Statistic of 320x200 autopilot run

Statistics: frames 9337
mkrealloctable: added 22167, symetry 613
calculate loop: tocalculate 5241302 avoided 3186884 mirrored 0
calculate:calculated 2054418 inside 300456

This means:
Aprox number of calculated pixels for every frame:220 (0.34%)
14% (1/6) of calculated pixels lies inside set
Aprox number of new lines/columns for every frame:2.37
Optimizations:
Aproximation :  99.49% of calculations
Symetry saved:  02.76% of remaining
Solid guessing: 60.80% of remaining
mirroring2:     0%     (no efective from mset. Slows down when active. Will be disabled)

Statistic of 640x480 autopilot run

Statistics: frames 2433
mkrealloctable: added 47405, symetry 1173
calculate loop: tocalculate 24968722 avoided 16780447 mirrored 0
calculate:calculated 8188275 inside 1122047

This means:
Aprox number of calculated pixels for every frame:3364 (01.09%)
13% (1/6) of calculated pixels lies inside set
Aprox number of new lines/columns for every frame:19.48
Optimizations:
Aproximation :  98.46% of calculations  (framerate is lower, step bigger..
                                         so this is normal)
Symetry saved:  02.47% of remaining
Solid guessing: 67.20% of remaining
mirroring2:     0%     (no efective from mset)

This shows that XaoS's optimization seems to perform well on most cases.
Solid guessing also perform as well as normally boundary detection (during
tests I did in xmand..)
Number of calculated points inside set seems to be low(1/6)
so I did following test:

Number of calculations spent inside/outside:
inside: 165858075
outside:46113860
3.5 times more inside than outside

This shows that some detection of points inside set will be helpfull..
It shoud slow down loop more than 2 times in case it will be sucessfull..


320x200 autopilot run profiles:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 51.58    261.31   261.31  5084736     0.05     0.05  mand_calc	 
            calculation loop still takes most of time
 39.96    463.79   202.48    45338     4.47     4.47  mkrealloc_table
            This function is not big problem since it takes too
	    much time just when called too often(framerate is high)
  3.87    483.42    19.63    22669     0.87    12.40  calculatenew
            Solid guesing if prety fast
  1.15    489.24     5.82    22669     0.26     0.26  moveoldpoints
            All this functions are not interesting for optimizing
  1.14    495.01     5.77    22669     0.25    21.87  do_fractal
  1.13    500.75     5.74    11018     0.52     0.52  do_autopilot
  <SNIP>

Tests how much spent by calling mand_calc function....
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  us/call  us/call  name    
 72.22      2.34     2.34                             mcount
 18.83      2.95     0.61        1 610000.00 900000.00  main
  8.95      3.24     0.29  5000000     0.06     0.06  test

  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 40.76      2.14     2.14                             mcount
 40.57      4.27     2.13  5000000     0.00     0.00  test
 18.67      5.25     0.98        1   980.00  3110.00  main

not more than 5 second of 280 (01.78%) This shows that I don't have to
optimize loop by integrating calculation loops..so I can keep clean
functions for calculating..

Calculating of new screen 320x200

Flat profile:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  us/call  us/call  name    
 60.87      0.14     0.14    20221     6.92     6.92  mand_calc
 34.78      0.22     0.08        1 80000.00 220000.00  calculatenew
 	Solid guessing is slow here, because of all optimizations
	done for zooming. Thats why XaoS is not very effective for
	calculating of new images. Maybe some special solid guessing
	loop should be done.
  4.35      0.23     0.01        2  5000.00  5000.00  mkrealloc_table
  <SNIP>

640x480 autopilot run


Flat profile:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 83.05    427.86   427.86  8480091     0.05     0.05  mand_calc
 	Framerate is lower, step is higher so calculation takes
	more time..
  8.72    472.78    44.92     4480    10.03    10.03  mkrealloc_table
  	As I said..does not take significant amount of time..because framerate
	is lower..
  6.11    504.24    31.46     2240    14.04   205.11  calculatenew
  	also this went up..
  0.83    508.53     4.29                             mcount
  0.58    511.51     2.98     5319     0.56     0.56  do_autopilot
  0.40    513.58     2.07     2240     0.92     0.92  moveoldpoints
  0.24    514.84     1.26     2240     0.56   226.67  do_fractal
  0.02    514.93     0.09    62109     0.00     0.00  tl_process_group
  <SNIP>

640x480 start:

Flat profile:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 71.43      0.75     0.75    90444     0.01     0.01  mand_calc
 24.76      1.01     0.26        1   260.00  1010.00  calculatenew
 	very similiar to 320x200 sample
  2.86      1.04     0.03                             mcount
  0.95      1.05     0.01        2     5.00     5.00  mkrealloc_table
  <SNIP>