CHAPTER OVERVIEW Fundamentals of Computer Architecture
8. Bringing It All Together The Hardware Engineers Perspective This chapter includes:
Assigning tasks to individual processor components; Micro-instructions; Instruction sets; The format of a program; The fetch-execute cycle; Executing programs in JASPer.
1 Slides for Fundamentals of Computer Architecture Mark Burrell, 2004 2
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
Recap
So what sort of tasks does the processor have to do?
It needs to be able to read bit patterns from, and write bit patterns to, memory. We saw how it could do this in the previous chapter; It needs to keep track of where we are within a program, so it knows what to do next; It needs to know what to do for a particular instruction; It needs to be able to perform arithmetic and logical operations; It needs some general purpose storage areas.
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004 3
Introducing Micro-Instructions
Data movement microinstructions
MAR [A];
ALU micro-instructions
ALUr = [ALUx]+[ALUy]
How can we go about obtaining the logical AND of the bit patterns stored in the A and B registers?
ALUx [A] ALUy [B] ALUr = [ALUx]&[ALUy] A [ALUr]
Test micro-instructions
if(PSR(z) == 1)
Processor control micro-instructions
halt
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
Introducing The Instruction Set
Our First Program
Our program in memory compare with the previous slide
If we were to write programs using micro-instructions
they would be extraordinarily long and very difficult to check that we havent introduced any mistakes. Therefore, we dont.
Instead, we group sets of micro-instructions together to form higher level instructions, known as assembly language instructions.
AND B,A
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004 5 Slides for Fundamentals of Computer Architecture Mark Burrell, 2004 6
The Fetch-Execute Cycle
Running the program The processor loads, or fetches, the first instruction from memory (stored in memory at location $0000) into the IR; Next it runs, or executes, this first instruction - it is the CU that does this, and as already seen, the A register takes the bit pattern $05; It then fetches the second instruction from memory and stores it in the IR; Next it executes this second instruction - the B register takes the bit pattern $03; etc. The only ways that the processor would cease to fetch the next instruction are: If the power to the processor is switched off; The halt microcode is executed; The processor reset button is pressed.
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004 7
Inside The Fetch-Execute Cycle
In our simple processor, the fetch cycle is defined as the following RTL sequence:
1 MAR [PC] 2 INC [PC] 3 PC [INC] 4 MDR [M[MAR]] 5 IR [MDR] 6 CU [IR(opcode)]
The execute cycle for the given opcode:
1 ALUx [A] 2 ALUy [B] 3 ALUr = [ALUx] + [ALUy] 4 A [ALUr]
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
Sign Extension
When we perform a data movement from the IR(operand) the data value is sign extended to 16 bits
Running The Program In JASPer
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
10
Chapter Summary
Tasks for individual registers
The PC is used to bookmark which instruction the processor is to execute next; The INC is used to add 1 to the PC; The MAR and MDR are used to access memory; A and B are general purpose registers; The IR is used to store the instruction; The ALUx and ALUy are the ALU inputs; The ALUr is the ALU output, and the PSR contains flags that are updated by the ALU.
Chapter Summary
Micro-instructions
The four sets of micro-instructions understood by our simple processor are
the data movement micro-instructions, the ALU micro-instructions, the test micro-instructions and the control micro-instructions;
All micro-instructions can be represented by an RTL description; Assembly language instructions can be defined using microinstructions.
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
11
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
12
Chapter Summary
Instruction sets
The number of instructions within an instruction set is limited by the width of the opcode; Our simple processor has an opcode width of eight bits and therefore we can have a maximum of 256 instructions in an instruction set. In reality we do not need this many to write useful programs.
Chapter Summary
The format of a program
The processor executes the machine codes of a program; Additionally we add mnemonics, operands and comments so that we can understand what the individual machine codes are to do.
The fetch-execute cycle
The processor runs programs by using the fetch-execute cycle; Each instruction in memory is in turn, fetched, placed in the IR, and then executed by the CU.
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
13
Slides for Fundamentals of Computer Architecture Mark Burrell, 2004
14
