Re: [xine-user] xine on set-top boxes: directfb

[Andrew Meredith <andrew@an...>]

I object !!

Hetz Ben Hamo wrote:
> 
> But today it's totally changed!
> You have support for Xv so you can run MPEG-2 at full speed like your Windows
> machine can

My two DVD enabled boxen (one laptop, one Athlon 1G workstation) run
DVDs considerably better under Linux than under Windows ... and the
laptop has a Windows only h/w accelerator !!

Just my 0.02 UKP worth ;)

Andrew Meredith
_______________________________________________________________
                  The Anvil Organisation Ltd.
                          Director
Tel: +44 (0) 1249 444240 | Email:              andrew@...
Fax: +44 (0) 1249 460560 | Web:           http://www.anvil.org/
Mob: +44 (0) 7802 389007 | SMS: andrew.meredith@...
_______________________________________________________________
   The box says Win95 or better .. Must run under Linux then!

Re: [xine-user] Xine

[Siggi Langauf <siggi@us...>]

Hi Dave,

first of all, a meaningful subject could help you getting faster answers
from this list...

On Wed, 4 Apr 2001 dave@... wrote:

[...]
> I'm running an Athalon 700 with 192MB ram. RH Linux 7.0. I have a
> Voodoo3 card and and ACER DVD CD Rom.

I guess you're installing xine from source, as special RPMs would be
needed for RH 7.

> Using KDE X windows but have also tried in Gnome. I can drag files
> (although to be honest I'm not sure if I'm dragging the correct one so I
> tried them all. I get several errors however on occasion, the rom spins
> up and the drive light goes crazy so it seems like its running, but no
> picture appears.

Hmmm, So you don't get _any_ video, I guess. Do you want to play DVDs? If
so:

0) RTFM (make sure /dev/dvd points to your drive, etc.)
1) insert DVD
2) start xine
3) press "DVD" button
4) press "play"

If your DVD is CSS "enhanced" (some people would call it "encrypted",
others "crippled"), you have to descramble it somehow. There are rumours
that there are a few special input plugins for this on the 'net, just have
a look at the "external xine resources" section on xine's homepage!

Regards,
	Siggi

Re: [xine-user] How to play vob files in command shell(not x-window)?

[Siggi Langauf <siggi@us...>]

On Wed, 4 Apr 2001, Cha, Seoung-yun wrote:

> I have hollywood+ decording card in my linux box.
> My linux box is kernel 2.2.16 RH7.0
> I want to play vob files in command shell, not x-window?
> Please, tell me how can i do?

Currently, I don't know any DVD (or vob file) player that works completely 
without X. 
If you only want to play without GUI (but still using X for video output),
you can use something like

xine -pfhq filename.vob

(see the man page for more info) Maybe, there's a console only player
based on xine within the next few months, but that won't help you now...

Regards,
	Siggi

[xine-user] laptop that works nice with xine

[Martin, Matthias <Matthias.Martin@ba...>]

If anyone of you is planning on buying a laptop: The Dell Latitude C800
rocks! It has Xv due to it using The ATI R128 M4....you get the idea. OpenGL
should work also, but haven't seen it yet. Well, most of these Dell
Notebooks are very LinuxFriendly(TM). 


Matthias Martin

Barco AG
Soodstrasse 55
8134 Adliswil
Switzerland

Tel: +41 (0) 1 712 60 46
Fax:+41 (0) 1 712 60 41

[xine-user] Re: SMP machines

[Bill Fink <billfink@ca...>]

OK, I patched video_out/video_out.c to calculate and display fps
(patch against 0.4.01 with Henry Worth's 2 and 2a patches is included
at the end of this message).  Here's a sample output with UP and Xv
(from the first section of video after the opening credits of the
Highlander DVD):

200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.0 fps
200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.0 fps
200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.0 fps
200 frames - 0 skipped - 0 discarded output in 8.4 secs at 23.9 fps
200 frames - 0 skipped - 0 discarded output in 8.4 secs at 23.8 fps
200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.2 fps
200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.1 fps
200 frames - 0 skipped - 0 discarded output in 8.4 secs at 23.9 fps
200 frames - 0 skipped - 0 discarded output in 8.4 secs at 24.0 fps
200 frames - 0 skipped - 0 discarded output in 8.3 secs at 24.0 fps

Since it wasn't dropping any frames, I decided to switch to Xshm using
xine's "-s" option, with the following result for the same section of
the Highlander DVD:

200 frames - 60 skipped - 0 discarded output in 8.4 secs at 16.8 fps
200 frames - 62 skipped - 0 discarded output in 8.3 secs at 16.6 fps
200 frames - 56 skipped - 0 discarded output in 8.4 secs at 17.2 fps
200 frames - 44 skipped - 0 discarded output in 8.3 secs at 18.7 fps
200 frames - 47 skipped - 0 discarded output in 8.3 secs at 18.4 fps
200 frames - 65 skipped - 0 discarded output in 8.3 secs at 16.2 fps
200 frames - 58 skipped - 0 discarded output in 8.4 secs at 16.9 fps
200 frames - 55 skipped - 0 discarded output in 8.3 secs at 17.5 fps
200 frames - 52 skipped - 0 discarded output in 8.3 secs at 17.8 fps
200 frames - 46 skipped - 0 discarded output in 8.4 secs at 18.3 fps

Next I was going to try it with an SMP kernel.  Unfortunately, however,
although I could successfully build an SMP kernel for my dual G4 PPC
(2x500 MHz), it wouldn't boot.  If I can manage later to create a usable
SMP kernel, I'll post a followup message with the xine performance
results.  But at least it shows a quantitative difference between
using Xv versus Xshm.

						-Bill



> On Mon, 9 Apr 2001, Bill Fink wrote:
> 
> > On 04/06/2001, Guenter Bartsch wrote:
> > 
> > > On Fri, 6 Apr 2001, Sven Heithecker wrote:
> > 
> > > does xine use more than one processor ?
> > 
> > provided that you have a kernel and a thread implementation that can
> > spread threads over multiple processors, xine should benefit from that.
> > However, the biggest part of the work, the mpeg video decoding, will still
> > be on one CPU.
> > 
> > Would be interesting to hear benchmark results from such a setup anyway.
> 
> I'd be willing to run a test on my dual G4 PPC system, but I have a
> question.  Is there a simple way to get xine 0.4.01 to report an effective
> frames per second rate?  I scanned the source but didn't see anything
> obvious.  I seem to recall this being discussed earlier on the list
> but the archive search feature doesn't seem to be too useful.
> 
> 						-Bill



xine-0.4.01 (with Henry Worth's 2 and 2a patches) fps patch:
------------------------------------------------------------------------
--- xine-0.4.01/video_out/.orig/video_out.c	Mon Apr  9 22:45:04 2001
+++ xine-0.4.01/video_out/.mod/video_out.c	Tue Apr 10 00:56:01 2001
@@ -71,6 +71,7 @@
   int                num_frames_delivered;
   int                num_frames_skipped;
   int                num_frames_discarded;
+  struct timeval     stats_time;
 
   int                width, height;
   int                bInitialized;
@@ -476,11 +477,23 @@
   pthread_mutex_unlock (&img->mutex);
 }
 
+void tvsub (struct timeval *tdiff, struct timeval *t1, struct timeval *t0) {
+
+	tdiff->tv_sec = t1->tv_sec - t0->tv_sec;
+	tdiff->tv_usec = t1->tv_usec - t0->tv_usec;
+	if (tdiff->tv_usec < 0) {
+		tdiff->tv_sec--;
+		tdiff->tv_usec += 1000000;
+	}
+}
+
 void vo_queue_frame (vo_image_buffer_t* img) {
 
   int32_t  diff;
   uint32_t cur_vpts;
   uint32_t pic_vpts ;
+  struct timeval prev_time, delta_time;
+  double deltat;
 
   pic_vpts = metronom_got_video_frame (img->PTS);
   img->PTS = pic_vpts;
@@ -538,8 +551,13 @@
    */
   
   if (gVO.num_frames_delivered>199) {
-    printf ("%d frames delivered, %d frames skipped, %d frames discarded\n", 
-	    gVO.num_frames_delivered, gVO.num_frames_skipped, gVO.num_frames_discarded);
+    prev_time = gVO.stats_time;
+    gettimeofday(&gVO.stats_time, NULL);
+    tvsub(&delta_time, &gVO.stats_time, &prev_time);
+    deltat = delta_time.tv_sec + ((double)delta_time.tv_usec) / 1000000;
+    printf ("%d frames - %d skipped - %d discarded output in %.1f secs at %.1f fps\n", 
+	    gVO.num_frames_delivered, gVO.num_frames_skipped, gVO.num_frames_discarded, deltat,
+	    (gVO.num_frames_delivered - gVO.num_frames_discarded - gVO.num_frames_skipped) / deltat);
 
     gVO.num_frames_delivered = 0;
     gVO.num_frames_discarded = 0;
@@ -1009,6 +1027,7 @@
     pthread_mutex_unlock ( &img->mutex );
   }
 
+  gettimeofday(&gVO.stats_time, NULL);
   gVO.num_frames_delivered = 0;
   gVO.num_frames_skipped   = 0;
   gVO.num_frames_discarded = 0;
@@ -1052,6 +1071,7 @@
   gVO.display_window_visible = 1;
 
 
+  gettimeofday(&gVO.stats_time, NULL);
   gVideoDriver = init_video_out_mga ();
   if (gVideoDriver) {
     printf ("Using Teletux Matrox G400 video extension for video output.\n");

Re: [xine-user] reproductible Floating point exception and debug coredump

[Daniel Caujolle-Bert <segfault@cl...>]

Hi Denis,

	To get rid of the "crash on debug build(tm)", apply this
patch (maybe more should be applied (audio/w32/ac3), but i haven't
time to browse the CVS tree). Anyway, that was fixed on CVS,
even it doesn't compile yet ;-)


Cheers.

-- 
73's de Daniel, F1RMB.

              -=- Daniel Caujolle-Bert -=- segfault@... -=-
                        -=- f1rmb@... (AMPR NET) -=-


----snip/snip
===================================================================
RCS file: /cvsroot/xine/xine/libmpeg2/Makefile.am,v
retrieving revision 1.13
retrieving revision 1.14
diff -u -r1.13 -r1.14
--- xine/libmpeg2/Makefile.am   2000/12/13 01:44:02     1.13
+++ xine/libmpeg2/Makefile.am   2001/03/07 23:31:50     1.14
@@ -10,7 +10,7 @@
 noinst_HEADERS = vlc.h mpeg2.h mpeg2_internal.h 
 
 debug:
-       $(MAKE) CFLAGS="$(DEBUG_CFLAGS)"
+       $(MAKE) CFLAGS="$(DEBUG_CFLAGS) @BUILD_LIB_STATIC@"
 
 mostlyclean-generic:
        -rm -f *~ \#* .*~ .\#*

Re: [xine-user] xine on set-top boxes: directfb

[Hetz Ben Hamo <hetz-home@co...>]

> In the past X was very bad at sharing the video card resources around so
> you couldn't run SVGAlib safely. And there was no clean way of starting it
> from X. Consequently games would have to require you switching virtual
> consoles at the least... not newbie friendly.

Yes, but today it's not needed any more - thanks to the VESA standard - you 
can switch resolutions on the fly (look at how SDL does that with XFree 4).

People seems to forget that you can't just ignore X - there are millions of 
applications that still need X - start from Commercial applications who uses 
motif and other things (GNOME, KDE etc). Sure you can use QT Embedded and 
fbdirect with GTK - but the point is - that you simply don't have drivers for 
the various chipset unless you got a very well supported card like Matrox 
G200 or G400. Go ahead and talk to nVidia to create a special FB driver with 
hardware acceleration - good luck!

And even that you do get somehow hardware accelerated drivers - then you're 
loosing 3D (DRI cannot run with frame buffer - this question has been asked 
in the DRI mailing list couple of times - all the developers say to the 
person who asks "good luck", cause it will be damn hard to do it unless you 
got some developers and financial support for that (or you can go the Mozilla 
way and have 3 years for a beta quality product - no flames, just an example).

So you loose 2D hardware acceleration with frame buffer (unless you write 
drivers for the various chipsets), 3D hardware, v4l support (the thing that 
fbdirect showing is rediculus in terms of performance and API - just try 2 
open 2 video windows - good luck), X compatibility (good like writing a layer 
which will translate Xlib calls to your frame buffer, brrrrr), and ofcourse 
you won't be able to run your stuff on *BSD and other unices - and for what? 

Yes, X was a total shit quality in the days of 3.x - I know this because I 
use Linux since 1995 (the days when you left your modem download slackware or 
Redhat 4.2 for 48 hours) and X was horrible - in terms of drivers, 
performance, applications and compatibility. But today it's totally changed! 
You have support for Xv so you can run MPEG-2 at full speed like your Windows 
machine can, you have good 3D with DRI (remember the days with GLX on XFree 
3.3.6? the special hack in LILO for DMA?), You got great things like QT and 
GTK, Dual head support and other stuff. True - it's not like all the 
capabilites of Windows and DirectX 8, but you see a lot more development then 
it used to be...

And for SVGALIB - the simplest thing that most people hated about it as that 
you need to become root in order to use it at all, and it's development is 
super-slow if not frozen.

Hetz,
A proud X user ;)

>
> There /are/ lots of good X11 replacements (BTW may I suggest having a look
> at http://www.directfb.org for an interesting one) available as you said
> and now that the DRM in the kernel mediates (I think) the video card
> resource problem, perhaps when 2.4.x becomes mainstream, we could see
> greater takeup.
>
> > On Sat, 7 Apr 2001, Guenter Bartsch wrote:
> > > Hi Folks,
> > >
> > > the upcoming version of xine will support scaling in software. This
> > > will not only enable fullscreen playback in Xshm (and even X11 remote)
> > > mode, but will also make it possible to have xine running on the
> > > framebuffer device, perhaps for a set-top box without X11.
> > >
> > > A very interesting project is
> > >
> > > http://directfb.org/
> > >
> > > which could provide the necessary widget for a xine-gui running on the
> > > framebuffer device.
> > >
> > > Anybody interested in working on this one?
> > >
> > > Cheers and have a nice weekend,
> > >
> > >    Guenter
> > >
> > > --
> > > time is a funny concept
> > >
> > >
> > > _______________________________________________
> > > xine-user mailing list
> > > xine-user@...
> > > http://lists.sourceforge.net/lists/listinfo/xine-user
> >
> > _______________________________________________
> > xine-user mailing list
> > xine-user@...
> > http://lists.sourceforge.net/lists/listinfo/xine-user