Add support for these opcodes, the conversion functions were already
there albeit need some new packing stuff.
Just like the tgsi version, piglit won't like it for all the same
reasons, so it's disabled (UP2H passes piglit arb_shader_language_packing
tests, albeit since PK2H won't due to those rounding differences I don't
know if that one works or not as the piglit test is rather difficult to
deal with).
Reviewed-by: Brian Paul <brianp@vmware.com>
Add support for these opcodes, the conversion functions were already
there albeit need some new packing stuff.
Just like the tgsi version, piglit won't like it for all the same
reasons, so it's disabled (UP2H passes piglit arb_shader_language_packing
tests, albeit since PK2H won't due those rounding differences I don't
know if that one works or not as the piglit test is rather difficult to
deal with).
The util functions handle the half-float conversion.
Note that piglit won't like it much due to:
a) The util functions use magic float mul conversion but when run inside
softpipe/llvmpipe, denorms are flushed to zero, therefore when the conversion
is from/to f16 denorm the result will be zero. This is a bug which should be
fixed in these functions (should not rely on denorms being available), but
will happen elsewhere just the same (e.g. conversion to f16 render targets).
b) The util functions use trunc round mode rather than round-to-nearest. This
is NOT a bug (as it is a d3d10 requirement). This will result of rounding not
representable finite values to MAX_F16 rather than INFINITY. My belief is the
piglit tests are wrong here but it's difficult to tell (generally glsl
rounding mode is undefined, however I'm not sure if rounding mode might need
to be consistent for different operations). Nevertheless, for gl it would be
better to use round-to-nearest, but using different rounding for GL and d3d10
is an unsolved problem (as it affects things like conversion to f16 render
targets, clear colors, this shader opcode).
Hence for now don't enable the cap bit (so the code is unused).
(Code is from imirkin, comment from sroland)
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Roland Scheidegger <sroland@vmvware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
If the tri is fully inside a scissor edge (or rather, we just use the
bounding box of the tri for the comparison), then we can drop these
additional scissor "planes" early. We do not even need to allocate
space for them in the tri.
The math actually appears to be slightly iffy due to bounding boxes
being rounded, but it doesn't matter in the end.
Those scissor rects are costly - the 4 planes from the scissor are
already more expensive to calculate than the 3 planes from the tri itself,
and it also prevents us from using the specialized raster code for small
tris.
This helps openarena performance by about 8% or so. Of course, it helps
there that while openarena often enables scissoring (and even moves the
scissor rect around) I have not seen a single tri actually hit the
scissor rect, ever.
v2: drop individual scissor edges, and do it earlier, not even allocating
space for them.
v3: help the compiler a bit with simpler code, suggested by Brian.
Reviewed-by: Brian Paul <brianp@vmware.com>
When we switched to 64bit rasterization, we could no longer use straight
aligned loads for loading the plane data. However, what the code actually
does for loading 3 planes, is 12 scalar loads + 9 unpacks, and then there's
another 8 unpacks for the transpose we need (!).
It would be possible to do the (scalar) loads of course already transposed
(at least saving the additional unpacks), however instead just use
(un)aligned vector loads, and recalculate the eo values, which is much less
instructions (note in case of the triangle_32_3_4 case, the eo values are
not even used, making the scalar loads + unpacks for them all the more
pointless).
This drops execution time of the triangle_32_3_4 function considerably,
albeit it doesn't really make a measurable difference (for small tris we're
essentially limited by vertex throughput in any case), for triangle_32_3_16
it's essentially noise (the loop is more costly than the initial code there).
(I'm thinking about just ditching storing the eo values in the plane data,
so could switch back to using aligned planes, however right now they are
still used in the other raster functions dealing with planes with scalar
code. Also not touching the ppc code, might not be that bad there in any
case.)
Reviewed-by: Brian Paul <brianp@vmware.com>
Enabling swrast on Android causes a link error because vtest is missing.
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The virgl reference counting of buffers is broken for prime fd buffers.
Each prime fd passed into virgl_drm_winsys_resource_create_handle creates
a new resource. The solution requires creating a separate hash table to
track flink names separately from prime handles.
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It is necessary to share the screen between mesa and gralloc to
properly ref count resources. This implements a hash lookup on
the file description to re-use an already created screen. This is
a similar implementation as freedreno and radeon.
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The change is necessary to avoid the following building error in android:
external/mesa/src/gallium/drivers/nouveau/nouveau_vp3_video_bsp.c: In function 'nouveau_vp3_bsp_next':
external/mesa/src/gallium/drivers/nouveau/nouveau_vp3_video_bsp.c:269:14: error: 'bsp_bo' undeclared (first use in this function)
assert(bsp_bo->size >= str_bsp->w0[0] + num_bytes[i]);
^
This matches the declaration of the variables in question.
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
This happens especially with exports and varying packing, where the last
bits aren't always filled in. We end up trying to do quad-wide stores,
which ends up being a lot of register moves that carefully preserve the
nop value. Instead don't do the stores.
total instructions in shared programs : 6131375 -> 6125267 (-0.10%)
total gprs used in shared programs : 910139 -> 895501 (-1.61%)
total local used in shared programs : 15328 -> 15328 (0.00%)
local gpr inst
helped 0 7442 4693
hurt 0 90 2687
Most of the helped/hurt instruction changes are by one or two ops
because can no longer do quad-wide stores in all cases.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
If an instruction has multiple defs, we have to do a lot more checks to
make sure that we can move it forward. Among other things, various code
likes to do
a, b = tex()
if () c = a
else c = b
which means that a single phi node will have results pointing at the
same instruction. We obviously can't propagate the tex in this case, but
properly accounting for this situation is tricky. Just don't try for
instructions with multiple defs.
This fixes about 20 shaders in shader-db, including the dolphin efb2ram
shader.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: mesa-stable@lists.freedesktop.org
It appears that the nvidia render engine is quite picky when it comes to
linear surfaces. It doesn't like non-256-byte aligned offsets, and
apparently doesn't even do non-256-byte strides.
This makes arb_clear_buffer_object-unaligned pass on both nv50 and nvc0.
As a side-effect this also allows RGB32 clears to work via GPU data
upload instead of synchronizing the buffer to the CPU (nvc0 only).
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> # tested on GF108, GT215
Tested-by: Nick Sarnie <commendsarnex@gmail.com> # GK208
Cc: mesa-stable@lists.freedesktop.org
Since we emulate clip-planes, the clip-vertex is used within the VS
itself (thanks to nir_lower_clip). So just ignore it as a VS output.
Fixes a boatload of piglit tests that were asserting on unknown
varying slot.
(Also unrelated spelling/typo fix.)
Signed-off-by: Rob Clark <robclark@freedesktop.org>
With glsl_to_nir we end up with local variables, instead of global, for
arrays.
Note that we'll eventually have to do something more clever, I think,
when we support multiple functions, but that will probably take some
work in a few places.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
With tgsi_to_nir we get this as a normal input with VARYING_SLOT_FACE.
But glsl_to_nir plus nir_lower_system_values this becomes an intrinsic.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
When using the "shared" vertex array configuration strategy, we bind
each of the buffers as a separate array. However there can be holes in
such vertex buffer lists, so just emit a disable for those.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: mesa-stable@lists.freedesktop.org
Teach the emitter that the two registers are sequential, and drop the
second arg entirely, in favor of a double-wide first argument.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
This largely leaves the existing image logic alone. When image support
is added this will have to be harmonized somehow.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Issue a MEM_BARRIER. No idea if this is sufficient. As there are no
tests for this, it'll have to do for now.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
(address, length) pairs are uploaded to the driver constbuf as well to
make these values available to the shaders.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com> (v1)
v1 -> v2: add defines for the various bits
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
In particular, AMDGPU_GEM_CREATE_CPU_GTT_USWC can affect even BOs created
in VRAM if they get evicted to GTT. In general there's no need to
restrict any of the flags to any particular domains.
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Failing to do this was resulting in the kernel driver unnecessarily
leaving open the possibility of CPU access to tiled BOs.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93862
(This change shouldn't be backported to stable branches, because
released versions of xf86-video-amdgpu unnecessarily try to map the
front buffer)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Very modest effect, but it's clearly the right thing to do.
total instructions in shared programs : 6131491 -> 6131398 (-0.00%)
total gprs used in shared programs : 910157 -> 910131 (-0.00%)
total local used in shared programs : 15328 -> 15328 (0.00%)
local gpr inst bytes
helped 0 55 85 85
hurt 0 26 20 20
Signed-off-by: Karol Herbst <nouveau@karolherbst.de>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Following shader-db results on GK110:
total instructions in shared programs : 6141510 -> 6131491 (-0.16%)
total gprs used in shared programs : 910187 -> 910157 (-0.00%)
total local used in shared programs : 15328 -> 15328 (0.00%)
local gpr inst bytes
helped 0 18 821 821
hurt 0 0 0 0
Signed-off-by: Karol Herbst <nouveau@karolherbst.de>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Add a debug option to select the LLVM SI Machine Scheduler.
R600_DEBUG=sisched
Signed-off-by: Axel Davy <axel.davy@ens.fr>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Just make sure that after we've submitted, we get to at least 5
(global) submits ago before we go on to do more. Prevents up to
seconds of lag with window movement in X with xcompmgr -c. There may
be useful tuning to do in the future, but for now this gets us
usability.
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
On an error return, the returned seqno will probably be unset, so we'd
lose track of what we've submitted so far for waiting on in the
future.
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
When RA fails, and we spill, we have to clean everything up before doing
RA again. We were forgetting to reset the hi/lo linked lists - at
least the hi list is guaranteed to still have pointers to now-deleted
RIG nodes.
Signed-off-by: Karol Herbst <nouveau@karolherbst.de>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: mesa-stable@lists.freedesktop.org
v2: drop inline keyword
drop radeon_llvm_dispose_kernel_module wrapper
v3: move definitions to .c file
use in radeonsi
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
The addition of spi_shader_col_format killed all color outputs
in precompiled shaders.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> (v1)
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com> (v1)
v2: also set the alpha func (trivial)
