editing zoom in view

[thunderchero]

Looks very peaceful (no invading fleets in other territories). Are you sure you develop the right mod (MUM)?

It is so hard to tell, in longer tests the AI would start eliminating the weakest. (once AI meets turn processing increases so much it is hard to test further)

I am starting to look at patch benefits vs turn processing on patches. here is an example what I am looking at.

my testing baseline
MUM, large, 40% density, impossible, max minors, borg and random off
I set test time to 45 min (baseline was 190 turns)
all tests use same starting seed

when I add this patch, Game will only complete 185 turns in 45 min

Code: Select all

NAME: Enabling Rebellions without Random Events
AUTHOR: Spocks-cuddly-tribble
DESC: This is the Enabling Rebellions without Random Events
DESC: patch file name: Enabling Rebellions without Random Events.patch
URL: 
TAG: 

# >>  = vanilla/original value

# <<  = new value

# Enabling Rebellions without Random Events

>> 0x0004b290 cc be 44
>> 0x0004b294 e1 be 44
>> 0x0004b298 d1 be 44
>> 0x0004b29c f6 be 44
>> 0x0004b2a0 0b bf 44
>> 0x0004b2a7 1e 00
>> 0x0004b2ba 66 3d 04 00
>> 0x0004b2c0 25 ff ff 00 00 ff 24 85 90 be 44 00 ba 2d 00
>> 0x0004b2d0 00
>> 0x0004b400 a3 c1 44
>> 0x0004b404 c4 c1 44
>> 0x0004b408 cb c1 44
>> 0x0004b40c d2 c1 44
>> 0x0004b410 d9 c1 44
>> 0x0004b596 11
>> 0x0004b59c ff 24
>> 0x0004b5a3 b9 2d 00 00 00 3b 0d 18 29 5a 00 0f 87 66 ff ff ff
>> 0x00057741 74 0f

<< 0x0004b290 2d 00 00
<< 0x0004b294 23 00 00
<< 0x0004b298 1e 00 00
<< 0x0004b29c 14 00 00
<< 0x0004b2a0 0a 00 00
<< 0x0004b2a7 10 27
<< 0x0004b2ba 8a c8 3c 04
<< 0x0004b2c0 80 3d 46 2b 5a 00 00 74 08 8b 14 8d 90 be 44
<< 0x0004b2d0 90
<< 0x0004b400 2d 00 00
<< 0x0004b404 23 00 00
<< 0x0004b408 1e 00 00
<< 0x0004b40c 14 00 00
<< 0x0004b410 0a 00 00
<< 0x0004b596 22
<< 0x0004b59c 8b 0c
<< 0x0004b5a3 3b 0d 18 29 5a 00 77 0e 80 3d 46 2b 5a 00 00 74 05
<< 0x00057741 90 90

so to me this patch would not be used...

[Spocks-cuddly-tribble]

I am starting to look at patch benefits vs turn processing on patches. here is an example what I am looking at.

when I add this patch, Game will only complete 185 turns in 45 min

- for repeatable test like this remove the 'BotF random engine fix' if used: viewtopic.php?p=56905#p56905
- do you have system events disabled for AI @ 8: viewtopic.php?p=24589#p24589

I did not expect sub_44BF30 (check empire systems for events) is cpu heavy....

The rebellion bias fix is needed to make the game harder when hitting low morale (default just turn off random events and your systems virtually never rebel). In ECM the federation end-game depends on this fix (to compensate their tech/economy advantages), for MUM not that of an issue.


Another unneeded code is sub_458720 write_TURNSTATUS (all calls should be nop-ed) but this would be a massive effort/patch with little effect.


A big problem for tactical combat lag could be my new AI commands. So you might be forced to make the massive AI fleets in MUM stupid again....

[thunderchero]

- for repeatable test like this remove the 'BotF random engine fix' if used: viewtopic.php?p=56905#p56905
- do you have system events disabled for AI @ 8: viewtopic.php?p=24589#p24589

neither patch is installed on MUM.

I was also thinking I should confirm on vanilla baseline test, not MUM

[Spocks-cuddly-tribble]

I am starting to look at patch benefits vs turn processing on patches.

so to me this patch would not be used...

I hear you and I'm with you. If possible my patches always try to make the game faster not slower (even when adding extra features).

I updated the patch in question: viewtopic.php?p=18755#p18755

Here is deviation only: (should be about 100 times faster + disables system events for AI systems = no extra patch needed)

Code: Select all

NAME: Enabling Rebellions without Random Events (ECM deviation update)

>> 0x0004b35b 23
>> 0x0004b35d 50
>> 0x0004b370 74

<< 0x0004b35b 22
<< 0x0004b35d 28
<< 0x0004b370 75


# 0044BF5B     2215 282B5A00    AND DL, [5A2B28] // player empires bitmask

# 0044BF70     75 34            JNZ SHORT 44BFA6 // is player (and alive)

[thunderchero]

I hear you and I'm with you. If possible my patches always try to make the game faster not slower (even when adding extra features).

I updated the patch in question: viewtopic.php?p=18755#p18755

Here is deviation only: (should be about 100 times faster + disables system events for AI systems = no extra patch needed)

Code: Select all

NAME: Enabling Rebellions without Random Events (ECM deviation update)

>> 0x0004b35b 23
>> 0x0004b35d 50
>> 0x0004b370 74

<< 0x0004b35b 22
<< 0x0004b35d 28
<< 0x0004b370 75


# 0044BF5B     2215 282B5A00    AND DL, [5A2B28] // player empires bitmask

# 0044BF70     75 34            JNZ SHORT 44BFA6 // is player (and alive)

this will either, drive us crazy or we will figure something out.

that patch must be effected by some other patch.

vanilla controlled test only applying that patch (original patch version) had no effect on turn processing.

mum controlled test with new patch 2 less turns. (it was worse)

vanilla controlled test
800 with large map patch and Dynamic Widescreen Zoom-In only (did not apply AI tweaks)
30 min 388 turns

I am starting to get a headache.......

edit; I need to run MUM test again... my bad, I might have had extra patch applied.

[Spocks-cuddly-tribble]

Here a fix to relocate stranded ships when betrayed by treaty partners (end/break treaty AI protection workaround):

[EDIT: Updated again!]

Move Fleets due to Treaty Fix for Contested Sectors & Stranded Fleets updated2.patch

(5.17 KiB) Downloaded 12 times

Outdated patch to remove: (or just use new patch and ignore conflict)

Move Fleets due to Treaty Fix Updated.patch

(5.11 KiB) Downloaded 31 times

Code: Select all

NAME: Move Fleets due to Treaty Fix for Contested Sectors & Stranded Fleets (updated #2)
DESC: Fixes the closed border check for contested sectors and moves fleets stranded due to broken treaties.
AUTHOR: Spocks-cuddly-tribble
URL: https://www.armadafleetcommand.com/onscreen/botf/viewtopic.php?p=57729#p57729

>> 0x00048139 31 d2
>> 0x0004813c 94
>> 0x00048142 3C 20 0F 83 39 02 00 00 88 C1 B8 01 00 00 00 D3 E0 8B 35 50 2B 5A 00 89 C1 21 F1 0F 84 3D 02 00 00 31 C0 8B 40 1C E8 93 2D 0D 00 89 C3 85 C0 0F 84 75 00 00 00 8A 8C 24 18 02 00 00 0F BF 03 31 D2 E8 98 0A 00 00 83 C0 38 31 D2 E8 1E 4D FF FF 8A 50 0E B8 01 00 00 00 D3 E0 39 C2 75 3F 83 BC 24 00 02 00 00 00 75 10 31 C2 31 C0 E8 BD 26 0D 00 89 84 24 00 02 00 00 BA 02 00 00 00 8B 84 24 00 02 00 00 E8 E5 24 0D 00 66 8B 13 66 89 10 89 C2 8B 84 24 00 02 00 00 E8 41 2C 0D 00 89 D8 E8 BA 29 0D 00 89 C3 85 C0 75 92
>> 0x0005df49 85 c0 75 1c bb b4 a4 57 00 ba 09 02 00 00 b8 90 a4 57 00 e8 4f 96 0a 00 85 c0 74 04 85 c9 74 e4 31 d2 66 8b
>> 0x0005e021 83 c4 74 5e 5a
>> 0x0005e06e 83 c4 74 5e 5a
>> 0x0018e218 20 ee

<< 0x00048139 33 c9
<< 0x0004813c 8c
<< 0x00048142 8A C8 B5 01 D2 E5 22 2D 50 2B 5A 00 0F 85 23 02 00 00 C1 E0 06 8B 80 2C 20 5A 00 E8 9E 2D 0D 00 85 C0 0F 84 0D 02 00 00 8B D8 8A CA 0F BF 03 33 D2 E8 A8 0A 00 00 80 78 2C 07 74 61 80 78 50 01 7D 5B 8D 50 38 52 E8 93 1E 03 00 85 C0 58 74 0E E8 19 4D FF FF B2 01 D2 E2 84 50 0E 74 3F 83 BC 24 00 02 00 00 00 75 10 33 D2 31 C0 E8 BD 26 0D 00 89 84 24 00 02 00 00 BA 02 00 00 00 8B 84 24 00 02 00 00 E8 E5 24 0D 00 66 8B 13 66 89 10 89 C2 8B 84 24 00 02 00 00 E8 41 2C 0D 00 89 D8 E8 BA 29 0D 00 89 C3 85 C0 75 82
<< 0x0005df49 eb 20 e8 20 cc 01 00 0f b6 41 36 50 52 e8 c5 a1 fe ff 58 5a e8 be a1 fe ff 83 c4 74 5e 5a 59 5b c3 90 0f b6
<< 0x0005e021 e9 25 ff ff ff
<< 0x0005e06e e9 d8 fe ff ff
<< 0x0018e218 30 eb


# sub_448D20 Move ships due to treaty (default peace & friendship between empires) 

# Fix move fleets from contested sectors and stranded (not if TF order 'build outpost' or with colony ships)

#00448D39     33C9              XOR ECX,ECX
#00448D3B     898C24 00020000   MOV [ESP+200],ECX

#00448D42     8AC8              MOV CL,AL
#00448D44     B5 01             MOV CH,1
#00448D46     D2E5              SHL CH,CL
#00448D48     222D 502B5A00     AND CH,[5A2B50]  // dead_races_BitMask
#00448D4E     0F85 23020000     JNZ 448F77     // exit race dead
#00448D54     C1E0 06           SHL EAX,6
#00448D57     8B80 2C205A00     MOV EAX,[EAX+5A202C]  // alienTFInfo
#00448D5D     E8 9E2D0D00       CALL 51BB00
#00448D62     85C0              TEST EAX,EAX
#00448D64     0F84 0D020000     JE 448F77     // exit no ships
#00448D6A     8BD8              MOV EBX,EAX
#00448D6C     8ACA              MOV CL,DL      // race territory
#00448D6E     0FBF03            MOVSX EAX,WORD [EBX]  //   // TF ID
#00448D71     33D2              XOR EDX,EDX
#00448D73     E8 A80A0000       CALL 449820          // TF entry
#00448D78     8078 2C 07        CMP BYTE [EAX+2C],7  // build outpost
#00448D7C     74 61             JE SHORT 448DDF      // don't move
#00448D7E     8078 50 01        CMP BYTE [EAX+50],1  // greatest colony ship production value
#00448D82     7D 5B             JGE SHORT 448DDF     // contains colony ships -> don't move
#00448D84     8D50 38           LEA EDX, [EAX+38]    // TF coordinates
#00448D87     52                PUSH EDX
#00448D88     E8 931E0300       CALL 47AC20  // check range stranded
#00448D8D     85C0              TEST EAX,EAX
#00448D8F     58                POP EAX  // TF coordinates
#00448D90     74 0E             JE SHORT 448DA0  // is stranded
#00448D92     E8 194DFFFF       CALL 43DAB0  // sector.lst entry
#00448D97     B2 01             MOV DL,1
#00448D99     D2E2              SHL DL,CL  // race territory
#00448D9B     8450 0E           TEST [EAX+E],DL  // check claim
#00448D9E     74 3F             JE SHORT 448DDF  // no claim of other race don't move TF

#00448DAA     33D2              XOR EDX,EDX

#00448DEA    ^75 82             JNZ SHORT 448D6E  // loop all TFs from alienTFInfo


#59040C treaty_addresses 0x18E218 20 EE -> 30 EB = share sub_45EB30 alliance_break_END for sub_45EE20 peace_break_END

#sub_45EB30 emp_alliance_break_END

#0045EB49     EB 20               JMP SHORT 45EB6B
#0045EB4B     E8 20CC0100         CALL 47B770   // PathFindGridMapTurnUpdate
#0045EB50     0FB641 36           MOVZX EAX,BYTE [ECX+36]
#0045EB54     50                  PUSH EAX
#0045EB55     52                  PUSH EDX
#0045EB56     E8 C5A1FEFF         CALL 448D20   // Move_ships_al_from_dl_due_to_treaty + stranded
#0045EB5B     58                  POP EAX
#0045EB5C     5A                  POP EDX
#0045EB5D     E8 BEA1FEFF         CALL 448D20   // Move_ships_al_from_dl_due_to_treaty + stranded
#0045EB62     83C4 74             ADD ESP,74
#0045EB65     5E                  POP ESI
#0045EB66     5A                  POP EDX
#0045EB67     59                  POP ECX
#0045EB68     5B                  POP EBX
#0045EB69     C3                  RETN
#0045EB6A     90                  NOP
#0045EB6B     0FB651 34           MOVZX EDX,BYTE [ECX+34]

#0045EC21    ^E9 25FFFFFF         JMP 45EB4B

#0045EC6E    ^E9 D8FEFFFF         JMP 45EB4B

[thunderchero]

Not sure how far I will take these tests, but found a very good patch.

The "Auto-Upgrade Fix for AI Ships Under Construction"

test control variables same as before
vanilla 800 with large map patch and Dynamic Widescreen Zoom-In only (did not apply AI tweaks)
30 min 388 turns

with auto patch applied I got to turn 388, 8 min and 55 sec quicker. I ended test at that turn due to I meet romulans.
(funny it picked that turn to find me)

here is screenshot of both maps
with patch

w_patch.jpg (384.82 KiB) Viewed 1997 times

without patch

wo_patch.jpg (389.05 KiB) Viewed 1997 times

The patch will still cause a 'slight' slowdown, but a human shouldn't notice it (even with MANY systems max hundredths of a second each turn).

tested new patch and no noticeable change.

[Spocks-cuddly-tribble]

Since the lazy AI in your above MUM game is not invading you might try this patch: (increase check flight turns away value)

Code: Select all

NAME: Invade check armed TskSh max X fight turns away (2->3)

>> 0x1DD6F 02
<< 0x1DD6F 03

# 0041E96E           push  2      ; invade fight turns away -> cf attack 3 0x1DD6F

The AI turn lag comes from idle/unused AI ships, so this might even help with this....

found a very good patch. with auto patch applied I got to turn 388, 8 min and 55 sec quicker.

In fact this is a very bad exploit workaround in a semi broken code (massive unneeded calculations loop).
But it's only used when researching new ships (not often = no turn-lag source). Same with AI fleet buildup fix (or my above treaty fix).

In your setup a lag-patch can pseudo-increase game speed and a faster code might slow you down.
Controlled = remove all pseudo random features -> too much work in BotF.
All patches changing game behavior (even random row-seeds) will distort any test results (too many in-game variables changed).
So you might get counterintuitive/biased results.

Even a single not upgraded AI ship might have the pontential to change the entire game time-line of subsequent events (plus all random value rows).

tested new patch and no noticeable change.

It skips all AI systems, so it can't cause any lag for just one player system, but it might change random value rows used by all random features (AI diplomacy, even basic stuff like pop growth).

[thunderchero]

In your setup a lag-patch can pseudo-increase game speed and a faster code might slow you down.
Controlled = remove all pseudo random features -> too much work in BotF.
All patches changing game behavior (even random row-seeds) will distort any test results (too many in-game variables changed).
So you might get counterintuitive/biased results.

Even a single not upgraded AI ship might have the pontential to change the entire game time-line of subsequent events (plus all random value rows).

first 2 line is over my pay grade (don't even bother to explain further)

I do understand patches change behavior, this goes back to my first comment "patch benefits vs turn processing"

the main thing I am looking at is if it increase turn processing that is when I look at "patch benefits vs turn processing"

btw I have maybe checked 20 or so patches so far and most have no change.

last test I went crazy and just added most of the "AI" patches (14 patches at once) and it changed behavior so much romulans found me 24 turns early in only 12 min

before that I wanted to test my patches "xy and build queue" and glad to see zero effect on behavior or turn processing.

[Spocks-cuddly-tribble]

first 2 line is over my pay grade (don't even bother to explain further)

I do understand patches change behavior,

It's not that hard to understand. Just note it's not only about the patch effects we player notice.

Computers are bad with random values, in the end they are all deterministic.
BotF is very bad, the used trick to simulate randomness is changing internal random seed rows each time a random value is generated e.g. all calls of:

507E50 Random_float_0_1_to_st0
507FF0 RandomInteger_to_eax

That's why you get always the same results when loading the same (unmodded) saved game and end X turns without player inputs.

Funny thing is unimportant stuff like random speech output in tactical combat (IIRC 3 choices) uses a better random engine by reading external system time (my random engine fix used the same idea).


Example for you test bias issue:

Even my updated rebellions fix generates each turn at least one random value changing the seeds via sub_507FF0. This value has no effect (unless your systems are rebellious) but all later random calls use changed seeds, at some point this affects all game features. Meaning you see a different game in your tests. Might be faster or slower depending on how well the AI does (ships, systems, range, diplomacy) but this doesn't mean the patch causes a significant game lag change.

xy and build queue -> does not affect real game behavior or use random features (just player gui data) = no change of later random features.


In the end we cause the turn lag by succeeding with:

- making the AI better in terms of obtaining more assets to compete with players
- creating a balance of power between the empires (see your MUM game) = hardest situation for AI codes

[thunderchero]

I did a little more testing on MUM.

MUM even during the first 60 sec of turn processing testing is slower.

I am starting to think it has something to do with my irregular galaxy code (when remove turn processing first 60 sec is much better)

problem is I thought that code was only used during galaxy generation.

[Spocks-cuddly-tribble]

The map size/shape in terms of number/density/positioning of star systems slows all larger maps down from the beginning (many AI codes scan the entire map each turn - see AI map data types). So ring galaxies from your larger maps oftentimes start a bit faster due to fewer stars and more restricted expansion directions. But worst part is the ship tasks after first contact and after share-range treaties.

The null_sub before AI_Units_Assign_Task and the pointless unfinished code in ships_to_tasks loop (nops) shows where Microprose ran out of time:

AI_empires_main_codes.png (16.64 KiB) Viewed 1852 times

ships_to_tasks_loop.png (25.31 KiB) Viewed 1852 times

There is still hope for a workaround to keep larger maps end-games semi playable with max 30-45 sec lag. But all previous patches caused inacceptable side effects. Truth to be told, I'm still surprised the BotF main code can handle larger galaxies and thousands of ships (no limit for dynamic core files and memory allocation).


Here is two new AI patches:

Code: Select all

NAME: Cancel Invade Fix for Ally Systems
DESC: 
AUTHOR: Spocks-cuddly-tribble
URL: 

>> 0x0001e0ca 8b 1d
>> 0x0001e0d0 01 d8 66 83 78 44 ff 74 0e 31 db 8a 58 4c 31 c0 66 89 f0 39 c3

<< 0x0001e0ca 03 05
<< 0x0001e0d0 66 83 78 44 ff 74 10 52 8b 50 4c 8b c6 e8 8e 12 01 00 5a 3C 01

#0041ECCA     0305 C8365A00      ADD EAX, [5A36C8]
#0041ECD0     66:8378 44 FF      CMP WORD [EAX+44],0FFFF
#0041ECD5     74 10              JE SHORT 41ECE7
#0041ECD7     52                 PUSH EDX
#0041ECD8     8B50 4C            MOV EDX, [EAX+4C]
#0041ECDB     8BC6               MOV EAX, ESI
#0041ECDD     E8 8E120100        CALL 42FF70
#0041ECE2     5A                 POP EDX
#0041ECE5     3C 01              CMP AL,1

Code: Select all

NAME: Engage order for AI fleets moving from staging point to invade, patrol, attack or harass target (default evade).
DESC: 
AUTHOR: Spocks-cuddly-tribble
URL: 

>> 0x0001d97a 31 db
>> 0x0001d97f 53 89 9c 24 40 01

>> 0x00033cb8 8b 44
>> 0x00033cbc 50 57

>> 0x00034fcb 52 8b 46 7c 50

>> 0x00035ed6 8b 54
>> 0x00035eda 52 31 c0 50


<< 0x0001d97a 6a 01
<< 0x0001d97f c6 84 24 40 01 00

<< 0x00033cb8 ff 74
<< 0x00033cbc 6a 01

<< 0x00034fcb 6a 01 ff 76 7c

<< 0x00035ed6 ff 74
<< 0x00035eda 6a 01 33 c0

>> 0x00008b2c ba 01 00 00 00 0f bf 01 89 79 54 e8 24 21 02 00 a8 80 74 07 c7 41 68 01 00 00 00
<< 0x00008b2c 42 89 79 54 8b 01 e8 29 21 02 00 a8 05 74 04 c6 41 2c 05 a8 80 74 04 c6 41 68 01

#invade
#0041E57A     6A 01                  PUSH 1
#0041E57F     C68424 40010000 00     MOV BYTE [ESP+140],0
#patrol
#004348B8     FF7424 14              PUSH DWORD [ESP+14]
#004348BC     6A 01                  PUSH 1
#attack
#00435BCB     6A 01                  PUSH 1
#00435BCD     FF76 7C                PUSH DWORD [ESI+7C]
#harass
#00436AD6     FF7424 18              PUSH DWORD [ESP+18]
#00436ADA     6A 01                  PUSH 1
#00436ADC     33C0                   XOR EAX,EAX

#0040972C     42              INC EDX
#0040972D     8979 54         MOV DWORD [ECX+54],EDI
#00409730     8B01            MOV EAX, [ECX]
#00409732     E8 29210200     CALL 42B860
#00409737     A8 05           TEST AL,5   // TF with colony or TT
#00409739     74 04           JZ SHORT 40973F
#0040973B     C641 2C 05      MOV BYTE [ECX+2C],5  // set evade
#0040973F     A8 80           TEST AL,80   // can cloak
#00409741     74 04           JE SHORT 409747
#00409743     C641 68 01      MOV BYTE [ECX+68],1  // set cloaked