---
###### tags: `Computer Architecture 2022`
---
<!-- * 編譯
> riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O3 -o HW2 HW2.c
* 反組譯並存成txt
> riscv-none-elf-objdump -d ./HW2 > asb.txt
* 看elf標頭
> riscv-none-elf-readelf -h ./HW2
* 看size
> riscv-none-elf-size ./HW2
* 執行code
> build/rv32emu HW2 -->
# Assignment 2-2: Decompress Run-Length Encoded List
## Problem description:
- This problem is based on [LeetCode 1](https://leetcode.com/problems/two-sum/) from [周士翔](https://hackmd.io/@zDmciYQATNm-8XeyJ5Th0Q/r1htullMj)
- I have
- Besides, I compared the original handwritten assembly code with unrolling handwritten assembly code. [Link](https://hackmd.io/Y4PD3GQjTO2FlA8KRGGb6g)
## C code
```cpp=
#include <stdio.h>
void twoSum(int* nums,int numSize, int target){
int i,j;
for(i=0;i<numSize;i++){
for(j=i+1;j<numSize;j++){
if(nums[i]+nums[j]==target)
printf("%d %d\n",i,j);
}
}
}
int main(){
int nums1[4] = {2,7,11,15}, target1 = 9, size1=4;
int nums2[3] = {3,2,4}, target2 = 6, size2=3;
int nums3[2] = {3,3}, target3 = 6, size3=2;
twoSum(nums1,size1,target1);
twoSum(nums2,size2,target2);
twoSum(nums3,size3,target3);
return 0;
}
```
## Assembly code
I modifed the assembly code, so it can be compile with `riscv-none-elf-gcc` and run in the `rv32emu`
```asm=
.file "twosum_.c"
.option nopic
.attribute arch, "rv32i2p1"
.attribute unaligned_access, 0
.attribute stack_align, 16
.text
.section .rodata
.align 2
.LC1:
.string "%d %d\n"
.text #command
.align 2
.globl twoSum
.type twoSum, @function
twoSum:
addi sp, sp, -4
sw ra, 0(sp)
mv t0, zero #int i=0
addi t1, t0, 1 #int j=i+1
mv a6, s0 #store nums1 address for i
mv a1, s0 #store nums1 address for j
loop2:
lw a2, 0(a6) #load nums[i]
lw a3, 4(a1) #load nums[j]
add a3, a2, a3 #nums1[i]+nums2[j]
beq a3, s1, print #if (nums1[i]+nums2[j]==target) go to print
addi a1, a1, 4 #point to next nums1[j] address
addi t1, t1, 1 #j++
blt t1, s2, loop2
loop1:
addi t0, t0, 1 #i++
addi a6, a6, 4 #point to next nums[i] address
mv a1, a6
addi t1, t0, 1 #j=i+1
bltu t0, s2, loop2 #if (i<numsize) go to loop2
print:
mv a1, t0 #move i
mv a2, t1
lui a5,%hi(.LC1)
addi a0,a5,%lo(.LC1)
call printf
lw ra, 0(sp)
addi sp, sp, 4
jr ra
.size twoSum, .-twoSum
.section .rodata
.align 2
.data
nums1: .word 2, 7, 11, 15
target1: .word 9
size1: .word 4
nums2: .word 3, 2, 4
target2: .word 6
size2: .word 3
nums3: .word 3, 3
target3: .word 6
size3: .word 2
.text
.align 2
.globl main
.type main, @function
main:
addi sp,sp,-80
sw ra,76(sp)
sw s0,72(sp)
addi s0,sp,80
la s0, nums1 #load nums1 address to s0
lw s1, target1 #load target1 to s1
lw s2, size1 #load size1 to s2
call twoSum
la s0, nums2
lw s1, target2
lw s2, size2
call twoSum
la s0, nums3
lw s1, target3
lw s2, size3
call twoSum
lw ra,76(sp)
lw s0,72(sp)
addi sp,sp,80
jr ra
#li a7, 10 #end
#ecall
.size main, .-main
.ident "GCC: (xPack GNU RISC-V Embedded GCC x86_64) 12.2.0"
```
## Instruction
## Result
### -O0
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O0 -o hw2/twosum_o0 -s hw2/twosum_.c`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | ------------- |
| 74932 | 2816 | 812 | 78560 | 132e0 | hw2/twosum_o0 |
<!--  -->
- ELF

- code
```asm=
00010184 <twoSum>:
10184: fd010113 addi sp,sp,-48
10188: 02112623 sw ra,44(sp)
1018c: 02812423 sw s0,40(sp)
10190: 03010413 addi s0,sp,48
10194: fca42e23 sw a0,-36(s0)
10198: fcb42c23 sw a1,-40(s0)
1019c: fcc42a23 sw a2,-44(s0)
101a0: fe042623 sw zero,-20(s0)
101a4: 0800006f j 10224 <twoSum+0xa0>
101a8: fec42783 lw a5,-20(s0)
101ac: 00178793 addi a5,a5,1
101b0: fef42423 sw a5,-24(s0)
101b4: 0580006f j 1020c <twoSum+0x88>
101b8: fec42783 lw a5,-20(s0)
101bc: 00279793 slli a5,a5,0x2
101c0: fdc42703 lw a4,-36(s0)
101c4: 00f707b3 add a5,a4,a5
101c8: 0007a703 lw a4,0(a5)
101cc: fe842783 lw a5,-24(s0)
101d0: 00279793 slli a5,a5,0x2
101d4: fdc42683 lw a3,-36(s0)
101d8: 00f687b3 add a5,a3,a5
101dc: 0007a783 lw a5,0(a5)
101e0: 00f707b3 add a5,a4,a5
101e4: fd442703 lw a4,-44(s0)
101e8: 00f71c63 bne a4,a5,10200 <twoSum+0x7c>
101ec: fe842603 lw a2,-24(s0)
101f0: fec42583 lw a1,-20(s0)
101f4: 000217b7 lui a5,0x21
101f8: 79078513 addi a0,a5,1936 # 21790 <__clzsi2+0x88>
101fc: 350000ef jal ra,1054c <printf>
10200: fe842783 lw a5,-24(s0)
10204: 00178793 addi a5,a5,1
10208: fef42423 sw a5,-24(s0)
1020c: fe842703 lw a4,-24(s0)
10210: fd842783 lw a5,-40(s0)
10214: faf742e3 blt a4,a5,101b8 <twoSum+0x34>
10218: fec42783 lw a5,-20(s0)
1021c: 00178793 addi a5,a5,1
10220: fef42623 sw a5,-20(s0)
10224: fec42703 lw a4,-20(s0)
10228: fd842783 lw a5,-40(s0)
1022c: f6f74ee3 blt a4,a5,101a8 <twoSum+0x24>
10230: 00000013 nop
10234: 00000013 nop
10238: 02c12083 lw ra,44(sp)
1023c: 02812403 lw s0,40(sp)
10240: 03010113 addi sp,sp,48
10244: 00008067 ret
00010248 <main>:
10248: fb010113 addi sp,sp,-80
1024c: 04112623 sw ra,76(sp)
10250: 04812423 sw s0,72(sp)
10254: 05010413 addi s0,sp,80
10258: 000217b7 lui a5,0x21
1025c: 79878793 addi a5,a5,1944 # 21798 <__clzsi2+0x90>
10260: 0007a603 lw a2,0(a5)
10264: 0047a683 lw a3,4(a5)
10268: 0087a703 lw a4,8(a5)
1026c: 00c7a783 lw a5,12(a5)
10270: fcc42423 sw a2,-56(s0)
10274: fcd42623 sw a3,-52(s0)
10278: fce42823 sw a4,-48(s0)
1027c: fcf42a23 sw a5,-44(s0)
10280: 00900793 li a5,9
10284: fef42623 sw a5,-20(s0)
10288: 00400793 li a5,4
1028c: fef42423 sw a5,-24(s0)
10290: 00300793 li a5,3
10294: faf42e23 sw a5,-68(s0)
10298: 00200793 li a5,2
1029c: fcf42023 sw a5,-64(s0)
102a0: 00400793 li a5,4
102a4: fcf42223 sw a5,-60(s0)
102a8: 00600793 li a5,6
102ac: fef42223 sw a5,-28(s0)
102b0: 00300793 li a5,3
102b4: fef42023 sw a5,-32(s0)
102b8: 00300793 li a5,3
102bc: faf42a23 sw a5,-76(s0)
102c0: 00300793 li a5,3
102c4: faf42c23 sw a5,-72(s0)
102c8: 00600793 li a5,6
102cc: fcf42e23 sw a5,-36(s0)
102d0: 00200793 li a5,2
102d4: fcf42c23 sw a5,-40(s0)
102d8: fc840793 addi a5,s0,-56
102dc: fec42603 lw a2,-20(s0)
102e0: fe842583 lw a1,-24(s0)
102e4: 00078513 mv a0,a5
102e8: e9dff0ef jal ra,10184 <twoSum>
102ec: fbc40793 addi a5,s0,-68
102f0: fe442603 lw a2,-28(s0)
102f4: fe042583 lw a1,-32(s0)
102f8: 00078513 mv a0,a5
102fc: e89ff0ef jal ra,10184 <twoSum>
10300: fb440793 addi a5,s0,-76
10304: fdc42603 lw a2,-36(s0)
10308: fd842583 lw a1,-40(s0)
1030c: 00078513 mv a0,a5
10310: e75ff0ef jal ra,10184 <twoSum>
10314: 00000793 li a5,0
10318: 00078513 mv a0,a5
1031c: 04c12083 lw ra,76(sp)
10320: 04812403 lw s0,72(sp)
10324: 05010113 addi sp,sp,80
10328: 00008067 ret
```
- Observation
- default compiler
- Register Usage
- s-type:s0
- a-type:a0, a1, a2, a3, a4, a5
- t-type:0
- other:sp, ra, zero
- Total:10
- branch instruction:11
- lines:108
### -O1
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O1 -o hw2/twosum_o1 -s hw2/twosum_.c`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | ------------- |
| 74852 | 2816 | 812 | 78480 | 13290 | hw2/twosum_o1 |
<!-- 
-->
- ELF

- code
```asm=
00010184 <twoSum>:
10184: 0cb05263 blez a1,10248 <twoSum+0xc4>
10188: fd010113 addi sp,sp,-48
1018c: 02112623 sw ra,44(sp)
10190: 02812423 sw s0,40(sp)
10194: 02912223 sw s1,36(sp)
10198: 03212023 sw s2,32(sp)
1019c: 01312e23 sw s3,28(sp)
101a0: 01412c23 sw s4,24(sp)
101a4: 01512a23 sw s5,20(sp)
101a8: 01612823 sw s6,16(sp)
101ac: 01712623 sw s7,12(sp)
101b0: 00058993 mv s3,a1
101b4: 00060a13 mv s4,a2
101b8: 00050913 mv s2,a0
101bc: 00000a93 li s5,0
101c0: 00021bb7 lui s7,0x21
101c4: 0400006f j 10204 <twoSum+0x80>
101c8: 00048613 mv a2,s1
101cc: 000b0593 mv a1,s6
101d0: 740b8513 addi a0,s7,1856 # 21740 <__clzsi2+0x88>
101d4: 328000ef jal ra,104fc <printf>
101d8: 00148793 addi a5,s1,1
101dc: 00440413 addi s0,s0,4
101e0: 00f98e63 beq s3,a5,101fc <twoSum+0x78>
101e4: 00078493 mv s1,a5
101e8: 00092783 lw a5,0(s2)
101ec: 00442703 lw a4,4(s0)
101f0: 00e787b3 add a5,a5,a4
101f4: ff4792e3 bne a5,s4,101d8 <twoSum+0x54>
101f8: fd1ff06f j 101c8 <twoSum+0x44>
101fc: 00490913 addi s2,s2,4
10200: 009b0e63 beq s6,s1,1021c <twoSum+0x98>
10204: 000a8b13 mv s6,s5
10208: 001a8a93 addi s5,s5,1
1020c: 01598863 beq s3,s5,1021c <twoSum+0x98>
10210: 00090413 mv s0,s2
10214: 000a8493 mv s1,s5
10218: fd1ff06f j 101e8 <twoSum+0x64>
1021c: 02c12083 lw ra,44(sp)
10220: 02812403 lw s0,40(sp)
10224: 02412483 lw s1,36(sp)
10228: 02012903 lw s2,32(sp)
1022c: 01c12983 lw s3,28(sp)
10230: 01812a03 lw s4,24(sp)
10234: 01412a83 lw s5,20(sp)
10238: 01012b03 lw s6,16(sp)
1023c: 00c12b83 lw s7,12(sp)
10240: 03010113 addi sp,sp,48
10244: 00008067 ret
10248: 00008067 ret
0001024c <main>:
1024c: fc010113 addi sp,sp,-64
10250: 02112e23 sw ra,60(sp)
10254: 000217b7 lui a5,0x21
10258: 74878793 addi a5,a5,1864 # 21748 <__clzsi2+0x90>
1025c: 0007a603 lw a2,0(a5)
10260: 0047a683 lw a3,4(a5)
10264: 0087a703 lw a4,8(a5)
10268: 00c7a783 lw a5,12(a5)
1026c: 02c12023 sw a2,32(sp)
10270: 02d12223 sw a3,36(sp)
10274: 02e12423 sw a4,40(sp)
10278: 02f12623 sw a5,44(sp)
1027c: 00300793 li a5,3
10280: 00f12a23 sw a5,20(sp)
10284: 00200713 li a4,2
10288: 00e12c23 sw a4,24(sp)
1028c: 00400713 li a4,4
10290: 00e12e23 sw a4,28(sp)
10294: 00f12623 sw a5,12(sp)
10298: 00f12823 sw a5,16(sp)
1029c: 00900613 li a2,9
102a0: 00400593 li a1,4
102a4: 02010513 addi a0,sp,32
102a8: eddff0ef jal ra,10184 <twoSum>
102ac: 00600613 li a2,6
102b0: 00300593 li a1,3
102b4: 01410513 addi a0,sp,20
102b8: ecdff0ef jal ra,10184 <twoSum>
102bc: 00600613 li a2,6
102c0: 00200593 li a1,2
102c4: 00c10513 addi a0,sp,12
102c8: ebdff0ef jal ra,10184 <twoSum>
102cc: 00000513 li a0,0
102d0: 03c12083 lw ra,60(sp)
102d4: 04010113 addi sp,sp,64
102d8: 00008067 ret
```
- Observation
- optimize compiler
- Register Usage
- s-type:s0 ~ s7
- a-type:a0, a1, a2, a3, a4, a5
- t-type:0
- other:sp, ra, zero
- Total:17
- branch instruction:15
- lines:88
### -O2
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O2 -o hw2/twosum_o2 -s hw2/twosum_.c`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | ------------- |
| 74848 | 2816 | 812 | 78476 | 1328c | hw2/twosum_o2 |
<!--  -->
- ELF

- code
```asm=
000100c4 <main>:
100c4: 000217b7 lui a5,0x21
100c8: 74878793 addi a5,a5,1864 # 21748 <__clzsi2+0x94>
100cc: 00c7a703 lw a4,12(a5)
100d0: fc010113 addi sp,sp,-64
100d4: 0007a883 lw a7,0(a5)
100d8: 0047a803 lw a6,4(a5)
100dc: 0087a683 lw a3,8(a5)
100e0: 02e12623 sw a4,44(sp)
100e4: 00200713 li a4,2
100e8: 00300793 li a5,3
100ec: 02010513 addi a0,sp,32
100f0: 00e12c23 sw a4,24(sp)
100f4: 00900613 li a2,9
100f8: 00400713 li a4,4
100fc: 00400593 li a1,4
10100: 02112e23 sw ra,60(sp)
10104: 03112023 sw a7,32(sp)
10108: 03012223 sw a6,36(sp)
1010c: 02d12423 sw a3,40(sp)
10110: 00f12a23 sw a5,20(sp)
10114: 00e12e23 sw a4,28(sp)
10118: 00f12623 sw a5,12(sp)
1011c: 00f12823 sw a5,16(sp)
10120: 0f4000ef jal ra,10214 <twoSum>
10124: 01410513 addi a0,sp,20
10128: 00600613 li a2,6
1012c: 00300593 li a1,3
10130: 0e4000ef jal ra,10214 <twoSum>
10134: 00c10513 addi a0,sp,12
10138: 00600613 li a2,6
1013c: 00200593 li a1,2
10140: 0d4000ef jal ra,10214 <twoSum>
10144: 03c12083 lw ra,60(sp)
10148: 00000513 li a0,0
1014c: 04010113 addi sp,sp,64
10150: 00008067 ret
00010214 <twoSum>:
10214: 0cb05063 blez a1,102d4 <twoSum+0xc0>
10218: fd010113 addi sp,sp,-48
1021c: 02812423 sw s0,40(sp)
10220: 03212023 sw s2,32(sp)
10224: 01312e23 sw s3,28(sp)
10228: 01412c23 sw s4,24(sp)
1022c: 01712623 sw s7,12(sp)
10230: 02112623 sw ra,44(sp)
10234: 02912223 sw s1,36(sp)
10238: 01512a23 sw s5,20(sp)
1023c: 01612823 sw s6,16(sp)
10240: 00100413 li s0,1
10244: 00058993 mv s3,a1
10248: 00060a13 mv s4,a2
1024c: 00450913 addi s2,a0,4
10250: 00021bb7 lui s7,0x21
10254: 04858a63 beq a1,s0,102a8 <twoSum+0x94>
10258: 00040b13 mv s6,s0
1025c: fff40a93 addi s5,s0,-1
10260: 00090493 mv s1,s2
10264: 00c0006f j 10270 <twoSum+0x5c>
10268: 00140413 addi s0,s0,1
1026c: 03345863 bge s0,s3,1029c <twoSum+0x88>
10270: 0004a703 lw a4,0(s1)
10274: ffc92783 lw a5,-4(s2)
10278: 00448493 addi s1,s1,4
1027c: 00e787b3 add a5,a5,a4
10280: ff4794e3 bne a5,s4,10268 <twoSum+0x54>
10284: 00040613 mv a2,s0
10288: 000a8593 mv a1,s5
1028c: 740b8513 addi a0,s7,1856 # 21740 <__clzsi2+0x8c>
10290: 00140413 addi s0,s0,1
10294: 264000ef jal ra,104f8 <printf>
10298: fd344ce3 blt s0,s3,10270 <twoSum+0x5c>
1029c: 001b0413 addi s0,s6,1
102a0: 00490913 addi s2,s2,4
102a4: fa899ae3 bne s3,s0,10258 <twoSum+0x44>
102a8: 02c12083 lw ra,44(sp)
102ac: 02812403 lw s0,40(sp)
102b0: 02412483 lw s1,36(sp)
102b4: 02012903 lw s2,32(sp)
102b8: 01c12983 lw s3,28(sp)
102bc: 01812a03 lw s4,24(sp)
102c0: 01412a83 lw s5,20(sp)
102c4: 01012b03 lw s6,16(sp)
102c8: 00c12b83 lw s7,12(sp)
102cc: 03010113 addi sp,sp,48
102d0: 00008067 ret
102d4: 00008067 ret
```
- Observation
- optimize compiler
- Register Usage
- s-type:s0 ~ s7
- a-type:a0 ~ a7
- t-type:0
- other:sp, ra
- Total:18
- branch instruction:14
- lines:87
### -O3
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O3 -o hw2/twosum_o3 -s hw2/twosum_.c`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | ---------- |
| 74848 | 2816 | 812 | 78476 | 1328c | hw2/twosum_o3 |
<!--  -->
- ELF

- code
```asm=
000100c4 <main>:
100c4: 000217b7 lui a5,0x21
100c8: 74878793 addi a5,a5,1864 # 21748 <__clzsi2+0x94>
100cc: 00c7a703 lw a4,12(a5)
100d0: fc010113 addi sp,sp,-64
100d4: 0007a883 lw a7,0(a5)
100d8: 0047a803 lw a6,4(a5)
100dc: 0087a683 lw a3,8(a5)
100e0: 02e12623 sw a4,44(sp)
100e4: 00200713 li a4,2
100e8: 00300793 li a5,3
100ec: 02010513 addi a0,sp,32
100f0: 00e12c23 sw a4,24(sp)
100f4: 00900613 li a2,9
100f8: 00400713 li a4,4
100fc: 00400593 li a1,4
10100: 02112e23 sw ra,60(sp)
10104: 03112023 sw a7,32(sp)
10108: 03012223 sw a6,36(sp)
1010c: 02d12423 sw a3,40(sp)
10110: 00f12a23 sw a5,20(sp)
10114: 00e12e23 sw a4,28(sp)
10118: 00f12623 sw a5,12(sp)
1011c: 00f12823 sw a5,16(sp)
10120: 0f4000ef jal ra,10214 <twoSum>
10124: 01410513 addi a0,sp,20
10128: 00600613 li a2,6
1012c: 00300593 li a1,3
10130: 0e4000ef jal ra,10214 <twoSum>
10134: 00c10513 addi a0,sp,12
10138: 00600613 li a2,6
1013c: 00200593 li a1,2
10140: 0d4000ef jal ra,10214 <twoSum>
10144: 03c12083 lw ra,60(sp)
10148: 00000513 li a0,0
1014c: 04010113 addi sp,sp,64
10150: 00008067 ret
00010214 <twoSum>:
10214: 0cb05063 blez a1,102d4 <twoSum+0xc0>
10218: fd010113 addi sp,sp,-48
1021c: 02812423 sw s0,40(sp)
10220: 03212023 sw s2,32(sp)
10224: 01312e23 sw s3,28(sp)
10228: 01412c23 sw s4,24(sp)
1022c: 01712623 sw s7,12(sp)
10230: 02112623 sw ra,44(sp)
10234: 02912223 sw s1,36(sp)
10238: 01512a23 sw s5,20(sp)
1023c: 01612823 sw s6,16(sp)
10240: 00100413 li s0,1
10244: 00058993 mv s3,a1
10248: 00060a13 mv s4,a2
1024c: 00450913 addi s2,a0,4
10250: 00021bb7 lui s7,0x21
10254: 04858a63 beq a1,s0,102a8 <twoSum+0x94>
10258: 00040b13 mv s6,s0
1025c: fff40a93 addi s5,s0,-1
10260: 00090493 mv s1,s2
10264: 00c0006f j 10270 <twoSum+0x5c>
10268: 00140413 addi s0,s0,1
1026c: 03345863 bge s0,s3,1029c <twoSum+0x88>
10270: 0004a703 lw a4,0(s1)
10274: ffc92783 lw a5,-4(s2)
10278: 00448493 addi s1,s1,4
1027c: 00e787b3 add a5,a5,a4
10280: ff4794e3 bne a5,s4,10268 <twoSum+0x54>
10284: 00040613 mv a2,s0
10288: 000a8593 mv a1,s5
1028c: 740b8513 addi a0,s7,1856 # 21740 <__clzsi2+0x8c>
10290: 00140413 addi s0,s0,1
10294: 264000ef jal ra,104f8 <printf>
10298: fd344ce3 blt s0,s3,10270 <twoSum+0x5c>
1029c: 001b0413 addi s0,s6,1
102a0: 00490913 addi s2,s2,4
102a4: fa899ae3 bne s3,s0,10258 <twoSum+0x44>
102a8: 02c12083 lw ra,44(sp)
102ac: 02812403 lw s0,40(sp)
102b0: 02412483 lw s1,36(sp)
102b4: 02012903 lw s2,32(sp)
102b8: 01c12983 lw s3,28(sp)
102bc: 01812a03 lw s4,24(sp)
102c0: 01412a83 lw s5,20(sp)
102c4: 01012b03 lw s6,16(sp)
102c8: 00c12b83 lw s7,12(sp)
102cc: 03010113 addi sp,sp,48
102d0: 00008067 ret
102d4: 00008067 ret
```
- Observation
- optimize compiler
- Register Usage
- s-type:s0 ~ s7
- a-type:a0 ~ a7
- t-type:0
- other:sp, ra
- Total:18
- branch instruction:14
- lines:87
### -Ofast
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -Ofast -o hw2/twosum_fast -s hw2/twosum_.c`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | --------------- |
| 74848 | 2816 | 812 | 78476 | 1328c | hw2/twosum_fast |
<!--  -->
- ELF

- code
```asm=
000100c4 <main>:
100c4: 000217b7 lui a5,0x21
100c8: 74878793 addi a5,a5,1864 # 21748 <__clzsi2+0x94>
100cc: 00c7a703 lw a4,12(a5)
100d0: fc010113 addi sp,sp,-64
100d4: 0007a883 lw a7,0(a5)
100d8: 0047a803 lw a6,4(a5)
100dc: 0087a683 lw a3,8(a5)
100e0: 02e12623 sw a4,44(sp)
100e4: 00200713 li a4,2
100e8: 00300793 li a5,3
100ec: 02010513 addi a0,sp,32
100f0: 00e12c23 sw a4,24(sp)
100f4: 00900613 li a2,9
100f8: 00400713 li a4,4
100fc: 00400593 li a1,4
10100: 02112e23 sw ra,60(sp)
10104: 03112023 sw a7,32(sp)
10108: 03012223 sw a6,36(sp)
1010c: 02d12423 sw a3,40(sp)
10110: 00f12a23 sw a5,20(sp)
10114: 00e12e23 sw a4,28(sp)
10118: 00f12623 sw a5,12(sp)
1011c: 00f12823 sw a5,16(sp)
10120: 0f4000ef jal ra,10214 <twoSum>
10124: 01410513 addi a0,sp,20
10128: 00600613 li a2,6
1012c: 00300593 li a1,3
10130: 0e4000ef jal ra,10214 <twoSum>
10134: 00c10513 addi a0,sp,12
10138: 00600613 li a2,6
1013c: 00200593 li a1,2
10140: 0d4000ef jal ra,10214 <twoSum>
10144: 03c12083 lw ra,60(sp)
10148: 00000513 li a0,0
1014c: 04010113 addi sp,sp,64
10150: 00008067 ret
00010214 <twoSum>:
10214: 0cb05063 blez a1,102d4 <twoSum+0xc0>
10218: fd010113 addi sp,sp,-48
1021c: 02812423 sw s0,40(sp)
10220: 03212023 sw s2,32(sp)
10224: 01312e23 sw s3,28(sp)
10228: 01412c23 sw s4,24(sp)
1022c: 01712623 sw s7,12(sp)
10230: 02112623 sw ra,44(sp)
10234: 02912223 sw s1,36(sp)
10238: 01512a23 sw s5,20(sp)
1023c: 01612823 sw s6,16(sp)
10240: 00100413 li s0,1
10244: 00058993 mv s3,a1
10248: 00060a13 mv s4,a2
1024c: 00450913 addi s2,a0,4
10250: 00021bb7 lui s7,0x21
10254: 04858a63 beq a1,s0,102a8 <twoSum+0x94>
10258: 00040b13 mv s6,s0
1025c: fff40a93 addi s5,s0,-1
10260: 00090493 mv s1,s2
10264: 00c0006f j 10270 <twoSum+0x5c>
10268: 00140413 addi s0,s0,1
1026c: 03345863 bge s0,s3,1029c <twoSum+0x88>
10270: 0004a703 lw a4,0(s1)
10274: ffc92783 lw a5,-4(s2)
10278: 00448493 addi s1,s1,4
1027c: 00e787b3 add a5,a5,a4
10280: ff4794e3 bne a5,s4,10268 <twoSum+0x54>
10284: 00040613 mv a2,s0
10288: 000a8593 mv a1,s5
1028c: 740b8513 addi a0,s7,1856 # 21740 <__clzsi2+0x8c>
10290: 00140413 addi s0,s0,1
10294: 264000ef jal ra,104f8 <printf>
10298: fd344ce3 blt s0,s3,10270 <twoSum+0x5c>
1029c: 001b0413 addi s0,s6,1
102a0: 00490913 addi s2,s2,4
102a4: fa899ae3 bne s3,s0,10258 <twoSum+0x44>
102a8: 02c12083 lw ra,44(sp)
102ac: 02812403 lw s0,40(sp)
102b0: 02412483 lw s1,36(sp)
102b4: 02012903 lw s2,32(sp)
102b8: 01c12983 lw s3,28(sp)
102bc: 01812a03 lw s4,24(sp)
102c0: 01412a83 lw s5,20(sp)
102c4: 01012b03 lw s6,16(sp)
102c8: 00c12b83 lw s7,12(sp)
102cc: 03010113 addi sp,sp,48
102d0: 00008067 ret
102d4: 00008067 ret
```
- Observation
- optimize compiler
- Register Usage
- s-type:s0 ~ s7
- a-type:a0 ~ a7
- t-type:0
- other:sp, ra
- Total:18
- branch instruction:14
- lines:87
### handwritten assembly code
<!-- In this part, I try to run the assembly code by [周士翔](https://hackmd.io/@zDmciYQATNm-8XeyJ5Th0Q/r1htullMj) on the `rv32emu` and compare it with `-Os` and `-Ofast` -->
- compile
`riscv-none-elf-gcc -march=rv32i -mabi=ilp32 -O0 -o hw2/handwritten hw22/handwritten.s`
- size
| text | data | bss | dec | hex | filename |
| ----- | ---- | --- | ----- | ----- | --------------- |
| 74680 | 2880 | 812 | 78372 | 13224 | ./hw22/handwritten |
<!--  -->
- ELF

- code
[In-the-above-part](##Assembly-code)
- Observation
- Register Usage
- s-type:s0 ~ s2
- a-type:a0, a1, a2, a3, a5, a6
- t-type:0
- other:sp, ra
- Total:11
- branch instruction:14
- lines:87
## CSR cycle
Using the `rv32emu --stats` to help me record the CSR cycle, and compare with the different compiler and handwritten assembly code.
### -O0
```
01
12
01
inferior exit code 0
CSR cycle count: 4251
```
### -O1
```
01
12
01
inferior exit code 0
CSR cycle count: 4087
```
### -O2
```
01
12
01
inferior exit code 0
CSR cycle count: 4077
```
### -O3
```
01
12
01
inferior exit code 0
CSR cycle count: 4077
```
### -Ofast
```
01
12
01
inferior exit code 0
CSR cycle count: 4077
```
### handwritten
```
01
12
01
inferior exit code 4
CSR cycle count: 3957
```
### Result
The handwritten assembly code had fewer CSR cycle when run in the `rv32emu`, besides, I found `printf` account for a great proportion of cycle.
So if the `printf` is removed, the program will reduce thousands of cycles.