From: Ryan McGovern <mcgoverp@...>
Date: 2006-07-15T00:27:56+09:00
Subject: Re: How to speed up ruby and make it as fast as possible

"I find this answer to be unsatisfactory.  I do not know C and neither do
many others who like Ruby but want better performance.  The knowledge
and syntax demands of programming in Ruby and C are worlds apart.  It is
not much different than telling someone to go code in (modern)
Assembler! "

Assembler is not that bad either, imho once you've learned how to 
program beyond just a specific language picking up another one is not 
that challenging.  And i agree learn c it's at the very least good for 
very speed important tasks, or interfacing directly with the os or 
writing ruby modules to do said interfacing. 

I mean try to beat this thing in speed *need an x86 compiler*

.model small

.stack 100h

.386

Bufsize= 800

bsize=800

.data

ten    DD 10

FACT DD Bufsize DUP(0)

INDEX dd 0

MULTI DD 0

LEN DD 1

.code

main    proc far

    mov ax,@data

    mov ds,ax

    mov ah,62h

    int 21h

    mov es,bx

    mov eax,0

    mov ebx,081h

    movzx ecx,BYTE PTR es:[80h]

L4:    mul ten

    movzx edx,BYTE PTR es:[ebx]

    and edx,000000fh

    add eax,edx

    inc ebx

    loop L4

;put code here

        MOV ESI,OFFSET FACT

         MOV [ESI],EAX

         DEC EAX

         MOV ECX,EAX

         CMP ECX,0

         JE  PEAX

Fac1:

     MOV EBX,0

     MOV ESI,OFFSET FACT

     MOV INDEX,0

Fac2:

     MOV EAX,[ESI]                    ;MOV FACT INTO EAX

     MUL ECX                         ;MUL BY COUNT

     ADD EAX,EBX                    ;ADD CONTENTS OF EBX TO 
ECX                    

     JNC Fac3                         ;NO CARRY => JUMP L3

     INC EDX                         ;OTHERWISE INC EDX

Fac3:

     MOV EBX,EDX                    ;MOV CONTENT OF EDX TO EBX

     MOV [ESI],EAX                    ;STORE MUL RESULT IN CURRENT ESI POS

     INC INDEX                    

     MOV EAX,INDEX                    

     CMP LEN,EAX                    ;COMPARE LEN, INDEX

     JNE Fac5                         ;NOT EQUAL => JUMP L2

     CMP EBX,0                    ;COMPARE FOR CARRY AFTER ADD ECX,EBX

     JZ Fac4                         ;NO CARRY => JUMP L4

     INC LEN                         ;INC LEN

     ADD ESI,4

     MOV [ESI],EBX

     JMP Fac4

Fac5:

     ADD ESI,4

     JMP Fac2

 

Fac4:

     LOOP Fac1



mov INDEX,0

mov ESI,OFFSET FACT

MK1:

mov eax,[esi]

push eax

inc INDEX

add esi,4

mov ecx,INDEX

cmp ecx, LEN

jnz MK1





;end code

PEAX:

    pop eax

    mov ebx,eax

    mov ecx,8

    cmp INDEX,0

    jz I1

L9:   

   

   

L3:

        mov edx,0f0000000h

    and edx,ebx

    push ecx

    mov cl,28

        shr edx,cl

    pop ecx

    mov ah,02h

    cmp dl,10

    jb L1

    add dl,37h

        jmp L2

L1:    or dl,30h

L2:    int 21h

    shl ebx,1

    shl ebx,1

    shl ebx,1

    shl ebx,1

    loop L3

    dec INDEX

    cmp INDEX,0

   

    jnz PEAX

   

    mov ah,4ch

    int 21h

I1:

    inc INDEX

    jmp L9



main    endp

end    main

Kent Sibilev wrote:
> I find this attitude is quite amazing. We all know that Ruby is not a
> best language in terms of performance. Even if we had Ruby 2.0 (YARV)
> delivered to us, Ruby's performance would still be a way behind those
> strong typed languges like ocaml or java. If you want to keep using
> Ruby, but sometimes you need a raw performance, you should learn C.
> It's not that bad, you know. If you always need a raw performance,
> then probably you are not using right tool for the job.
>
> just my 2 cents.
>
> On 7/14/06, Jamal Mazrui <Jamal.Mazrui@fcc.gov> wrote:
>> -----Original Message-----
>> From: list-bounce@example.com [mailto:list-bounce@example.com] On Behalf
>> Of Reggie Mr
>> Sent: Friday, July 14, 2006 6:56 AM
>>
>> >  Write it C.
>>
>> I find this answer to be unsatisfactory.  I do not know C and neither do
>> many others who like Ruby but want better performance.  The knowledge
>> and syntax demands of programming in Ruby and C are worlds apart.  It is
>> not much different than telling someone to go code in (modern)
>> Assembler!
>>
>> I think the more that Rubyists acknowledge and try to directly solve the
>> language's performance hindrance rather than rationalizing it away, the
>> better the prospects for the success of our beloved language.  Of
>> course, better algorithms can produce better performance, but that
>> argument is also true for the languages with which Ruby compares
>> unfavorably in performance.  Please do not claim that Ruby's adoption by
>> others is irrelevant:  a good idea is worth spreading, and the more
>> users, the better library support and learning resources available to us
>> all.
>> Also, please do not tell me to rewrite Ruby, myself, if I care about
>> performance.  As stated, I do not know C or have the interest and time
>> to devote to such an undertaking.  I do promote Ruby to friends and
>> colleagues in ways that I can.  We should be able to have a legitimate
>> desire for language improvements without being characterized as
>> ignoramuses, nay-sayers or lazy spectators.
>>
>> Regards,
>> Jamal
>>
>>
>
>