Project

General

Profile

Actions

Emulator Issues #182

closed

disappearing enemies from OpenGL draw distance Starfox Assault (JIT only)

Added by the.ryoku.kitsune about 16 years ago.

Status:
Fixed
Priority:
Normal
Assignee:
-
Category:
JIT
% Done:

0%

Operating system:
N/A
Issue type:
Bug
Milestone:
Regression:
No
Relates to usability:
No
Relates to performance:
No
Easy:
No
Relates to maintainability:
No
Regression start:
Fixed in:

Description

What steps will reproduce the problem?

  1. get to [boss] cut scene in Starfox assault
  2. skip/play cut scene models will be fine
  3. on player control models are distorted [bones/rigs or animations
    corrupted but used]
  4. second boss [first level] off center or out of place (clipping threw the
    ground) impossible to beat
  5. after an enemy (boss or level enemy) gets to far away into draw distance
    fade they disappear completely and do not return

What is the expected output? What do you see instead?
Expected to see the models as they appeared on the console. using the
emulator the boss models (presumably complex rigs/animations) boss models
are distorted or vanish for no reason

What version of the product are you using? On what operating system?
Dolphin SVC r 438 X64, Windows Xp x64

Please provide any additional information below.

Issues discovered while playing Starfox: Assault
Due to the nature of the Open GL plug in

A) when enemies get to far away into the opacity distance falloff they
cease to exist
B) There are flickering models when something is just outside the drop off
mark.
C) If a model goes outside the [scripted] drop off to far they cease to be
a game entity, IE boss models moving far into distance disappear.

I recommend adding a user adjustable draw distance, for performance and
stability of games.


Related issues 3 (0 open3 closed)

Has duplicate Emulator - Emulator Issues #457: Starfox Assault Gamestopping BugsDuplicate

Actions
Has duplicate Emulator - Emulator Issues #979: EFB overflow/leak on StarFox AssaltDuplicate

Actions
Has duplicate Emulator - Emulator Issues #2087: Star Fox Assault first boss disappearsDuplicate

Actions
Actions #1

Updated by hrydgard about 16 years ago

I think most of these problems are caused by OptimizeQuantizers = True. Try it with
False.

Actions #2

Updated by the.ryoku.kitsune about 16 years ago

negative on that, i forgot to mention that i tried it in both boolean values.

Actions #4

Updated by Leoharantes almost 16 years ago

Well, the game worked fine with the optimize quantizer = false in the older versions
like 177, but since about rev 300 or less the problem remains, with that setting
turned on or off

Actions #5

Updated by MrMister1991 almost 16 years ago

The issue still persists in Revision 1590, haven't tried Starfox Assault in the
latest revisions yet but I assume the same problem exists in those as well until
someone reports it. AKA Boss still impossible to defeat.

Actions #6

Updated by lpfaint99 almost 16 years ago

Issue 457 has been merged into this issue.

Actions #7

Updated by Xtreme.Starfox over 15 years ago

Optimize Quantizers Off Fixes the Aparoid clipping into the ground Problem and also
the On Foot Missions Player Skeleton, but it still doesn't fix it from Vanishing
later as it gets too far from the Drawing Distance, so it's still broken there.

I attached a Picture of Fixed On foot playing in Multiplayer.

Actions #8

Updated by ChaosCode about 15 years ago

seems to be jit related as the IL version works.

Actions #10

Updated by Anonymous over 14 years ago

Issue 979 has been merged into this issue.

Actions #11

Updated by Anonymous over 14 years ago

This is also very old. Any chance it has been fixed?

Actions #12

Updated by nash679 over 14 years ago

I will check this out sometime

Actions #13

Updated by skidau almost 14 years ago

Issue 2087 has been merged into this issue.

Actions #14

Updated by Anonymous almost 14 years ago

Please respond if this issue is still valid, or it will be closed.

Actions #15

Updated by NeoBrainX almost 14 years ago

  • Status changed from New to Fixed

Already asked for re-confirmation, got no response => marking fixed...

Actions #16

Updated by skidau almost 14 years ago

  • Status changed from Fixed to Accepted
  • Category set to jit
  • Operating system N/A added
Actions #17

Updated by MofoMan2000 over 13 years ago

I'm not having this problem anymore in r7216. I just now beat the Aparoid.

Actions #18

Updated by MofoMan2000 over 13 years ago

Wait, I used D3D9 backend, I'll try OpenGL once I get back from classes…

Actions #19

Updated by MofoMan2000 over 13 years ago

Yeah, I didn't see this problem in the OpenGL backend, and I just played through the entire first mission. I think this issue can be closed finally.

Actions #20

Updated by Anonymous over 13 years ago

  • Status changed from Accepted to Fixed
Actions #21

Updated by MofoMan2000 over 13 years ago

GAH, it's still a problem in JIT, but with JITIL it's fine.

Actions #22

Updated by NeoBrainX over 13 years ago

  • Status changed from Fixed to Accepted
Actions #23

Updated by jacobmp92 over 13 years ago

GAH, it's still a problem in JIT, but with JITIL it's fine.

That's how it's been this whole time. ;)

Actions #24

Updated by MofoMan2000 over 13 years ago

… I'm not stupid, I swear it…

Fuck. I'm stupid sometimes.

Actions #25

Updated by the.ryoku.kitsune over 13 years ago

the game is unavailable to me now so I can't test any updates. my image is gone and my disc is scratched so I'm out of the loop on this issue. please coordinate with other users on this issue for testing purposes.

Actions #26

Updated by LM1234 over 13 years ago

  • Status changed from Accepted to Fixed
Actions #27

Updated by degasus about 11 years ago

  • Status changed from Fixed to Questionable

The "fix" ( revision 0fcf1d6197ff ) disables the jit64 faddsx instruction completely. Does anybody know what is broken in this instruction? imo we should try to fix this instruction instead of just disabling it.

So, first step would also be to check is this issue is still valid when reverting this "fix".

Actions #28

Updated by degasus about 11 years ago

  • Status changed from Questionable to Accepted

After hours of debugging:
The difference between Interpreter and Jit64 in faddsx is that the Interpreter checks for subnormal by doubleToFloat convertion:
https://code.google.com/p/dolphin-emu/source/browse/Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_FPUtils.h#76

Jit64 itself doesn't check this:
https://code.google.com/p/dolphin-emu/source/browse/Source/Core/Core/Src/PowerPC/JitCommon/Jit_Util.cpp#387

At least disable this line recreate this issue.

As this affects all float instructions, disable just one of them because of a visual issue isn't the way to go.

btw, the affected call in jit64:
PC 8015ed00: 5.047743e-40(37c5fc6c00000000) + 5.123296e-39(37fbe4d500000000) = 0.000000e+00(0)(interpreter) <-> 5.628070e-39(37fea46280000000)(jit64)

btw2: the smallest normal float is 1.175e-38, so all args are subnormal.

Actions #29

Updated by flacs about 11 years ago

It seems like it is a general problem that the Gekko's FPSCR register is only updated in the interpreter and the JITs don't update their respective CPU (on x86, the MXCSR register).

Actions #30

Updated by degasus about 11 years ago

I wouldn't call this a fix as it's just another workaround, but it still works fine:
http://pastie.org/8407740

Actions #31

Updated by JMC4789 almost 11 years ago

This issue happens in D3D and OGL in Master 4.0-370. Using JIT for now, will do more testing later.

Actions #32

Updated by degasus almost 11 years ago

JMC: There is no need to test here. This issue is well known to happen because of denormal floats. Just see my comment #28

btw: this is fixed in the ppc_fp branch

Actions #33

Updated by JMC4789 almost 11 years ago

I just wanted to let everyone know that it seems to be only bug remaining in this game. I played the rest of the game in JIT and there were no problems.

The crashes mentioned in the Wiki + other issues mentioned here are no longer issues. Considering how old this issue is, it's not really a surprise, and I'm sure other people have tested.

This game, at least from my quick tests on Console, should be 5 stars after this is fixed :D

Actions #34

Updated by degasus almost 11 years ago

  • Status changed from Accepted to Fixed

as ppc_fp is merged in b863e4067771 , this issue is fixed now.

Actions

Also available in: Atom PDF