Add NASM - x64

[DutchOrange-1]

Language Name

Assembly - X64

Docker Image

Locally run with NASM

Sample Implementation

; V1 - NASM 
; Takes about 9.774s - Note, my machine takes 600s to run a python script of the same formula. 
; Very much un-optomised, will fix in the comming days
; Result is just printed to the output. 
; Ran on Linux mint

extern printf
global main

section .data
    itterations  dd     1000000000    ; Number of itterations
    denom   dd 0        ;Denominator temp  ;esi
    nume     dd 0       ;Numerator temp   ;edi
    temp    dd 0.0
    result  dq 0.0696969   ; Result 
    four    dd 4.0 ; Last caculation
    fmt db "Result = %.16f", 10, 0

section .text


main:
    mov rbp, rsp; for correct debugging


    mov rbp, rsp; for correct debugging
    mov rbp, rsp; for correct debugging
    mov rcx , [itterations] ; How many counts
    xor rax, rax            ; Just clearing it. 
    xor rdx, rdx
    xorps xmm0, xmm0
    xorps xmm1, xmm1
    xorps xmm2, xmm2
    
    
    loopy:
    ; First do numerator:
    mov  rax, 1
    CALL even_test ; Numerator updates - rdi
    
    ; Now Denominator: rsi
    mov  rsi, rdx   ; Move k into rax
    add  rsi, rsi   ; Multiply by 2
    add  rsi, 1     ; Add one
    
    ; Now Combine them esi/edi
    ;fld  dword [esi]  ; Move to ST0 
    cvtsi2sd xmm0, rdi     ; xmm0 = float(esi) ; rdi = numerator
    cvtsi2sd xmm1, rsi     ; xmm1 = float(edi) ; rsi = denominator 
    
    divsd xmm0, xmm1       ; xmm0 = xmm0 / xmm1 ; s-single, d-double
    addpd xmm2, xmm0       ; Sum them
    
    ; Logic code
    inc rdx  ; Used for caculations - k
    dec rcx  ; Used to exit
    jnz loopy
    jmp Print_and_exit
    
    
Print_and_exit:
    push rbp
    movsd xmm0, xmm2
    addpd xmm0, xmm0 
    addpd xmm0, xmm0  
    mov rdi, fmt
    mov rax, 1
    call printf      ; Call the C library function
    
    xor rax, rax     ; Return 0 from main
    pop rbp          ; Restore stack
    ret
    
power_loop:             ; Multiplies ST0 with ST1, and stores in ST0 
  FMUL ST0, ST1
  dec   rcx
  jnz   power_loop ; Jump back 
  ret
  
 even_test: ; Insted of (-1^)x, it can be simplified 
     AND rax, rdx ; 0 = even , 1 = odd
     jz  make_pos
     jmp make_neg
 
     make_neg:
     mov rdi, -1
     ret
     
     make_pos:
     mov rdi, 1
     ret

Why add this language?

Its the foundation of them all, and should be made #1

Implementation Rules

• Single-threaded (no concurrency/parallelism)
• Uses the Leibniz formula for calculating pi
• Reads rounds from rounds.txt
• Outputs pi with 16 decimal places

[DutchOrange-1]

I improved it, bringing it down to 3.58 s.
(I should note that when running the fastest C version of the calculations on my machine, it takes about 5.8 s to complete, indicating the larger time is due to the OS / Hardware.)

Also, is this Repo still active? I see the last update was 3 months ago. Or is there a new one?

[DutchOrange-1]

Have an updated x64 version; it runs in less than 1.5s locally. As discussed above, this is a hardware issue and should run much faster on the Server.

Let me know how I can be added to the database.