Apparently, Geminilake requires you to whack a chicken bit to select
either compute or tessellation mode for barriers. The recommendation
is to switch between them at PIPELINE_SELECT time.
We may not need to do this all the time, but I don't know that it hurts
either. PIPELINE_SELECT is already a pretty giant stall.
This appears to fix hangs in tessellation control shaders with barriers
on Geminilake. Note that this requires a corresponding kernel change,
drm/i915: Whitelist SLICE_COMMON_ECO_CHICKEN1 on Geminilake.
in order for the register write to actually happen. Without an updated
kernel, this register write will be noop'd and the fix will not work.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Memtrace aubs are similar to classic aubs, with the major
difference being how command submission is serialized (as register
writes instead of a high-level submit message). Some internal
tools generate or consume only memtrace aubs.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
This was never enabled in secondary buffers because hiz_enabled was
never set to true for those.
If the app provides a framebuffer in the inheritance info when beginning
a secondary buffer, we can determine if HiZ is enabled and therefore
allow the PMA optimization to be enabled within the command buffer.
This improves performance by ~13% on an internal benchmark on Skylake.
v2: Use anv_cmd_buffer_get_depth_stencil_view().
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
If we have a color attachment, but its writes are masked, this would
have still returned true. This is inconsistent with how HasWriteableRT
in 3DSTATE_PS_BLEND is set, which does take the mask into account.
This could lead to PixelShaderHasUAV not being set in 3DSTATE_PS_EXTRA
if the fragment shader does use UAVs, meaning the fragment shader may
not be invoked because HasWriteableRT is false. Specifically, this was
seen to occur when the shader also enables early fragment tests: the
fragment shader was not invoked despite passing depth/stencil.
Fix by taking the color write mask into account in this function. This
is consistent with how things are done on i965.
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes hangs seen due to the lock not being released here.
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Older OpenGL defines two equations for converting from signed-normalized
to floating point data. These are:
f = (2c + 1)/(2^b - 1) (equation 2.2)
f = max{c/2^(b-1) - 1), -1.0} (equation 2.3)
Both OpenGL 4.2+ and OpenGL ES 3.0+ mandate that equation 2.3 is to be
used in all scenarios, and remove equation 2.2. DirectX uses equation
2.3 as well. Intel hardware only supports equation 2.3, so Gen7.5+
systems that use the vertex fetcher hardware to do the conversions
always get formula 2.3.
This can make a big difference for 10-10-10-2 formats - the 2-bit value
can represent 0 with equation 2.3, and cannot with equation 2.2.
Ivybridge and older were using equation 2.2 for OpenGL, and 2.3 for ES.
Now that Ivybridge supports OpenGL 4.2, this is wrong - we need to use
the new rules, at least in core profile. That would leave Gen4-6 doing
something different than all other hardware, which seems...lame.
With context version promotion, applications that requested a pre-4.2
context may get promoted to 4.2, and thus get the new rules. Zero cases
have been reported of this being a problem. However, we've received a
report that following the old rules breaks expectations. SuperTuxKart
apparently renders the cars red when following equation 2.2, and works
correctly when following equation 2.3:
https://github.com/supertuxkart/stk-code/issues/2885#issuecomment-353858405
So, this patch deletes the legacy equation 2.2 support entirely, making
all hardware and APIs consistently use the new equation 2.3 rules.
If we ever find an application that truly requires the old formula, then
we'd likely want that application to work on modern hardware, too. We'd
likely restore this support as a driconf option. Until then, drop it.
This commit will regress Piglit's draw-vertices-2101010 test on
pre-Haswell without the corresponding Piglit patch to accept either
formula (commit 35daaa1695ea01eb85bc02f9be9b6ebd1a7113a1):
draw-vertices-2101010: Accept either SNORM conversion formula.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
Previously, we were flagging the instruction state buffer for capture
but not surface state or dynamic state. We want those captured too.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Some older versions of the Vulkan driver didn't properly tag dynamic
state as needing to be captured. Also, this prevents crashes when
looking at dumps on older kernels.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
We were walking the sections, printing the batches, and then freeing
them in one pass. If the batch happens to reference any earlier
sections (which it almost certainly will since it's at the end), we will
access freed memory.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
This reverts commit 9cd60fce9c.
Above commit caused 2000+ piglit tests to assert fail. Disabling
the align1 mode on gen10 for now to avoid failures.
Cc: Matt Turner <mattst88@gmail.com>
Cc: Rafael Antognolli <rafael.antognolli@intel.com>
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Tested-by: Rafael Antognolli <rafael.antognolli@intel.com>
This should shut up some Valgrind errors during pre-regalloc
scheduling. The errors were harmless since they could only have led
to the estimation of the bank conflict penalty of an instruction
pre-regalloc, which is inaccurate at that point of the program
compilation, but no less accurate than the intended "return 0"
fall-back path. The scheduling pass is normally re-run after regalloc
with a well-defined grf_used value and accurate bank conflict
information.
Fixes: acf98ff933 "intel/fs: Teach instruction scheduler about GRF bank conflict cycles."
Reported-by: Eero Tamminen <eero.t.tamminen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
The weight_vector_type constructor was inadvertently assuming C++17
semantics of the new operator applied on a type with alignment
requirement greater than the largest fundamental alignment.
Unfortunately on earlier C++ dialects the implementation was allowed
to raise an allocation failure when the alignment requirement of the
allocated type was unsupported, in an implementation-defined fashion.
It's expected that a C++ implementation recent enough to implement
P0035R4 would have honored allocation requests for such over-aligned
types even if the C++17 dialect wasn't active, which is likely the
reason why this problem wasn't caught by our CI system.
A more elegant fix would involve wrapping the __SSE2__ block in a
'__cpp_aligned_new >= 201606' preprocessor conditional and continue
taking advantage of the language feature, but that would yield lower
compile-time performance on old compilers not implementing it
(e.g. GCC versions older than 7.0).
Fixes: af2c320190 "intel/fs: Implement GRF bank conflict mitigation pass."
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104226
Reported-by: Józef Kucia <joseph.kucia@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Vulkan spec doesn't specify that VK_REMAINING_ARRAY_LAYERS is allowed
in the passed VkClearRect struct.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
We still have gpu hangs on Cannonlake when using push constants, so
disable them for now until we have a proper fix for these hangs.
v2: Add warning message when creating context too.
Signed-off-by: Rafael Antognolli <rafael.antognolli@intel.com>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
According to the RENDER_SURFACE_STATE internal documentation, the
R32G32B32_FLOAT restriction is marked "IVB" only. We choose to apply
it to Ivybridge and Baytrail, but not Haswell.
Apparently fixes KHR-GL46.texture_size_promotion.functional on Haswell.
Changes these tests from crashing to skipping on Haswell:
- KHR-GL46.direct_state_access.textures_storage_multisample_2d_rgb32f
- KHR-GL46.direct_state_access.textures_storage_multisample_3d_rgb32f
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Unfortunately, in aubinator and aubinator_error_decode we don't always
know how many of a given state we have, so we must guess. One day,
we'll come up with a way to annotate the batch to solve this problem.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
The shared framework can now do everything that aubinator_error_decode
ever did and more. It's time to make the switch.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Previously, if a group was nested in another group such that it didn't
start on a dword boundary, we would decode it as if it started at the
start of its first dword. This changes things to work even more in
terms of bits so that we can properly decode these structs. This
affects MOCS, attribute swizzles, and several other things.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
We can write to the same output but in different components, like
in this example:
layout(location = 0, component = 0) out ivec2 dEQP_FragColor_0;
layout(location = 0, component = 2) out ivec2 dEQP_FragColor_1;
Therefore, they are not two different outputs but only one.
Fixes:
dEQP-VK.glsl.440.linkage.varying.component.frag_out.*
v3:
- Remove FRAG_RESULT_MAX.
- Add const and use sizeof (Ian).
- Do three-pass to set properly the locations of fragment
outputs when having arrays (Jason).
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Push constants on Intel hardware are significantly more performant than
pull constants. Since most Vulkan applications don't actively use push
constants on Vulkan or at least don't use it heavily, we're pulling way
more than we should be. By enabling pushing chunks of UBOs we can get
rid of a lot of those pulls.
On my SKL GT4e, this improves the performance of Dota 2 and Talos by
around 2.5% and improves Aztec Ruins by around 2%.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
In Vulkan, we don't support classic pull constants and everything the
client asks us to push, we push. However, for pushed UBOs, we still
want to fall back to conventional pulls if we run out of space.
Push constants work in terms of 32-byte chunks so if we want to be able
to push UBOs, every thing needs to be 32-byte aligned. Currently, we
only require 16-byte which is too small.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
In order to do this we have to modify push constant set up to handle
ranges. We also have to tweak the way we handle dirty bits a bit so
that we re-push whenever a descriptor set changes.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
There are several places where we look up opcodes in an array of stages.
Assert that the we don't end up going out-of-bounds.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
We want to call brw_nir_analyze_ubo_ranges immedately after
anv_nir_apply_pipeline_layout and it badly wants constants. We could
run an optimization step and let constant folding do it but that's way
more expensive than needed. It's really easy to just handle constants
in apply_pipeline_layout.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
This rewires the logic for assigning uniform locations to work in terms
of "complex alignments". The basic idea is that, as we walk the list of
instructions, we keep track of the alignment and continuity requirements
of each slot and assert that the alignments all match up. We then use
those alignments in the compaction stage to ensure that everything gets
placed at a properly aligned register. The old mechanism handled
alignments by special-casing each of the bit sizes and placing 64-bit
values first followed by 32-bit values.
The old scheme had the advantage of never leaving a hole since all the
64-bit values could be tightly packed and so could the 32-bit values.
However, the new scheme has no type size special cases so it handles not
only 32 and 64-bit types but should gracefully extend to 16 and 8-bit
types as the need arises.
Tested-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
The testing for this extension is currently very poor. The CTS tests
only test accessing UBOs and SSBOs at dynamic offsets so none of our
constant-offset paths get triggered at all. Also, there's an assertion
in our handling of nir_intrinsic_load_uniform that offset % 4 == 0 which
is never triggered indicating that nothing every gets loaded from an
offset which is not a dword. Both push constants and the constant
offset pull paths are complex enough, we really don't want to ship
without tests. We'll turn the extension back on once we have decent
tests.
This addresses a long-standing back-end compiler bug that could lead
to cross-channel data corruption in loops executed non-uniformly. In
some cases live variables extending through a loop divergence point
(e.g. a non-uniform break) into a convergence point (e.g. the end of
the loop) wouldn't be considered live along all physical control flow
paths the SIMD thread could possibly have taken in between due to some
channels remaining in the loop for additional iterations.
This patch fixes the problem by extending the CFG with physical edges
that don't exist in the idealized non-vectorized program, but
represent valid control flow paths the SIMD EU may take due to the
divergence of logical threads. This makes sense because the i965 IR
is explicitly SIMD, and it's not uncommon for instructions to have an
influence on neighboring channels (e.g. a force_writemask_all header
setup), so the behavior of the SIMD thread as a whole needs to be
considered.
No changes in shader-db.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This makes the dataflow propagation logic of the copy propagation pass
more intelligent in cases where the destination of a copy is known to
be undefined for some incoming CFG edges, building upon the
definedness information provided by the last patch. Helps a few
programs, and avoids a handful shader-db regressions from the next
patch.
shader-db results on ILK:
total instructions in shared programs: 6541547 -> 6541523 (-0.00%)
instructions in affected programs: 360 -> 336 (-6.67%)
helped: 8
HURT: 0
LOST: 0
GAINED: 10
shader-db results on BDW:
total instructions in shared programs: 8174323 -> 8173882 (-0.01%)
instructions in affected programs: 7730 -> 7289 (-5.71%)
helped: 5
HURT: 2
LOST: 0
GAINED: 4
shader-db results on SKL:
total instructions in shared programs: 8185669 -> 8184598 (-0.01%)
instructions in affected programs: 10364 -> 9293 (-10.33%)
helped: 5
HURT: 2
LOST: 0
GAINED: 2
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Currently the liveness analysis pass would extend a live interval up
to the top of the program when no unconditional and complete
definition of the variable is found that dominates all of its uses.
This can lead to a serious performance problem in shaders containing
many partial writes, like scalar arithmetic, FP64 and soon FP16
operations. The number of oversize live intervals in such workloads
can cause the compilation time of the shader to explode because of the
worse than quadratic behavior of the register allocator and scheduler
when running out of registers, and it can also cause the running time
of the shader to explode due to the amount of spilling it leads to,
which is orders of magnitude slower than GRF memory.
This patch fixes it by computing the intersection of our current live
intervals with the subset of the program that can possibly be reached
from any definition of the variable. Extending the storage allocation
of the variable beyond that is pretty useless because its value is
guaranteed to be undefined at a point that cannot be reached from any
definition.
According to Jason, this improves performance of the subgroup Vulkan
CTS tests significantly (e.g. the runtime of the dvec4 broadcast test
improves by nearly 50x).
No significant change in the running time of shader-db (with 5%
statistical significance).
shader-db results on IVB:
total cycles in shared programs: 61108780 -> 60932856 (-0.29%)
cycles in affected programs: 16335482 -> 16159558 (-1.08%)
helped: 5121
HURT: 4347
total spills in shared programs: 1309 -> 1288 (-1.60%)
spills in affected programs: 249 -> 228 (-8.43%)
helped: 3
HURT: 0
total fills in shared programs: 1652 -> 1597 (-3.33%)
fills in affected programs: 262 -> 207 (-20.99%)
helped: 4
HURT: 0
LOST: 2
GAINED: 209
shader-db results on BDW:
total cycles in shared programs: 67617262 -> 67361220 (-0.38%)
cycles in affected programs: 23397142 -> 23141100 (-1.09%)
helped: 8045
HURT: 6488
total spills in shared programs: 1456 -> 1252 (-14.01%)
spills in affected programs: 465 -> 261 (-43.87%)
helped: 3
HURT: 0
total fills in shared programs: 1720 -> 1465 (-14.83%)
fills in affected programs: 471 -> 216 (-54.14%)
helped: 4
HURT: 0
LOST: 2
GAINED: 162
shader-db results on SKL:
total cycles in shared programs: 65436248 -> 65245186 (-0.29%)
cycles in affected programs: 22560936 -> 22369874 (-0.85%)
helped: 8457
HURT: 6247
total spills in shared programs: 437 -> 437 (0.00%)
spills in affected programs: 0 -> 0
helped: 0
HURT: 0
total fills in shared programs: 870 -> 854 (-1.84%)
fills in affected programs: 16 -> 0
helped: 1
HURT: 0
LOST: 0
GAINED: 107
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
This should allow the post-RA scheduler to do a slightly better job at
hiding latency in presence of instructions incurring bank conflicts.
The main purpuse of this patch is not to improve performance though,
but to get conflict cycles to show up in shader-db statistics in order
to make sure that regressions in the bank conflict mitigation pass
don't go unnoticed.
Acked-by: Matt Turner <mattst88@gmail.com>
Unnecessary GRF bank conflicts increase the issue time of ternary
instructions (the overwhelmingly most common of which is MAD) by
roughly 50%, leading to reduced ALU throughput. This pass attempts to
minimize the number of bank conflicts by rearranging the layout of the
GRF space post-register allocation. It's in general not possible to
eliminate all of them without introducing extra copies, which are
typically more expensive than the bank conflict itself.
In a shader-db run on SKL this helps roughly 46k shaders:
total conflicts in shared programs: 1008981 -> 600461 (-40.49%)
conflicts in affected programs: 816222 -> 407702 (-50.05%)
helped: 46234
HURT: 72
The running time of shader-db itself on SKL seems to be increased by
roughly 2.52%±1.13% with n=20 due to the additional work done by the
compiler back-end.
On earlier generations the pass is somewhat less effective in relative
terms because the hardware incurs a bank conflict anytime the last two
sources of the instruction are duplicate (e.g. while trying to square
a value using MAD), which is impossible to avoid without introducing
copies. E.g. for a shader-db run on SNB:
total conflicts in shared programs: 944636 -> 623185 (-34.03%)
conflicts in affected programs: 853258 -> 531807 (-37.67%)
helped: 31052
HURT: 19
And on BDW:
total conflicts in shared programs: 1418393 -> 987539 (-30.38%)
conflicts in affected programs: 1179787 -> 748933 (-36.52%)
helped: 47592
HURT: 70
On SKL GT4e this improves performance of GpuTest Volplosion by 3.64%
±0.33% with n=16.
NOTE: This patch intentionally disregards some i965 coding conventions
for the sake of reviewability. This is addressed by the next
squash patch which introduces an amount of (for the most part
boring) boilerplate that might distract reviewers from the
non-trivial algorithmic details of the pass.
The following patch is squashed in:
SQUASH: intel/fs/bank_conflicts: Roll back to the nineties.
Acked-by: Matt Turner <mattst88@gmail.com>
