compiler generates the assembly code. • Usually, the compiler first converts the program (the syntax tree) into an intermediate form of pseudo code that closely parallels assembly. • It then maps the pseudo code into assembly. • This is because, assembly is specific to an architecture – hence this step is not portable across different machines. Converting to pseudo code however is portable. • Hence, compiler writers can simply port their compilers to different OSes by changing this one last step.
Code generation • The statement A = B + 4 is now generated into specific ISA: ld [B], %r1 add %r1, 4, %r2 st %r2, %r0, [A]
Compilation: end of our discussion • So far we looked at the process of compilation. • For more information, you should take a course on Compilers/Translators (offered sometimes in Spring by Dr. Okie) (or) read a very popular textbook in Compilers called the “Dragon Book”
Summary • Compiler: • Programs are compiled into machine code • Operating system: • executes a program (program in execution is called a process). • Execution starts when a user passes the program as an argument to a “method/function call” called exec implemented by the OS. • Next how does the control unit interact with the OS? • How does it execute the machine code that the OS wants it to execute?
Next. • Once machine code is generated, linked and loaded, it is executed by the CPU. • Next: The all important: Fetch – execute cycle. • Defines how instructions are executed. • Every commercial architecture out there (from Intel to ARM) follows this cycle.