Dissidia 012: Duodecim Final Fantasy - ULUS10566

[Spheric Rain]

Just a little information from my test. I tried using r3345 and r3427 to run this game and below is the result and the video configuration that I've used.

r3345
Only GE Graphics, Use vertex cache, Use Shaders, Enable VAO, Enable saving GE screen, Enable decoding of indexed textures. Ignore invalid memory access.
   
   
The game is playable with the graphical bug as you've seen. It ran smoothly in a constant 29-31 FPS.

r3427
Only GE Graphics, Enable saving GE screen, Disable optimised VertexInfo. Ignore invalid memory access.
   
The game is playable only until the end of that cutscene. It ran smoothly with 29-31 FPS, but stopped working and paused in a white screen after the cutscene.

I've tried other video configurations but it all either made the thing worse or made it unplayable.

[DragonNeos]

This game was tested with r3732. Starting with r3728, the emulator just displays a black screen with these lines being repeated in the Logger:

Code:

03:29:34  WARN hle.SysMemUserForUser - ffst_thread - malloc cannot allocate partition=0, name='ModuleMgr-TestInfo', type=PSP_SMEM_Low, size=0x4770, addr=0x00000000, maxFreeMem=0x46EC00, totalFreeMem=0x46F300
03:29:34 ERROR hle.ModuleMgrForUser - ffst_thread - Failed module allocation of size 0x00004770 for 'DISC0:/PSP_GAME/USRDIR/DATA/MODULE/LIBSUPPREACC.PRX' (maxFreeMemSize=0x0046EC00)

Using the External Software Renderer in r3727, the character models don't seem to be fully rendered as parts of their body are missing. Some animation effects seem to be either too dark or too light depending on the attack. Some black triangles appear on the ground throughout the entire battle. The error with too many NOP instructions being executed can occur either after a battle or cutscene.

[gid15]

This game was tested with r3732. Starting with r3728, the emulator just displays a black screen with these lines being repeated in the Logger:

Code:

03:29:34  WARN hle.SysMemUserForUser - ffst_thread - malloc cannot allocate partition=0, name='ModuleMgr-TestInfo', type=PSP_SMEM_Low, size=0x4770, addr=0x00000000, maxFreeMem=0x46EC00, totalFreeMem=0x46F300
03:29:34 ERROR hle.ModuleMgrForUser - ffst_thread - Failed module allocation of size 0x00004770 for 'DISC0:/PSP_GAME/USRDIR/DATA/MODULE/LIBSUPPREACC.PRX' (maxFreeMemSize=0x0046EC00)

Oops, should be fixed in r3733...

[DragonNeos]

This game is able to load again starting with r3733. However, the emulator seems to have a problem getting through the first cutscene in Story Mode. After the first cutscene ends, the emulator freezes and displays a white screen. Normally, the white screen is supposed to be a short transition between cutscenes.

[DragonNeos]

This game was tested with revision 4e46e0a. There seems to be internal memory issues as the game doesn't load and the emulator displays the message "Critical error: Check console output for details." with the following text appearing in the log:

Code:

06:03:38  WARN hle.SysMemUserForUser - GUI - malloc cannot allocate partition=2, name='ThreadMan-RootMem', type=PSP_SMEM_Addr, size=0x4000, addr=0x08800000, maxFreeMem=0x48F300, totalFreeMem=0x48F300

It should be noted that this issue only occurs if there aren't any system files placed in the flash0 folder. The emulator will be able to load this game if system files are placed in the flash0 folder, but there are audio issues (some stem from libatrac3plus.prx) and the console window will display "Exception: java.lang.OutOfMemoryError" after a certain point (e.g., during the tutorial) and become unresponsive.

[gid15]

Code:

06:03:38  WARN hle.SysMemUserForUser - GUI - malloc cannot allocate partition=2, name='ThreadMan-RootMem', type=PSP_SMEM_Addr, size=0x4000, addr=0x08800000, maxFreeMem=0x48F300, totalFreeMem=0x48F300

Oops, sorry, this should now be fixed in 6bac677.

[DragonNeos]

This game was tested with revision 8f3d95c. I seem to have encountered another rare crash with the profiler as I attempt to close the emulator:

Code:

Exception in thread "GUI" java.lang.ArrayIndexOutOfBoundsException: 65
    at jpcsp.Allegrex.Common.disasmVDMVSMVTM(Common.java:911)
    at jpcsp.Allegrex.Instructions$237.disasm(Instructions.java:9537)
    at jpcsp.Allegrex.Common.disasmJUMP(Common.java:732)
    at jpcsp.Allegrex.Instructions$92.disasm(Instructions.java:3572)
    at jpcsp.Allegrex.compiler.CodeInstruction.disasm(CodeInstruction.java:724)
    at jpcsp.Allegrex.compiler.Profiler.logCode(Profiler.java:183)
    at jpcsp.Allegrex.compiler.Profiler.logCode(Profiler.java:195)
    at jpcsp.Allegrex.compiler.Profiler.logCode(Profiler.java:161)
    at jpcsp.Allegrex.compiler.Profiler.logCodeBlock(Profiler.java:140)
    at jpcsp.Allegrex.compiler.Profiler.exit(Profiler.java:119)
    at jpcsp.Emulator.exit(Emulator.java:107)
    at jpcsp.MainGUI.exitEmu(MainGUI.java:2795)
    at jpcsp.MainGUI.formWindowClosing(MainGUI.java:1760)
    at jpcsp.MainGUI.access$000(MainGUI.java:102)
    at jpcsp.MainGUI$2.windowClosing(MainGUI.java:387)
    at java.awt.AWTEventMulticaster.windowClosing(Unknown Source)
    at java.awt.AWTEventMulticaster.windowClosing(Unknown Source)
    at java.awt.Window.processWindowEvent(Unknown Source)
    at javax.swing.JFrame.processWindowEvent(Unknown Source)
    at java.awt.Window.processEvent(Unknown Source)
    at java.awt.Component.dispatchEventImpl(Unknown Source)
    at java.awt.Container.dispatchEventImpl(Unknown Source)
    at java.awt.Window.dispatchEventImpl(Unknown Source)
    at java.awt.Component.dispatchEvent(Unknown Source)
    at java.awt.EventQueue.dispatchEventImpl(Unknown Source)
    at java.awt.EventQueue.access$300(Unknown Source)
    at java.awt.EventQueue$3.run(Unknown Source)
    at java.awt.EventQueue$3.run(Unknown Source)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.security.ProtectionDomain$1.doIntersectionPrivilege(Unknown Source)
    at java.security.ProtectionDomain$1.doIntersectionPrivilege(Unknown Source)
    at java.awt.EventQueue$4.run(Unknown Source)
    at java.awt.EventQueue$4.run(Unknown Source)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.security.ProtectionDomain$1.doIntersectionPrivilege(Unknown Source)
    at java.awt.EventQueue.dispatchEvent(Unknown Source)
    at java.awt.EventDispatchThread.pumpOneEventForFilters(Unknown Source)
    at java.awt.EventDispatchThread.pumpEventsForFilter(Unknown Source)
    at java.awt.EventDispatchThread.pumpEventsForHierarchy(Unknown Source)
    at java.awt.EventDispatchThread.pumpEvents(Unknown Source)
    at java.awt.EventDispatchThread.pumpEvents(Unknown Source)
    at java.awt.EventDispatchThread.run(Unknown Source)

As mentioned in the Dissidia: Final Fantasy Playable Demo thread, there are various issues with how the graphics are displayed and frequent NOP issues that can force the user to have to restart the emulator.

I'm not sure why using the file libatrac3plus.prx causes character voices to not play before a battle begins and after a battle ends so I provided a DEBUG log of hle.sceAtrac3plus.

Another issue is how the game is not able to autosave after a battle. The game encounters a problem and brings the save prompt in order to allow the user to be able to save progress. If internal PSP files are stored in the flash0 directory, there aren't any save issues. I provided a DEBUG log of hle.sceUtility for both situations (internal PSP files test is contained within the hle.sceAtrac3plus log file). The selection box as shown in the screenshots seem to be a little too large in terms of width in comparison to the PSP implementation.

As mentioned in a previous post, the game is not able to continue after the first cutscene. The emulator crashes during the white screen transition (requiring Task Manager to close the emulator) as the game is supposed to go into the second cutscene. Many warnings and error messages appear in the log:

Code:

09:44:47  WARN       ge - GUI - Texture in swizzled VRAM not supported 0x04288000 (appears as soon as first cutscene begins)
...
09:46:27  WARN       ge - GUI - Unknown/unimplemented video command [unk 0xef](hex=00FEDCBC,int=16702652,float=-146703155441043020000000000000000000000.000000) at 0x04134780 (this message and the following appear during the white screen transition)
...
09:46:27  WARN       ge - GUI - sceGuLightSpot(0,X,X,-1.07409408E8) invalid argument value
...
09:46:27  WARN       ge - GUI - Unknown ztst function 186
...
09:46:27  WARN       ge - GUI - Invalid texture address 0x220CC000 for texture level 0
09:46:27 ERROR   memory - GUI - read32 - Invalid memory address: 0x1523B7E0 PC=0x88000000
...
09:46:27 ERROR       ge - GUI - Unknown color type 1
...

[gid15]

Code:

Exception in thread "GUI" java.lang.ArrayIndexOutOfBoundsException: 65
    at jpcsp.Allegrex.Common.disasmVDMVSMVTM(Common.java:911)
    at jpcsp.Allegrex.Instructions$237.disasm(Instructions.java:9537)
    at jpcsp.Allegrex.Common.disasmJUMP(Common.java:732)
    at jpcsp.Allegrex.Instructions$92.disasm(Instructions.java:3572)
    at jpcsp.Allegrex.compiler.CodeInstruction.disasm(CodeInstruction.java:724)
    at jpcsp.Allegrex.compiler.Profiler.logCode(Profiler.java:183)

This should now be fixed in fb20f9c.

As mentioned in a previous post, the game is not able to continue after the first cutscene. The emulator crashes during the white screen transition (requiring Task Manager to close the emulator) as the game is supposed to go into the second cutscene. Many warnings and error messages appear in the log:

Is it the same behavior with the external software renderer?

[DragonNeos]

The External Software Renderer can reach the second cutscene but its performance is slow. On the first cutscene, the right portion of the screen is flickering constantly. The audio that plays in this cutscene is in sync. As you can see with the following screenshots, many things are not being displayed in the second cutscene. Character models have vertical geometric lines that appear on their faces. The audio in this cutscene is not in sync with what is being displayed on screen. On the fourth screenshot, the text appears to be slightly cut off from the bottom.

[DragonNeos]

Autosave Issue:
As mentioned in this post, the game has an issue with autosaving after any event (after a battle, changing settings, etc.). The issue started with revision 9009c2e. The game can autosave on the build before this one with revision 596803c.

  Log_r596803c(64bit)_INFO (hle.sceUtility DEBUG).zip (Size: 241.84 KB / Downloads: 142)
  Log_r7bda5b7(64bit)_INFO (hle.sceUtility DEBUG).zip (Size: 246.68 KB / Downloads: 167)

OpenGL Story Mode Cutscene Crash Issue:
Starting with revision 3587 (bd71f28), the emulator will become unresponsive after the first cutscene. The game is able to get to the second cutscene on the build before this one with revision 3586 (d237620).

  Log_r3586(64bit)_INFO (Compiler DEBUG).7z (Size: 1.24 MB / Downloads: 154)
  Log_r7bda5b7(64bit)_INFO (Compiler DEBUG).7z (Size: 1.02 MB / Downloads: 177)

OpenGL Graphics Issue: (only affects ATI/AMD graphics cards)
Starting with revision 3163 (0dbfc3e), the game produces improper hair textures that extend past a character's head and other graphical glitches (even-numbered screenshots). On the previous build with revision 3162 (1b99afe), none of the aforementioned issues appear although the rendering of the models is not correct with glaring areas of missing graphics (odd-numbered screenshots). I have performed a frame capture of both builds (5th Screenshot - r3162, 6th Screenshot - r3163) with the Compiler set to 3000 and have enabled "Disable optimized VertexInfo reading (may improve compatibility)".

  Log_r3162-1b99afe(64bit)_INFO (Compiler DEBUG).zip (Size: 1.68 MB / Downloads: 157)
  Log_r3163-0dbfc3e(64bit)_INFO (Compiler DEBUG).zip (Size: 1.68 MB / Downloads: 142)

  JPCSP v0.6 r3162-1b99afe D012FF Frame Capture.7z (Size: 1.75 MB / Downloads: 150)
  JPCSP v0.6 r3163-0dbfc3e D012FF Frame Capture.7z (Size: 1.76 MB / Downloads: 181)

OpenGL Color Mask Issue: (only affects ATI/AMD graphics cards)
Ever since the option "Enable the shader implementation for the "Color Mask" (more accurate but slower)" was implemented in revision 2167 (aa7db04), it has never worked properly with my graphics card (ATI Radeon HD 4200). Many graphical issues appear whenever this option is enabled. An example of what Color Mask looks like on my computer is shown on the 3rd Screenshot of the Dissidia: Final Fantasy Playable Demo thread.

NOP Issue:
The current value of 5000 is too low as many games or applications easily pass this number and force the emulator to stop. On VideoEngine.java, I propose for the value to be changed to a higher amount such as 250000 for a temporary workaround until a proper solution is made.

Performance Issue:
The performance in cutscenes is bad (about 10 FPS) even with the Compiler set to 50. The audio seems to be tied to the performance as the character's mouth movement and audio are not in sync. Some battle stages also have bad performance.