Lecture overview

1
Lecture overview

• In this lecture I will cover
• introduction to a simple processor and its
  machine language
• introduce the general problem of language
  processing

2
Introduction to Simpletron

• In order to illustrate what we will be talking
  about I will first introduce you to a simulation
  for a very simple computer system called
  Simpletron.
• Simpletron is the GUI front end which allows you
  to interact with the simulation of a simple
  processor which executes Simple Machine Language
  (SML). The simulation itself is called
  SimpleVirtualMachine (SVM for short).

3

• SVM effectively executes SML programs by
  simulating the operation of the simple processor
  and memory.
• Programmers view of SVM looks at the structure of
  the SVM in terms of the operations defined by the
  SML instructions

4
SimpleVirtualMachine - Programmers view
5

• Basic unit of size for information transfer and
  storage is a word
• word size is 4 hexadecimal digits
• instructions (4 digits in size) fetched from
  memory at location specified by program counter
  into instruction register
• instruction word is decoded into
• operation code (first 2 digits) and
• operand (second 2 digits) which gives location
  (memory address) of data to be used
• memory size is 0x100 words (i.e. hex 100 256)
  addressed from 00 to FF

6

• program counter is incremented to point to next
  memory location
• operation code is then executed - involving use
  of accumulator for most instructions
• SML programs always start executing with the
  instruction held at location 00 in memory i.e.
  programs are loaded into memory at consecutive
  positions starting from 00
• accumulator - holds operand data for operations
  and results of operations - but temporary store
  i.e. data overwritten if not stored into memory
  location

7

• Example instruction operation codes in SML
• 20 - loads a word from memory location specified
  in operand into accumulator
• 30 - adds word from location specified in operand
  to word in accumulator, leaving the result in
  accumulator
• 21 - stores word value in accumulator into
  location specified in operand

8

• Thus assuming 2 values represented by X and Y
  exist in locations 10 and 11 respectively and
  TOTAL value is held at location 12 then the
  following example high level language code
• TOTAL X Y
• may be translated into the following SML code in
  order for the high level language actions to be
  executed on the Simpletron processor
• 2010
• 3011
• 2112

9
SML instructions

• The numeric code is the 2 digit value that is
  actually used to represent SML instructions -
  mnemonics is just to help us talk about them

10
Consequences of choice of word size and structure

• We have to remind ourselves of the structure of
  an instruction word in SML,
• ltdigits for instruction code, address of operandgt
• Number of digits used for instruction code
  determines the maximum number of different
  instructions that can be recognised by the
  virtual machine e.g. 2 digits gives values in
  range 00 to FF (maximum possible is 162 256
  instructions).

11

• Similarly the number of digits used for address
  determines the maximum possible range of
  locations that can addressed - e.g. 2 digits only
  permits 00-FF address locations to be identified
  (256 locations), and thus 256 in the maximum size
  of memory that can be addressed. Code example

12

• size of word in digits also determines the range
  of integer values that can be used as data i.e.
  instructions only operate on words, thus number
  of digits in word determine the range of numbers
  that can be operated on e.g. 4 digits gives -FFFF
  to FFFF. Code example

13
High level languages and low level languages

• computer languages are used to define algorithms
  in a precise and structured form
• high level languages use constructs, instructions
  and data components that are similar to normal
  human thought e.g.
• TOTAL X Y

14

• machine languages use constructs, instructions
  and data components that produce operations by a
  physical computer e.g.
• 2010
• 3011
• 2112
• General problem HLL and ML are different and HLL
  program needs to be converted into computer
  operations that are specified by ML - and many
  computer operations are needed to execute a HLL
  statement
• this is the case with any real hardware computer
  except the processor and the machine language
  code are considerably more complex

15
Language processors

• A language processor is a program that accepts a
  high level language program as input and converts
  it into a sequence of operations that is or can
  be executed by the computer
• in general 3 approaches to language processing
• translation (or compilation)
• interpretation
• compilation/interpretation

16
Compiler/translator and interpreter analogies

• Imagine the computer as a group of people who
  understand English and who can carry out various
  instructions given to them in English
• English acts like their machine language
• We will imagine that the solution to various
  problems have been written in Russian which thus
  acts like the HLL

17

• A translator or compiler is like someone who
  takes a manual of instructions written in Russian
  and translates them into a manual in English. An
  English reader of the manual can then carry out
  the operations as specified in the manual.
• An interpreter is like someone who takes a manual
  of instructions written in Russian and reads one
  Russian instruction and immediately carries out
  that instruction, before proceeding to the next
  instruction in the manual

18

• Since both translators/compilers and interpreters
  are actually software that run on some machine it
  would be more accurate to think of them actually
  as manuals of instruction written in English that
  are being carried out
• the compiler/translator manual specifies how to
  translate Russian manuals into English manuals
• the interpreter manual specifies for any Russian
  instruction what action should be carried out

19
Compilers or Translators

• Compilers (translators) convert all of the HLL
  program file (called source code) into the
  sequence of machine instructions (called the
  object code) that are necessary to carry out the
  actions specified by the HLL program on the some
  specific machine (called the target machine)
• the sequence of machine language instructions are
  held in a file (called an executable file)

20

• executable file can then be loaded into memory of
  target computer and then executed
• machine instructions are executed directly by
  target machine
• languages usually compiled include C, C,
  Pascal, Cobol, Fortran

21
Interpreters

• interpreters take a single HLL statement from the
  program, one at a time
• it analyses the statement to find out what action
  is required by that statement
• the interpreter then itself carries out the
  actions required by the source code program by
  executing instructions within the interpreter
  code
• languages usually interpreted include LISP,
  Prolog, JavaScript

22
some trade-offs

• speed of execution compiled code just executes
  (fast), interpreted code is repeatedly analysed
  then executed (slow)
• code complexity and size of language processor
  compilers analyse whole program and generate
  machine instructions (large and complex),
  interpreters analyse a statement at a time and
  execute actions instructions within interpreter
  code (smaller and simpler)

23
Compiler-Interpreters

• Compiler-interpreters converts the whole of a HLL
  program file into a sequence of instructions
  (called intermediate code) held in a file. The
  file is then executed by an interpreter for the
  intermediate code.
• Intermediate code instructions are selected to be
  easy to analyse (i.e. find out what machine
  action they require) and simple to execute (fewer
  machine instructions required to execute them)

24
Java approach

• Java traditionally used compiler-interpreter
  approach (javac and java)
• intermediate code produced by javac called
  bytecode because each intermediate code operation
  is specified by a single byte of data
• the Java bytecode interpreter is the Java virtual
  machine (JVM)
• java compiler is large and complex, while the
  interpreter is small - the code executes slower
  than compiled code but faster than pure
  interpreted code - but Just-in-Time and hot-spot
  compilers help

25
Just-In-Time Compilers

• JIT Java compiler will convert Java intermediate
  code(bytecode) for a segment of code (normally
  single method) into a sequence of machine
  instructions in a temporary file in memory
• these machine instructions will then be directly
  executed by hardware
• when the method finishes executing, the method
  that is next to be executed in the byte code is
  JIT compiled
• the byte code is compiled just in time for it
  to be executed