Re: fallback mode vs llvmpipe rendering



On Mon, 2012-06-25 at 16:16 +0200, Olav Vitters wrote:
> On Mon, Jun 25, 2012 at 10:05:10AM -0400, Adam Jackson wrote:
> > On Mon, 2012-06-25 at 09:05 -0400, Matthias Clasen wrote:
> > > One example is ppc - no llvm support on that architecture...
> > 
> > Or s390 (anymore) or arm (effectively).  Honestly I'm coming to resent
> 
> How often is software rendering needed in those cases (s390, ppc, arm)?

s390s don't have video cards, but do run Xvnc.  So, always.

ppc tends to have a PCI bus.  Modern blade kinds of machines tend to
have server GPUs, which pretty much uniformly don't have 3D engines.
Every PCI PPC Mac had a 3D-capable GPU, although nobody ever got around
to doing a 3D driver for imstt(4), and the ATI Rage 128 is no longer
supported in Mesa, so low-end kit might struggle.  The high end GPU for
that class of machines was somewhere around a GeForce 6600, which
certainly works on PC machines; the caveat is big-endian is basically
non-existent on client hardware anymore so it's probably untested.

ARM tends to have non-PCI integrated graphics, and the drivers tend to
be non-free, and even the non-free ones tend to be terrible.  So,
depressingly often.

- ajax

Attachment: signature.asc
Description: This is a digitally signed message part



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]