graphics.library 40.25 beta series - Page 7

gulliver · 05 April 2010, 21:58

I am uploading it to The Zone!

DONE!

hit · 05 April 2010, 22:07

Quote:

Originally Posted by gulliver

I am uploading it to The Zone!

DONE!

thnx gulliver, lets hope Cosmos can now add some more cool stuff

08 April 2010, 15:34

Quote:

Originally Posted by Cosmos

superlayers.library from CGX is running on OCS/ECS/AGA Workbench ??

Yes, it worked on my classic amiga.that was short before year 2000.maybe you can try out and install latest CGX 3 to see if that work.try MCP opaque window move and you see big speedup.But the best is when you can look how that work and you can create a own faster layer library.A CGX driver for AGA is not need.But the agacgx is also intresting, maybe here can run SDL on AGA.

Ratte · 09 April 2010, 00:49

Quote:

Originally Posted by Cosmos

- Four 'jsr -$1B0(a6)' replaced by faster 'bsr.w R_LockLayerRom' (Cosmos)
- Four 'jsr -$1B6(a6)' replaced by faster 'bsr.w R_UnLockLayerRom' (Cosmos)
- Fourteen 'jsr -$2F4(a6)' replaced by faster 'bsr.w R_GetDisplayInfoData' (Cosmos)
- One 'jsr -$2E8(a6)' replaced by faster 'bsr.w R_AddDisplayInfoData' (Cosmos)
- Four 'jsr -$2EE(a6)' replaced by faster 'bsr.w R_SetDisplayInfoData' (Cosmos)
- Three 'jsr -$2DC(a6)' replaced by faster 'bsr.w R_NextDisplayInfo' (Cosmos)
- Ten 'jsr -$84(a6)' replaced by faster 'addq.b #1,$127(a6)' (Cosmos)

It isnt a good way to disable the jumptables.
It will be impossible to enhance single functions by 3rdparty software.
CGFX / P96 and other tools and also the philosophy of AmigaOS are broken by your way.
Jumptables arnt senseless ...

StingRay · 09 April 2010, 01:28

Quote:

Originally Posted by wawa

@korodny: calm down, why are you so upset?

He doesn't sound upset to me and I actually agree with everything he said! So far I didn't see a real explanation why P96 has "to die". Cosmos decided and that's how it should be now? How is that helping? I for one won't switch from a perfectly fine working P96 system to something that runs really unstable on my A4k (CGX). Unity? Dictatorship more like since the people have no choice anyway! For me one of the major reasons to not even try this updated graphics.library since I do not like this approach at all!

wawa · 09 April 2010, 02:05

@stingray: well, i think this is just his posting style. anyway i dont think dictatorship is in question here or anywere else in the community other than in the case people are going to welcome it willingly.

as for reason to replace gfx system i would name just the sole possibility to update and or develop new hardware drivers.

see the threads here:
http://www.amiga.org/forums/showthread.php?t=52261
and here:
http://www.amiga.org/forums/showthread.php?t=40173

09 April 2010, 04:35

@Cosmos

Please keep up your efforts! This is a worthwhile project.

Cosmos · 09 April 2010, 05:33

>Cosmos decided and that's how it should be now

We must choose one ! Two RTG system are too much complicated for users and developpers...

>How is that helping

Simplicity !

>Unity?

With complexicity, WE (all of us) lost time for nothing...

>For me one of the major reasons to not even try this updated graphics.library since I do not like this approach at all!

When finish, you'll boot directly in RTG (if gfx card detected) without any HD : so, you will change your mind !

>It isnt a good way to disable the jumptables.

jmptable for these functions are not requied = we have a slowdown for nothing... I know perfectly what I'm doing and what I'm not doing ! Only these functions are now bsr (with R_Disownblitter, R_Ownblitter and R_WaitBlit)

Ratte : I really don't know how our system is working with zillions patchs : for example, R_RectFill is patched TREE times with VisualPrefs, Birdie and others I don't remember on my HD...

Cosmos · 09 April 2010, 05:35

>Four 'jsr -$1B0(a6)' replaced by faster 'bsr.w R_LockLayerRom'
>Four 'jsr -$1B6(a6)' replaced by faster 'bsr.w R_UnLockLayerRom'

These functions are patched by P96 with the same code (4.w => $xxx(a6)) !

So, we lost precious fastram for nothing...

StingRay · 09 April 2010, 09:27

Quote:

Originally Posted by wawa

@stingray: well, i think this is just his posting style. anyway i dont think dictatorship is in question here or anywere else in the community other than in the case people are going to welcome it willingly.

Maybe it is his writing style, maybe not. Fact is, he decided that people have to use CGX now, if they like it or not. Unity? Yes. Free choice? No!

Quote:

Originally Posted by wawa

as for reason to replace gfx system i would name just the sole possibility to update and or develop new hardware drivers.

I don't have a problem with a completely new RTG system, in fact, I'm all up for it since I think that would be the best solution. But replacing one obsolete RTG system (last official P96 update is from 1999 IIRC) with another one that is even commercial and forcing people to use it is anything but a good solution. My A4k runs P96 fine (contrary to CGX as already said) and WinUAE supports P96 as well so why should I switch to CGX if the only reason is "unity"?

Quote:

Originally Posted by Cosmos

We must choose one ! Two RTG system are too much complicated for users and developpers...

How is "Cosmos decides that P96 must die" WE? YOU decided, no one else.
As said, I'm up for a completely new RTG system, developed from scratch, everything else is not going to work, not for me at least.

Quote:

Originally Posted by Cosmos

Simplicity !

Why not P96 then? Same simplicity!

Quote:

Originally Posted by Cosmos

When finish, you'll boot directly in RTG (if gfx card detected) without any HD : so, you will change your mind !

I won't change my mind about this. You FORCE me to use CGX, I say NO!
So far I still didn't see any valid reason why I should switch from P96 to CGX and so I'll happily continue to use P96 on my A4k and UAE.

Ratte · 09 April 2010, 09:40

Quote:

Originally Posted by Cosmos

>Four 'jsr -$1B0(a6)' replaced by faster 'bsr.w R_LockLayerRom'
>Four 'jsr -$1B6(a6)' replaced by faster 'bsr.w R_UnLockLayerRom'

These functions are patched by P96 with the same code (4.w => $xxx(a6)) !

So, we lost precious fastram for nothing...

P96 patched the functions with new offsets inside the jumptables.
You are disabling jumptables, its like 198x-democoding style with direct jumps to fixed systemoffsets.
This is against coding rules ...
Better safe some cpu-cycles in intensive code-parts, not in a simple indirect jumps.

mrodfr · 09 April 2010, 09:43

Quote:

Originally Posted by Cosmos

When finish, you'll boot directly in RTG (if gfx card detected) without any HD : so, you will change your mind !

I don't know the projects/roadmaps/choices, from Cosmos, about the RTG part and the use of CGX or P96.

Now, IMHO:

If the graphics.library (in ROM) allow to have also CGX screenmodes (and of course the old PAL screenmodes for compatibility and use of the Amiga on TV screens) immediately when swithing the AMIGA on, I don't see problems.

But the important fact is if P96 software is installed on the boot partition and a P96 monitors is present in devs, the amiga should switch to P96 during the boot.

After the workbench completely booted, if on screenmodes prefs, the users will see CGX screenmodes possible (in ROM) and P96 screenmode (from software) and PAL or NTSC screenmodes (in ROM) and could choose easily CGX or P96 screenmodes or PAL screenmodes, that will be the must !! ;-).

StingRay · 09 April 2010, 09:57

Quote:

Originally Posted by Ratte

Better safe some cpu-cycles in intensive code-parts, not in a simple indirect jumps.

Couldn't agree more!

Cosmos · 09 April 2010, 11:36

>P96 patched the functions with new offsets inside the jumptables

I'm talking about the code, not the offsets !

The beta 6 is working 100% with P96 or CGX

>You are disabling jumptables, its like 198x-democoding style with direct
>jumps to fixed systemoffsets. This is against coding rules ...

No. I repeat : R_LockLayerRom and UnlockLayerRom from P96 ARE THE SAME than in the original graphics.library v40.24

Watch by yourself ! 10000000000000000000000000000000000000000000000000% same

Oups... not 100% identical : P96 take 24 bytes, and the v40.24 (or v40.25 beta) take 22 bytes !

So P96 patch the graphics.library with SLOWER code !! I'm not sure that a 6-bytes-fastram-access ($x,a6) is faster than 4-bytes-chip-access (4.w,a6)...

Now Ratte, you understand why I call these functions by 'bsr.w' : it's 100% ok with coding rules...

Cosmos · 09 April 2010, 11:46

>Why not P96 then?

P96 or Cgx are identical when we are on the Workbench...

I choose CGX because of the BVision/CyberVision/GRex/3dfx...

>As said, I'm up for a completely new RTG system, developed from scratch, everything else is not going to work, not for me at least

Why not => so, must be 100% compatible with P96/CGX !

StingRay · 09 April 2010, 12:13

Quote:

Originally Posted by Cosmos

So P96 patch the graphics.library with SLOWER code !! I'm not sure that a 6-bytes-fastram-access (lea $x,a6) is faster than 4-bytes-chip-access (4.w,a6)...

You say the P96 patch is slower but in the same sentence you say you are not sure if your approach is faster. Not bad... I won't start any coding related discussions here but you might want to remember that fast ram is called fast ram for a reason!

Quote:

Originally Posted by Cosmos

P96 or Cgx are identical when we are on the Workbench : we need to choose only ONE RTG system !

Apparently they are not identical, P96 works stable on my A4k whereas CGX does not! Also, I still read "we"... "We" didn't choose, you did!

Quote:

Originally Posted by Cosmos

I choose CGX because of the BVision/CyberVision/GRex/3dfx...

Could you please eloborate a bit more? Since this is still no answer for me.

Quote:

Originally Posted by Cosmos

Why not => so, must be 100% compatible with P96/CGX !

Yes, and it should be free and probably open source but I don't think that this will ever happen. It would be the best solution though IMHO.

Cosmos · 09 April 2010, 12:32

>I won't start any coding related discussions here but you might want to remember that fast ram is called fast ram for a reason!

Hum... 2 bytes take 2 bytes in the precious code cache... 4 ko only on 040 and 8 ko only on 060...

>Could you please eloborate a bit more? Since this is still no answer for me.

CGX support BVision/CyberVision/GRex/3dfx and P96 no !

>P96 works stable on my A4k whereas CGX does not!

CGX is very stable here... Anyone have bugs with CGX 4 ???

>Also, I still read "we"... "We" didn't chose, you did!

Someone need to take a decision... To choose between many solutions... To show a direction...

Mad-Matt · 09 April 2010, 12:51

CGX v4 on Bvision is rock solid. No issues where I can say it was cgx that messed up. P96 on my mediator/voodoo is a mess and quite often boot stalls when it should open a sceen or a grey screen will open. p96 is also measurably slower as it doesnt have hardly seem to have any hardware acceleration apart from basic lines.

Im guessing cgx is also many developers rtg of choice since most rtg softwre will look for presence of cybergraohx and p96 relias very much on is cgx emu to make stuff work. Can only think of one or two bits of software that was 096 only (as cgx doesnt have a native emu)

wawa · 09 April 2010, 12:58

i think we ahve some serious misunderstandings here.

i believe cosmos wants to establish the new graphics.library along with an original with rtg system included. to do that he has to choose a system to be compatible to and it is cgx for a good reason since p96 is already compatible to it. also a good thing is if the new system patches the underlying old function by function while development like afaos imposes aros functions over respective aos functions in patched libraries.

none is going to force anyone to buy cgx or none is gonna take your p96 away!
i agree it should be open source to make sence, otherwise it will end like both previous standards. but aomeone must coordinate the effort.

Cosmos · 09 April 2010, 13:32

>Fourteen 'jsr -$2F4(a6)' replaced by faster 'bsr.w R_GetDisplayInfoData'

@Ratte : I checked CGX, and, you're right for this one ! (P96 don't patch it)

So, the beta 7 will have :

Fourteen 'bsr.w R_GetDisplayInfoData' replaced by 'jsr -$2F4(a6)'

09 April 2010, 11:36	#134
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>P96 patched the functions with new offsets inside the jumptables I'm talking about the code, not the offsets ! The beta 6 is working 100% with P96 or CGX >You are disabling jumptables, its like 198x-democoding style with direct >jumps to fixed systemoffsets. This is against coding rules ... No. I repeat : R_LockLayerRom and UnlockLayerRom from P96 ARE THE SAME than in the original graphics.library v40.24 Watch by yourself ! 10000000000000000000000000000000000000000000000000% same Oups... not 100% identical : P96 take 24 bytes, and the v40.24 (or v40.25 beta) take 22 bytes ! So P96 patch the graphics.library with SLOWER code !! I'm not sure that a 6-bytes-fastram-access ($x,a6) is faster than 4-bytes-chip-access (4.w,a6)... Now Ratte, you understand why I call these functions by 'bsr.w' : it's 100% ok with coding rules... Last edited by Cosmos; 16 July 2011 at 18:12.

09 April 2010, 11:46	#135
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>Why not P96 then? P96 or Cgx are identical when we are on the Workbench... I choose CGX because of the BVision/CyberVision/GRex/3dfx... >As said, I'm up for a completely new RTG system, developed from scratch, everything else is not going to work, not for me at least Why not => so, must be 100% compatible with P96/CGX ! Last edited by Cosmos; 09 April 2010 at 12:10.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
PUAE 2.5.1 beta series	GnoStiC	support.OtherUAE	3	07 February 2016 20:09
WinUAE 2.6.1 beta series	Toni Wilen	support.WinUAE	21	19 June 2013 11:54
PUAE 2.3.3 beta series	GnoStiC	support.OtherUAE	95	05 January 2012 19:18
dos.library v40.4 beta series	Cosmos	Coders. General	85	19 July 2011 01:49
WinUAE 2.3.1 beta series	Toni Wilen	support.WinUAE	90	23 February 2011 21:17

05 April 2010, 21:58	#121
gulliver BoingBagged Join Date: Aug 2007 Location: The South of nowhere Age: 46 Posts: 2,358	I am uploading it to The Zone! DONE!

09 April 2010, 02:05	#126
wawa Registered User Join Date: Aug 2007 Location: berlin/germany Posts: 1,054	@stingray: well, i think this is just his posting style. anyway i dont think dictatorship is in question here or anywere else in the community other than in the case people are going to welcome it willingly. as for reason to replace gfx system i would name just the sole possibility to update and or develop new hardware drivers. see the threads here: http://www.amiga.org/forums/showthread.php?t=52261 and here: http://www.amiga.org/forums/showthread.php?t=40173

09 April 2010, 04:35	#127
zoneuser Posts: n/a	@Cosmos Please keep up your efforts! This is a worthwhile project.

09 April 2010, 05:33	#128
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>Cosmos decided and that's how it should be now We must choose one ! Two RTG system are too much complicated for users and developpers... >How is that helping Simplicity ! >Unity? With complexicity, WE (all of us) lost time for nothing... >For me one of the major reasons to not even try this updated graphics.library since I do not like this approach at all! When finish, you'll boot directly in RTG (if gfx card detected) without any HD : so, you will change your mind ! >It isnt a good way to disable the jumptables. jmptable for these functions are not requied = we have a slowdown for nothing... I know perfectly what I'm doing and what I'm not doing ! Only these functions are now bsr (with R_Disownblitter, R_Ownblitter and R_WaitBlit) Ratte : I really don't know how our system is working with zillions patchs : for example, R_RectFill is patched TREE times with VisualPrefs, Birdie and others I don't remember on my HD...

09 April 2010, 05:35	#129
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>Four 'jsr -$1B0(a6)' replaced by faster 'bsr.w R_LockLayerRom' >Four 'jsr -$1B6(a6)' replaced by faster 'bsr.w R_UnLockLayerRom' These functions are patched by P96 with the same code (4.w => $xxx(a6)) ! So, we lost precious fastram for nothing...

09 April 2010, 12:32	#137
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>I won't start any coding related discussions here but you might want to remember that fast ram is called fast ram for a reason! Hum... 2 bytes take 2 bytes in the precious code cache... 4 ko only on 040 and 8 ko only on 060... >Could you please eloborate a bit more? Since this is still no answer for me. CGX support BVision/CyberVision/GRex/3dfx and P96 no ! >P96 works stable on my A4k whereas CGX does not! CGX is very stable here... Anyone have bugs with CGX 4 ??? >Also, I still read "we"... "We" didn't chose, you did! Someone need to take a decision... To choose between many solutions... To show a direction...

09 April 2010, 12:51	#138
Mad-Matt Longplayer Join Date: Jan 2005 Location: Lincoln / UK Age: 45 Posts: 1,853	CGX v4 on Bvision is rock solid. No issues where I can say it was cgx that messed up. P96 on my mediator/voodoo is a mess and quite often boot stalls when it should open a sceen or a grey screen will open. p96 is also measurably slower as it doesnt have hardly seem to have any hardware acceleration apart from basic lines. Im guessing cgx is also many developers rtg of choice since most rtg softwre will look for presence of cybergraohx and p96 relias very much on is cgx emu to make stuff work. Can only think of one or two bits of software that was 096 only (as cgx doesnt have a native emu)

09 April 2010, 12:58	#139
wawa Registered User Join Date: Aug 2007 Location: berlin/germany Posts: 1,054	i think we ahve some serious misunderstandings here. i believe cosmos wants to establish the new graphics.library along with an original with rtg system included. to do that he has to choose a system to be compatible to and it is cgx for a good reason since p96 is already compatible to it. also a good thing is if the new system patches the underlying old function by function while development like afaos imposes aros functions over respective aos functions in patched libraries. none is going to force anyone to buy cgx or none is gonna take your p96 away! i agree it should be open source to make sence, otherwise it will end like both previous standards. but aomeone must coordinate the effort.

09 April 2010, 13:32	#140
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	>Fourteen 'jsr -$2F4(a6)' replaced by faster 'bsr.w R_GetDisplayInfoData' @Ratte : I checked CGX, and, you're right for this one ! (P96 don't patch it) So, the beta 7 will have : Fourteen 'bsr.w R_GetDisplayInfoData' replaced by 'jsr -$2F4(a6)'

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)