----- Original Message ----- > On Thu, Apr 21, 2011 at 11:39 AM, Marek Olšák <mar...@gmail.com> > wrote: > > --- > > configure.ac | 5 +++++ > > 1 files changed, 5 insertions(+), 0 deletions(-) > > > > diff --git a/configure.ac b/configure.ac > > index 8e9f73f..84a75c4 100644 > > --- a/configure.ac > > +++ b/configure.ac > > @@ -1778,6 +1778,11 @@ AC_ARG_ENABLE([gallium-r300], > > [build gallium r300 @<:@default=disabled@:>@])], > > [enable_gallium_r300="$enableval"], > > [enable_gallium_r300=auto]) > > +if test "x$enable_gallium_r300" != xno; then > > + if test "x$MESA_LLVM" = x0; then > > + AC_MSG_ERROR([LLVM is required to build r300g]) > > + fi > > +fi > > if test "x$enable_gallium_r300" = xauto; then > > GALLIUM_DRIVERS_DIRS="$GALLIUM_DRIVERS_DIRS r300" > > gallium_check_st "radeon/drm" "dri-r300" > > -- > > 1.7.4.1 > > How does this affect architectures on which r300g works but don't > have > LLVM JIT support? See [1]. (I'm assuming that LLVMpipe requires JIT > support in LLVM). r300g works on Alpha, SPARC, and probably others > that don't have JIT support. > > I think the purpose of this patch is to simply prevent downstream > packagers from screwing up SWTCL users, but on, say, Alpha and SPARC, > I don't think SWTCL cards are possible. Does LLVMpipe in addition to > r300g do any good for HWTCL cards?
The Mesa state tracker uses SWTNL for GL selection/feedback regardless of the driver. Some SPECviewperf viewsets and CAD apps use it. So using LLVM speeds up selection/feedback for all gallium drivers. We have only tested LLVM with x86/x86_64. So indeed, using it/requiring it on other platforms is not advisable. Jose _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
