[Savage40] 3D Support for SuperSave IX/C SDR

Wed Oct 6 11:01:19 PDT 2004

--- Hassan Aurag <aurag at cae.com> wrote:

> The clone I was talking about is indeed xinerama off.
> 
> But I just tried xinerama on and the problem is there too sometimes.
> This
> time it happened with a java app. 
> 
> As a side note, it always happens on crt2. Sometimes switching and
> killing
> the app fixes it. When that happens, It looks as if I'm having a few
> mouse
> cursors on the screen. It always happens also (coincidence??) when I
> move
> the mouse too quickly after doing something to screen 2, eg: move a
> window
> to it or click a link on the browser....
> 
> In some rare cases, I noticed a tearing of the screens like all
> colours get
> f**ked up ....
> 
> Is there a more verbose mode I could turn to help in tracing this?
> 

Try again with the most recent version of xorg cvs.  I fixed a bug in
the supersavage tile setup last night.  The cursor problems sound like
they are tile related so the xorg update should fix that.  Also I've
gotten a report of problems with XaaScreenToScreenCopy on crtc2 for
supersavage users.  I haven't gotten a chance to look into yet, but you
might try adding the XaaNoScreenToScreenCopy option to your setup.  you
can also try disabling various XAA function using the XaaNO... options
to see which ones are giving you trouble.

Alex

> 
> 
> -----Original Message-----
> From: Alex Deucher [mailto:agd5f at yahoo.com] 
> Sent: Tuesday, October 05, 2004 6:00 PM
> To: Hassan Aurag; savage40 at probo.com
> Subject: RE: [Savage40] 3D Support for SuperSave IX/C SDR
> 
> 
> 
> --- Hassan Aurag <aurag at cae.com> wrote:
> 
> > Hi alex,
> > 
> > I downloaded the latest stuff (xorg cvs) and just wanted to say
> thanks
> > again. It's really cool.
> > 
> > I have one question  and one bug report though ;-)
> > 
> > 1- question: how does the caluclation goes as resolution/depth vs 
> > video memory to get both my screens dri'ed.
> 
> The DRI doesn't work with "regular" xinerama multi-head which is what
> I
> committed to xorg cvs.  xinerama and multi-head have issues with the
> DRI
> that I won't get into here.  The DRI will eventually work with
> dualhead once
> I merge my savage MergedFB code.  It will follow the behavior of
> mergedfb in
> the radeon and sis drivers in xorg cvs.
> 
> Unfortunately since the DRI does not allocate back and depth buffers
> per
> drawable, it has pretty hefty memory requirements.  you need 
> virtualx res * virtualy res * bytes/pixel * 3 = bytes required for 3d
> and
> that's just for the front, back, and depth buffers.  you also need
> space for
> hwcursor and BCI buffer. It's nice to have some space for textures as
> well,
> although they can be stored in AGP memory if need be.
> 
> so for a 1024x768 desktop at 16 bpp you have:
> 1024*768*2 bytes*3 buffers = ~4.6 Megs
> with dualhead (two desktops of 1024x768 @ 16 bpp) you'd need:
> 2 screens*1024*768*2 bytes*3 buffers = ~9.2 Megs
> 
> eventually once we get a new memory manager in the drm we can look at
> allocating per drawable which will alleviate the heavy requirements.
> 
> > 2- Bug report: I have been experiencing this problem for a while
> now 
> > and not just with cvs xorg but also with previous versions too. 
> > Mozilla will sometimes hang X and the only way to get out is to 
> > actually switch to a
> > console then kill mozilla or any mozilla based browser (epiphany,
> > firefox
> > ....). The switching (to console) always takes huge amounts of
> time,
> > but
> > there does not seem to be any real activity going on. And everytime
> > it was
> > when my browser was on screen 1. I'm using dual head with clone
> > option and
> > not spanning option. I am attaching a zipped version of
> > Xorg.0.log.old to
> > see if this helps. 
> 
> I'm not sure I understand exactly what clone mode you are talking
> about
> here.  Clone support isn't available in xorg yet (that'll come with
> mergedfb).  it's either dual or single head.  Maybe you are running
> dualhead
> without xinerama.  in that case gnome or kde starts on both heads,
> which I
> suppose could be seen as a cloned desktop.  that said, you here are a
> few
> things you might try:
> 
> - set AGPMode to 2x or 4x.  Some people report problems in 1x mode.
> 
> - set the shadowstatus option on both devices.  some chips hang when
> reading
> the regular status register.
> 
> - try disabling certain XAA routines. use the XaaNo... options on
> each
> screen.  See the xorg.conf or XF86Config man pages.
> 
> 
> Alex
> 
> > 
> > -----Original Message-----
> > From: savage40-bounces at probo.com
> [mailto:savage40-bounces at probo.com]
> > On
> > Behalf Of Alex Deucher
> > Sent: Tuesday, September 28, 2004 12:39 PM
> > To: Martin Steigerwald; savage40 at probo.com
> > Subject: Re: [Savage40] 3D Support for SuperSave IX/C SDR
> > 
> > 
> > 
> > --- Martin Steigerwald <Martin at lichtvoll.de> wrote:
> > 
> > > 
> > > Hello,
> > > 
> > > Does the most recent Savage driver support the 3D hardware of
> this 
> > > graphics controller?
> > 
> > Yes.  Although you'll need to use xorg cvs.  xorg 6.8.0/1 did not 
> > include 3d support.  It will also support dualhead once I commit my
> 
> > latest patch.
> > 
> > In order to build you'll need to check out xorg cvs: 
> > http://www.freedesktop.org/XOrg/CvsPage
> > then add:
> > #define BuildDevelDRIDrivers YES
> > to your xc/config/cf/host.def
> > then build the tree (change to xc and run 'make World'). Install
> the 
> > newly built tree (change to xc and run 'make install' as root). 
> > finally build and
> > install the savage drm change to xc/extras/drm/linux and run
> 'make',
> > then
> > copy the resulting savage.o (2.4 kernel) or savage.ko (2.6 kernel)
> to
> > the
> > apporiate place in your kernel modules tree (usually
> > /lib/modules/$KERNEL_VERSION/kernel/drivers/char/drm/) and run
> > 'depmod -a'.
> > you'll need a copy of your current kernel source installed.  on 2.4
> > you'll
> > also have to run 'make dep' on your kernel tree or the drm won't
> > build.
> > 
> > good luck.
> > 
> > Alex
> > 
> > 
> > > 
> > > 0000:01:00.0 0300: 5333:8c2e (rev 05)
> > > 0000:01:00.0 VGA compatible controller: S3 Inc. SuperSavage IX/C
> > SDR
> > > (rev
> > > 05) (prog-if 00 [VGA])
> > >         Subsystem: IBM ThinkPad T23 (2647-4MG)
> > >         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV-
> VGASnoop-
> > > ParErr- Stepping- SERR- FastB2B-
> > >         Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium
> > > >TAbort-
> > > <TAbort- <MAbort- >SERR- <PERR-
> > >         Latency: 248 (1000ns min, 63750ns max), Cache Line Size:
> > 0xff
> > > 
> > > (1020 bytes)
> > >         Interrupt: pin A routed to IRQ 11
> > >         Region 0: Memory at c0100000 (32-bit, non-prefetchable) 
> > > [size=512K]
> > >         Region 1: Memory at e8000000 (32-bit, prefetchable) 
> > > [size=64M]
> > >         Region 2: Memory at e4000000 (32-bit, prefetchable) 
> > > [size=64M]
> > >         Region 3: Memory at e0000000 (32-bit, prefetchable) 
> > > [size=32M]
> > >         Capabilities: [dc] Power Management version 2
> > >                 Flags: PMEClk- DSI+ D1+ D2+ AuxCurrent=0mA
> > > PME(D0-,D1-,D2-,D3hot-,D3cold-)
> > >                 Status: D0 PME-Enable- DSel=0 DScale=0 PME-
> > >         Capabilities: [80] AGP version 2.0
> 
=== message truncated ===

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 