Jason Ekstrand [Wed, 18 Jan 2017 02:38:39 +0000 (18:38 -0800)]
nir/gcm: Add a real concept of "progress"
Now that the GCM pass is more conservative and only moves instructions
to different blocks when it's advantageous to do so, we can have a
proper notion of what it means to make progress.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4636>
Jason Ekstrand [Wed, 18 Jan 2017 02:38:38 +0000 (18:38 -0800)]
nir/gcm: Move block choosing into a helper function
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4636>
Jason Ekstrand [Wed, 18 Jan 2017 02:38:37 +0000 (18:38 -0800)]
nir/gcm: Use an array for storing the early block
We are about to adjust our instruction block assignment algorithm and we
will want to know the current block that the instruction lives in. In
order to allow for this, we can't overwrite nir_instr::block in the
early scheduling pass.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4636>
Jason Ekstrand [Wed, 18 Jan 2017 02:38:36 +0000 (18:38 -0800)]
nir/gcm: Loop over blocks in pin_instructions
Now that we have the new block iterators, we can simplify things a bit.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4636>
Jason Ekstrand [Wed, 18 Jan 2017 02:38:35 +0000 (18:38 -0800)]
nir/dominance: Better handle unreachable blocks
v2: Fix minor comments (Ken)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4636>
Daniel Schürmann [Sun, 19 Apr 2020 15:17:04 +0000 (16:17 +0100)]
aco: use v_subrev_f32 for fsub with an sgpr operand in src1
This fixes an accidentally introduced regression.
Fixes: 9be4be515f2a08b9c9e5ae1fc4c5dc9a830c2337 ('aco: implement 16-bit nir_op_fsub/nir_op_fadd')
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4633>
Daniel Stone [Sun, 19 Apr 2020 11:57:04 +0000 (12:57 +0100)]
CI: Disable Lima jobs due to lab unhealthiness
The BayLibre LAVA host appears to be down.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4631>
Daniel Stone [Thu, 16 Apr 2020 22:31:47 +0000 (23:31 +0100)]
ci/windows: Make Chocolatey installs more reliable
Chocolatey installs depend on downloading binaries from SourceForge,
which is an unreliable host: container builds often fail because it
cannot pick up winflexbison.
Add a loop to retry chocolatey installs if any installs have failed, and
ensure Python is in the accessible PowerShell path rather than relying
on the path being externally refreshed.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4631>
Arcady Goldmints-Orlov [Thu, 16 Apr 2020 21:28:12 +0000 (16:28 -0500)]
nir: Lower returns correctly inside nested loops
Inside nested flow control, nir_lower_returns inserts predicated breaks
in the outer block. However, it would omit doing this if the remainder
of the outer block (after the inner block) was empty. This is not
correct in the case of loops, as execution just wraps back around to the
start of the loop, so this change doesn't skip the predication inside
loops.
Fixes: 79dec93ead6e (nir: Add return lowering pass)
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2724
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4603>
Jason Ekstrand [Thu, 16 Apr 2020 20:33:43 +0000 (15:33 -0500)]
anv: Apply any needed PIPE_CONTROLs before emitting state
Push constants in particular can get picked up by the hardware at weird
times that happen *before* 3DPRIMITIVE. Therefore, we need to flush
before we emit all our state to ensure that any data they may pick up is
in memory in time. This fixes an app which does vkCmdCopyBuffers
immediately followed by a vkCmdBeginRenderPass and vkCmdDraw which uses
the destination of the copy as a UBO which we push.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4601>
Jason Ekstrand [Thu, 16 Apr 2020 20:31:15 +0000 (15:31 -0500)]
anv: Move vb_emit setup closer to where it's used in flush_state
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4601>
Albert Astals Cid [Wed, 26 Feb 2020 22:05:51 +0000 (23:05 +0100)]
Fix promotion of floats to doubles
Use the f variants of the math functions if the input parameter is a
float, saves converting from float to double and running the double
variant of the math function for gaining no precision at all
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3969>
Connor Abbott [Fri, 17 Apr 2020 17:31:56 +0000 (19:31 +0200)]
ir3/ra: Fix off-by-one issues with live-range extension
The intersects() function assumes that inside each instruction values
always die before they are defined, so that if the end of one range is
the same instruction as the beginning of the next then they don't
intersect. However, this isn't the case for values that become live at
the beginning of a basic block, which become live *before* the first
instruction, or instructions that die at the end of a basic block which
die after the last instruction.
For example, imagine that we have two values, A which is defined earlier
in the block and B which is defined in the last instruction of the block
and both die at the end of the basic block (e.g. are used in the next
iteration of a loop). We would compute a range for A of, say, (10, 20)
and for B of (20, 20) since each block's end_ip is the same as the ip of
the last instruction, and RA would consider them to not interfere.
There's a similar problem with values that become live at the beginning.
The fix is to offset the block's start_ip and end_ip by one so that they
don't correspond to any actual instruction. One way to think about this
is that we're adding fake instructions at the beginning and end of a
block where values become live & die. We could invert the order, so that
values consumed by each instruction are considered dead at the end of
the previous instruction, but then values that become dead at the
beginning of the basic block would incorrectly have an empty live range,
with a similar problem at the end of the basic block if we try to say
that values are defined at the beginning of the next instruction. So
the extra padding instructions are unavoidable.
This fixes an accidental infinite loop in the shader for
dEQP-VK.spirv_assembly.type.scalar.u32.switch_vert.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4614>
Lionel Landwerlin [Fri, 17 Apr 2020 17:42:41 +0000 (20:42 +0300)]
util/sparse_free_list: manipulate node pointers using atomic primitives
Probably doesn't fix anything but those should be accessed in an
atomic way just like the head pointer.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: e4f01eca3b3cd1 ("util: Add a free list structure for use with util_sparse_array")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4613>
Timothy Arceri [Thu, 12 Mar 2020 23:17:08 +0000 (10:17 +1100)]
glsl: only set stage ref when uniforms referenced in stage
This updates the NIR uniform linker to behave like the GLSL IR
linker and fixes a number of CTS tests once we enable the NIR
linker for glsl.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Tue, 31 Mar 2020 02:49:30 +0000 (13:49 +1100)]
glsl: pull mark_array_elements_referenced() out into common helper
We will reuse this helper in the NIR linker in the following
patches.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Tue, 10 Mar 2020 01:55:53 +0000 (12:55 +1100)]
glsl: fix block index in NIR uniform linker
We only want to set the index for the first block of an array. Also
add a comment about why we do not break here.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Thu, 5 Mar 2020 23:27:44 +0000 (10:27 +1100)]
glsl: error check max user assignable uniform locations
This adds the error check to the NIR uniform linker.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Thu, 5 Mar 2020 03:46:26 +0000 (14:46 +1100)]
glsl: fix explicit locations for the glsl linker
We already reserved explicit locations in the GLSL linker.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Thu, 5 Mar 2020 03:51:20 +0000 (14:51 +1100)]
Revert "glsl: fix resizing of the uniform remap table"
This reverts commit
e0aa0a839f9c168784a1f50013c83877cc876094.
Instead we fix it correctly in the following patch.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Wed, 26 Feb 2020 02:17:02 +0000 (13:17 +1100)]
glsl: tidy up uniform storage value count code in NIR linker
This makes the code cleaner and better reflects what the existing
glsl IR linker does possibly fixing subtle bugs.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Thu, 6 Feb 2020 23:55:03 +0000 (10:55 +1100)]
glsl: fix struct offsets in the nir uniform linker
This change properly applies layouts to structs of uniforms in a
similar way to the GLSL IR linker.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Timothy Arceri [Fri, 13 Mar 2020 03:18:27 +0000 (14:18 +1100)]
nir: add matrix_layout to nir_variable data
This will be used by the following patch.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4623>
Lionel Landwerlin [Fri, 17 Apr 2020 06:51:22 +0000 (09:51 +0300)]
anv: skip writing perfcntr in results on Gen12+
We were not capturing the register already so don't bother writing the
delta in the results (we were previously doing a delta between two 0
values).
v2: Fix unused function warning
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4586>
Lionel Landwerlin [Thu, 16 Apr 2020 10:03:47 +0000 (13:03 +0300)]
intel/perf: Enable MDAPI queries for Gen12
We're missing the cases for gen12 leading to those metrics going
missing.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 15b7b56eb2fb41 ("intel/perf: add TGL support")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4586>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:40:12 +0000 (15:40 -0400)]
pan/bit: Add fp16 min/max tests
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 15:40:59 +0000 (11:40 -0400)]
pan/bit: Add constants test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 21:00:31 +0000 (17:00 -0400)]
pan/bit: Add fexp2_fast test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:59:41 +0000 (16:59 -0400)]
pan/bit: Add fexp2_fast interp
Kind of a hack and not at all how the h/w does it.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:45:22 +0000 (16:45 -0400)]
pan/bit: Add FMA_MSCALE test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:21:14 +0000 (16:21 -0400)]
pan/bit: _MSCALE interp
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 19:25:27 +0000 (15:25 -0400)]
pan/bit: Add BI_TABLE test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 19:25:17 +0000 (15:25 -0400)]
pan/bit: Add log2 helper interp
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 17:19:13 +0000 (13:19 -0400)]
pan/bit: Add FMA_REDUCE test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 17:14:51 +0000 (13:14 -0400)]
pan/bit: Add BI_REDUCE_FMA interp
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:55:02 +0000 (12:55 -0400)]
pan/bit: Add frexp_log test
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:50:48 +0000 (12:50 -0400)]
pan/bit: Add FREXP interp support
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:52:18 +0000 (15:52 -0400)]
pan/bi: Lower special ops to 32-bit
We don't have 16-bit tables. We could probably do a bit better to avoid
so many conversions but hey.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:52:03 +0000 (15:52 -0400)]
pan/bi: Round constants to 32-bit
We can only access lo/hi at 32-bit intervals.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:40:00 +0000 (15:40 -0400)]
pan/bi: Dump extra bits for disasm
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:37:21 +0000 (15:37 -0400)]
pan/bi: Pack MAX.v2f16
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:21:08 +0000 (15:21 -0400)]
pan/bi: Pack ADD.v2f16
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Fri, 17 Apr 2020 19:20:50 +0000 (15:20 -0400)]
pan/bi: Structify add and min/max fp16 ADD
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 15:45:49 +0000 (11:45 -0400)]
pan/bi: Workaround constant packing errata
Incomplete fix.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 14:39:42 +0000 (10:39 -0400)]
pan/bi: Try to reuse constants in ALU
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 00:20:37 +0000 (20:20 -0400)]
pan/bi: Handle st_vary with <4 components
Still no writemasks.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 00:20:16 +0000 (20:20 -0400)]
pan/bi: Fix vec2/3 handling
Otherwise we get moves from null.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 23:50:24 +0000 (19:50 -0400)]
pan/bi: Implement flog2
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:52:21 +0000 (18:52 -0400)]
pan/bi: Add fexp2 implementation
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:51:55 +0000 (18:51 -0400)]
pan/bi: Fix lower_combine swizzle rewrite
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:36:28 +0000 (18:36 -0400)]
pan/bi: Fix packing with low-nibble-set on hi constant
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:32:31 +0000 (18:32 -0400)]
pan/bi: Fix packing with multiple constants
Need to use bottom nibble of the 64, not the half.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:29:36 +0000 (18:29 -0400)]
pan/bi: Fix bi_get_immediate with multiple imms
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Wed, 15 Apr 2020 00:09:00 +0000 (20:09 -0400)]
pan/bi: Ensure CONSTANT srcs have types
So the next commit is valid.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 22:29:08 +0000 (18:29 -0400)]
pan/bi: Split src/dest index printing
So we can handle constant printing correctly.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:59:19 +0000 (16:59 -0400)]
pan/bi: Add fexp2_fast packing
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:44:44 +0000 (16:44 -0400)]
pan/bi: Pack FMA_MSCALE
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:31:34 +0000 (16:31 -0400)]
pan/bi: Structify FMA_MSCALE
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 20:13:53 +0000 (16:13 -0400)]
pan/bi: Add _MSCALE flag for FMA/ADD
So we can bias by exponents.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 19:25:04 +0000 (15:25 -0400)]
pan/bi: Add log2_help packing
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 17:14:29 +0000 (13:14 -0400)]
pan/bi: Pack ADD_FREXPM
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 17:14:05 +0000 (13:14 -0400)]
pan/bi: Add bi_pack_fma_2src helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:54:17 +0000 (12:54 -0400)]
pan/bi: Add frexp_log packing
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:37:29 +0000 (12:37 -0400)]
pan/bi: Add log_frexpe op to IR
As part of BI_FREXP
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:34:42 +0000 (12:34 -0400)]
pan/bi: Add FLOG2_U op to disassembler
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:33:08 +0000 (12:33 -0400)]
pan/bi: Add op for ADD_FREXPM
Used in log2. Needs a new class as well due to scheduling silliness.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:22:28 +0000 (12:22 -0400)]
pan/bi: Add special op for exp2
Needs some extra help but basically exp2_fast
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 16:21:25 +0000 (12:21 -0400)]
pan/bi: Add BI_TABLE for fast table accesses
Used to implement SPECIAL ops. Separate class since they are faster
which means you can pair them with actual work on FMA.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Tue, 14 Apr 2020 00:03:48 +0000 (20:03 -0400)]
pan/bi: Disable FMA scheduling for CONVERT
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Alyssa Rosenzweig [Mon, 13 Apr 2020 22:28:29 +0000 (18:28 -0400)]
pan/bi: Add disasm for ADD.i8
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4615>
Jason Ekstrand [Thu, 9 Apr 2020 22:09:10 +0000 (17:09 -0500)]
spirv,nir: Move the SPIR-V vector insert code to NIR
This also makes spirv_to_nir a bit simpler because the new
nir_vector_insert helper automatically handles a constant component
selector like nir_vector_extract does.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
Jason Ekstrand [Thu, 9 Apr 2020 22:03:37 +0000 (17:03 -0500)]
spirv: Call nir_builder directly for vector_extract
The nir_builder helper already handles checking if the component
selector is an immediate and returns an undef in the OOB case.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
Jason Ekstrand [Fri, 17 Apr 2020 14:52:25 +0000 (09:52 -0500)]
nir/builder: Handle any bit-size selector in nir_extract
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
Jason Ekstrand [Thu, 9 Apr 2020 21:58:58 +0000 (16:58 -0500)]
spirv: Error if OpCompositeInsert/Extract has OOB indices
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
Jason Ekstrand [Wed, 8 Apr 2020 16:14:40 +0000 (11:14 -0500)]
spirv,nir: Add a better vector_insert
The old one in spirv_to_nir was besel'ing the whole vector for every
component. If we think about this as a vector operation, we can do it
way more efficiently.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
Jason Ekstrand [Thu, 9 Apr 2020 21:44:33 +0000 (16:44 -0500)]
spirv: Handle OOB vector extract operations
We use vtn_vector_extract to handle vector component level derefs. This
makes us gracefully handle the case where your vector component is OOB
and give you an undef. The SPIR-V working group is still working out
whether or not this is technically legal but it's very little code for
us to handle it so we may as well.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4495>
D Scott Phillips [Fri, 17 Apr 2020 17:28:06 +0000 (10:28 -0700)]
util/sparse_array: don't stomp head's counter on pop operations
By temporarily storing the new_head by a uint32_t, we wipe out the
counter section of the head pointer.
Fixes: e4f01eca ("util: Add a free list structure for use with util_sparse_array")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4612>
Danylo Piliaiev [Thu, 16 Apr 2020 13:34:25 +0000 (16:34 +0300)]
st/mesa: Update shader info of ffvp/ARB_vp after translation to NIR
We must update stp->Base.info after translation and before
st_prepare_vertex_program is called, because inputs_read
may become outdated after NIR optimization passes.
For ffvp/ARB_vp inputs_read is populated based on declared
attributes without taking their usage into consideration.
When creating shader variants we expect that their inputs_read
would match the base ones for input mapping to work properly.
Cc: <mesa-stable@lists.freedesktop.org>
Fixes: 8a0dd0af3f1a6c0310a08daf4220132ec6815b31
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2758
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4598>
Samuel Pitoiset [Wed, 15 Apr 2020 06:49:32 +0000 (08:49 +0200)]
aco: fix exporting the viewport index if the fragment shader needs it
It's like the layer, it has to be exported via the pos and also
as a varying if the fragment shader reads it.
Fixes dEQP-VK.draw.shader_viewport_index.fragment_shader_*
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4564>
Samuel Pitoiset [Wed, 15 Apr 2020 06:37:40 +0000 (08:37 +0200)]
radv/llvm: fix exporting the viewport index if the fragment shader needs it
It's like the layer, it has to be exported via the pos and also
as a varying if the fragment shader reads it.
Fixes dEQP-VK.draw.shader_viewport_index.fragment_shader_*
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4564>
Samuel Pitoiset [Thu, 16 Apr 2020 14:58:13 +0000 (16:58 +0200)]
radv: set missing SHARED_VGPR_CNT for NGG VS and ACO
shuffle is implemented with shared VGPRs with ACO and Wave64.
Fixes dEQP-VK.subgroups.shuffle.framebuffer.subgroupshuffle*_vertex
with Wave64.
Fixes: c24d9522dae ("radv: Enable ACO for NGG VS/TES, but disable NGG for ACO GS.")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4595>
Samuel Pitoiset [Thu, 16 Apr 2020 14:15:45 +0000 (16:15 +0200)]
radv: fix geometry shader primitives query with ACO on GFX10
Fixes
dEQP-VK.query_pool.statistics_query.*.geometry_shader_primitives.*.
Fixes: c24d9522dae ("radv: Enable ACO for NGG VS/TES, but disable NGG for ACO GS.")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4593>
Ian Romanick [Tue, 14 Apr 2020 17:30:53 +0000 (10:30 -0700)]
intel/compiler: Fixup operands in fs_builder::emit() that takes array
The versions that take a specific number of operands will do various
fixups depending on the platform and the opcode. However, the version
that takes an array of sources did not. This makes all version operate
similarly.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Ian Romanick [Tue, 29 Oct 2019 23:56:39 +0000 (16:56 -0700)]
intel/compiler: CSEL can do saturate
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Ian Romanick [Wed, 30 Oct 2019 00:02:46 +0000 (17:02 -0700)]
intel/compiler: Only GE and L modifiers are commutative for SEL
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Ian Romanick [Tue, 14 Apr 2020 00:23:25 +0000 (17:23 -0700)]
intel/compiler: Silence unused parameter warning in update_inst_scoreboard
src/intel/compiler/brw_fs_scoreboard.cpp: In function ‘void {anonymous}::update_inst_scoreboard(const fs_visitor*, const ordered_address*, const fs_inst*, unsigned int, {anonymous}::scoreboard&)’:
src/intel/compiler/brw_fs_scoreboard.cpp:793:45: warning: unused parameter ‘shader’ [-Wunused-parameter]
793 | update_inst_scoreboard(const fs_visitor *shader, const ordered_address *jps,
| ~~~~~~~~~~~~~~~~~~^~~~~~
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Ian Romanick [Mon, 13 Apr 2020 23:55:43 +0000 (16:55 -0700)]
intel/compiler: Silence unused parameter warning in fs_live_variables::setup_one_read
src/intel/compiler/brw_fs_live_variables.cpp: In member function ‘void brw::fs_live_variables::setup_one_read(brw::fs_live_variables::block_data*, fs_inst*, int, const fs_reg&)’:
src/intel/compiler/brw_fs_live_variables.cpp:56:67: warning: unused parameter ‘inst’ [-Wunused-parameter]
56 | fs_live_variables::setup_one_read(struct block_data *bd, fs_inst *inst,
| ~~~~~~~~~^~~~
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Ian Romanick [Mon, 13 Apr 2020 23:49:11 +0000 (16:49 -0700)]
intel/compiler: Silence unused parameter warnings in vec4_tcs_visitor
In file included from src/intel/compiler/brw_vec4_tcs.cpp:31:
src/intel/compiler/brw_vec4_tcs.h: In member function ‘virtual void brw::vec4_tcs_visitor::emit_urb_write_header(int)’:
src/intel/compiler/brw_vec4_tcs.h:74:43: warning: unused parameter ‘mrf’ [-Wunused-parameter]
74 | virtual void emit_urb_write_header(int mrf) {}
| ~~~~^~~
src/intel/compiler/brw_vec4_tcs.h: In member function ‘virtual brw::vec4_instruction* brw::vec4_tcs_visitor::emit_urb_write_opcode(bool)’:
src/intel/compiler/brw_vec4_tcs.h:75:57: warning: unused parameter ‘complete’ [-Wunused-parameter]
75 | virtual vec4_instruction *emit_urb_write_opcode(bool complete) { return NULL; }
| ~~~~~^~~~~~~~
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4582>
Jason Ekstrand [Wed, 15 Apr 2020 18:12:02 +0000 (13:12 -0500)]
intel/blorp: Delete an unused enum
This was lying around from back when BLORP write to fs_visitor directly.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4606>
Jason Ekstrand [Sat, 4 Apr 2020 01:20:53 +0000 (20:20 -0500)]
anv: Emit pushed UBO bounds checking code in the back-end compiler
This commit fixes performance regressions introduced by
e03f9652801ad7
in which we started bounds checking our push constants. This added a
LOT of shader code to shaders which use the robustBufferAccess feature
and led to substantial spilling. The checking we just added to the FS
back-end is far more efficient for two reasons:
1. It can be done at a whole register granularity rather than per-
scalar and so we emit one SIMD8 SEL per 32B GRF rather than one
SIMD16 SEL (executed as two SELs) for each component loaded.
2. Because we do it with NoMask instructions, we can do it on whole
pushed GRFs without splatting them out to SIMD8 or SIME16 values.
This means that robust buffer access no longer explodes our register
pressure for no good reason.
As a tiny side-benefit, we're now using can use AND instead of SEL which
means no need for the flag and better scheduling.
Vulkan pipeline database results on ICL:
Instructions in all programs:
293586059 ->
238009118 (-18.9%)
SENDs in all programs:
13568515 ->
13568515 (+0.0%)
Loops in all programs: 149720 -> 149720 (+0.0%)
Cycles in all programs:
88499234498 ->
84348917496 (-4.7%)
Spills in all programs:
1229018 -> 184339 (-85.0%)
Fills in all programs:
1348397 -> 246061 (-81.8%)
This also improves the performance of a few apps:
- Shadow of the Tomb Raider: +4%
- Witcher 3: +3.5%
- UE4 Shooter demo: +2%
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4447>
Jason Ekstrand [Sat, 4 Apr 2020 02:25:40 +0000 (21:25 -0500)]
intel/cfg: Add first/last_block helpers
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4447>
Connor Abbott [Fri, 17 Apr 2020 13:28:10 +0000 (15:28 +0200)]
tu: Use tu_cs_add_entries() with non-render-pass secondaries
Even though vkCmdRenderPassBegin() isn't allowed inside a secondary
command buffer, vkCmdDispatch() is, and we emit an IB with compute
dispatches, which means that if the secondary command buffer records a
vkCmdDispatch() then we'll have an IB inside an IB, which is illegal.
Fixes hangs in e.g.
dEQP-VK.api.command_buffers.record_simul_use_secondary_one_primary.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4605>
Ilia Mirkin [Thu, 16 Apr 2020 02:27:11 +0000 (22:27 -0400)]
mesa: add interaction between compute derivatives and variable local sizes
This is an added interaction in NV_compute_shader_derivatives added in
Sep 2019.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4583>
Karol Herbst [Thu, 16 Apr 2020 14:01:47 +0000 (16:01 +0200)]
st/mesa: properly guard fallback_copy_texsubimage aginst failed maps
Fixes random crashes in some packed_pixel GLES CTS tests
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4592>
Pierre-Eric Pelloux-Prayer [Thu, 16 Apr 2020 12:52:29 +0000 (14:52 +0200)]
radeonsi: don't assume ctx is always a threaded_context
Fixes: dcb1e8fef8a ("radeonsi: use thread_context::bytes_mapped_limit")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4587>
Daniel Stone [Thu, 16 Apr 2020 13:56:18 +0000 (14:56 +0100)]
Revert "CI: Disable Windows/VS2019 builds"
DNS is now fixed.
This reverts commit
460b8b1758d953b2b820443615d73ccdb1455b5e.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4591>
Ilia Mirkin [Thu, 16 Apr 2020 18:23:22 +0000 (14:23 -0400)]
docs: update for recently-added nvc0 features
Also sort while we're at it.
And add NV_pixel_buffer_object.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4599>
Ilia Mirkin [Sat, 11 Apr 2020 07:04:01 +0000 (03:04 -0400)]
nv50,nvc0: update with latest caps
One notable change is that DRAW_INFO_START_WITH_USER_INDICES is enabled.
An audit of the code indicates that it should work, and a number of
piglit tests exercising glMultiDrawElements continue to function.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4520>
Jason Ekstrand [Wed, 15 Apr 2020 21:38:56 +0000 (16:38 -0500)]
intel/batch_decoder: Stop printing to stdout
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4597>
Jason Ekstrand [Wed, 15 Apr 2020 21:06:03 +0000 (16:06 -0500)]
anv: Report correct SLM size
Fixes: d787a2d0 "anv: Implement VK_KHR_pipeline_executable_properties"
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4597>
projects / mesa.git / log