Optimizer compile time constants problem

1
2
Page 1 of 2

Avra

Schreiberling

Gender:
Location: Belgrade, Serbia
Age: 53
Homepage: rs.linkedin.com/in…
Posts: 653
Registered: 07 / 2002

Subject:

Optimizer compile time constants problem

· Posted: 08.06.2014 - 23:55 · #1

Is it me missing something obvious or OptimiserRev and OptimiserBuild always show zero?

Code

WriteLn(Serout, 'OptimiserRev = ' + IntToStr(OptimiserRev));
WriteLn(Serout, 'OptimiserBuild = ' + IntToStr(OptimiserBuild));

F9, Ctrl+F9, and Ctrl+Shift+F9 give me the same result.

btw. CompilerBuild and all other compile time constants give correct result.

Merlin

Administrator

Gender:
Age: 24
Posts: 1409
Registered: 03 / 2005

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 10:34 · #2

Hi Avra.

Do you see the optimiser screen flash up when you run any of your options?

The values are substituted by the optimiser when it is run. If it is not run, the values are zero.

I believe that fields are also zeroed if the optimiser is run in smartlinker mode only, but I am not 100% sure about that. Perhaps Rolf can confirm?

Avra

Schreiberling

Gender:
Location: Belgrade, Serbia
Age: 53
Homepage: rs.linkedin.com/in…
Posts: 653
Registered: 07 / 2002

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 10:55 · #3

Quote by Merlin

Do you see the optimiser screen flash up when you run any of your options?

Yes, from the flash up screen I could see that optimizer has saved about 26%.
AvrCo v5.04.73

Gunter

Administrator

Gender:
Location: Frankfurt Main / Germany
Posts: 1697
Registered: 02 / 2003

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 11:25 · #4

Hi Avra and Merlin,

seems not to be a bug: my AVRco shows
OPTIMISERREV .EQU 0302h ; const
OPTIMISERBUILD .EQU 0B00h ; const
in 5.04.72 and in 5.04.73

Gunter

Merlin

Administrator

Gender:
Age: 24
Posts: 1409
Registered: 03 / 2005

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 11:31 · #5

Hi Avra.

I also have v5.04.73.

I tried compiling a demo and looking at the ASM files. These constants are set correctly. To confirm, if you look at the ASM file, are the variables set? If so, then I guess it must be a compiler issue. If not, if you can send me your project via pm I'll compile here and see what is going on.

btw.

Have you enough frame space? More efficient on frame is

Code

  Write(Serout, 'OptimiserRev   = ');
  WriteLn(SerOut IntToStr(OptimiserRev));
  Write(Serout, 'OptimiserBuild = ');
  WriteLn(SerOut, IntToStr(OptimiserBuild));

Avra

Schreiberling

Gender:
Location: Belgrade, Serbia
Age: 53
Homepage: rs.linkedin.com/in…
Posts: 653
Registered: 07 / 2002

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 12:28 · #6

Quote by Merlin

if you look at the ASM file, are the variables set?

This is from my ASM file:

Code

COMPILERREV             .EQU    1F8h            ; const
COMPILERBUILD_Y         .EQU    00Eh            ; const
COMPILERBUILD_M         .EQU    006h            ; const
COMPILERBUILD_D         .EQU    001h            ; const
COMPILEYEAR             .EQU    00Eh            ; const
COMPILEMONTH            .EQU    006h            ; const
COMPILEDAY              .EQU    009h            ; const
COMPILEHOUR             .EQU    00Bh            ; const
COMPILEMINUTE           .EQU    023h            ; const
PROJECTBUILD            .EQU    02Fh            ; const
OPTIMISERREV            .EQU      0302h         ; const
OPTIMISERBUILD          .EQU      0B00h         ; const

The only odd thing is a little different number format.

This is from serial simulator output:

Code

43 6F 6D 70 69 6C 65 72 52 65 76 20 20 20 20 3D
C o m p i l e r R e v =
20 35 30 34 0D 0A 43 6F 6D 70 69 6C 65 72 42 75
5 0 4 . . C o m p i l e r B u
69 6C 64 20 20 3D 20 30 31 2E 30 36 2E 32 30 31
i l d = 0 1 . 0 6 . 2 0 1
34 0D 0A 43 6F 6D 70 69 6C 65 54 69 6D 65 20 20
4 . . C o m p i l e T i m e
20 20 3D 20 30 39 2E 30 36 2E 32 30 31 34 20 31
= 0 9 . 0 6 . 2 0 1 4 1
31 3A 33 35 0D 0A 50 72 6F 6A 65 63 74 42 75 69
1 : 3 5 . . P r o j e c t B u i
6C 64 20 20 20 3D 20 34 37 0D 0A 4F 70 74 69 6D
l d = 4 7 . . O p t i m
69 73 65 72 52 65 76 20 20 20 3D 20 30 0D 0A 4F
i s e r R e v = 0 . . O
70 74 69 6D 69 73 65 72 42 75 69 6C 64 20 3D 20
p t i m i s e r B u i l d =
30 0D 0A 54 45 45 45 45 45 45 45 45 45 53 54 21
0 . . T E E E E E E E E E S T !

I have prepared the simplest project and I will send it to you and Rolf.

UPDATE:
Simple project sent to PM.

Quote

Have you enough frame space? More efficient on frame is...

That's the first thing I try. Sim shows that limit was never reached, and I have just tested your example with an empty project and stack and frame at 100, and the problem is still the same.

Attachments


Filename:	system administration controls.png
Filesize:	33.3 KB
Title:
Download counter:	88

Merlin

Administrator

Gender:
Age: 24
Posts: 1409
Registered: 03 / 2005

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 13:00 · #7

Hi Avra.

Thanks for the sources. I can see what the problem is, but it is an issue with the compiler, not the optimiser. I have contacted you and Rolf via pm to explain the issue. Hopefully Rolf can resolve it for you.

Administrator

Gender:
Location: Germany
Age: 24
Homepage: e-lab.de
Posts: 5558
Registered: 03 / 2002

Subject:

Re: Optimizer compile time constants problem

· Posted: 09.06.2014 - 14:53 · #8

Avra, Merlin,

sorry, there is no way to replace these zeros at compile time, because the
compiler doesn't know anything abot the Optimiser. See my pm.

rolf

1
2
Page 1 of 2

Optimizer compile time constants problem

Attachments

Registered users in this topic