On Sat, 2015-10-17 at 15:24 -0400, Ilia Mirkin wrote: > "compute" in this context is "initialize the compute engine so that > kernels may be executed", not "convert the llvm ir bitcode that > clover > hands us into nv50 ir". The former has actually been around for > years, > Samuel just fixed up a few fermi-specific bits.
Can't we use LLVM IR-> TGSI -> nv IR for that? I understand there is more work to be done (for one we'd also need a nouveau variant of libclc). Having a Fermi card in my nb, change from "it will crash even if someone did the work" to "it's feasible if you find someone with plenty of time on his/her hands" is still nice. Jan > > On Sat, Oct 17, 2015 at 3:11 PM, Jan Vesely <jan.ves...@rutgers.edu> > wrote: > > Does this mean it should be possible to hook up clover with > > nouveau? > > > > Jan > > > > On Fri, 2015-10-16 at 19:22 +0200, Samuel Pitoiset wrote: > > > Compute support was not enabled by default because weird effects > > > on 3D state happened, but I can't reproduce them anymore. > > > > > > This also enables MP performance counters by default on Fermi. > > > > > > Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com> > > > --- > > > src/gallium/drivers/nouveau/nvc0/nvc0_query.c | 3 +-- > > > src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 7 +------ > > > 2 files changed, 2 insertions(+), 8 deletions(-) > > > > > > diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_query.c > > > b/src/gallium/drivers/nouveau/nvc0/nvc0_query.c > > > index c81b85a..80f311b 100644 > > > --- a/src/gallium/drivers/nouveau/nvc0/nvc0_query.c > > > +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_query.c > > > @@ -291,7 +291,6 @@ nvc0_screen_get_driver_query_info(struct > > > pipe_screen *pscreen, > > > count += NVE4_HW_SM_QUERY_COUNT; > > > } else > > > if (screen->base.class_3d < NVE4_3D_CLASS) { > > > - /* NVC0_COMPUTE is not always enabled */ > > > count += NVC0_HW_SM_QUERY_COUNT; > > > } > > > } > > > @@ -358,7 +357,7 @@ > > > nvc0_screen_get_driver_query_group_info(struct > > > pipe_screen *pscreen, > > > count++; > > > } else > > > if (screen->base.class_3d < NVE4_3D_CLASS) { > > > - count++; /* NVC0_COMPUTE is not always enabled */ > > > + count++; > > > } > > > } > > > } > > > diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c > > > b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c > > > index ba53d10..d6a4ac5 100644 > > > --- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c > > > +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c > > > @@ -561,12 +561,7 @@ nvc0_screen_init_compute(struct nvc0_screen > > > *screen) > > > switch (screen->base.device->chipset & ~0xf) { > > > case 0xc0: > > > case 0xd0: > > > - /* Using COMPUTE has weird effects on 3D state, we need to > > > - * investigate this further before enabling it by default. > > > - */ > > > - if (debug_get_bool_option("NVC0_COMPUTE", false)) > > > - return nvc0_screen_compute_setup(screen, screen > > > ->base.pushbuf); > > > - return 0; > > > + return nvc0_screen_compute_setup(screen, screen > > > ->base.pushbuf); > > > case 0xe0: > > > return nve4_screen_compute_setup(screen, screen > > > ->base.pushbuf); > > > case 0xf0: > > > > _______________________________________________ > > mesa-dev mailing list > > mesa-dev@lists.freedesktop.org > > http://lists.freedesktop.org/mailman/listinfo/mesa-dev > >
signature.asc
Description: This is a digitally signed message part
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev