Login
Username:

Password:

Remember me



Lost Password?

Register now!
Sections
Who's Online
58 user(s) are online (47 user(s) are browsing Forums)

Members: 0
Guests: 58

more...
Support us!
Recent OS4 Files
OS4Depot.net



« 1 2 3 4 (5) 6 7 8 ... 30 »


Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2008/4/10 13:57
From Northern Ireland
Posts: 901
anyone else getting a gfx memory leak with this? After several videos I am sitting at 87% GFX RAm used. It is not being released...

_________________
AmigaOne X1000.
Radeon R9 270X

http://www.tinylife.org.uk/
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@kas1e

Thank you.

Progress on vo_comp output driver:
* Back-fill routine now implemented for vo_comp
* Routine to find max screen resolution implemented.
* Alignment changed from 32bit to 64bit..

Other stuff:
* Striped down p96_pip (don't think it works, but anyway its there)
* Linux paths fix for mplayer.
* Added MikeJT SDL full screen fix.

_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@ddni

whit this?

is that whit vo_comp or anything?

_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Not too shy to talk
Joined:
2008/5/2 12:14
From Sweden
Posts: 499
I did some benchmarks on my X1000 with RadeonHD6xxx with a file 720p/H264 file that is too slow to play no matter what:

MUI-mplayer, cgx_wpa:
BENCHMARKs: VC: 28.337s VO: 13.846s A: 0.383s Sys: 0.025s = 42.591s
BENCHMARK%: VC: 66.5325% VO: 32.5089% A: 0.8999% Sys: 0.0586% = 100.0000%

This new mplayer, cgx_wpa:
BENCHMARKs: VC: 25.724s VO: 13.728s A: 0.635s Sys: 0.047s = 40.133s
BENCHMARK%: VC: 64.0952% VO: 34.2052% A: 1.5817% Sys: 0.1179% = 100.0000%

This new mplayer beta2, cgx_wpa:
BENCHMARKs: VC: 25.643s VO: 13.716s A: 0.364s Sys: 0.048s = 39.771s
BENCHMARK%: VC: 64.4765% VO: 34.4873% A: 0.9147% Sys: 0.1215% = 100.0000%

This new mplayer beta2, comp:
BENCHMARKs: VC: 25.503s VO: 4.524s A: 0.357s Sys: 0.048s = 30.432s
BENCHMARK%: VC: 83.8022% VO: 14.8666% A: 1.1729% Sys: 0.1583% = 100.0000%

The comp driver is the fastest of them all with the shortest playback time and lowest VO time. The cgx_wpa driver in the MUI-mplayer is slightly faster than this new mplayer, but not by much. But it's h264 codec is a bit slower so it still ends up at the bottom of these.

Btw: How to I get OSB fonts to load properly? It looks in ~/.mplayer/subfont.ttf. I created the .mplayer/, and later ~/.mplayer/ but it still can't find it.

_________________
Software developer for Amiga OS3 and OS4.
Develops for OnyxSoft and the Amiga using E and C and occasionally C++
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@Deniil

Quote:
file that is too slow to play no matter what:


Its the audio output that is the problem.

use the "-ao" option to set audio output to SDL.

Quote:
Btw: How to I get OSB fonts to load properly? It looks in ~/.mplayer/subfont.ttf. I created the .mplayer/, and later ~/.mplayer/ but it still can't find it.


you don't as I just fixed that.

And the error messages in mplayer is misleading, use snoopy to see where really looks for it.

EDIT:
corrected the error message for next version.


Edited by LiveForIt on 2014/4/6 18:33:40
_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2006/11/29 8:09
From Lyon, France
Posts: 992
@LiveForIt

Let's hope p96_pip works with your version on MPLayer in order fot us to have only ONE version for every system. And we'll also need a Non AltiVec version

I don not have the money right now (next month I hope) but I shoulg dive to everyone involved : Kas1e, LiveForIt (you) and Joerg (and maybe some I do forget ).

_________________
--
AmigaONE X1000 1,8Ghz, 2GB, Sapphire Radeon HD 7700, with Amiga OS 4.1 FE
http://www.amiga-ng.org
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@joerg

Quote:
The cgx_* modes are MorphOS/CyberGraphX code and of course not usable on AmigaOS (AmigaOS doesn't have a cybergraphics.library, except for a very minimal emulation for ancient AmigaOS 3.x/m68k software only).


its probably just a wrapper, what ever goes into CyberGraphX end up in Picasso96, I can change and see what happens if I change to ARGB.

The wrapper is JIT compiled and most of the work is done by function it calls natively, I don't think it going to do a major difference.

What I wrote earlier on is wrong, it does not use a bitmap at all, and just use a memory buffer, it uses the Cybergraphic equivalent to IP96->p96WritePixelArray(), to draw the graphics.

It should be calling Picasso96 function direct, but it's not important.

Maybe the the best thing to do is use some macro magic or if defines, its not like the source code is flooded whit CyberGraphX stuff.

Stupid to have two different versions of the source code to maintain, that will be just be more work in the end.


Edited by LiveForIt on 2014/4/7 3:10:55
Edited by LiveForIt on 2014/4/7 3:11:39
_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@K-L

Quote:
(and maybe some I do forget )


Maybe MakeJT, he did some work on SDL.

_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@kas1e and @fab

Quote:
It didn't have gui code , so we need just reuse it as well.


Maybe I'm being overcritical but it has some GUI event code there.

I'm not saying we don't need some event code for the window.

But if you have a different GUI's maybe there should be different event codes as well.

That's what I'm thinking.

That's way I think it should be in GUI/ maybe even create a file called GUI/AmigaOS_no_gui.c, where all window handling goes, Opening, closing window, events.


Edited by LiveForIt on 2014/4/7 3:37:51
Edited by LiveForIt on 2014/4/7 3:40:32
Edited by LiveForIt on 2014/4/7 3:42:38
Edited by LiveForIt on 2014/4/7 4:30:58
_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@all
beta3 out, try it and see how it works.

https://dl.dropboxusercontent.com/u/69 ... veForIt_mplayer_beta3.lha


Progress on vo_comp output driver:
* Back-fill routine now implemented for vo_comp
* Routine to find max screen resolution implemented.
* Alignment changed from 32bit to 64bit..

Other stuff:
* Striped down p96_pip (don't think it works, but anyway its there)
* Linux paths fix for mplayer.
* Added MikeJT SDL full screen fix.
* New cgx_wpa driver whit ARGB (should be faster)

_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2007/9/11 11:31
From Russia
Posts: 5390
@LiveForIT
Quote:

Progress on vo_comp output driver


On my peg2/radeon9250, vo_comp works almost the same by speed as cgx_wpa, just a little faster (cgx_wpa give 45% cpu loading on some video, vo_comp give 41-42% cpu loading). overlay driver give 26% of cpu loading on same video.

In compare with vo_comp from beta2, its the same for my config.

Quote:

Striped down p96_pip (don't think it works, but anyway its there)


P96 driver works in window mode, just events didn't handles (i.e. can't make fullscreen by dbl-click, can't press "esc"). Running it with -fs in command line (so to fullscreen), make it crash:

Quote:

Stack trace:
(0x6203AE80) control()+0x7c (section 1 @ 0x319E4)
(0x6203AEA0) config_video_out()+0x88 (section 1 @ 0x1CE18)
(0x6203AEE0) config()+0x260 (section 1 @ 0x9F314)
(0x6203AF50) vf_config_wrapper()+0x8c (section 1 @ 0x6C760)
(0x6203AF70) mpcodecs_config_vo()+0x638 (section 1 @ 0x69AA0)
(0x6203B000) init_vo()+0x1f8 (section 1 @ 0x184CE0)
(0x6203B060) decode()+0x468 (section 1 @ 0x1852AC)
(0x6203B140) decode_video()+0x74 (section 1 @ 0x65F68)
(0x6203B180) update_video()+0x804 (section 1 @ 0xC03C)
(0x6203B250) main()+0x2b10 (section 1 @ 0x10040)
(0x6203CD00) native kernel module newlib.library.kmod+0x000020a4
(0x6203CD70) native kernel module newlib.library.kmod+0x00002d0c
(0x6203CF10) native kernel module newlib.library.kmod+0x00002ee8
(0x6203CF50) _start()+0x170 (section 1 @ 0x16C)
(0x6203CF90) native kernel module dos.library.kmod+0x0002533c
(0x6203CFC0) native kernel module kernel+0x0006acdc
(0x6203CFD0) native kernel module kernel+0x0006ad5c


Quote:

New cgx_wpa driver whit ARGB (should be faster)


Seems its actually slower.

Is your beta3 still altivec binary or generic ? Why i ask, is because now in beta3 your cgx_wpa the same by speed as my cgx_wpa on generic binary. While, in your beta2 it was on 10% faster (becaus of altivec). If now its generic binary, then its probably indeed a bit faster (on 2-3%), but if that altivec binary still, then something wrong and it slower than before on 10%.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2006/11/29 8:09
From Lyon, France
Posts: 992
@kas1e

Are you sure P96_PIP works with LiveForIt's Beta 3? When entering -vo p96_pip I only get : "Error opening/initializing the selected video_out (-vo) device".

And vo_comp still doesn't work with my Radeon 9000 Pro (grey screen and 100% CPU). Weird since it works for you.

Regarding Altivec, wehen looking a the debug output of MPlayer I can see this : "[swscaler @ 0x591c3bb0]ALTIVEC: Color Space ARGB"

_________________
--
AmigaONE X1000 1,8Ghz, 2GB, Sapphire Radeon HD 7700, with Amiga OS 4.1 FE
http://www.amiga-ng.org
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@kas1e

Quote:
Seems its actually slower.


Its faster on X1000, so the information Hans gave is not 100% correct. Correct for X1000, but wrong for Pegasus, so I should have two versions then.

Anyway be nice if some get some reports from AmigaONE-XE users, maybe DMA is broken on Pegasus.

_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2006/11/20 16:26
From Norway
Posts: 2734
@K-L

Quote:
When entering -vo p96_pip I only get : "Error opening/initializing the selected video_out (-vo) device".


try

-vo pip


_________________
(NutsAboutAmiga)

Basilisk II for AmigaOS4
AmigaInputAnywhere
Excalibur
and other tools and apps.
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2006/11/29 8:09
From Lyon, France
Posts: 992
@LiveForIt

Thanks

Here are the results with this new version on my AmigaOne XE G4 1,26 Ghz and Radeon 9000 Pro (AGP) :

Beta3 PIP :

BENCHMARKs: VC: 18.722s VO: 6.711s A: 0.000s Sys: 0.729s = 26.162s
BENCHMARK%: VC: 71.5600% VO: 25.6534% A: 0.0000% Sys: 2.7866% = 100.0000%

It's now the faster version of all the versions I've tested so far.

CGX_WPA :

BENCHMARKs: VC: 20.684s VO: 26.026s A: 0.000s Sys: 0.699s = 47.410s
BENCHMARK%: VC: 43.6287% VO: 54.8971% A: 0.0000% Sys: 1.4741% = 100.0000%

(solwer than the 1st version you released since I got :

BENCHMARKs: VC: 20.206s VO: 21.766s A: 0.000s Sys: 0.692s = 42.664s
BENCHMARK%: VC: 47.3599% VO: 51.0180% A: 0.0000% Sys: 1.6222% = 100.0000%
)

SDL :

BENCHMARKs: VC: 20.890s VO: 31.473s A: 0.000s Sys: 1.045s = 53.408s
BENCHMARK%: VC: 39.1133% VO: 58.9302% A: 0.0000% Sys: 1.9565% = 100.0000%

SDL is still really slow on non Radeon HD systems and results of every version is about the same with this video output on my A1.

Comp

Still doesn't work on my Radeon 9000 Pro on AGP (grey screen and 100%CPU usage). I don't know why. I doubt this card doesn't support compositing since Radeon HD Boing Ball from Hans works.

_________________
--
AmigaONE X1000 1,8Ghz, 2GB, Sapphire Radeon HD 7700, with Amiga OS 4.1 FE
http://www.amiga-ng.org
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2007/9/11 11:31
From Russia
Posts: 5390
@K-L
Are you sure that with p96_pip its all not placebo or some sideeffects ? Why i ask, because p96 driver we use in both versions are the same. And by changelogs LiveForIt show, he didn't add anything to it, instead of making it works.

Also i do test right now his beta3 and my beta6 : p96_pip driver works 1:1 the same by speed (even if his version are from altivec binary, and my from generic). At least on pegasos2 it is like this, and should be like this , as driver code are the same.

More of it, my should be been a very little faster, because by fixing white strips it mean disabling of compositing, which reduce transfers a bit as well.

Can you just grab beta3 from liveforit, my beta6, then hard reboot for everytest you do, and compare it then. My bet it will be 1:1 the same. + - a little which will depends on moon phase.

Quote:

Still doesn't work on my Radeon 9000 Pro on AGP (grey screen and 100%CPU usage). I don't know why. I doubt this card doesn't support compositing since Radeon HD Boing Ball from Hans works.


Maybe some more init code missing in vo_comp for now..

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2006/11/29 8:09
From Lyon, France
Posts: 992
@Kas1e

No placebo since I realize some benchmarks at each version.

Here are results from Beta 6 :

MUI-Mplayer Non Altivec *Beta 6* P96_PIP :

BENCHMARKs: VC: 21.728s VO: 7.111s A: 0.000s Sys: 0.634s = 29.472s
BENCHMARK%: VC: 73.7227% VO: 24.1271% A: 0.0000% Sys: 2.1502% = 100.0000%

MUI-Mplayer Non Altivec *Beta6* CGX_WPA :

BENCHMARKs: VC: 23.133s VO: 24.921s A: 0.000s Sys: 0.811s = 48.865s
BENCHMARK%: VC: 47.3395% VO: 51.0004% A: 0.0000% Sys: 1.6601% = 100.0000%

Well Beta6 is slower but I think your version is Non AltiVec. The 1st version (2011) is still the fastest on my system.

For those who want to test :

MPlayer -benchmark -vo *output video you want to test* -nosound yourvideo.avi

_________________
--
AmigaONE X1000 1,8Ghz, 2GB, Sapphire Radeon HD 7700, with Amiga OS 4.1 FE
http://www.amiga-ng.org
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Quite a regular
Joined:
2006/11/29 8:09
From Lyon, France
Posts: 992
@LiveForIt

I've asked for X1000 tests with Radeon HD 4650

Here are the results :

Mplayer LiveForIt Beta3 Altivec Compositing :

BENCHMARKs: VC: 9.244s VO: 3.974s A: 0.000s Sys: 0.353s = 13.571s
BENCHMARK%: VC: 68.1167% VO: 29.2807% A: 0.0000% Sys: 2.6026% = 100.0000%

MPlayer LiveForIt Beta 3 Altivec cgx_wpa:

BENCHMARKs: VC: 9.406s VO: 3.831s A: 0.000s Sys: 0.297s = 13.534s
BENCHMARK%: VC: 69.5020% VO: 28.3040% A: 0.0000% Sys: 2.1940% = 100.0000%

Mplayer LiveForIt Beta 3 AltiVec SDL:

BENCHMARKs: VC: 9.445s VO: 6.279s A: 0.000s Sys: 0.258s = 15.983s
BENCHMARK%: VC: 59.0963% VO: 39.2866% A: 0.0000% Sys: 1.6171% = 100.0000%

@Kas1e

The same with Beta6 :

MUI-Player NON AltiVec SDL :

BENCHMARKs: VC: 14.134s VO: 6.458s A: 0.000s Sys: 0.175s = 20.767s
BENCHMARK%: VC: 68.0603% VO: 31.0993% A: 0.0000% Sys: 0.8404% = 100.0000%

MUI-Mplayer NON AltiVec CGX_WPA :

BENCHMARKs: VC: 14.072s VO: 15.899s A: 0.000s Sys: 0.180s = 30.151s
BENCHMARK%: VC: 46.6709% VO: 52.7307% A: 0.0000% Sys: 0.5984% = 100.0000%


_________________
--
AmigaONE X1000 1,8Ghz, 2GB, Sapphire Radeon HD 7700, with Amiga OS 4.1 FE
http://www.amiga-ng.org
   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Just can't stay away
Joined:
2012/8/14 12:51
From France
Posts: 1494
@K-L
Yes, compositing is very fast on X1000 for non-streamed videos.

But, personaly, i'm unable to show a streamed video, it "hangs" every few seconds. It's strange because the CPU consumption is very low.

I don't understand how TommySammy succeed to use smtube with Compositing Mplayer ???

   Report Go to top

Re: Mplayer Altivec test version for G4 and PA6T-1682M
Home away from home
Joined:
2007/9/11 11:31
From Russia
Posts: 5390
@K-L
Probably just differences bettwen altivec and genertic versions yep. Once will sort all bugs with muimplayer will compile altivec one as well

@LiveForIT

Btw, as for putting events handling out of drivers, that imho not very good idea, as we better need to follow original mplayer logic: they have it in all their drivers as i see. So to avoid any mess and making things different imho better just do it as it done in mplayer : gui/mui , gui/reaction, and events are in drivers as it now (that minimal set of course, not whole).

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites
   Report Go to top


« 1 2 3 4 (5) 6 7 8 ... 30 »



[Advanced Search]


Powered by XOOPS 2.0 © 2001-2016 The XOOPS Project