search

hrydgard / ppsspp

A PSP emulator for Android, Windows, Mac and Linux, written in C++. Want to contribute? Join us on Discord at https://discord.gg/5NJB6dD or just send pull requests / issues. For discussion use the forums at forums.ppsspp.org.
https://www.ppsspp.org
Other
11.02k stars 2.15k forks source link

Remaining issues in Dan Ball Senki W #4786

Open daniel229 opened 10 years ago

daniel229 commented 10 years ago

A video lead you to know the issue 1and 2 http://www.youtube.com/watch?v=KTlrWd7hKYc

normal 03

Camera stop 04

it display now,but some time not correct. 01

sometime not correct 02

log

08:16:216 AskaMainThre I[SCEGE]: GLES\Framebuffer.cpp:732 Resizing FBO for 00178000 : 256 x 512 x 3
08:16:219 AskaMainThre W[G3D]: GLES\TextureCache.cpp:336 Render to texture with different strides 512 != 256
08:16:219 AskaMainThre W[G3D]: GLES\TextureCache.cpp:339 Render to texture with different formats 1 != 3
08:16:220 AskaMainThre I[G3D]: GLES\ShaderManager.cpp:140 Linked shader: vs 94 fs 95
08:16:220 AskaMainThre W[G3D]: GLES\TextureCache.cpp:407 Render to area containing texture at 00178000 +240x0
08:16:221 AskaMainThre I[G3D]: GLES\ShaderManager.cpp:140 Linked shader: vs 94 fs 97
08:16:247 AskaMainThre W[G3D]: GLES\TextureCache.cpp:381 Ignoring possible render to texture at 04178000 +0x272 / 480x512
08:16:250 AskaMainThre I[G3D]: GLES\ShaderManager.cpp:140 Linked shader: vs 28 fs 99
raven02 commented 10 years ago

For 2nd issue , any savestate can shared?

daniel229 commented 10 years ago

move to right side,then go outside. 01

savedate here,rename jpg to rar npjh50618data0001

daniel229 commented 10 years ago

Updated two issue to top post. 4-missing background. 5-video not played.

raven02 commented 10 years ago

Have this improved lately?

daniel229 commented 10 years ago

No improvement.

daniel229 commented 10 years ago

Test with 0.9.6-436 For the video part,the audio can be heard, but display nothing. Log shows this,and audio stuttering when these messages posting. AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 319900 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 508000 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 712820 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 926000 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 1135000 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 1314740 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 1553000 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 1732740 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 1941740 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 2121480 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 2313760 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 2593820 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 2852980 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 3099600 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 3367120 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 3634640 dts: 8948940 AskaMainThre I[ME]: HLE\sceMpeg.cpp:1172 Audio end reach. pts: 3881260 dts: 8948940

daniel229 commented 10 years ago

FPS issue fixed by https://github.com/hrydgard/ppsspp/pull/5174 03

daniel229 commented 10 years ago

https://github.com/hrydgard/ppsspp/issues/3110 Frontier Gate Boost+ also got FPS issue fixed.these two games have several issues the same.

unknownbrackets commented 10 years ago

Huh, that's very interesting.

The camera stopping thing is still happening, right? AFAIK the video is a depth issue.

-[Unknown]

daniel229 commented 10 years ago

Yes camera is still stopping.

unknownbrackets commented 10 years ago

Does the black box thing still happen?

-[Unknown]

daniel229 commented 10 years ago

still happen when screen transition.block transfer not help 01 02

unknownbrackets commented 10 years ago

Have any of the changes helped specifically 2, 4, and 5? Is 1 affected by jit on/off?

-[Unknown]

daniel229 commented 10 years ago

1,2,4,5 no change at all.with jit off,Ican not move.

jit off 01

unknownbrackets commented 10 years ago

Well, then there's definitely some cpu bug here, even if it's the interpreter that's wrong. If you switch the CONDITIONAL_DISABLE defines in CompVFPU.cpp, and use jit, does it still have the above problem? If not, does it happen switching them in CompFPU.cpp?

-[Unknown]

daniel229 commented 10 years ago

change #define CONDITIONAL_DISABLE ; to #define CONDITIONAL_DISABLE { fpr.ReleaseSpillLocks(); Comp_Generic(op); return; } in CompVFPU.cpp,the jit without anyproblem.

daniel229 commented 10 years ago

It gets problem in CompFPU.cpp

daniel229 commented 10 years ago

the problem is 

void Jit::Comp_FPU2op(MIPSOpcode op) {

unknownbrackets commented 10 years ago

Which one inside there? Change for example case 13: to case 13: DISABLE; and it will disable trunc.w.s. abs, mov, and neg should be pretty hard to be wrong, sqrt, trunc.w.s, and cvt.s.w are the likely ones.

Does the camera thing go away if you have the CONDITIONAL_DISABLE swapped in CompVFPU.cpp and use jit? It could be a totally different instruction causing that problem.

-[Unknown]

daniel229 commented 10 years ago

Yes,disable case 13,the problem presents.CONDITIONAL_DISABLE swapped in CompVFPU.cpp does not help the camera problem.

unknownbrackets commented 10 years ago

Does it affect anything to revert this change? https://github.com/hrydgard/ppsspp/commit/cbf1df9

-[Unknown]

daniel229 commented 10 years ago

both jit and non-jit can't move.

unknownbrackets commented 10 years ago

So basically this and #4268 don't agree about that change, seems this likes it. I'm not sure why interpreter is not behaving the same way...

-[Unknown]

unknownbrackets commented 10 years ago

Does this game crash for you with simulate block transfers on? http://forums.ppsspp.org/showthread.php?tid=5201&pid=86714#pid86714

The black box thing seems really weird. It's hard to catch those things in the GE debugger, but I wonder what the vertex type / positions are.

I guess the camera stop thing is likely to be a cpu bug as well, if this uncovered an interpreter cpu bug...

-[Unknown]

daniel229 commented 10 years ago

Is that Danball Senki Boost,it does not crash for me,but I find a graphical issue with simulate block transfers on.

simulate block transfers on 01

simulate block transfers off 02

before the black box drawing. 03 04

daniel229 commented 10 years ago

Gedebugger about the graphical issue 06

unknownbrackets commented 10 years ago

Oh. I was confused. Thought they were the same game.

Hmm Boost seems like a format mismatch... hard to say. W definitely looks like that texture ought to display. I wonder if it is depth related. Maybe minz and maxz clamp before depth test or affect clipping... are the vertex addresses? Or it could be stencil.

-[Unknown]

daniel229 commented 10 years ago

Dan Ball Senki W random hangs in battle with jit off.

unknownbrackets commented 10 years ago

Hmm, well, same question then - if you disable CompFPU or etc. does this happen even with jit on? Have to narrow it down.

-[Unknown]

daniel229 commented 10 years ago

Disable either CompFPU or CompVFPU would not hang.Disable both,it hangs. I find other bug that the character face to another way when talking due to Disable codes in void Jit::Comp_Vcmp(MIPSOpcode op) {

daniel229 commented 10 years ago

The hangup is due to disable both void Jit::Comp_Vcmp(MIPSOpcode op) { and void Jit::Comp_FPUComp(MIPSOpcode op) {

unknownbrackets commented 10 years ago

Hmm, so there must be a wrong NAN/INF/-0.0 comparison hiding in those somewhere...

-[Unknown]

daniel229 commented 10 years ago

Change all CONDITIONAL_DISABLE; to DISABLE;,and leave the one invoid Jit::Comp_Vcmp(MIPSOpcode op) {,still experience the camera problem.

daniel229 commented 10 years ago

The video is showing now since https://github.com/hrydgard/ppsspp/pull/6737 but missing audio and skipped in about 2 second. The missing audio due to https://github.com/hrydgard/ppsspp/pull/6384 ,revert it,the Level 5 logo works fine, the intro video has sound though,but just for about 2 second and then skipped. revert https://github.com/hrydgard/ppsspp/pull/5381 the intro video works fine now.

debug log for the missing audio(rename jpg to rar) ppsspplog

02

daniel229 commented 9 years ago

The second issue fixed in https://github.com/hrydgard/ppsspp/commit/a04af187ea055699cba72db093a78dead9656c72 .

unknownbrackets commented 9 years ago

So, just to recap:

  1. Camera issue remains, and is not caused by jit.
  2. Square problem is now fixed.
  3. FPS is now correct.
  4. Backgrounds still do not show correctly in menus, even with block transfers enabled.
  5. Videos now display.
  6. Jit disabled no longer causes T formation or lack of movement.
  7. Intro video does not play audio (affected by #6737.)
  8. Video skips after 2 seconds (affected by #5381.)
  9. Simulate UMD delays helps the camera issue (item 1.)

-[Unknown]

daniel229 commented 9 years ago

Remove issues that fixed in the top.

unknownbrackets commented 9 years ago

It helps the camera issue? Hmm. Well, that explains why jit doesn't affect it. It might be a separate timing issue that simulate UMD delays only accidentally helps. Hmm.

So the main issues are video related, timing related, and something to do with block transfers still.

Does the background show correctly if you open the menu, close it, and open it again? And if you move, is it wrong until you close and open it again? Or how wrong is it?

-[Unknown]

daniel229 commented 9 years ago

Background is randomly happen. close it, and open it again it may get different results.

right 01

wrong 1 02

wrong 2 03

DanSenki commented 8 years ago

Prior to the current build, are there any more problems to the game other than the black model outlines and the skipping cutscenes?

Battles work perfectly fine. npjh50618_00015

npjh50618_00000

daniel229 commented 8 years ago

@DanSenki That model looks fine here.what device did you play on and which version of PPSSPP?If you changed any setting,I recommend trying the defaults. 01

unknownbrackets commented 8 years ago

I'd be interested to know what setting causes that.

-[Unknown]

DanSenki commented 8 years ago

Woah woah woah how did that work.

(While I did some fixes that work temporarily, when I switch between windows, the graphics glitch.)

Are there still any issues on cutscenes though?

DanSenki commented 8 years ago

npjh50618_00001 npjh50618_00000

Even on default, this still happens.

DanSenki commented 8 years ago

uljm05990_00000 npjh50618_00016 npjh50618_00017

UPDATE: Restoring defaults not only doesn't fix anything, it makes the screen shake when pressing the menu.

unknownbrackets commented 8 years ago

What device, video card, operating system version, etc. is this on? Are you using the latest git build?

-[Unknown]

DanSenki commented 8 years ago

Windows 10 Home, with Intel HD Graphics, with a 1.0.1 build.

Graphic Card shouldn't be an issue, though. It ran BOOST almost perfectly.

Oh shit. uh, ignore the BOOST screenshot.

DanSenki commented 8 years ago

screenshot_10 screenshot_11

I think I found a source for the skipping cutscenes. Still no trace for the outlines?

daniel229 commented 8 years ago

The cutscenes would be played on this build.The graphical problem seems like a driver problem? http://www.filedropper.com/ppsspp-v101-216-gb9fbaed-windows

DanSenki commented 8 years ago

Unless Level-5 made the graphics more obnoxious than in the original game, I can't really say it's a driver problem or compatibility.