Curse of the SDL

[commodorejohn]

Yeah, but how many people have one? I didn't even have an RTG card until fairly recently.

[matthey]

Clickboom's quake doesn't do warp3d. I belive that with warp 3d support result would similar, ie. sldquake / glquake

Hardware 3D support gives a significant speedup. Here are my results with CSMK3 060@75MHz and Mediator with Voodoo 4 but no Cyberpatcher type patcher...

640x480x16
-----------
Frank Wille's Quake 3-4 fps
Clickboom Quake060 4-5 fps
QuakeGL 20-25 fps (looks best too)

640x480x8
----------
Frank Wille's Quake 4-5 fps
Clickboom Quake060 5-6 fps
QuakeGL No 8 bit hardware 3D support on Voodoo 3+

Hardware 3D acceleration does matter. It's probably more important with a slow bus.

It is 060 version, it much faster than original exe with 040 also.

It's not that most compilers do a good job of optimizing for the 060 but rather that 040 compiled code doesn't run very well on the 060.

Bern is author of SDL version of Netdurf, so in this case 3d tests are useless.

Yea, 3D hardware acceleration isn't going to help much with Netsurf unless it can be used like kas1e does for his disk mag which is fast and nice on the Voodoo.

Maybe 80% todays classic amigas has at least 030 and 64mb ram? Maybe 50% has 040/060 with at least 64mb ram? 40% has a somekind of RTG card? 20% has a mediator or similar with at least 8mb ram on their garphics card?

Those are plausible guesses of what power classic Amiga users use. Hopefully the new fpga Amigas will help out those numbers and give some additional 3D accelerated hardware options.

[utri007]

Good target system would be 030/040, 64mb ram and 2mb graphicscard/AGA.

There is plenty of app/games for 020/8mb and AGA. There is even more apps/games for systems like 060, 64mb ram and 16mb voodoo than those specs that I just suggested.

It seems that the games are either 020/8mb or 060/128mb/3d, middle level amigas has completely forgotten

There is no point to argue, how thing should do with modern systems, because most of us don't have 3D and 68k Amigas isn't modern hardware. Thats why SDL sucks with 68k Amigas

[commodorejohn]

Yeah. Speaking for myself, I have an A3000 with a 28MHz G-Force 040 accelerator, 8MB accelerator RAM, 8MB motherboard fast RAM, 2MB chip RAM, and a 2MB RTG card. Aside from perhaps a simple sound card and maybe a 40MHz 040, this is as much as I'm likely to ever have, as the Amiga is just a hobby for me, not something for which I could justify the prices high-end accelerators with lots of RAM command. Of course, I don't think people should be obligated to target my system, but it would be nice to see some more Amiga development for something less than the creme de la creme (to say nothing of the poor neglected unexpanded OCS Amiga.)

[DavidF215]

Nobody who have common sense doesn't dream to play halflife with his/her 68k amiga.

Ah sure. It might work with 8 bit graphics.

Bernd_afa: have written a game engine for amiblitz and storm mesa, but nobody write a fast working game for it, and Thilo choose his own 2d engine that can do some blits on 2d Card.

Is your engine publicly available? If so, then where? I wouldn't mind doing some experimenting with it.

[utri007]

You should add more ram to your system. I do understand that amiga is hobby machine, but hobbies cost. Does it require GVP sims or can you use standard 72 pin sims? It would be about 200$ to get 64mb to your accelerator with GVP sims, but I'm sure that you woun't repent it.

[commodorejohn]

You should add more ram to your system. I do understand that amiga is hobby machine, but hobbies cost. Does it require GVP sims or can you use standard 72 pin sims? It would be about 200$ to get 64mb to your accelerator with GVP sims, but I'm sure that you woun't repent it.

1MB GVP SIMMs only. I'd have to swap out the accelerator entirely to get it above 8MB. $200 is within reason for the occasional purchase, but the fact is that I just don't use my Amiga for enough to justify spending that money on a RAM expansion instead of, say, buying a few games or saving towards another interesting old system. 18MB should be plenty for a lot of purposes, certainly for a lot of non-3D gaming, but with so many programs that are just PC ports via SDL that bring with them PC expectations of RAM capacity and CPU horsepower, not so much.

[unusedunused]

Bern is author of SDL version of Netdurf, so in this case 3d tests are useless.

I am not the author of netsurf SDL.Artur do a alot and great work on netsurf to add features that netsurf sdl version(in mains source do not have).In newest netsurf artur use agar GUI for GUI in some gadgets

http://libagar.org/

, i do not like the mix, i have told Artur, because i fear some hang problems due to message lost.I am not familar with that

But the sdl version can better update with new core, and if the standard core not run on amiga, i help Artur.

All other do Artur alone in great work.He have done lots work to bring a browser to amiga.

when no Artur was here, i only compile from time to time the sdl version and release it and before no java script is in, i do not do any work to enhance it.

[unusedunused]

Ah sure. It might work with 8 bit graphics.


Is your engine publicly available? If so, then where? I wouldn't mind doing some experimenting with it.

its in sourceforge for amiblitz 3.it work in amiblitz 3, there is a demo in the amiblitz3/sourcecodes/includes/gl.bb2 include.

amiblitz execute the example if the file is not include and is use standalone.
see here the syntax of that example program.its easy to bring 2d games to 3d, because you can choose a virtual coordinate system.I decide to use stormmesa, but native PPC stormmesa do not work with new OS4 warp3d.because OS4 devs remove a feature in warp3d that stormmesa need.If it work on MOS, i dont know, because no intrestet user have MOS and want test.

The code below produce this testimage.later version should load milkshape objects make out of 2d easy 3d with depthmap etc.but because stormmesa is not go to a standard GL API, i give up, because i dont want do special native versions.and thilo do continue his 2d engine  

CNIF #__include=0
  ;Amiblitz 2 test for opengl
  Syntax 2  ;the relax declare mode
  optimize 7

  WBStartup


  .begin
  width.l=320:height.l=200
  gl_init{width,height,16,0}  ;the screen open
  gl_2dview{640,480}          ;need to set coord leftedge=0 rightedge=640 and hold size of pics
                              ;depent on this resolution
                              ;topedge=0 and buttom edge=480
                              ;on default ogl use leftedge=-1 rightedge+1 top +1 buttom -1
  gl_load {0,"data/pattern"}
;  Stop
  gl_load {1,"data/ball64x64",0}
  gl_load {2,"data/reflect.jpg",0}
  gl_load {3,"data/times32.png",0}

  gl_sphere {4,40,0,15,15}            ;create the sphere at beginning for more speed.But possible on all place
  gl_cylinder{5,110,30,50,0,15,15}    ;a cylinder with texture 0
  gl_cylinder{6,110,30,50,2,15,15}    ;another cylinder for the reflection with texture2
  gl_quad{7,50,50,50,0}
  gl_quad{8,50,50,50,2}

  .startvals
  zval.f=0:xval.f=320:yval.f=240:trans.f=255:yvallight.f=yval:rot.f=6:tsize.f=1:rot.f=84
  Repeat
    ev.l = Event
    Select ev
    Case #IDCMP_RAWKEY
      Select EventCode
      Case $4e : xval+10
      Case $4f : xval-10
      Case $4c : yval-10
      Case $4d : yval+10
      Case $13 : rot.f+6
      Case $45 : ev=#IDCMP_CLOSEWINDOW
      End Select

      Select EventVanillaKey
      Case @"q" : zval-10
      Case @"a" : zval+10
      Case @"t" : trans.f+10
      Case @"g" : trans.f-10
      Case @"m" : tpos.l+2
      Case @"Q" : yvallight.f-10
      Case @"A" : yvallight+10
      Case @"s" : tsize.f*1.01
      Case @"S" : tsize.f*0.99
      ;Case @"r" : rot.f+6
      End Select
    End Select
    FlushEvents
    ;If light Then glEnable_ #GL_LIGHTING: Else glDisable_ #GL_LIGHTING


    ;this example Show all what is possible with that 3dlib
    gl_blit{0,xval,yval} ;simple blit a image
    gl_light{ 0,320,yvallight,0,255,255,160} ;switch on a yellow spotlight
    gl_color{255,150,150}         ;tint texture red
        ;blit this image
    gl_texcoord{0,tpos,0,128*tsize,128*tsize} ;for scroll the texture with m key
    gl_blit{0,xval+128,yval}      ;and blit


    gl_light{0}                   ;now use the default light.
    gl_blit{1,200,yval,zval,-1,0,0,rot} ;another image but rotatable -1 mean make that not transparent

    ;now blit a reflective object
    gl_blit{7,xval-128,yval-128,zval,-1,rot+15,rot+15,0}
    gl_texmode{3}       ;now switch on the spherical mode for the reflect effect
    gl_blit{8,xval-128,yval-128,zval,255,rot+15,rot+15,0} ;blit the reflection map

    gl_light{ 0,220,yvallight,0,255,255,160} ;switch on a yellow spotlight for the font

    gl_usefont {3,32,32,0.3} ;use picture 3 with 32*32 tile and scale the font by 0.5
    gl_textcolor {220,220,220,200} ;we dont want the color of the bitmap
    gl_print{"test",48,200}        ;print 2 lines
    gl_nprint{" 2. Part"}          ;print 2 lines
    gl_textcolor {0,200,255}       ;change color again
    gl_print{"This is a line scale to 0.4",65,230,0.4}
    gl_light{0}                    ;default light

    gl_color{160}           ;make 2. ball a little darker only 1 color mean r g and b is set to 160
    gl_scale{2.4,1.0,1.0}   ;size the ball in x
    gl_blit{1,300,yval,zval,trans,0,0,rot} ;make the ball transparent choose with g and t
    gl_blit{4,300,100,0,-1,rot}

    gl_color{255,200,0}
    gl_scale{0.3,1,1}
    gl_blit{4,320,250,zval,-1,-1,rot,-1}     ;blit another sphere but diffrent color and size


    gl_color{140,70,140}                     ;blit a reflective cone
    gl_scale{0.5,1,1}
    gl_blit{5,180,370,zval,-1,-1,rot,-1}
    gl_scale{0.5,1,1}
    gl_texmode{3}
    gl_blit{6,180,370,zval,255,-1,rot,-1}

    gl_texmode{3}
    gl_blit{6,320,270,zval,150,-1,rot,-1}    ;blit the transparency cone

    gl_nprint{"Key r=Rotate "+Str$(rot),360,0}
    gl_nprint{"m=Move Texture",360}
    gl_nprint{"s=Size Texture",360}
    gl_nprint{"Crsr = move",360}
    gl_nprint{"q    = z +",360}
    gl_nprint{"a    = z -",360}
    gl_nprint{"Q    = Light y +",360}
    gl_nprint{"A    = Light y -",360}


    Delay_ 1
    gl_show {}  ;show the render and clear all realtime effects (it call gl_reset)
  Until ev=#IDCMP_CLOSEWINDOW
  End
CEND

[Cammy]

Here is an old video from AmiTopiaTV which has a clip of me at the start playing Day of the Tentacle on my 8MB 68020 A1200 through WarpSDL. Just to show that an AGA-native SDL library can work, and that it's not too slow at all (this version of SCUMMVM runs far faster than any other, DOTT isn't playable even on my 030 unless I use WarpSCUMM).

[youtube]E9uGOOv6uxA[/youtube]

Video: http://www.youtube.com/watch?v=E9uGOOv6uxA
WarpSDL: http://www.algonet.se/~chaozer/warpsdl.shtml

[Crumb]

I am not the author of netsurf SDL.Artur do a alot and great work on netsurf to add features that netsurf sdl version(in mains source do not have).In newest netsurf artur use agar GUI for GUI in some gadgets

http://libagar.org/

, i do not like the mix, i have told Artur, because i fear some hang problems due to message lost.I am not familar with that

But the sdl version can better update with new core, and if the standard core not run on amiga, i help Artur.

I think both of you are doing a good job but I wonder why you don´t adapt OS4 Reaction/ClassAct code as that would probably bring a notable improvement in usability and a good speed up in GUI too. In addition to that I would get rid of nasty ixemul and would use libc2 or libnix instead.

Uploading dependencies sources to Aminet would be helpful too I guess, perhaps that way somebody gets more interested in helping.

[unusedunused]

Is it a problem? There are 3D gfx cards for Amiga.

this give me the idea, to create for the SDL display a Warp3d context and implement for SDL hardware alpha blit function a function that paint a rectangle with texture in warp3d.
or look at other SDL implementation that use for this opengl.
so the Limit is not SDL.When a natami have a alpha blit function, natami can do too all blits lots faster with the blitter.and the CPU can init in the time the blitter work, the next alpha blit.so all work in same way as amiga blitter but with alpha channel

As i write before, SDL is not bad written, it have less calling overhead in blit functions as P96 and CGX.
Only on Amiga there is no hardware alpha support, and all need done in software.

On classic is 256 pixel texure limit, but the blit can split in diffrent calls automatic.

maybe i should ask natami devs, if i can get a natami to add SDl and hardware alpha support for natami. ;-)

But i am really lazy, maybe somebody else can do it.

Here is an old video from AmiTopiaTV which has a clip of me at the start playing Day of the Tentacle on my 8MB 68020 A1200 through WarpSDL. Just to show that an AGA-native SDL library can work, and that it's not too slow at all (this version of SCUMMVM runs far faster than any other, DOTT isn't playable even on my 030 unless I use WarpSCUMM).

[youtube]E9uGOOv6uxA[/youtube]

Video: http://www.youtube.com/watch?v=E9uGOOv6uxA
WarpSDL: http://www.algonet.se/~chaozer/warpsdl.shtml

the good thing on scumm is, that not the whole content of the display need redraw with 20 or more fps /sec(SDL can check to only redraw changed parts), and scumm games are very old.most games do not more as 320*240

Remember when the last lucas arts adventure with scummvm is release, and how much CPU power was here.

because when lucasarts scumm script system was program, PC are not fast.

but look when most SDL games are written.

SDL support alpha in Hardware, so if maybe the natami support a OS function that can blit alpha in hardware with blitter, then SDL can run lots faster.

SDL blit have less call overhead as P96.

I forget to write before

sdlquake and quake68k use only software render

[unusedunused]

Conclusion:

SDLquake is 37%-22% less speedy than clickboom's quake

Such a low spec machines every frame is important. Good result any way, I would ques that it is much more worse.

then try the test.let somebody else start sdlquake or quake68k in random order.
Now you need tell in 20 tries correct what version is faster.If you can tell this in 100% of the tries if the game run in 3 fps or 5 fps, then you are really good.

I know from my own tests, it doesnt matter, the game feel too slow.so you can see a classic is not able to do games in 640*480 16 bit

and quake is a good case test, remember quake have only 8 bit screen.When you use a 16 bit screen all go slower

@matthey

have you use for sdlquake the version i upload ?
the old version is lots slower.

Artur have use for all his games new SDl

[unusedunused]

I think both of you are doing a good job but I wonder why you don´t adapt OS4 Reaction/ClassAct code as that would probably bring a notable improvement in usability and a good speed up in GUI too. In addition to that I would get rid of nasty ixemul and would use libc2 or libnix instead.

Uploading dependencies sources to Aminet would be helpful too I guess, perhaps that way somebody gets more interested in helping.

Reaction GUI look very ugly, itix have done years ago a MUI interface for netsurf.thats the better choose, and we begin to port MOS MUI netsurf and it work a little.

But netsurf team change the backend interface lots, so itix version do not work with newer netsurf core.

itix does not want change code to new backend interface and give up netsurf for MOS.
I can understand him, its really not nice what netsurf developers do, they often change their API

So netsurf 68k use SDL, because SDL is mainted by a netsurf developer and all backend API changes he add in the SDL code.so amiga version can more easy keep upto date.
and when netsurf developers get not the idea to change backend api lots, then netsurf 68k have a MUI interface. ;-)

but netsurf dev change backend API, so netsurf 68k use SDL. ;-(

and only when java and doom work in netsurf, i have motivation to do more time spend on netsurf.

[utri007]

Reaction is not that ugly and it would be much easier task than start from scrats with MUI

Why not make better screen support? 8 bit screens? Amiga fonts? That should be possible wit SDL also. It would make lots of more useability to it and with Reaction/MUI gui it would feel like native amiga program.

[unusedunused]

Reaction is not that ugly and it would be much easier task than start from scrats with MUI

the reaction flat non shaded GUI elements from reaction i find more ugly as the current GUI elements in Netsurf
But when use zune MUI programs can look nicer.there is a MUI GUI for netsurf too here.
and OS4 reaction work diffrent as OS3 reaction.and OS3 reaction is a dead end.no enhance possible.MUI can enhance in zune.

also reaction can program really bad, class render is done in input device task.this mean if something fail to render in netsurf class, the whole amiga is dead, because when input device crash, no program can get mouse messages, or keyboard messages.

debuggers can not work to show correct errors.
all in all i see Reaction as a bad design.I do not like MUI too, because i want a GUI system that need not compile in the program code, and i want a GUI Editor, but at least MUI does not render in input device task.

Why not make better screen support? 8 bit screens? Amiga fonts? That should be possible wit SDL also. It would make lots of more useability to it and with Reaction/MUI gui it would feel like native amiga program.

the answer is simple. no sense see to spend alot work for this.and for something that make not fun, money is maybe a motivation.

Here can look on OS4, user pay for lots that is on other systems free(for example firefox and other bounties).But on 68k nobody do bounty, but whine when netsurf not run on a non gfx card system faster as on a fast PC.

I have the feeling some classic users want not pay any cent to upgrate their hardware.
But they want that developers work day and night for free, so that software that run on fast hardware run on their slow hardware too, so they need not put any cent to upgrade the amiga.

Instead the developer should spend work for free to strip the software down so it work on a Amiga.

what advantage should bring to use amiga fonts.I see only lots more work.
the fonts look ugly for inet pages.and noticable faster it not get.

I have written in older post, compare the netsurf load time with internal fonts(internal fonts work  lots faster as amiga fonts) and truetype fonts.
If it really get faster with internal fonts, its possible to disable the antialiasing of the fonts.this cost most speed.

but as i told before, in real world pages, font render time is not very speed critical.most time is need for CSS layout, image decode and image show.

Software development cost lots time, so a good developer need to choose, time to add a feature and the usability of a code.
because time is short, then time need spend for features that bring a really good and usefull enhancement.