This was already happening because blorp happens to flush at the end of
every call, but we have been talking about removing that at some point,
and this would surely get overlooked.
v2 (Kenneth Graunke): Rebase on latest master. Note that we did remove
the other flush, and this change actually did get overlooked!
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
intel_flush() now did nothing except call through (and
intel_batchbuffer_flush() does the no-op check, too!)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
df06745c5a made it so that we didn't
allocate extra uniform space for unused clip planes, which also
incidentally made us not allocate any space at all, which we were relying
on for this no-uniforms case. Instead of putting the knowledge of this
special HW exception into the thing that normally preallocates prog_data
for us, just allocate it here.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=68766
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Since we can have per-pixel lod we should also honor the filter per-pixel
(in fact we didn't honor it per quad neither in the multiple quad case).
Do this by running the linear path and simply beating the weights into shape
(the sample with the higher weight is the one which should have been chosen
with nearest filtering hence adjust filter weight to 1.0/0.0 based on that).
If all pixels use nearest filter (either min and mag) then still run just a
nearest filter as this is way cheaper (probably around 4 times faster for 2d,
more for 3d case) and it should be relatively rare that pixels really need
different filtering. OTOH if all pixels would require linear don't do anything
special since the linear path with filter adjustments shouldn't really be all
that much more expensive than ordinary linear, and we think it's rare that
min/mag filters are configured differently so there doesn't seem much value
in trying to optimize this further.
This does not yet fix the AoS path (though currently AoS is only used for
single quads hence it could be considered less broken, just never honoring
per-pixel filter decision but doing it per quad).
v2: simplify code a bit (unify min linear and min nearest cases)
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
While a sqrt here and there shouldn't hurt much (depending on the cpu) it is
possible to completely omit it since rho is only used for calculating lod and
there log2(x) == 0.5*log2(x^2). Depending on the exact path taken for
calculating lod this means we get a simple mul instead of sqrt (in case of
nearest mip filter in fact we don't need to replace the sqrt with something
else at all), only in some not very useful path this doesn't work (combined
brilinear calculation of int level and fractional lod, accurate rho calc but
brilinear filtering seems odd).
Apart from being faster as an added bonus this should increase our crappy
fractional accuracy of lod, since fast_log2 is only good for ~3bits and this
should increase accuracy by one bit (though not used if dimension is just one
as we'd need an extra mul there as we never had the squared rho in the first
place).
v2: use separate ilog2_sqrt function if we have squared rho.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
This is just preparation for per-pixel (or per-quad in case of multiple quads)
min/mag filter since some assumptions about number of miplevels being equal
to number of lods no longer holds true.
This change does not change behavior yet (though theoretically when forcing
per-element path it might be slower with different min/mag filter since the
code will respect this setting even when there's no mip maps now in this case,
so some lod calcs will be done per-element just ultimately still the same
filter used for all pixels).
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Fixes "Missing break in switch" defect reported by Coverity.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
The downstream android kernel driver is "kgsl", the upstream drm/kms
driver is called "msm". Since libdrm_freedreno handles the differences
between the two, we need to load the same thing for either device.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
We need to set the flag on all the .xyzw components that are written by
the instruction, not just on .x. Otherwise a later use of rN.y (for
example) will not trigger the appropriate sync bit to be set.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Seems like most/all instructions have some restrictions about const src
registers. In seems like the 2 src (cat2) instructions can take at most
one const, and the 3 src (cat3) instructions can take at most one const
in the first 2 arguments. And so on. Handle this properly now.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
GLSL 1.30 doesn't allow precision qualifiers on sampler types,
but in GLSL ES, sampler types are also allowed. This seems like
an oversight (since the intention of including these in GLSL 1.30
is to allow compatibility with ES shaders).
Currently, Mesa allows "default" precision qualifiers to be set for
sampler types in GLSL (commit d5948f2). This patch makes it follow
GLSL ES rules and also allow declaring sampler variables with a
precision qualifier in GLSL 1.30 (and later). e.g.
uniform lowp sampler2D sampler;
This fixes a shader compilation error in Khronos OpenGL conformance
test "depth_texture_mipmap".
V2: Update comments.
Signed-off-by: Ian Romanick <idr@lists.freedesktop.org>
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Ian Romanick <idr@lists.freedesktop.org>
Cc: <mesa-stable@lists.freedesktop.org>
Previously, we allocated space in brw_vs_prog_data's params and
pull_params arrays for MAX_CLIP_PLANES vec4s---even when it wasn't
necessary.
On a 64-bit architecture, this used 0.5 kB of space (8 clip planes *
4 floats per plane * 8 bytes per float pointer * 2 arrays of pointers =
512 bytes). Since this cost was per-vertex shader, it added up.
Conveniently, we already store the number of clip plane constants in the
program key. By using that, we can allocate the exact amount of space
needed. For the common case where user clipping is disabled, this means
0 bytes.
While we're here, mention exactly what code requires this extra space,
since it wasn't obvious.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
When validating draw parameters move check for 0 draw count last
(drawing with count 0 is not an error), so that other parameters (e.g.: the
primitive type) are validated and the correct errors (if applicable) are
generated.
>From the OpenGL 3.3 spec page 33 (page 48 of the PDF):
"[Regarding DrawArraysOneInstance, in terms of which other draw operations
are defined:]
If count is negative, an INVALID_VALUE error is generated."
This patch also changes the bahavior of MultiDrawElements to perform the draw
operation if some primitive's index counts are zero.
Signed-off-by: Fabian Bieler <fabianbieler@fastmail.fm>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
MAD will be generated directly from ir_triop_fma, so this assertion
checks that all ir_expressions are usable.
Reviewed-by: Paul Berry <stereotype441@gmail.com>
MSAA was tested by one user on RS690 and it works for him with color
compression (CMASK) disabled. Our theory is that his chipset lacks CMASK RAM.
Since we don't have hardware documentation about which chipsets actually have
CMASK RAM, I had to take a guess based on the presence of HiZ.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
In particular noone is interested in the vertex count, so drop that,
and also drop the duplicated num_primitives_generated /
so.primitives_storage_needed variables in drivers. I am unable for now to figure
out if primitives_storage_needed in SO stats (used for d3d10) should
increase if SO is disabled, though the equivalent num_primitives_generated
used for OpenGL definitely should increase. In any case we were only counting
when SO is active both in softpipe and llvmpipe anyway so don't pretend there's
an independent num_primitives_generated counter which would count always.
(This means the PIPE_QUERY_PRIMITIVES_GENERATED count will still be wrong just
as before, should eventually fix this by doing either separate counting for this
query or adjust the code so it always counts this even if SO is inactive depending
on what's correct for d3d10.)
Reviewed-by: Brian Paul <brianp@vmware.com>
SEQ and SNE are not native i915 instructions, so they each generate at
least 3 instructions. If both operands are uniforms or constants, we
get 5 instructions like:
U[1] = MOV CONST[1]
U[0].xyz = SGE CONST[0].xxxx, U[1]
U[1] = MOV CONST[1].-x-y-z-w
R[0].xyz = SGE CONST[0].-x-x-x-x, U[1]
R[0].xyz = MUL R[0], U[0]
This code is stupid. Instead of having the individual calls to
i915_emit_arith generate the moves to utemps, do it in the caller. This
results in code like:
U[1] = MOV CONST[1]
U[0].xyz = SGE CONST[0].xxxx, U[1]
R[0].xyz = SGE CONST[0].-x-x-x-x, U[1].-x-y-z-w
R[0].xyz = MUL R[0], U[0]
This allows fs-temp-array-mat2-index-col-wr and
fs-temp-array-mat2-index-row-wr to fit in hardware limits (instead of
falling back to software rasterization).
NOTE: Without pending patches to the piglit tests, these tests will now
fail. This is an unrelated, pre-existing issue.
v2: Copy most of the body of the commit message into comments in the
code. Suggested by Eric.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
The command is submitted once the event has been triggered, but it might not
have completed yet. Therefore, we have to add it to deps in order to wait on it.
Signed-off-by: Niels Ole Salscheider <niels_ole@salscheider-online.de>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
I previously fixed this partly in 9e8400f4c9,
however I didn't go far enough in testing it, now when I parse a TGSI shader
with arrays in it my iterator can see the ArrayID set to the proper value.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Now that we use a fixed set of register classes, we can set up the
register set and conflict graphs once, at context creation, rather than
on every VS compile. This is obviously less expensive, and also what
we already do in the FS backend.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
We're soon going to be calling brw_alloc_reg_set() from outside of the
visitor, where we don't have the precomputed "max_grf" variable handy.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
For now, nothing else can get allocated over them. That may change at
some point in the future.
This also means that base_reg_count can be computed without knowing the
number of registers used for the payload, which is required if we want
to allocate the register set once at context creation time.
See commit 551e1cd44f, which implemented
virtually identical code in the FS backend.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
Arrays, structures, and matrices use large VGRFs of arbitrary sizes.
However, split_virtual_grfs() breaks those down into VGRFs of size 1.
For reference, commit 5d90b98879 is the
analogous change to the FS backend.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
(From a suggestion by Francisco Jerez)
If an enum represents a bitfield of flags, e.g.:
enum E {
A = 1,
B = 2,
C = 4,
D = 8,
};
then C++ normally prohibits statements like this:
enum E x = A | B;
because A and B are implicitly converted to ints before OR-ing them,
and an int can't be stored in an enum without a type cast. C, on the
other hand, allows an int to be implicitly converted to an enum
without casting.
In the past we've dealt with this situation by storing flag bitfields
as ints. This avoids ugly casting at the expense of some type safety
that C++ would normally have offered (e.g. we get no warning if we
accidentally use the wrong enum type).
However, we can get the best of both worlds if we override the |
operator. The ugly casting is confined to the operator overload, and
we still get the benefit of C++ making sure we don't use the wrong
enum type.
v2: Remove unnecessary comment and unnecessary use of "enum" keyword.
Use static_cast.
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
We never noticed that this field was uninitialized because it is only
used in an error path that reports internal Mesa errors.
But it's silly to have it around anyway because &brw->ctx is
equivalent.
Should fix Coverity defect CID 1063351: Uninitialized pointer field
(UNINIT_CTOR) /src/mesa/drivers/dri/i965/brw_vec4_emit.cpp: 148
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
