From the Vulkan spec with KHX extensions:
"If queries are used while executing a render pass instance that has
multiview enabled, the query uses N consecutive query indices
in the query pool (starting at query) where N is the number of bits
set in the view mask in the subpass the query is used in.
How the numerical results of the query are distributed among the
queries is implementation-dependent. For example, some implementations
may write each view's results to a distinct query, while other
implementations may write the total result to the first query and write
zero to the other queries. However, the sum of the results in all the
queries must accurately reflect the total result of the query summed
over all views. Applications can sum the results from all the queries to
compute the total result."
In our case we only really emit a single query (in the first query index)
that stores the aggregated result for all views, but we still need to manage
availability for all the other query indices involved, even if we don't
actually use them.
This is relevant when clients call vkGetQueryPoolResults and pass all N
queries to retrieve the results. In that scenario, without this patch,
we will never see queries other than the first being available since we
never emit them.
v2: we need the same treatment for timestamp queries.
v3 (Jason):
- Better an if instead of an early return.
- We can't write to this memory in the CPU, we should use
MI_STORE_DATA_IMM and emit_query_availability (Jason).
v4 (Jason):
- No need to take the value to write as parameter, just hard code it to 0.
Fixes test failures in some work-in-progress CTS multiview+query tests.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Since setup of ALLOW_RGB10_CONFIGS was moved to i965's own
brw_config_options.xml, this was hard-coded to false and
could not be overriden by drirc. Add some parsing into
i965's private screen->optionCache to enable drirc again.
Fixes: b391fb26df ("dri_util: remove ALLOW_RGB10_CONFIGS option (v2)")
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Marek Olšák <marek.olsak@amd.com>
Cc: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
This was handled for VS, but not for GS.
Fixes for gallium drivers using nir:
spec@arb_gpu_shader5@arb_gpu_shader5-xfb-streams-without-invocations
spec@arb_gpu_shader5@arb_gpu_shader5-xfb-streams*
spec@arb_transform_feedback3@arb_transform_feedback3-ext_interleaved_two_bufs_gs*
spec@ext_transform_feedback@geometry-shaders-basic
spec@ext_transform_feedback@* use_gs
spec@glsl-1.50@execution@geometry@primitive-id*
spec@glsl-1.50@execution@geometry@tri-strip-ordering-with-prim-restart gl_triangle_strip *
spec@glsl-1.50@transform-feedback-builtins
spec@glsl-1.50@transform-feedback-type-and-size
v2: don't call st_translate_program_stream_output) for TCS
v3: drop scanning patch outputs as TCS can't output xfb
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Tested-by: Karol Herbst <kherbst@redhat.com>
if no destination:
a) convert _RET instructions to non _RET variants if no dst
b) set src0 to undefined if it's a READ, this should get DCE then.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The normal ssa renumbering isn't sufficient for LDS queue access,
this uses two stacks, one for the lds queue, and one for the
lds r/w ordering.
The LDS oq values are incremented in their use in a linear
fashion.
The LDS rw values are incremented in their definitions and used
in the next lds operation to ensure reordering doesn't occur.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
So LDS ops have to be SLOT_X,
and LDS OQ reads have read port restrictions so we try
and force those into only having one per slot and avoiding
bank swizzles.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This tries to avoid an lds queue read getting scheduled separately
from an lds ret read, the non-sb code uses the same style of hammer,
this isn't foolproof.
We can do better, but it's a bit tricky, as you have to scan ahead
and either schedule more lds oq moves and more lds reads and that
could lead to you running out of space anyways.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This adds support for tracking the lds oq read/writes
so can avoid scheduling other things in between.
This patch just adds the tracking and assert to show
problems.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
You have to schedule LDS_READ_RET _, x and MOV reg, LDS_OQ_A_POP
in the same basic block/clause. This makes sure once we've issues
and MOV we don't add another block until we balance it with an
LDS read.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
We need to convert these to the hw special registers.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This handles parsing the LDS ops and queue accessess.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This fixes bad interactions with the LDS special values.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
For LDS read/write ordering we use the LDS_RW value, reads
will wait on previous writes.
For LDS read/read from LDS queue ordering we use the LDS_OQ
values, we define two for now, though initially we'll just
support OQA.
Also add the check for the lds oq values
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It's rare to have a final alu clause on normal shaders (exports)
but tess shaders write to LDS as their output, so we see some
alu clauses, and the CF_END get put in the wrong place.
This makes sure to update last_cf correctly.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This adds support for GDS ops to sb backend.
This seems to work for atomics and tess factor writes.
Acked-By: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Some tess shaders were doing MOVA_INT _, c0.x on cayman, and then
hitting an assert in sb_bc_finalize.cpp:translate_kcache.
This makes sure the toplevel kcache tracker gets updated,
and the clause gets fixed up.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This field is ignored for tf writes so should be 0.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Looks like the decompress does not handle invalid encodings well,
which happens with random memory. Of course apps should not use it
with random memory, but they are allowed to ....
Fixes: 44fcf58744 "radv: Disable DCC for GENERAL layout and compute transfer dest."
Reviewed-by: Dave Airlie <airlied@redhat.com>
Now that we have two of these, we're duplicating a bunch of this logic.
The next commit will add more logic, which would make the duplication
seem worse.
This ends up setting EXEC_OBJECT_CAPTURE on the batch, which isn't
necessary (it's already captured), but it should be harmless.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Having a boolean for "we're using malloc'd shadow copies for all
buffers" is cleaner than having a cpu_map pointer for each. It was
okay when we had one buffer, but this is more obvious.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Previously the dataflow propagation algorithm would calculate the ACP
live-in and -out sets in a two-pass fixed-point algorithm. The first
pass would update the live-out sets of all basic blocks of the program
based on their live-in sets, while the second pass would update the
live-in sets based on the live-out sets. This is incredibly
inefficient in the typical case where the CFG of the program is
approximately acyclic, because it can take up to 2*n passes for an ACP
entry introduced at the top of the program to reach the bottom (where
n is the number of basic blocks in the program), until which point the
algorithm won't be able to reach a fixed point.
The same effect can be achieved in a single pass by computing the
live-in and -out sets in lock-step, because that makes sure that
processing of any basic block will pick up the updated live-out sets
of the lexically preceding blocks. This gives the dataflow
propagation algorithm effectively O(n) run-time instead of O(n^2) in
the acyclic case.
The time spent in dataflow propagation is reduced by 30x in the
GLES31.functional.ssbo.layout.random.all_shared_buffer.5 dEQP
test-case on my CHV system (the improvement is likely to be of the
same order of magnitude on other platforms). This more than reverses
an apparent run-time regression in this test-case from my previous
copy-propagation undefined-value handling patch, which was ultimately
caused by the additional work introduced in that commit to account for
undefined values being multiplied by a huge quadratic factor.
According to Chad this test was failing on CHV due to a 30s time-out
imposed by the Android CTS (this was the case regardless of my
undefined-value handling patch, even though my patch substantially
exacerbated the issue). On my CHV system this patch reduces the
overall run-time of the test by approximately 12x, getting us to
around 13s, well below the time-out.
v2: Initialize live-out set to the universal set to avoid rather
pessimistic dataflow estimation in shaders with cycles (Addresses
performance regression reported by Eero in GpuTest Piano).
Performance numbers given above still apply. No shader-db changes
with respect to master.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104271
Reported-by: Chad Versace <chadversary@chromium.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
