for who using for,while , this show you how it's working

[freepurebasic]

The difference between for and while in any programming language :

Code: Select all

 howmannytimes=10000


 
;check if timeGetTime_() it's work 
startscript = timeGetTime_()



stopscript = timeGetTime_()
wasnothing= stopscript-startscript


startscript = timeGetTime_()

Delay(howmannytimes)

stopscript = timeGetTime_()
washowmannytimesmilisecconds= stopscript-startscript
;^----------------i was ckecking if timeGetTime_() was work
;you must get 0 and howmannytimes (min and max)





startscript = timeGetTime_()

 For x=1 To howmannytimes;wait "howmannytimes" milisecconds
   Delay(1);this is 1 milliseccond KNOWN TIME  
 Next

stopscript = timeGetTime_()
wasfor= stopscript-startscript



startscript = timeGetTime_()

 k=0
 While(k<howmannytimes) ;wait "howmannytimes" milisecconds
   Delay(1);this is 1 milliseccond KNOWN TIME
   k+1
 Wend

stopscript = timeGetTime_()
waswhile=stopscript-startscript


Debug "Results:"
Debug "void execution:" + Str(wasnothing)
Debug "Delay(howmannytimes) proc's execution:" + Str(washowmannytimesmilisecconds)
Debug "for execution:" + Str(wasfor)
Debug "while execution:" + Str(waswhile)

the same results in php or any script interpreter/pre-compiler / compiler
if you don't believe just TRY YOURSELF

[c4s]

Code: Select all

Debug "Results:"
Debug "void execution:" + Str(wasnothing)
Debug "Delay(howmannytimes) proc's execution:" + Str(washowmannytimesmilisecconds)
Debug "for execution:" + Str(wasfor)
Debug "while execution:" + Str(waswhile)

First Rule For Performance Tests Using PureBasic:
Disable the debugger!

[Ramihyn_]

You want to learn about resolution/precision of Delay() and TimeGetTime().

[freepurebasic]

not really wasn't my intention to do this ... you wrong understand,
the facts will be right this : you must get a number bigger for "for 's execution" than "while execution" wich is great !!!

questo erat demonstrandum

be sure you have buyed purebasic , here i have not so i don't know if work similary to freebasic (freebasic.net)
is no need to disable debugger because the script is relative with anyone language of this fk world .. so modify it for c++ and test it ,on windows api , you will get the same results ..

[Ramihyn_]

I dont understand half of what you say, but your measuring is totally bogus like that. Let's ignore the whole "loop" for a moment and do the following:

Code: Select all

FileID.i = CreateFile(#PB_Any, "delayunrolled.pb")

For f.i = 1 To 10000
  WriteStringN(FileID, "Delay(1)")
Next

CloseFile(FileID)

Run the file, load the delayunrolled.pb, add your time measuring like at the top:

Code: Select all

startscript = timeGetTime_()

and at the bottom

Code: Select all

stopscript = timeGetTime_()
tdiff = stopscript-startscript

MessageRequester("Blah...", "unrolled : " + Str(tdiff))

Compile it as an executable and run it. Run it again, run it while your machine is heavily in use, run it again. Compare the results, do the same with a small number of repetitions like 50, do it with a large number like 30000. Compare the numbers - what did you expect to get and what did you get?

[breeze4me]

Case 1: Debugger off.

Results:
void execution:0
Delay(howmannytimes) proc's execution:10000
for execution:19531
while execution:19531

Case 2: Debugger on.

Results:
void execution:0
Delay(howmannytimes) proc's execution:10000
for execution:19531
while execution:19532

the facts will be right this : you must get a number bigger for "for 's execution" than "while execution" wich is great !!!

no differences.

[luis]

Delay(1) is not guaranteed to give 1 ms delay.

Anyway, what this code should demonstrate ?

Especially considering you are putting a considerable delay inside the loops and so actually leveling the for/next and while/wend performances ?

BOH. I don't get it.

Moreover, where is the question ?

[Thorium]

I dont get what you want to tell us.

[freepurebasic]

there are people(MAYBE not you,not sure) that make confusion between for and while , so was no question was a demo that for is slow and while is speeding up.

giving this example i said : this is equal for any "language" (script interpreter/pre compiler/compiler) the "for" is slower than "while".

just all

i'll prepare you for php too if you want to see it (winbinder) ,works in win 7 too ,better .
this is php cgi solution ,so you can load a library and access a function as well ,in PHP !

and how i said before : if you adapt this code for any language you will have the prove the while sentence writed so large is faster like execution than a for sentence.

that's all puppies !

[freepurebasic]

breeze4me thank you man !!!!!!!!!!!!!!!
thank you !!!!!

Code: Select all

for execution:19531
while execution:19532

try to bigger processors (over dual core) you will get other data

yes!

now try it for 3000 000 times you will understand it

then try to modify a script with graphics ,3d too , i sow some examples here, try to convert all for cycles to while cycles and see the difference using your : Windows Task Manager / performance (xp,vista,7)
or Windows Task Manager / performance .. resource monitor

and see your processor when running a "for" based script and then see "while" based (the same graphic script)

[luis]

So you are saying, as a general rule, not knowing how any language can specifically implement a certain construct, that a "for/next" is always slower then a "while/wend" ?

I don't see why.

And specifically:

Code: Select all

For k = 1 To 3
 Debug k
Next

; For k = 1 To 3
  MOV    dword [v_k],1
_For1:
  MOV    eax,3
  CMP    eax,dword [v_k]
  JL    _Next2
; Debug k
; Next
_NextContinue2:
  INC    dword [v_k]
  JNO   _For1
_Next2:


k = 1
While k <= 3
 Debug k
 k + 1
Wend
 
; k = 1
  MOV    dword [v_k],1 
; While k <= 3
_While3:
  MOV    ebx,dword [v_k]
  CMP    ebx,3
  JG    _Wend3
; Debug k
; k + 1
  INC    dword [v_k]
; Wend
  JMP   _While3
_Wend3:

I disagree, and I don't know where you took this strange notion from.


Your test on my PC

Code: Select all

[16:35:36] Results:
[16:35:36] void execution:0
[16:35:36] Delay(howmannytimes) proc's execution:10000
[16:35:36] for execution:10001
[16:35:36] while execution:10000

Anyway, the test is meaningless because as I said you leveled the timing of the two loops by putting a delay inside. So the time spent in the loop code become irrelevant.

[freepurebasic]

nice but is wrong .... you are wrong.
can you did it again ?

anyway nice asm solution but isn't close

Code: Select all

I said you leveled the timing of the two loops by putting a delay inside.

=

Code: Select all

what you are seeing there :delay(1) is one miliseccond known time so 10000 times for 1 is 10000 known time is that you need to see to know the api function really works in the compiler you choose to calibrate the script,to verify if method choosen can deliver the answer you need using the same line of the method.

so having 0 and max-of-loops * 1 milliseccond of delay can you mesure using the same method the speed of the script/the program execution (not only in purebasic but in php,mysql,javascript,c++,pascal,fortran,freebasic,dev c,bloodshed c,etc etc etc etc)

using the same method





your asm script secvences are two : one for "for" and the seccond for "while"

but do you calculate the processor ticks ?????
i see a script but no answer and no calculation there
and depence where you tested : in a virtual machine or a processor on windows ???



uh oh uh oh

[Thorium]

nice but is wrong .... you are wrong.
can you did it again ?

You dont know what you are talking about.
Fix your time messuring first.

[freepurebasic]

Thorium with his hammer ...anyway you dance very well
Continue like this ...

Fix your time messuring first.

hmmm ... is already calculated you need JUST to look CAREFULLY the results on screen to make decisions.....
Delay(howmannytimes) it's the measure !!!

..api windows / purebasic measure i don't interesting if Delay works really , what i interest is- using Delay in all cases like measure to obtain the same result: "While cycle" is speeding up , "for cycle" is slower than "while" ! this i need to know 100% to make a decision.

so my advice is : use "while" instread "for" in ANY language!

[breeze4me]

Code: Select all

;DisableDebugger
CompilerIf #PB_Compiler_Debugger
  End
CompilerEndIf

SetPriorityClass_(GetCurrentProcess_(), #REALTIME_PRIORITY_CLASS)
SetThreadPriority_(GetCurrentThread_(), #THREAD_PRIORITY_TIME_CRITICAL)


#code = 0  ; While ... Wend
;#code = 1  ; For ... Next

Global.q t1, t2, t3, t4, t5, t6, maxfreq
Global.d d1, d2, d3
Global howmannytimes, k


howmannytimes = 10000000

QueryPerformanceFrequency_(@maxfreq)


!align 4
!align 4
!align 4
!align 4
QueryPerformanceCounter_(@t1)
!align 4
!align 4
!align 4
!align 4
CompilerIf #code
  For k = 1 To howmannytimes
  Next
CompilerElse
  k = 0
  While (k < howmannytimes)
    k = k + 1
  Wend
CompilerEndIf
QueryPerformanceCounter_(@t2)

!align 4
!align 4
!align 4
!align 4
QueryPerformanceCounter_(@t3)
!align 4
!align 4
!align 4
!align 4
CompilerIf #code
  For k = 1 To howmannytimes
  Next
CompilerElse
  k = 0
  While (k < howmannytimes)
    k = k + 1
  Wend
CompilerEndIf
QueryPerformanceCounter_(@t4)

!align 4
!align 4
!align 4
!align 4
QueryPerformanceCounter_(@t5)
!align 4
!align 4
!align 4
!align 4
CompilerIf #code
  For k = 1 To howmannytimes
  Next
CompilerElse
  k = 0
  While (k < howmannytimes)
    k = k + 1
  Wend
CompilerEndIf
QueryPerformanceCounter_(@t6)


t2 - t1
t4 - t3
t6 - t5

d1 = t2 / maxfreq * 1000
d2 = t4 / maxfreq * 1000
d3 = t6 / maxfreq * 1000

CompilerIf #code
  MessageRequester("Result: For ... Next", StrD(d1) + " ms" + #LF$ + StrD(d2) + " ms" + #LF$ + StrD(d3) + " ms")
CompilerElse
  MessageRequester("Result: While ... Wend", StrD(d1) + " ms" + #LF$ + StrD(d2) + " ms" + #LF$ + StrD(d3) + " ms")
CompilerEndIf

[/size]

My result:
(CPU: AMD Athlon II X4 630 Propus 2.8GHz)

10,000,000 loops of "While ... Wend" ; 23 ~ 28 ms
10,000,000 loops of "For ... Next" ; 24 ~ 26 ms

Thus, the difference is negligible.

so my advice is : use while instread for in ANY language!

My conclusion is that the "While" and "For" are almost the same speed in PureBasic.
But I don't know about the others.

---------------------------
Result: While ... Wend
---------------------------
26.4243081174 ms
23.8658823957 ms
28.5692734691 ms

23.3605109029 ms
28.5634068017 ms
26.8682192849 ms

28.5983274411 ms
23.3795077307 ms
25.2926000372 ms

24.5992158221 ms
28.5745814063 ms
23.6133363319 ms

24.4567396136 ms
25.4708349804 ms
27.0366764491 ms

24.8836095090 ms
24.8872412555 ms
25.7912667671 ms



---------------------------
Result: For ... Next
---------------------------
24.9914444434 ms
26.5922065514 ms
24.9819460295 ms

25.0149111130 ms
26.6491970348 ms
24.1136792525 ms

24.7863904491 ms
26.5977938537 ms
24.3670634117 ms

25.1557111309 ms
26.6821621184 ms
24.7327523470 ms

24.5000412064 ms
26.6033811560 ms
24.7665555259 ms

24.7307967912 ms
26.5980732188 ms
24.4743396158 ms