PPSSPP 1.0 - Regressions and bugs to fix before release

[hrydgard]

Alright, time to do 1.0 finally, I'm aiming for end of January.

Please report anything that has gotten worse since the last release in this issue so we can fix it up before we release 1.0.

Also bugs in new features, like if there are problems with the new savestate UI :)

Features:

• Low latency audio implemented for Windows and Android
• Adhoc branch merged

Bugs:

• Prince of Persia Rival Swords black screen: #7314
• x86 SIMD regressions: https://github.com/hrydgard/ppsspp/issues/7292
• Resizing the window must work correctly in D3D, and new D3D crashes from the last attempt
• FIXED: #6898 (on screen buttons partly disappear in non-buffered rendering)
• #7324 Android screen resolution issue

Minor:

• Savestate screenshots are sometimes black

[LunaMoo]

Pause screen kind of looks cluttered now with save/load text repeated all over the place. Maybe the screenshot itself could be used as a load button to at least remove one row? I think it would feel kind of natural thing to try pressing if no load button would exist:3.

Anyway it's not much of an issue, just new thing I'm gonna get used to one way or another and the little screenshots are quite useful, thanks for implementing it. Oh and ~ a happy new year.:)

[hrydgard]

I agree about the clutteredness, regard this as the first draft. All ears for improvements, your idea isn't bad :)

Happy new year!

[Joelgp83]

Don't know if this counts as a regression, but its definitely something that needs to be worked out for an official release.

First off, running Linux Mint 17.1, and using the ppsspp-sdl git unstable branch from https://launchpad.net/~ppsspp/+archive/ubuntu/testing, if that matters at all.

The up/down/left/right arrows (and by extension, the dpad on my moga pro power 2 controller in Mode B / HID) do not navigate the menu ui properly.

Lets say you go into settings. Hitting down on the arrow keys or down on the dpad takes you back one screen up, instead of moving the selection highlight. The up arrow seems to act like the enter key.

At first I thought the error was limited to some bad mapping of the moga dpad, but nope, it affects the arrow keys on the default keymap, even when the controllers not connected.

By the way, the moga pro power 2 mode B works flawlessly in game, with all buttons except select. If you want to do controller testing, you can pair the controller to the system with pairing code 1234.

Hope this helps. You guys are awesome for making this, as always. Happy New Year!

[hrydgard]

@Joelgp83 , just fixed the moga dpad in Mode A, can you retry on your setup?

Hm wait, my fix only works on Android, this is mode B so does not apply. Weird.

[chaosjr7]

The Mana Khemia Issue still exists I think, I will check the latest GIT, to be sure..If it helps and is still there like I think it might be , The unthrottle Button held and the quote "END" Key Tapped a bit, seems to be a decent workaround for now. I have (U) Version of Mana Khemia so my "END" key is 'Cross'. Good Luck. See Ya Later.

The Bug is Still there and the Temporary workaround still works too. Oh and that workaround isn't an instant fix either, it takes a bit of time to go through.. But it does get out of the problem. Good Luck Again, and Goodbye for now.

[vsub]

About Mana Khemia(usa version),no matter how many times I tried with different settings(0.9.9.1-1338 x86),this was the only way to fix it

1.Either use this code that forces the game to run at 60fps(I always keep that code enabled for this game anyway because it looks too slow without it(I played the PS2 version too many times and it looks low in the PSP without this code))

_C1 60 FPS _L 0x2001C238 0x0E26C5AB

2.Or Press and hold the key for rapid-file and press and hold the key for the x button immediately when the last enemy dies...in the results screen you have extremely small amount of time to start mashing the x button(rapid-fire)and if you wait in the results screen too long,the game won't allow you to continue.

Here is a save before the boss battle(you can easily kill it because this is at least second playthrough) http://www63.zippyshare.com/v/43650213/file.html

[unknownbrackets]

Okay, with that save, I think I know the cause. But is this an actual regression? Did it ever work?

The game is running at that point at 20 fps, which I think is incorrect. As I recall, 20 fps in this game is generally caused by sceCtrl delays. It calls sceCtrl twice to grab input and iirc discards one. So, I suppose it works when you mash X enough because you eventually get lucky and get the press and release both in events it doesn't discard, or something to that effect.

But even on a real PSP this game seems to run at inconsistent FPS and drops to 20 sometimes. We're not accurate here, so it is a bug, but it's been this way for as long as I know.

-[Unknown]

[vsub]

Hmm I guess I was wrong about the second method. Even if all menus appear,when you mash the x button(rapid-fire)it will actually continue the game after 1-2 second of holding the rapid fire and the x button.

And yes,this problem didn't exist before(just for a test I tried something really old(0.9.1-542)and it was working fine there)

[Joelgp83]

@hrydgard, I have work so i wont be able to test the android-moga changes for awhile. Also, I have no idea when these new changes will propogate to the ppa i am using.

Finally, i must mention that linux mint sees the controller in mode a as a computer/phone, not a joypad. Mode b is properly recognised as a HID joypad.

[sum2012]

@hrydgard can we fix #6898 non buffered rendering ,button disappear bug ?

[hrydgard]

Hm, I thought I fixed that, or a similar problem, a while ago... I'll have another look because yeah, it's pretty important.

Yeah, I got repro. Tekken 6 on my Zenfone exhibits it, also on PC. Let's see...

There we go, fixed.

[unknownbrackets]

There's apparently a bug in Yu-Gi-Oh games on ARM only:

http://forums.ppsspp.org/showthread.php?tid=6466&pid=99587#pid99587

There's also this, but I think it's a bug in the real game. Maybe 9 GB would make it work better.

http://forums.ppsspp.org/showthread.php?tid=1395&pid=99701#pid99701

-[Unknown]

[trnzaddict]

I don't know if this counts as regressions, but After Burner Climax WAS working on earlier versions. Now it hangs after displaying the Havok video and also has strobing FMV. Also Resistance:Retribution which did crash PPSSPP on earlier revisions now doesn't, but is now plagued with graphical bugs on both back ends as soon as you start walking around. That didn't happen before.

[hrydgard]

@trnzaddict , thanks. could you create separate issues with screenshots for those two, unless they are already reported?

[sum2012]

My game are a bit differance :After Burner Black Falcon EUR version As a fast test, work in -v0.9.6-120-g291d25f don't work in -v0.9.6-691-g4f2f8f0 I cannot search opened "After Burner" issue. Anyway ,I sleep now

[daniel229]

Afterburner Black Falcon thread https://github.com/hrydgard/ppsspp/issues/6794

[Dhezta]

Can you fix Final Fantasy II games? its very slow motion when enter the world map.......thanks...

[007flawless]

GTA Vice City Stories and Liberty City Stories still have Serious Issues on PowerVR GPUs.. Main Character, Pedestrians and basically all Humans are Invisible except in Cutscenes.. Also Traffic and all Vehicles are Invisible.. Among other disappearing Objects like some Trees.. I wish this would be looked Into.. But Thanks for all the effort PPSSPP Devs

[Joelgp83]

Finally got a chance to test your latest moga changes on Android, as well as do more linux testing.

On Android, Mode A works flawlessly in-ui and in-game now, thank you.

Moga on Linux (ppsspp-sdl, can't get ppsspp-qt to recognize anything) is still a different story. I've moved and expanded on the details to a new issue, #7300.

[hrydgard]

@Joelgp83 Thanks for your testing! Yes, please move to a separate issue as it's not a regression and not critical (moga on linux is not a common usecase by any measure, I would imagine...)

[Joelgp83]

I figured you guys, as moderators, were the only ones who could directly rip posts from a thread and make it its own thread. Or did you just want me to create new thread + delete what's here? Or something else?

I do agree about the usecase, but I've also found it ironic that I'm able to get more of my linux games to work out of the box with it than an x360 controller. I also guess you could simply link in some of your pre-existing moga code from the android codepath to add, with a minimum of effort, autodetection / autoconfiguration of such devices. Then again I could be wrong. Either way, you're still right, it needs to be in its own thread.

[hrydgard]

I can't "rip posts" anymore than you can - by copy-pasting it into a new issue.

Well, no, as with Android we get the correct buttons without having to do much except some manual dpad management, while in mode B we get whatever button Linux has assigned which seem crazy.

[Joelgp83]

I see. My apologies for the misunderstanding. I'll make a thread and copy/paste it myself, then.

-edit- Done. Anyone interested who read the original long post with screenshots should head to the other issue, I've updated it with some other moga bugs I've found.

[coolmisty45]

hi i play wwe 2k14 it force closed after 3 mins of play, but the ppsspp from playstore didnt force close when i play the game -plz fix

[hrydgard]

@coolmisty45 thanks for reporting! Does this also happen in the Windows version of PPSSPP?

@Joelgp83, great, thanks.

[coolmisty45]

@hrydgard i dont think so but i havent played to it to much on windows computer

[Joelgp83]

@hrydgard, got a chance to sideload and test v9.9.1-1442 on a current-gen Amazon FireTV. Results were, well, less than optimal. The FireTV supports controllers, and is beefy enough to theoretically play games very well, but I am practically unable to navigate the ui with the standard FireTV remote control with any reliability, and attempting to use my moga (mode B, since FireOS is forked android I could sideload moga Pivot to enable Mode A, but did not have time) to navigate results in the same ui controls issues as linux mint had. In short, I can't launch games or map controls or change settings, which means I can't test real-world performance either. :(

[LunaMoo]

@hrydgard Seems that new savestate UI on pause screen(https://github.com/hrydgard/ppsspp/commit/6879f05eb069c87ea19badaca7035e90881a44c1) broke ability to store savestate slot into game config(slots from general config are remembered just fine).

[Some1Else46]

Not sure if this is even worth the time looking into, but #5333 still seems to be an issue for me with the latest iOS git build from the buildbot on iOS 8.1.1.

[hrydgard]

@LunaMoo , are you sure? The savestate UI saves the slot the same way as before..

[LunaMoo]

@hrydgard well maybe it's just me, but changed slot after I create config is not saved when I leave the game:|. For example to reproduce it, run the game change to slot 2, exit game, open it again and slot 2 should be highlighted. Then create custom config for that game, change it to other slot ie slot 3, and leave the game again, after returning to it, it'll continue to have slot 2 selected, and it'll forever start from that slot as long as custom config exists.

Also when I was testing it on fresh build in new folder just now, to make sure it's not any weird setting leftover, I found out that games which have their custom config created during the first run(before exiting at least once) do not populate "recent" list, seems not related to new ui. Maybe @Bigpet could have a look at that.:3

[sum2012]

Just a small note,I test Yu-Gi-Oh 6 crash with edit card with Android version (LG nexus 5)

[unknownbrackets]

Oh, cool, when did it start/end? Is it that same BKPT thing?

-[Unknown]

[sum2012]

I have to go to work.I cannot test deep now. But I test disable JIT can temp solve the problem

[sum2012]

@unknownbrackets git bisect result,this commit https://github.com/hrydgard/ppsspp/commit/74d8a9bdba9b8348e60e8a6b40b01da26b5eb020

anyway, v0.9.9.1-1486-g7c3bc2b Android log https://gist.github.com/sum2012/7a0d797f86f186782ed6

[unknownbrackets]

Worst case I can think of is it's fiddling with instruction bits directly without rewriting the entire thing and we're jumping into space. But in that case, was it just luck that it worked before?

Otherwise, we're probably not invalidating or clearing properly? I guess?

-[Unknown]

[sum2012]

@unknownbrackets it does not crash in windows,so that I do not think that it rated to invaliding and clearing.

[tiamat999]

Edf 2 portable speed up?

[Bigpet]

Just as a point of discussion since Minamoto-kun brought it up in the forums.

Fix Symbian regressions? (I don't really care since I don't have a symbian device but I think either dropping Symbian support or fixing these should be done)

stasb-48 listed on the forums: (currently still inquiring about details, since the changes after don't seem related)

PPSSPP-v0.9.8-1149 - last version were save/load is work PPSSPP-v0.9.9.1-49 - last version were on screen control buttons not cut off PPSSPP-v0.9.9.1-186 - last version were videos are played

[sum2012]

Hope today we can solve Online Chess Kingdoms black screen in latest version. v0.9.8-46-g8b92dce work

[sum2012]

oop,I tested before #6924

[vnctdj]

@hrydgard What about improvements on the new savestate UI ? Will it be ready for v1.0 ? :)

[sum2012]

@hrydgard Don't forget update font atlas,thanks

[sergiobenrocha2]

@hrydgard , maybe could you include #3318 ? It stuck at all and becomes unresponsive now in git version, it seems in old versions you could press start button to skip access page.

[hrydgard]

@sum2012 don't worry, I will :) Will do it just before release to make sure the latest translations are in.

[sum2012]

Hope to solve #7386

[thedax]

https://github.com/hrydgard/ppsspp/issues/7316 Fixing this would be lovely, since it's a regression.

[hrydgard]

@thedax done.

[thedax]

@hrydgard: Thank you.

[psennermann]

Well as regression since last release there would also be #7336 , but at the moment it doesn't seem so easy to understand why ppsspp has got so much slowdown on PoverVR devices...