| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
Mark variable MAYBE_UNUSED to avoid unused-but-set-variable warning in
release build.
Signed-off-by: Grazvydas Ignotas <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
Reviewed-by: Emil Velikov <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is mostly for variables that are only used in asserts and cause
unused-but-set-variable warnings in release builds. Could just use
UNUSED directly, but MAYBE_UNUSED should be less confusing and is
similar to what the Linux kernel has.
And yes __attribute__((unused)) can be used on variables on both GCC 4.2
(oldest supported by mesa) and clang 3.0 (just some random old version,
not sure what's the minimum for mesa).
Signed-off-by: Grazvydas Ignotas <[email protected]>
Reviewed-by: Francisco Jerez <[email protected]>
Reviewed-by: Chad Versace <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
Reviewed-by: Emil Velikov <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
combineLd/St would combine, i.e. :
st u32 # g[$r2+0x0] $r2
st u32 # g[$r2+0x4] $r3
into:
st u64 # g[$r2+0x0] $r2d
But this is only valid if r2 contains an 8 byte aligned address,
which is not guaranteed for compute shaders
This commit checks for src0 dim 0 not being indirect when combining
loads / stores as combining indirect loads / stores may break alignment
rules.
Signed-off-by: Hans de Goede <[email protected]>
Reviewed-by: Ilia Mirkin <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
Currently we were two restrictive, and would insert an output move in
cases like: MOV OUT[0], IN[0].xyzw
Loosen the restriction to allow the current instruction to appear in the
neighbor list but only at it's current possition.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
| |
Normally the offset in the group would be the same, but not always. For
example, in a sam(w) which only writes the 4th component.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This *seems* like a hw bug, and maybe only applies to certain a4xx
variants/revisions. But setting the SRGB bit in sampler view state
(texconst0) causes invalid alpha for ASTC textures. Work around this
setting up a second texture state and using that to sample alpha
separately.
This way, srgb->linear conversion happens in hw *prior* to
interpolation.
This fixes 546 dEQP tests: dEQP-GLES3.functional.texture.*astc*srgb*
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
| |
Better workaround in the following patch.
This reverts commit 899bd63acefd49a668e11c42d2ad92fa55aa157d.
|
|
|
|
| |
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
| |
This reverts commit ccdcf91104a5f07127b5b8d8570b5c4bbcf86647.
It breaks most KDE apps, because DRI doesn't support the RGBA component
ordering.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95071
|
|
|
|
|
|
|
|
|
|
|
| |
Allows the build to work when the python3 binary is not "python3".
v2: remove x bit from the script at Emil's suggestion
Signed-off-by: Jonathan Gray <[email protected]>
Reviewed-by: Emil Velikov <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When uploading a linear, void-extent, ASTC LDR block on Skylake, we are
required to flush to zero the UNORM16 channel values that would be
denormalized. This is specifically required for the values: 1, 2, and 3.
Fixes the 14 failing tests in:
dEQP-GLES3.functional.texture.compressed.astc.void_extent_ldr.*
v2: Split out flushing function (Kristian Høgsberg)
v3: Map with READ instead of INVALIDATE (Kenneth Graunke)
Signed-off-by: Nanley Chery <[email protected]>
Acked-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
src/intel/genxml/gen_pack_header.py requires python3.
v2: check for python3.5 as well
Signed-off-by: Jonathan Gray <[email protected]>
Reviewed-by: Emil Velikov <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
| |
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94181
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is equivalent of 73b01e2711ff45a1f313d5372d6c8fa4fe55d4d2
for blorp.
v2 (Ken): No need to call _mesa_format_has_color_component() now
that the number of components is gotten from
_mesa_base_format_component_count().
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Iago Toral Quiroga <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
| |
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
| |
In case blorp needs to configure it will be just as if render or
compute pipeline had configured it.
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
| |
Reviewed-by: Topi Pohjolainen <[email protected]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the past, BLORP has clobbered all BRW_NEW_* state flags, to trigger
re-emission of the entire 3D pipeline on the next draw. However, there
are some packets BLORP simply leaves alone, so there's no need to
re-emit them. Trying to reduce the set of dirty bits flagged after
BLORP runs is tricky.
Instead, we introduce a BRW_NEW_BLORP flag. This should be set on any
atom which emits a packet that BLORP also emits. When BLORP runs, it
will flag BRW_NEW_BLORP, causing those packets to get re-emitted.
This also makes it easy to avoid re-emitting specific atoms - we can
simply drop the BRW_NEW_BLORP flag on those.
To start, we assume that all packets need to be re-emitted. This is the
safest approach and closest to the existing code's behavior. Many of
these are obviously not required, and can be dropped in subsequent
patches.
Signed-off-by: Topi Pohjolainen <[email protected]>
Signed-off-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is identical to the blorp version which only differs in case
fragment shader isn't used. In that case blorp would reset batch
buffer address to zero.
This is not really needed, and having blorp to use base state
address setup that is compatible with normal upload allows one to
skip resetting it.
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
| |
Signed-off-by: Topi Pohjolainen <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The ds_bpermute instruction allows threads to transfer data directly
to or from the vgprs of other threads. These instructions use the LDS
hardware to transfer data, but do not read or write LDS memory.
DDX BEFORE: | DDX AFTER:
|
v_mbcnt_lo_u32_b32_e64 v2, -1, 0 | v_mbcnt_lo_u32_b32_e64 v2, -1, 0
v_mbcnt_hi_u32_b32_e64 v2, -1, v2 | v_mbcnt_hi_u32_b32_e64 v2, -1, v2
v_lshlrev_b32_e32 v4, 2, v2 | v_and_b32_e32 v2, 60, v2
v_and_b32_e32 v2, 60, v2 | v_lshlrev_b32_e32 v2, 2, v2
v_lshlrev_b32_e32 v3, 2, v2 | ds_bpermute_b32 v3, v2, v0
s_mov_b32 m0, -1 | ds_bpermute_b32 v0, v2, v0 offset:4
ds_write_b32 v4, v0 | s_waitcnt lgkmcnt(0)
s_waitcnt lgkmcnt(0) |
v_or_b32_e32 v0, 1, v2 |
v_lshlrev_b32_e32 v0, 2, v0 |
ds_read_b32 v1, v3 |
ds_read_b32 v0, v0 |
s_waitcnt lgkmcnt(0) |
|
LDS: 1 blocks | LDS: 0 blocks
Reviewed-by: Michel Dänzer <[email protected]>
Acked-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
We're trying to move to more of the new style intrinsics with include
the correct target name, and map directly to ISA instructions.
v2:
- Only do this with LLVM 3.8 and newer.
Reviewed-by: Michel Dänzer <[email protected]>
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
| |
The range metadata tells LLVM the range of expected values for this intrinsic,
so it can do some additional optimizations on the result.
Reviewed-by: Michel Dänzer <[email protected]>
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
| |
Reviewed-by: Michel Dänzer <[email protected]>
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Although Gen9 samples from most HDR ASTC surfaces of correctly,
there currently are no software workarounds to fix the incorrect
sampling that occurs in others of certain color endpoint modes.
With this change, we are no longer failing the 14 tests from:
dEQP-GLES3.functional.texture.compressed.astc.endpoint_value_hdr_cem_15.*
Signed-off-by: Nanley Chery <[email protected]>
Acked-by: Kenneth Graunke <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
| |
Add explicit inline and non-inline versions of CompleteDrawContext
to make gcc happy.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
v2: use _mm_cmpunord_ps for vIsNaN
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
|
|
| |
Switch boolean template arguments to typename template arguments of type
std::integral_constant<bool, VALUE>.
This allows the template argument unroller to easily be extended to enums.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
Reduces sorting cost
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
- Plus some minor code refactoring
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
For example, R10G10B10A2_UNORM.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
v2: varient -> variant
Reviewed by: George Kyriazis <[email protected]>
|
|
|
|
|
|
| |
v2: use alternate logic version in swr_check_render_cond
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
| |
Fixes a whole bunch of dEQP-GLES3.functional.fragment_ops.random.* (now
they all pass)
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
| |
Pull in RB_BLEND_* fixes.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
| |
Part of fixing piglit EXT_framebuffer_multisample/sample-coverage inverted
(there is also a bug with RCL tiled blits)
Cc: "11.1 11.2" <[email protected]>
|
|
|
|
|
|
| |
There's no reason we couldn't do non-MSAA full resolution tile buffer
load/stores, but we would have claimed buffer overflow was being
attempted. Nothing does this currently.
|
|
|
|
|
| |
I noticed this as a problem with ET:QW traces emitting coverage code when
the framebuffer was supposed to be single sampled.
|
|
|
|
|
|
|
|
|
|
| |
This was a bug from the MSAA enabling. Tests for surfaces with
nr_samples==1 instead of 0 (generally GL renderbuffers) would incorrectly
fail out.
Fixes the ARB_framebuffer_sRGB piglit tests other than srgb_conformance.
Cc: "11.1 11.2" <[email protected]>
|
|
|
|
|
|
|
|
|
| |
I had made the previous blit fix non-MSAA only because I was thinking
about how the hardware infers stride from the RENDERING_CONFIG packet.
However, I'm also inferring the stride for both MSAA src and dst in
vc4_render_cl.c from the width argument in the ioctl.
Fixes 15 EXT_framebuffer_multisample piglit tests.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On Broadwell, I get the following shader-db statistics:
Tessellation Control Shaders:
total instructions in shared programs: 57327 -> 57012 (-0.55%)
instructions in affected programs: 27334 -> 27019 (-1.15%)
helped: 45
HURT: 0
total cycles in shared programs: 265692 -> 255188 (-3.95%)
cycles in affected programs: 263122 -> 252618 (-3.99%)
helped: 184
HURT: 26
Tessellation Evaluation Shaders:
total instructions in shared programs: 23236 -> 23157 (-0.34%)
instructions in affected programs: 2791 -> 2712 (-2.83%)
helped: 27
HURT: 0
total cycles in shared programs: 151858 -> 149704 (-1.42%)
cycles in affected programs: 151858 -> 149704 (-1.42%)
helped: 101
HURT: 114
Geometry Shaders:
Orbital Explorer goes from 6442 -> 6356 instructions.
Two Shadow of Mordor shaders increase by a single instruction.
Signed-off-by: Kenneth Graunke <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
|