Final Fantasy XIII-2 [BLUS30776]
Started by GrantKane




4 posts in this topic
GrantKane
RPCS3 Tester


2
2,670 posts 1,777 threads Joined: Aug 2017
07-15-2017, 02:34 PM -
#1
RPCS3 v0.0.2-6-7528a16 Alpha
White screen only.
Quote:·F 0:00:06.547069 {PPU[0x1000000] Thread (main_thread) [0x0087fa64]} MEM: Access violation writing location 0x0
MarioSonic2987
RPCS3 Tester


4
296 posts 105 threads Joined: Aug 2017
07-15-2017, 07:41 PM -
#2
Game's title should be renamed to Final Fantasy XIII-2.
Desktop: Intel Core i9-9900KF, Nvidia GeForce RTX 2080 (OpenGL, DirectX12 & Vulkan), Windows 11
Laptop: Intel Core i7-7700HQ, Nvidia GeForce GTX 1060 (OpenGL, DirectX12 & Vulkan), Windows 11
Asinine
Moderator
*****


4
2,060 posts 83 threads Joined: Aug 2017
12-21-2017, 08:31 PM -
#3
In preparation for upcoming changes to the compatibility database, this thread is being moved to intro. Due to other threads for the same media (Disc or PSN) already being in that category.

If you find a specific game ID that does not perform like the others for the same game and in the same media type (PSN or Disc) even when tested with the same build and settings then please discuss it in the relevant thread(s) and PM a moderator so we can look into it.

Tested with: RPCS3 v0.0.4-6226-17bfb3be Alpha

Requires PPU interpreter (fast)
Asinine
Moderator
*****


4
2,060 posts 83 threads Joined: Aug 2017
06-19-2018, 11:14 AM -
#4
Ingame as of my report here https://forums.rpcs3.net/thread-196850.html
badman79
Member


0
4 posts 0 threads Joined: Jun 2022
06-04-2022, 09:46 PM -
#5
Hi guys.

I have some feedback from some brief testing.

I agree with Ingame.  I can get past the title screen.

Also referencing this post https://forums.rpcs3.net/thread-203091.html

But am replying to this one as I think this one is in the right place for rules?

Testing was done on 0.0.22-13655-5ad4b624 Alpha | master

First change was to switch PPU to interpreter, leaving it on LLVM causes a non random game crash during the chapter 1 opening cut scene, the crash happens if you try to skip the scene as well.  Due to the performance loss I have played the game on LLVM after that part of the game and didnt find more problems related to it.

Second change was to switch from Vulkan to Opengl rendering API.  This is to partially fix a rendering problem with the chain stagger gauge.  I will attach some screenshots.  In Vulkan, the box is see through, Opengl fixes the black background in all but one fight I have tested (the second tutorial fight remains broken, but every other fight is fixed I tested), however there is also meant to be a red bar that appears to represent the progress of the stagger meter, none of the compatibility settings make it consistently appear, however on 3 fights (out of 50 or so) it randomly worked.  All 3 screenshots are from the same enemy just in case questions are asked about it been a different fight.

I tried accurate SPU xfloat, SPU interpreter, Precise ZCULL, Write color buffers, Strict rendering mode and disable Vertex cache, sadly none of them make it render 100%.  Just it worked randomly a few times.  But the black background works simply by switching to Opengl.


In addition to the mog clock which appears when a enemy spawns in the field has rendering issues, I will guess its the same type of problem as I think its missing a background which breaks it, it basically has reduced visibility and only shows the green instead of amber overlay over part of it.  This one also not fixed by any options and has yet to randomly work like when the red background appeared in the first issue.

Another observation that is whilst switching to interpreter has a clear performance impact, same with switching to opengl, all the options under additional settings, as well as zcull and xfloat have practically no performance impact, same with increasing driver delay, so I would consider having those pre configured for safety first.

Finally an observation on the cpu type for the LLVM compiler, I noticed an issue with the ATB in the game that its timing was out of sync (abilities show as charged/ready before the bar gets there), I noticed skylake was the configured cpu type, post haswell Intel CPU features have been chaotic, so I dropped it back to 'x86-64-v3' based on this documentation here https://clang.llvm.org/docs/UsersManual.html for haswell feature set, and it did fix that issue.  Sadly it didnt fix the chapter 1 cut scene LLVM crash.

Also the reason I referenced the other post.  There is a weird issue which I initially assumed to be messing up the DLC PLG installs, its hard to explain but basically the game has a thing called the histaria crux that appears at end of chapter 1, it has portals/tiles you can navigate on like a map layout thingy to choose to where to go next, if you have the DLC installed then 3 will appear to the left of your starting position in separate positions, however what happened instead they appeared on top of each other and so its not only cosmetically wrong but you can only select the top one, luckily by fluke the game has an alternative view point which allowed you to select all of them, I was convinced this was a DLC pkg problem, but then I found the above referenced post where this apparently happens after chapter 2 in the main game as well, and the guy progressed by using the alternate view.  I am not convinced this one is an emulator issue, but tagged it at end of this post for benefit of others.


Attached Files Thumbnail(s)
           


Forum Jump:


Users browsing this thread: 1 Guest(s)