Just another Purebasic Optimizer - New release - JaPBo v1.14

[Deeem2031]

I finished Version 1.05:

What's new? Nothing, it's just a bug-fix release.

edit:

Version 1.06:

-replace SYS_StringEqual call with a faster routine if possible

that means all commands like "If String.s" will run faster

e.g.:

Code: Select all

Procedure a(x) 
  Protected a,b,c 
  a = x 
EndProcedure 

#r = 40000000 

s.s = "" 

st1 = ElapsedMilliseconds() 
For i = 1 To #r 
  a(2) 
  If s 
  EndIf 
Next 
t1 = ElapsedMilliseconds()-st1 

MessageRequester("",Str(t1))

without opt: ca. 1030 ms
with opt: ca. 375 ms

[ebs]

Deeem2031,

I think I've found a bug in your latest optimizer (v1.05). When compiling when program, I get the following error:

---------------------------
jaPBe - Assembler error
---------------------------
Opt_Purebasic.asm [1168]:
push dword _s4
error: undefined symbol.

---------------------------
OK
---------------------------

Looking through the file Opt_Purebasic.asm, I noticed that every other reference is "_S4" - with an uppercase S, not a lowercase s.
Just this one line is "_s4". Doesn't FAsm consider "_s4" and "_S4" different variables?

I found one other occurrence of this. Most of the time, the asm statement is:
PUSH dword [PB_StringBase]
but a few times it is:
PUSH dword [pb_stringbase]

Again, notice the upper- vs. lowercase issue.

I edited the Opt_Purebasic file to change all these items to uppercase, and I no longer get any errors. Can you check into this and see if this is a bug?

Thanks,
Eric

[Deeem2031]

Yes, this is a bug but i can't locate them because I haven't got a code which produce this error. So it would be helpful, if you post a code or, if you don't want to, the log-file.

[ebs]

Deeem2031,

The log file is over 130 KB. Can you give me an e-mail address, so I can send it to you? If it would help, I can post just the parts that show the bug.

Regards,
Eric

[Deeem2031]

The part that shows the bug should be enouth to fix it. I only need the info in which routine the bug is. Something like "-> replace ClearLoop (Version 1.0)" before the bug appears, show this.

Btw. the email-addr mostly stand under the posts of the user. -> webmaster@deeem2031.de

[ebs]

Btw. the email-addr mostly stand under the posts of the user. -> webmaster@deeem2031.de

Of course - I should have noticed that!

Here's one section of the log file that changes "_S4" to "_s4":

Code: Select all

>Found _procedure120
-> replace ClearLoop (Version 2.0)
-> replace Esi with Esp (RemoveCheck) (Version 1.0)
--> Found Esi-Command: "push esi"
--> Found Esi-Command: "pop esi"
-> remove unused register (Version 1.1)
--> check register ebx
---> register won't be used -> remove it
Remove: " PUSH ebx" (Pos: 8704832)
Remove: " POP ebx" (Pos: 8708648)
--> check register ecx
---> register is Ecx -> remove it
Remove: " PUSH ecx" (Pos: 8704880)
Remove: " POP ecx" (Pos: 8708600)
--> check register ebp
---> register won't be used -> remove it
Remove: " PUSH ebp" (Pos: 8704928)
Remove: " POP ebp" (Pos: 8708552)
--> check register esi
---> register won't be used -> remove it
Remove: " PUSH esi" (Pos: 8704976)
Remove: " POP esi" (Pos: 8708504)
--> check register edi
---> register won't be used -> remove it
Remove: " PUSH edi                                                                                                                                                                                                                                                           " (Pos: 8705024)
Remove: " POP edi" (Pos: 8708456)
Modify: " PUSH dword _S4" -> "push dword _s4" (Pos: 8705320)
Modify: " PUSH dword 268435456" -> "push dword 268435456" (Pos: 8705368)
Modify: " PUSH dword 374" -> "push dword 374" (Pos: 8705424)
Modify: " PUSH dword 350" -> "push dword 350" (Pos: 8705472)
Modify: " PUSH dword 95" -> "push dword 95" (Pos: 8705520)
Modify: " PUSH dword 375" -> "push dword 375" (Pos: 8705568)
-> removed 5 unused register
--> modify stack
-> replace Esi with Esp (Version 1.0)
--> Modify Esi-commands
--> Remove Esi-commands

and one showing the "PB_StringBase" problem:

Code: Select all

>Found _procedure44
-> replace ClearLoop (Version 2.0)
-> replace Esi with Esp (RemoveCheck) (Version 1.0)
--> Found Esi-Command: "push esi"
--> Found Esi-Command: "pop esi"
-> remove unused register (Version 1.1)
--> check register ebx
---> register won't be used -> remove it
Remove: " PUSH ebx" (Pos: 9098792)
Remove: " POP ebx" (Pos: 9100944)
--> check register ecx
---> register is Ecx -> remove it
Remove: " PUSH ecx" (Pos: 9098840)
Remove: " POP ecx" (Pos: 9100896)
--> check register ebp
---> register won't be used -> remove it
Remove: " PUSH ebp" (Pos: 9098888)
Remove: " POP ebp" (Pos: 9100848)
--> check register esi
---> register won't be used -> remove it
Remove: " PUSH esi" (Pos: 9098936)
Remove: " POP esi" (Pos: 9100800)
--> check register edi
---> register won't be used -> remove it
Remove: " PUSH edi                                                                                                                                                                                                                                                           " (Pos: 9098984)
Remove: " POP edi" (Pos: 9100752)
Modify: " PUSH dword [PB_StringBase]" -> "push dword [pb_stringbase]" (Pos: 9099280)
-> removed 5 unused register
--> modify stack
-> replace Esi with Esp (Version 1.0)
--> Modify Esi-commands
--> Remove Esi-commands

Regards,
Eric

[Deeem2031]

Thx, fixed version: www.deeem2031.de/PB/D_Optimizer.zip

[ebs]

Deeem2031,

Wow - that's what I call fast service!
I won't be able to try out the new version until tomorrow, but I'll let you know if it works on my program.

Thanks,
Eric

[ricardo]

For more infos just take a look to the german forum: http://forums.purebasic.com/german/view ... 6423#56423

And the people that don't understand german?????

Every day more, looks like need to learn german to participate in this forums

[va!n]

JaPBo v1.07 released!

@ricardo:
quote: "And the people that don't understand german?????"

For this people we have this english thread to give Deeem2031 any bugreport, wishes, feedback

[ebs]

Deeem2031,

I just tried v1.10 and my program compiled without error.

Thanks,
Eric

[ricardo]

JaPBo v1.07 released!

@ricardo:
quote: "And the people that don't understand german?????"

For this people we have this english thread to give Deeem2031 any bugreport, wishes, feedback

Yes, but my question was because the 'more info' link was in German.

I hope you can explian more detailed about this tool in English. Thanks.

[va!n]

@ricardo:
ahhh sorry.. now i understand i just only wrote it, because interested people can take a look and follow the german thread too! But all "important" news from the german forum will posted here in english too
So there should be nothing you must miss when dont take a look to the german thread.Hope this is ok for you.. .

[ricardo]

@ricardo:
ahhh sorry.. now i understand i just only wrote it, because interested people can take a look and follow the german thread too! But all "important" news from the german forum will posted here in english too
So there should be nothing you must miss when dont take a look to the german thread.Hope this is ok for you.. .

Thanks!!

[dmoc]

Edit: added times for default #N = 5999999

Great utils, thanks! But something strange trying them with "Test.pb". First, I'm on an old 233MHz notebook so I have reduced the loop counter to 1,000,000. Next I run different combo's of both optimizers and here are the results ("D" for Deeem's, "M" for Meier's)...

Code: Select all

M D  #N=1000000 #N=5999999
0 0   15215  90229
0 1   14995  90105
1 0    4740   29444
1 1    4909   29325

Notice last one increases Guess it will pay to experiment or is this just an anomaly?

PS: Also getting a dos window with the latest (so not seeing any progress bar)