Revision 2006...

[andutrache]

yes really odd but it still doesn't work even on rev 2009.
ATI Radeon 6970 here, and btw im NOT getting Nvidia just to play on Jpcsp
On an all msg displayed console it fills up with this when i press run:

34731 [user_main] DEBUG hle.ThreadManForUser - End of Interrupt
34748 [user_main] DEBUG hle.sceCtrl - hleCtrlExecuteSampling
34748 [user_main] DEBUG hle.ThreadManForUser - Triggering Interrupt VBLANK(0x1E)
34748 [user_main] DEBUG hle.ThreadManForUser - End of Interrupt
34765 [user_main] DEBUG hle.sceCtrl - hleCtrlExecuteSampling
34765 [user_main] DEBUG hle.ThreadManForUser - Triggering Interrupt VBLANK(0x1E)
34765 [user_main] DEBUG hle.ThreadManForUser - End of Interrupt
34782 [user_main] DEBUG hle.sceCtrl - hleCtrlExecuteSampling
34782 [user_main] DEBUG hle.ThreadManForUser - Triggering Interrupt VBLANK(0x1E)
34786 [user_main] DEBUG hle.ThreadManForUser - End of Interrupt

[MaXiMu]

Some Resolutions modes not work or experimental by me using nvidia card in this test.

And filter bilineal is work experimental

[dista_bagus]

on r2009 still can't play games.grey.
anyone tried it with r2009,is it work.
only work that r2005.

[gid15]

Something really odd is going on here...
I just tested the demo sample with r2009 and it's working just fine.
Plus, the filters seem to be working fine, as it's possible to distinguish the appearence of the bilinear blur from the sharpness of not using any filter at all...
Is this an ATI exclusive issue?

The problem is related to the change in sceDisplay: initializing the OpenGL context with samples is causing a problem on AMD. When removing "withSamples(16)", it is working again...

There is also a warning into the LWJGL documentation for this feature:

"WARNING: Some pixel formats are known to cause troubles on certain buggy drivers. Example: Under Windows, specifying samples != 0 will enable the ARB pixel format selection path, which could trigger a crash."

http://www.lwjgl.org/javadoc/org/lwjgl/o...ormat.html

[gid15]

Something really odd is going on here...
I just tested the demo sample with r2009 and it's working just fine.
Plus, the filters seem to be working fine, as it's possible to distinguish the appearence of the bilinear blur from the sharpness of not using any filter at all...
Is this an ATI exclusive issue?

The problem is related to the change in sceDisplay: initializing the OpenGL context with samples is causing a problem on AMD. When removing "withSamples(16)", it is working again...

There is also a warning into the LWJGL documentation for this feature:

"WARNING: Some pixel formats are known to cause troubles on certain buggy drivers. Example: Under Windows, specifying samples != 0 will enable the ARB pixel format selection path, which could trigger a crash."

http://www.lwjgl.org/javadoc/org/lwjgl/o...ormat.html

A fix for this problem is now available in r2010.
Also r2011 includes 2 fixes for shaders on AMD.

[andutrache]

Reporting with r2012 :
1) Games boot again
2) The Bilinear/None filtering option doesn't save -> always defaults to none ;
3) Changing Resolutions destroys almost all games (they only show a little part of a corner on the screen), you need something to prevent it from going offscreen & such;
4) Savegames now work perfectly with the crypto engine enabled i saved and loaded successfully in all my games;
5) Shaders are as ugly as they were in r2005
6) OLD Problem : The Resize menu where you select 1x,1,5x,2x...etc doesn't keep the aspect ratio of the psp resolution so i need to resize manually after that.
7) If i go to full screen i cant revert back and have to forcekill the emu, also Media Engine Movies do not work in full screen.

Well, at least for me, the emu is looking good, keep up the good work and thanks for the fixes .

[Hykem]

@gid: Yeah... I suspected it had something to do with that. It was the only absolute change I introduced. Unfortunately, I had to means to test it in other system.
Thanks for removing that and for providing more fixes for the shaders under AMD. Great job!

@andutrache:
1.-
2.- I decided not to allow saving this in the Settings.properties right in r2006 because I was afraid it could be an unstable feature. Turns out I was right, but for the wrong cause.
Saving the filter status and the selected resolution should be easy to add.
3.- Yes, this is normal. The only resolutions that should not corrupt the graphical display are the native ones 480x272, 960x544 and 1440x816. The others need a more complex rescaling, so they're still a TODO feature.
4.-
5.- Any improvements with dynamic shader generation?
6.- Hmm... Could you send me some screenshots, with any test application running, demonstrating the resizing?
7.- This may be also normal. Full screen still needs more improvements, like the removal of the window's decorations before the full maximization. Unfortunately, Java makes this a quite hard task because it envolves disposing the current window, setting it's undecorated parameter to true and then build the window again. I'm currently looking for better ways to implement a workaround for this.

@Vegerunks: Are you still experiencing low FPS after gid's fixes in r2014?

[Kai3213]

5.- Any improvements with dynamic shader generation?

Enabling the option often results in a crash right before the 3d graphics are displayed. Tested on rev 2014

Edit: Scratch that, only when the saving GE and or Only GE options are enabled but it crashes from time to time..

[andutrache]

@ Hykem : I have attached screens of P3P demonstrating issues 3 & 6.
Also -> 5 improvements yes speed goes up to 60 fps but graphics are still destroyed.
I used P3P for tests .

First pic is 2.5x scaling, second 2x then 1.5x after that the Normal 1x, then 960x544 resolution, then 2.5x corrected (look at moon and selection circle), the last one is 1440x816.

[Vegerunks]

@Vegerunks: Are you still experiencing low FPS after gid's fixes in r2014?

No Sir, I am not Ever since 2012 or whatever Revision it was I have got a slightly faster FPS and now some Maps on Dissidia run at 100% FPS.