Skip to content

Instructions

:material-circle-edit-outline: 约 576 个字 :material-image-multiple-outline: 25 张图片 :material-clock-time-two-outline: 预计阅读时间 6 分钟

Memory Layout

alt text The reason for using iteration to replace recursion: - The size of the stack is limited - The computer is vulnerable - Registers and addresses are stored and protected in the stack - Too much recursion leads to stack overflow

Frame Pointer: Located at the bottom of the stack, for faster addressing

Tip

Use the "nickname"

alt text

W type operator:
* Single word

alt text

Tip

For all the jumping operations, beq and bne are enough

jal, jalr

Warning

The MIPS instruction set is different from RISC-V, don't mix the two.

alt text

ASCII: Abundant at first (English, Latin)
Unicode: 32-bit character set (standard)
* Widely used in programming languages

Byte/Halfword/Word Operations

alt text

Warning

All the operations are presented in 32-bit in the experiment section.

alt text
Reduce the usage of saved registers
* In embedded development, every bit of memory matters
* Although embedded development is difficult, it is the trend. Some students will be forced by life... (Quoted from Prof. Ma De) 😂

Immediate number and address

If the number is too large, larger than 12-bit
lui:
alt text
contains storage of 20-bit

Why 20-bit?

when you want to add addi💻

alt text

  • SB-type *Not suitable for Chinese students : 指令地址一定要对齐(half-word alignment: preparation for ARM ISA) 因此没有第一位 alt text
  • UJ FORMAT *:
    alt text
    When the address is too big, use lui

Collection of Instructions

alt text

Synchronization in RISC-V

  • nobody can interrupt the atomic operation
    alt text lr.d and sc.d
  • lr.d is used to load the value of the address into the register, thus the CPU remembers the operation lr.d
  • sc.d is used to store the value of the register into the address, lr.d and sr.d are a pair of operations, and the CPU will check if the value of the address has been changed during the operation. If it has not been changed, the operation is successful. Ohterwise its not.

Example

r20->7 then lr.d s2 (r20)
s2=7
CPU remebers the procedure
sc.d rd (r20) s2
the value on the (r20) address might be changed because of multi-core structure, when encountered this situation, 1 will be stored in rd(unsuccessful, the value of s2 cannot be stored in that memory).

alt text No interruption is allowed
in cpu, one paticular address need to be locked, so that no other cpu can access it.
alt text

Translating and staring a program

alt text
alt text
alt text
when learning compiling
.dll is a dynamic link library, which is a collection of small programs, any of which can be called when needed by a larger program that is running in the computer. The small program that lets the larger program communicate with a specific device such as a printer or scanner is often packaged as a DLL program (usually referred to as a DLL file).
alt text
Starting Java Applications
the virtual machine of Java allows Java to be run on any platform.
alt text

A C Sort Example To Pull It All Together

alt text
alt text

Arrays Versus Pointers

alt text(comilers are smart enough to optimize the code, to change the array to pointer and vice versa)
alt text

Real Stuff: MIPS Instrcuctions

all 32-bit instructions
alt text the difference:
alt text
reverse version of RISC-V
the store instruction is the same as the load instruction
not successful in the market😂

Real Stuff: The Intel x86 ISA

8080: 8-bit
8086: 16-bit
Further evolution:
* AMD64(64)
regs have 32bits and 64bits
Too complex!!!
Mac: ARM, 外星科技！！！

Implementing IA-32

  • All things taught in this course is RV64I
    There is extentions: M,A,F,D,C

Fallacies

  • Poerful instruction -> higher performance No!!!
  • Backward compatibility -> ISA doesn't change No!!!