Instruction Set II › ~comp212 › lec2008 › lec-07... · Comp 212 Computer Org & Arch 1 Z. Li,...

Comp 212 Computer Org & Arch 1 Z. Li, 2008

COMP 212 Computer Organization & Architecture

COMP 212 Fall 2008

Lecture 7

Instruction Set II


• Quiz

– Fill in your student number only, do NOT write down your name

– Open book, but NO calculator, NO discussions,

– Relax and have fun


Instruction Set Overview


Instruction Set

• What is an Instruction Set ?

– Complete collections of all instructions that can be understood by

certain family of CPU

» E.g. X86 instruction set, MIPS instruction set, ARM instruction set,

PowerPC instruction set

– It is in binary form called Machine Code

– For programmer, we have Assembly Code representation.


Instruction Elements• Recap of Instruction Cycles

– CPU can not directly operates on memory

– Data/Instructions have to be loaded into Registers

• Elements of an Instruction

– Operation Code:

» Do what ? ADD, LOAD, MOVE..etc.

– Source Operand reference:

» On what ? Register/Mem location, …etc.

– Result/Target Operand Reference:

» To where ? Registers/mem locations…etc.

– Next Instruction Reference

» What is the next PC ?Comp 212 Computer Org & Arch 6 Z. Li, 2008

Machine code format

• Instruction word has unique bit patterns, e.g. 16 bit

instruction can be partitioned into:

– 4-bit Opcode: what to do

– 6-bit Operand ref 1:

– 6-bit Operand ref 2:

– E.g.: ADD R1, R2 = 1001 000001 000010

• Assembly Code

– Give it a human readable form


Where are the operands ?• Operands can be in:

– Registers inside CPU:

» eg. ADD R1, R2, add values in R1 and R2, store in R1

– Main memory

» eg. LOAD R1, [R2]: load memory location stored in R2 into R1.

– IO Device (memory mapped)

• Addressing mode: how to specify the location ?

– Immediate (within the instruction)

– Register

– Memory Location indicated by a register

– Will cover in more detail in the next lecture.


Number of Addresses in Instruction

• 4-addresses

– 2 source operands addr, 1 destination operand addr, and 1 next

instruction addr,

» e.g. ADD R1, R2, R3, R4 ; %% R1=R2+R3; PC=R4.

– But rarely implemented.

• 3-addresses

– 2 source operands, 1 destination operands

» E.g. ADD R1, R2, R3 ; %% R1=R2+R3; PC=PC+1

– No good, requires long instruction word to implement.


Number of Addresses in Instruction

• 2-addresses

– 1 source and destination operand, 1 source:

» e.g. ADD R1, R2; %% R1=R1+R2; PC=PC+1;

» E.g. MOVE R2, R3; %%% R2 = R3;

– Common in modern instruction set implementations.

• 1-addresses

– Have an implicit operands, e.g. AC register

» E.g. ADD R1 ; %% AC=AC+R1; PC=PC+1

– Also quite common

• 0-address.

– Stack operation (in more details later), all addresses implicit.


How many addresses is good ?

• More addresses:

– More complex (powerful also ?) instructions

– Fewer instructions per program

– Can specify more registers

– Longer instruction takes longer to fetch/execute.

• Fewer addresses:

– Less complex instructions

– More instructions per program

– But faster instruction fetch and execution.


How many addresses is good ?

• Example– 3 vs 2 vs 1 Operands


Instruction Set Design

• Instruction Format

– Length in bits, 16, 32, or 64 bits ?

• Registers:

– Number of registers can be referenced by the instructions and their usage, e.g. AC, PC

vs general purpose registers, Rx.

• Addressing:

– Modes of reference operand (in more detail later)

• Data types:

– What to support and how they operate, e.g. int, int32, float, double.

• Operation repertoire:

– How many operations to provide, how they relate to each other.

– Limited to 2n operations for n-bit opcode.


Operand (Data) Types

• Addresses

– How to interpret addresses ?

• Numbers:

– Integer, Floating point,

– Decimal (BCD):

» need 4-bit per decimal number, i.e., 0000 = 0, 0001=1, …, 1000=9, 1001=9.

• Characters:

– To store text strings,

– e.g., ASCII code, 7-bit defines 128 characters.

• Logical Data:

– For a given n-bit data, considered to be n elements

– Support bit wise operations. .


ASCII (informational)


Instruction Types

• Typically, instructions can be grouped into the following

types:

– Data transfer and IO

– Arithmetic

– Logical

– Conversion

– System Control

• The instruction set on MIPS will be covered in more detail

for a mini project


(1) Data Transfer Instructions

• Need to specify» Source, Destination

» Amount of data to be transferred

• Pentium Data Transfer Instructions:


(2) Arithmetic

• Numerical computing:

– Addition, subtraction, division, multiplication

• Pentium Arithmetic Instructions:


(3) Logical

• Bit-wise operations on operand

• Pentium Logical Instructions:

– AND – bit wise and operation

– SHL/SHR: bit wise shift


(3) Logical

• Basic Logical Operations:

– AND, OR and XOR on P and Q:

– Shift operations:


(4) System Controls

• Basic Logical Operations:

– JMP: set PC to the value specified in JMP

– JE/Z: JMP if certain register specified equal to zero.

– CALL: save all program execution context, i.e. all registers, and jump

to a segment of program. When complete, load context and continue

the execution.

– Branch operation:

» IF R1=0 JMP to 1000h, ELSE JMP to 1200h. See HW.1Q.4


Lec07-Stack


Stack and Function Calls

• Stack is a data structure implemented in hardware that

– Supports FILO operations,

» ie. Data are PUSHed into stack in order of say, 1, 2, 3, 4, but when POP,

goes up by the order of 4, 3, 2, 1.

» What is the pop result if we have:

» PUSH 1O

» PUSH 15

» PUSH 16

» POP

» Stack would have 10<15<16, when pop, the TOP of the stack is 16, so 16 is

out.



• What is the use of Stack ?

• One important application is Procedural control

– A procedure is a collection of instruction sequence that do certain

job. .e.g compute X+Y and store in Z.

– It can be CALLed multiple times

– Stack is used to remember the right PC address for programs to

return to the right sequence.



• CALL

Push the next PC into

stack,

• RETRUN

Pop the Stack value to

PC.



• For the sequence of execution above:– 4000: CALL Proc 1 (at 4500) : PUSH 4101 to Stack– JMP to 4500 – 4600: CALL Proc 2 (at 4800) : PUSH 4601 to Stack– JMP to 4800– Proc 2 RETRUN: POP Stack: PC= 4601– At 4650 CALL Proc 2 (at4800): PUSH 4651 to Stack– JMP to 4800– Proc 2 RETURN: POP: PC=4651– Proc 1 RETURNL PoP: PC=4101


Lec-07: Addressing

Endian Issue

Addressing Modes


Little vs Big Endian

• What order do we read numbers that occupy more than

one byte

• e.g. (numbers in hex to make it easy to read)

• 12345678 can be stored in 4x8bit locations as follows


Byte Order (example)

• How bytes are stored within a word (4 byte) ?

• Consider 4 byte BCD word for “12345678”, its storage in memory:

– Address Value (1) Value(2)

– 184 12 78

– 185 34 56

– 186 56 34

– 187 78 12

• Shall we read it in Big Endian (1) or Little Endian (2) way ?


Byte Order Names

• The problem is called Endian: How to arrange bytes in a

given word structure.

• The system on the left has the least significant byte in

the lowest address, this is called big-endian

• The system on the right has the least significant byte

in the highest address, this is called little-endian


C Data Structure – Word is 8 bytes


Standard for Endian

• Pentium (80x86), VAX are little-endian

• IBM 370, Moterola 680x0 (Mac), and most RISC are

big-endian

• Internet is big-endian

– Makes writing Internet programs on PC more awkward!

– WinSock provides htoi and itoh (Host to Internet & Internet to

Host) functions to convert


Addressing Modes

• Immediate Address

• Direct (Memory) Address

• Indirect (Memory) Address

• Register (direct)

• Register Indirect

• Displacement (Indexed)

• Stack


Immediate Addressing

• Operand is part of instruction

– e.g. ADD 5: Add 5 to contents of accumulator, 5 is operand

• No memory reference to fetch data

• Fast

• But Limited range:

– E.g. 32 bit instruction, 4 bit opcode, only 28 bit operand

OperandOpcode


Direct Addressing Diagram

• Address field contains address of operand

• Effective address (EA) = address field (A)

• e.g. ADD A

– Add contents of cell A to accumulator

– Look in memory at address A for operand

• Single memory reference to access data

• No additional calculations to work out effective address

• Still limited address space, usually less than word

length in addressing space:

– Eg.. 32bit instruction, 4bit opcode, address space: 228


Indirect Addressing

• Memory cell pointed to by address field

contains the address of (pointer to) the

operand

• EA-Effective Address = (A)

– Look in A, find address (A) and look there for

operand

• e.g. ADD (A)

– Add contents of cell pointed to by contents of A

to accumulator


Indirect Addressing

• Large address space, as the true address is

stored in memory, can have potential 2n where n

= word length

• May be nested, multilevel, cascaded

– e.g. Effective Address EA = ((A))

– A->A=0000 0000:[0100 0001]:

– 0100 0001: [0101 1110] => EA=0101 1110

• Multiple memory accesses to find operand

• Hence slower

1000 0110

0000 0010

0000 0010 0110 0010

0110 0010

1000 01101000 0110


Register Addressing

• No memory access, very fast execution

• Direct Addressing:

– operand specify which register has the operand

• The operand is inside register


Register Indirect Addressing

• Indirect Addressing via Registers

• Effective Address is stored inside

register

– EA = (R)

• Operand is in memory cell pointed to

by contents of register R

• Large address space (2n)

• One fewer memory access than

indirect (mem) addressing


Displacement Addressing

• Address field hold two values:

– EA = A + (R), or ( R) +A

– A = base value

– R = register that holds displacement

– or vice versaRegister ROpcode

Instruction

Memory

OperandPointer to Operand

Registers

Address A

+


Relative Addressing

• A version of displacement addressing

• R = Program counter, PC

• EA = A + (PC)

• i.e. get operand from A cells from current location

pointed to by PC

• c.f locality of reference & cache usage


Base-Register Addressing

• A holds displacement

• R holds pointer to base address

• R may be explicit or implicit

• e.g. segment registers in 80x86


Indexed Addressing

• A = base

• R = displacement

• EA = A + R

• Good for accessing arrays

– EA = A + R

– R++


Combinations

• Postindex

• EA = (A) + (R)

• Preindex

• EA = (A+(R))

• (Draw the diagrams)

EA = (A) + ( R )

AR


Stack Addressing

• Operand is (implicitly) on top of stack

• e.g.

– ADD

» Pop 2 operands and add them together


Lec 07 Instruction Set Examples

(informational)


Pentium Addressing Modes• Virtual or effective

address is offset into

segment

– Base address from

segment register

– Effective or virtual

address

– Linear hardware

address =

Base+Effective

addresses


Pentium Addressing Modes

– LA = linear address

– SR = seg register

– PC = program counter

– R = register

– S = scaling factor

– B = base register

– I = index register

– A = immediate addr in instruction


PowerPC Addressing Modes• Load/store architecture

– Indirect» Instruction includes 16 bit displacement to be added to base register (may be GP

register)» Can replace base register content with new address

– Indirect indexed» Instruction references base register and index register (both may be GP)» EA is sum of contents


PowerPC Addressing Modes• Branch address

– Absolute» Unconditional jumps, addr from 24 bit immediate within the instruction» Conidtional jumps, 16 bit immediate addr» Effective Addr = extended 24/16 bit addr

– Relative» EA = (PC) + Immediate from instruction

– Indirect» From link/count registers

• ALU operations:– Fixed point:

» Operands in registers or part of instruction– Floating point is register only


Instruction Length issues• What is a good instruction length ?

• Affected by and affects:

– Opcode length:

» more instructions if longer

– Addressing mode:

– Memory size / organization:

» Longer instruction, larger addressing space

» Wasteful, 64bit insturction vs 32 bit instruction, almost double the size of program.

– Bus structure:

» want to have integer number of transfers to fetch an instruction

– CPU

» Bottleneck if fetch is slower than execution, shorter instruction can be a solution.

• Trade off between powerful instruction repertoire and saving space


Allocation of Bits

• Number of addressing modes

– Need some explicit signaling on this

• Number of operands

– Range and flexibility tradeoff: single operand, more instructions.

• Register versus memory

– 8~32 registers are desirable.

• Address range and granularity

– Number of bits used for address limits the range


Pentium Instruction Format


Pentium Instruction – Variable lenth• Opcode (1~2 bytes)

– Operations, also direction of operation to/from mem

• ModR/m (0~1 byte)

– Whether operand is in mem/reg.

• SIB (1 byte)

– Scale – index –base: scale (2 bits), index reg (3 bits) and base reg (3 bits),

for addressing.

• Displacement (1~4 bytes)

– Number of bits used for address limits the range

• Immediate (1~4 bytes)

– Operand immediate.


Summary

• Addressing

– Endian issue, an annoyance need to take care of

– Immediate, Direct, In-direct, Base+Offset, and combinations

• Stack

– FILO data structure

– Used in program control, e.g. system calls push/pop PCs.

• Instruction Set:

– Design issues: length, number of operands, …,etc

– Types:

» Data movement

» Arithmetic and Logical

» Program Control

Instruction Set II › ~comp212 › lec2008 › lec-07... · Comp 212 Computer Org & Arch 1 Z. Li,...

Documents

Transcript of Instruction Set II › ~comp212 › lec2008 › lec-07... · Comp 212 Computer Org & Arch 1 Z. Li,...