English Amiga Board

English Amiga Board (http://eab.abime.net/index.php)
-   support.FS-UAE (http://eab.abime.net/forumdisplay.php?f=122)
-   -   FS-UAE 1.1.1 screenmode problem when P96 active (http://eab.abime.net/showthread.php?t=63823)

PHabermehl 05 April 2012 15:02

FS-UAE 1.1.1 screenmode problem when P96 active
 
3 Attachment(s)
Hello Frode,

as soon as I activate P96 - even without using a P96 screen - display of other screens than WB gets corrupted.

It worked on all versions <=1.0.2, 1.1.0 also was defective.

Find config attached, I just changed amiga model from SUPER to A1200/020 to activate/deactivate P96emu.
A1200/020 with manually activated P96 fails also.

Oh, and I could just confirm that the issue disappears when I remove SetPatch from my Startup-Sequence. Again, it worked fine with releases < 1.1.0 and it still works with P96 disabled.

Help, please :mad

Peter

nexusle 05 April 2012 15:29

I haven't this problem.

Your config file seems ok, but what's with the string: "fullscreen_mode=fullscreen"? You don't need this, if you use "fullscreen = 1"...

You can also try with "uae_gfxcard_size = 32" config option

FrodeSolheim 05 April 2012 15:35

Quote:

Originally Posted by PHabermehl (Post 811661)
Hello Frode,
as soon as I activate P96 - even without using a P96 screen - display of other screens than WB gets corrupted.
It worked on all versions <=1.0.2, 1.1.0 also was defective.
Find config attached, I just changed amiga model from SUPER to A1200/020 to activate/deactivate P96emu.
A1200/020 with manually activated P96 fails also.
Oh, and I could just confirm that the issue disappears when I remove SetPatch from my Startup-Sequence. Again, it worked fine with releases < 1.1.0 and it still works with P96 disabled.
Help, please :mad
Peter

I haven't seen this problem, but I'll try to reproduce it (first step in fixing it). Could you tell med what kickstart / workbench version this is, and also the version of SetPatch? (open a shell in Workbench and run SetPatch manually -it will print the version number).

Quote:

Originally Posted by nexusle (Post 811666)
but what's with the string: "fullscreen_mode=fullscreen"? You don't need this, if you use "fullscreen = 1"...

fullscreen_mode=fullscreen actually did do something useful, but in version 1.1.1 "fullscreen" is the default value for fullscreen_mode on all platforms. fullscreen_mode=window (the old default on Linux/Mac) uses a fake fullscreen, by creating a borderless window covering the entire screen.

PHabermehl 05 April 2012 15:58

Okay, it is:

Kick 40.63
WB 40.42
Setpatch 40.16

Foul 05 April 2012 16:33

here is the latest one on Aminet : http://aminet.net/package/util/boot/SetPatch_43.6b

All the version i got (just rename to Setpatch)

http://foulcap.free.fr/Amiga/SetPatch40.16
http://foulcap.free.fr/Amiga/SetPatch40.17
http://foulcap.free.fr/Amiga/SetPatch43.6
http://foulcap.free.fr/Amiga/SetPatch44.2
http://foulcap.free.fr/Amiga/SetPatch44.13
http://foulcap.free.fr/Amiga/SetPatch44.16
http://foulcap.free.fr/Amiga/SetPatch44.38
http://foulcap.free.fr/Amiga/SetPatch44.39

:great

PHabermehl 05 April 2012 17:14

Je ne parle plus de Francais, mais "Document non trouvé" n'est pas bon quand j'essaye de telecharger v.44.39...

Thanks for your help, but could you check the link, please?

I meanwhile confirmed that v.43.6b and v44.38 show the same behaviour as 40.16.

Best regards,
Peter

Foul 05 April 2012 17:46

fixed/fixé ;)

43.6 is latest for Workbench 3.1, others are for OS 3.5/3.9

FrodeSolheim 05 April 2012 18:17

Quote:

Originally Posted by PHabermehl (Post 811661)
Hello Frode,
as soon as I activate P96 - even without using a P96 screen - display of other screens than WB gets corrupted.

Thank you for your help on PM -I got the problem reproduced now. Not surprisingly at all, the reason was the (temporary) ugly BEAMCON0-hack in v1.1 (I "knew" releasing a devel version with that one would come back and bite me somehow ;-) -but that was the price of getting the P96 preview "out there" for initial testing).

Quote:

Originally Posted by Foul (Post 811689)
fixed/fixé ;)
43.6 is latest for Workbench 3.1, others are for OS 3.5/3.9

Thank you, these will probably come in handy :)

PHabermehl 05 April 2012 21:24

Excellent international cooperation :-)

FrodeSolheim 05 April 2012 22:29

Quote:

Originally Posted by PHabermehl (Post 811725)
Excellent international cooperation :-)

Indeed :) -I tested a bit more with your startup sequence. With BEAMCON0 hack disabled, it is still a bit weird - lots of bits in BEAMCON0 is set (12, 11, 9, 8, 7, 3) when starting sysinfo, existing UAE code logs a warning, and autoscaling custom limits return 0, 0, 0, 0, so the result is that the entire display with overscan etc is shown (but no graphic corruption).

Code:

L: Display Size: 724x444 Offset: 2x6
L: First: 29 Last: 312 MinV: 28 MaxV: 251 Min: 26
L: >  2  6 724 444
L: BEAMCON0 v = 7048 (0x041b88)
L:  - bit 12
L:  - bit 11
L:  - bit 9
L:  - bit 8
L:  - bit 7
L:  - bit 3
L: warning: 1B88 written to BEAMCON0 PC=00F8F6D4
L: BEAMCON0 0020 -> 1b88 PC=%08x
L: BEAMCON0=1B88 VTOTAL=023D  HTOTAL=0081
L:  HSSTOP=0015 HBSTRT=0001  HBSTOP=0021
L:  VSSTOP=000E VBSTRT=023E  VBSTOP=0016
L:  HSSTRT=000B VSSTRT=0007 HCENTER=004B
L: NTSC mode dblscan V=47.4564Hz H=27240.0000Hz (130x574+1) IDX=11 (NTSC)
L: >  0  0  0  0

By trial and error, it seems that an important factor is the following line in your Startup-Sequence:
Code:

C:FastIPrefs nowait nogauge nofullwb
Without this line, the system behaves "normally"...

PHabermehl 05 April 2012 22:35

c'mon ... actually I already expected something like this. But I need FastIPrefs for allocating pens 4-7 for correct colors of MagicWorkbench and the background image :-)

and ... it worked on my real amiga, it worked on fs-uae <1.1.0. so, there must be an interference with the p96 emulation...

Seems like I have to upgrade to os3.9 and newicons soon :rolleyes

Just wanted to emulate my former A4k and began to remember all the hacks and patches I used to use. Have you noticed nickprefs to eleminate the floppy drive click? fs-uae behaves absolutely authentic. While booting my virtual hd, the "empty" drives click, and as soon as nickprefs is loaded, the drive click is gone. exactly as it was 20 years ago on my real amiga :-)

peter

nexusle 05 April 2012 22:56

@Peter:

Warum nutzt Du nicht "FullPalette"?! Damit kannst Du JEDE Farbe der Palette fest zuweisen. *

* Why don't you use "FullPalette"? With that you can assign ANY color palette set.

FrodeSolheim 05 April 2012 23:07

2 Attachment(s)
Quote:

Originally Posted by PHabermehl (Post 811737)
c'mon ... actually I already expected something like this. But I need FastIPrefs for allocating pens 4-7 for correct colors of MagicWorkbench and the background image :-)

and ... it worked on my real amiga, it worked on fs-uae <1.1.0. so, there must be an interference with the p96 emulation...r

Well, as I said, the graphic corruption itself was due to a hack I did for P96.

It will work again in 1.1.2 (but autoscaling won't work in this case). I also tested with WinUAE for reference (to make sure I haven't introduced any other bugs).

WinUAE 2.3.3: with autoscaling filter, only a black screen
WinUAE 2.4.0: with autoscaling filter, the entire amiga display is shown with overscan etc (same as the result with the FS-UAE I now have here).
WinUAE - latest unversioned beta/zip from today, with autoscaling filter, the entire screen is shown squashed to the left half of the screen while the right half is black.

I have added screenshot of the result of starting SysInfo on this setup with todays latest WinUAE.zip, and config + WinUAE logs attached, in case this is relevant for Toni's work on autoscaling.

EDIT: just to clarify, WinUAE (all versions) without autoscaling filter shows the amiga display just fine.

PHabermehl 05 April 2012 23:51

Quote:

Originally Posted by nexusle (Post 811741)
@Peter:

Warum nutzt Du nicht "FullPalette"?! Damit kannst Du JEDE Farbe der Palette fest zuweisen.

Well, I just tried to do again what I did 20 years ago... I never claimed that it makes any sense :D

And I don't know what was state-of-the-art later. When I used real Amiga Hardware, FastIPrefs was very popular, among many other hacks.

BTW, don't blame me for answering in English, but this is the ENGLISH Amiga Board and I don't want to exclude anybody...

Best regards
Peter

nexusle 05 April 2012 23:56

Yeah... And why you answered in french above?! ;)

PHabermehl 05 April 2012 23:59

Quote:

Originally Posted by nexusle (Post 811759)
Yeah... And why you answered in french above?! ;)

Because I can... Naaaw... actually a good point, got me :)

prowler 06 April 2012 00:16

C'mon, guys. Please stick to English, or provide an English translation as well.

weiju 06 April 2012 01:44

Quote:

Originally Posted by prowler (Post 811768)
C'mon, guys. Please stick to English, or provide an English translation as well.

My French is pretty rusty, but I think it was about a broken link to version 44.39 of SetPatch.
That said, it seems the link to 43.6 is broken :)

Wei-ju

Toni Wilen 06 April 2012 10:57

Ok, this is promoted mode (Promoted to 31KHz).

There are 2 problems:

Horizontal display end position (DIWSTOP) is _much_ farther than real end of display, for some reason it is after horizontal blanking which makes no sense. This can be compensated so that display looks better but..

Problem 2: Both horizontal DIWSTRT and DIWSTOP do not match bitmap's edges at all (lots of empty space in both sides, I assume it has something to do with mode promotion, "normal" modes are exactly matched), autoscale can't position this correctly.

-> Better option is to disable mode promotion (sys:prefs/icontrol)

PHabermehl 06 April 2012 15:53

Quote:

Originally Posted by Toni Wilen (Post 811818)

-> Better option is to disable mode promotion (sys:prefs/icontrol)

Toni,

Many thanks for your investigations - never would have found that on my own.

Deactivated mode promotion and it works now!

Thanks and happy easter holidays,
Peter


All times are GMT +2. The time now is 06:36.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, vBulletin Solutions Inc.

Page generated in 0.04644 seconds with 11 queries