Alyssa Rosenzweig [Thu, 21 Jul 2022 19:59:22 +0000 (15:59 -0400)]
pan/bi: Introduce TEXC_DUAL psuedoinstruction
There are two "shapes" of TEXC in the IR:
* Regular texturing. This TEXC writes a single set of staging registers.
* Dual texturing. This TEXC writes two sets of staging registers.
Currently we model both with a 2-destination TEXC, with a null second
destination for the usual case where dual texturing isn't used. This is awkward.
To make the "shapes" of instructions more predictable, make TEXC only write a
single set of staging registers (like the hardware instruction) and split off a
TEXC_DUAL pseudoinstruction for the second case, lowered late.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 19:43:29 +0000 (15:43 -0400)]
pan/bi: Use bi_emit_split_i32 for trimming vecs
This ensures the shape of the split is correct (no nulls in the destination).
This should not affect the register allocation, though it does "cost" extra
variable names.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 19:40:05 +0000 (15:40 -0400)]
pan/bi: Don't remove dests in DCE
Removing dests without removing instructions only makes sense for certain
pseudo-instructions, but it makes the IR needlessly complicated for all
instructions. There's no real reason to do so, we can signal this in a different
way instead.
No shader-db changes.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 18:53:35 +0000 (14:53 -0400)]
pan/bi: Register allocate BLEND dest on Valhall
On Bifrost, BLEND writes to the link register, acting like a function call. On
Valhall, BLEND does not write anything. But the BLEND instruction in our IR is a
pseudo-instruction with Bifrost semantics, expanding to a multi-instruction
sequence on Valhall. So it's not worth special casing Valhall in instruction
selection.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 18:47:49 +0000 (14:47 -0400)]
pan/bi: Improve register printing
Use "r0" style to denote machine registers, consistent with the assembly. Use
instead "nr0" for NIR registers so they don't collide. This reduces noise for
the average case (either register allocated or almost SSA).
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 18:45:49 +0000 (14:45 -0400)]
pan/bi: Clean up destination printing
Now we can trust in I->nr_dests, avoiding a bunch of special cases and incorrect
printing for instructions without a destination. To compensate for those
instructions being hard to scan now, indent.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 18:21:55 +0000 (14:21 -0400)]
pan/bi: Remove trivial dest[0] = bi_null()
It's already NULL and will interfere with the more comprehensive src/dest
overwork we're doing.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 15:53:23 +0000 (11:53 -0400)]
pan/bi: Add validation for nr_srcs/nr_dests
Now that we set nr_srcs/nr_dests accurately, assert as much in the validator.
This pass will be deleted later in the series, but having it here is expected to
be useful for bisection, in case there are cases missed. Certainly running the
CTS at this point in the series is helpful to prove completeness of the
beginning of this series.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 16:57:57 +0000 (12:57 -0400)]
pan/bi: Use nr_{srcs,dests} in bi_foreach_{src,dest}
This avoids crawling sources that must be null.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 20:08:59 +0000 (16:08 -0400)]
pan/bi: Consider nr_srcs/dests in CSE
Now that these are meaningful for COLLECT and SPLIT.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 16:04:23 +0000 (12:04 -0400)]
pan/bi: Model 3rd source for ATEST
ATEST on Valhall takes atest-param as an explicit encoded source. The source
also exists on Bifrost, but it's implied in the encoding. Rather than treating
it as a "ghost" source, model it in the IR explicitly so the count is correct.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 15:56:44 +0000 (11:56 -0400)]
pan/bi: Set I->nr_dests, I->nr_srcs
The builder is the primary producer of instructions, and generally the shape of
an instruction is fixed at build-time. It must set nr_dests/nr_srcs
appropriately. Likewise, when we modify sources later, we need to update
nr_srcs/nr_dests to keep everything consistent (and keep the tests passing).
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Fri, 22 Jul 2022 18:32:41 +0000 (14:32 -0400)]
pan/va: Split out compare instructions
The different combine modes form different instructions from each other and in
particular from the two-source version on Bifrost. Model them as such so we can
represent the relevant Valhall-specific lowering/optimizations accurately in the
compiler.
This requires updating the unit tests to use the new names since there's not
much point keeping around the aliases.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Fri, 22 Jul 2022 18:22:16 +0000 (14:22 -0400)]
pan/bi: Model [IF]CMP_{OR,AND,MULTI} ops
Valhall-style comparisons are 3-source, allowing multiple comparisons to be
combined without bitwise arithmetic. For example, the sequence
FCMP.f32.eq t, z, w
FCMP_AND.f32.eq dest, x, y, t
calculates (z == w) && (y == t) in 2 instructions (would be 3 on Bifrost).
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Thu, 21 Jul 2022 20:48:36 +0000 (16:48 -0400)]
pan/bi: Consider all dests in helper_block_update
If an instruction has multiple destinations and *any* of them are needed by
helper invocations, we should keep helper invocations alive. This is a bug fix.
Consider the GLSL:
first = texture(sampler, ...);
float res = texture(sampler, vec2(first.y)).x + first.x;
Corresponding to the IR:
first = ...
x, y, z, w = SPLIT first
second = TEX y, y
x', y', z', w' = SPLIT second
FADD res, x, x'
Here, x is not required by helper invocations (the coordinates to TEX) while y
is required. If we only look at only the first destinations, we incorrectly
decide that first is not required and fail to set the .skip bit, leading to
incorrect results.
Fixes:
5febeae58e0 ("pan/bi: Emit collect and split")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Alyssa Rosenzweig [Wed, 31 Aug 2022 20:54:02 +0000 (16:54 -0400)]
pan/bi: Fix out-of-bounds write in va_lower_split_64bit
...with dual source blending. Fixes shaders/dolphin/smg.1.shader_test
There are more IR sources than Valhall machine sources here.
Fixes:
b48933d6411 ("pan/va: Include BLEND for va_swap_12")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17794>
Yonggang Luo [Tue, 30 Aug 2022 18:05:35 +0000 (02:05 +0800)]
gallium/hud: Fixes compile error in hud/hud_context.c:
../../src/gallium/auxiliary/hud/hud_context.c:1017:1: error: static declaration of 'access' follows non-static declaration
These error will shown when any header #include <io.h>
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18333>
Thomas H.P. Andersen [Mon, 29 Aug 2022 23:14:07 +0000 (01:14 +0200)]
r600: fix warnings for missing-braces on clang
[2624/3754] Compiling C++ object src/gallium/drivers/r600/sfn/tests/libr600_test.a.p/sfn_test_shaders.cpp.o
../src/gallium/drivers/r600/sfn/tests/sfn_test_shaders.cpp:3066:27: warning: suggest braces around initialization of subobject [-Wmissing-braces]
r600_shader_key key = {0};
^
{}
1 warning generated.
[2711/3754] Compiling C++ object src/gallium/drivers/r600/libr600.a.p/sfn_sfn_nir_lower_64bit.cpp.o
../src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp:910:33: warning: suggest braces around initialization of subobject [-Wmissing-braces]
nir_const_value val[4] = {0};
^
{}
../src/gallium/drivers/r600/sfn/sfn_nir_lower_64bit.cpp:1328:38: warning: suggest braces around initialization of subobject [-Wmissing-braces]
nir_ssa_scalar channels[4] = { 0 };
^
{}
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18307>
Thomas H.P. Andersen [Mon, 29 Aug 2022 22:35:57 +0000 (00:35 +0200)]
d3d12: fix warnings for missing-braces on clang
[3485/3754] Compiling C++ object src/gallium/drivers/d3d12/libd3d12.a.p/d3d12_screen.cpp.o
../src/gallium/drivers/d3d12/d3d12_screen.cpp:56:65: warning: suggest braces around initialization of subobject [-Wmissing-braces]
static GUID OpenGLOn12CreatorID = { 0x6bb3cd34, 0x0d19, 0x45ab, 0x97, 0xed, 0xd7, 0x20, 0xba, 0x3d, 0xfc, 0x80 };
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
{ }
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18307>
Mike Blumenkrantz [Fri, 2 Sep 2022 13:42:45 +0000 (09:42 -0400)]
llvmpipe: don't assume pipe_context is always available in flush_frontbuffer
Fixes:
91dcadf9569 ("llvmpipe: finish rendering before flushing frontbuffer resources.")
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18392>
Alyssa Rosenzweig [Thu, 1 Sep 2022 19:11:36 +0000 (15:11 -0400)]
pan/bi: Fix dual texturing with uniforms
The GLSL code sequence:
texture2D(tex0, u_coords) + texture2D(tex1, u_coords)
will be optimized to
TEXC_DUAL tex0/tex1, u_coords, #texture_descriptor
If this optimization happens after lowering FAU, the resulting TEXC instruction
is unschedulable: both the uniform and the constant descriptor fight for the
same FAU slot.
However, if this optimization happens before lowering FAU, then the FAU lowering
will move the descriptor into a register, complicating the dual texturing fixup
in RA.
To fix this interaction, fuse dual texturing before lowering FAU and keep
texture descriptors as constants when lowering FAU of TEXC.
Fixes scheduling failure in piglit drawoverhead -test 3 with uniform reordering.
Fixes:
a4d3a296477 ("pan/bi: Enable dual texture fusing pass")
Fixes:
6b2eda6b729 ("pan/bi: Reorder pushed uniforms to avoid moves")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18378>
Erik Faye-Lund [Wed, 31 Aug 2022 13:13:28 +0000 (15:13 +0200)]
mesa/st: do not blit when using compressed fallback
If we're using the blit-path, we don't update the compressed image, which
will be needed if an application try to download the texture-image again
afterwards.
I previously fixed this for the memcpy-codepath, but we actually need to
fall back for the blit code-path as well. So let's move this early-out
earlier to catch both.
Fixes:
8f446322e1b ("mesa/st: do not use memcpy when using compressed fallback")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18342>
Lionel Landwerlin [Fri, 2 Sep 2022 06:27:31 +0000 (09:27 +0300)]
anv/hasvk: tweak loading failure messages
We don't want to print out too many :
MESA: error: ../src/intel/vulkan/anv_device.c:769: anv does not support Intel(R) HD Graphics (HSW GT1); use hasvk (VK_ERROR_INCOMPATIBLE_DRIVER)
whenever anv is not able to load on a HSW device. Similarly hasvk
should not print error on anything gfx9+.
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/18208>
Kenneth Graunke [Wed, 31 Aug 2022 07:41:51 +0000 (00:41 -0700)]
anv: Remove anv_batch_emit_reloc and just open-code it
We don't need the relocation offsets anymore, and just want to pin the
BO, and combine the address into a uint64_t. We can just open code
those two things; it's actually less code.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 07:03:20 +0000 (00:03 -0700)]
anv: Inline write_reloc into the only remaining caller
This is writing an address and clflushing, but it's not really about
execbuf relocations.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 06:50:13 +0000 (23:50 -0700)]
anv: Drop offset from anv_reloc_list_append
No longer used.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 01:47:17 +0000 (18:47 -0700)]
anv: Make a helper function for pinning a state pool's BOs
A bit less duplicated code, though with all the success checking, it
doesn't actually save us a whole lot.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 02:55:53 +0000 (19:55 -0700)]
anv: Delete has_bindless_images and has_bindless_samples flags
These are always true now.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 02:01:33 +0000 (19:01 -0700)]
anv: Delete has_a64_buffer_access flag
It's always true.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 00:17:53 +0000 (17:17 -0700)]
anv: Delete relocation support from batch submission
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 23:12:57 +0000 (16:12 -0700)]
anv: Delete wrapper BOs for relocations
These were only used in the non-softpin case.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 00:07:33 +0000 (17:07 -0700)]
anv: Delete anv_reloc_list_add()
We don't need the offset to write a relocation at any longer, so all
it does is call anv_reloc_list_add_bo() at this point. Just use that.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 23:09:05 +0000 (16:09 -0700)]
anv: Delete softpin checks
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 22:12:27 +0000 (15:12 -0700)]
anv: Delete use_relocations flag
There are no relocations.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 01:22:02 +0000 (18:22 -0700)]
anv: Drop state pool relocation munging
Now that the state pool's center is always 0, this is not needed.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 00:47:34 +0000 (17:47 -0700)]
anv: Delete "back" allocation from anv_block_pool
This was only used with relocations.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 22:33:31 +0000 (15:33 -0700)]
anv: Delete relocation support from anv_block_pool
We no longer use relocations.
v2: silence fedora builder warning
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Wed, 31 Aug 2022 01:09:42 +0000 (18:09 -0700)]
anv: Delete "back" allocation from state pool
This was only used with relocations, which no longer happen.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 23:02:38 +0000 (16:02 -0700)]
anv/tests: Don't use relocations in a test case
We won't support relocations shortly.
v2: Deal with softpin padding requirement (Lionel)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Fri, 2 Sep 2022 08:02:25 +0000 (11:02 +0300)]
anv/tests: remove back allocation tests
We'll remove driver code for this in the following commits.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 22:57:19 +0000 (15:57 -0700)]
anv: Delete shader constants UBO from descriptor sets
We now always softpin and use the load_global_constant case, so there's
no need to set up a UBO for NIR constants.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 22:09:36 +0000 (15:09 -0700)]
anv: Delete batch buffer growing code.
This was only needed on Haswell and older due to the kernel command
parser not allowing us to chain batches. anv no longer support this.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 21:59:00 +0000 (14:59 -0700)]
anv: Delete image param support.
This was only used prior to Skylake, which anv no longer supports.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 21:50:51 +0000 (14:50 -0700)]
anv: Drop checks for version 8 or 9
anv no longer supports versions below this.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Kenneth Graunke [Tue, 30 Aug 2022 21:54:02 +0000 (14:54 -0700)]
anv: Fail to create a device on ver < 9
These are now only supported by hasvk.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Wed, 3 Aug 2022 09:38:39 +0000 (12:38 +0300)]
anv: remove unused gfx7 code
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Wed, 3 Aug 2022 08:58:09 +0000 (11:58 +0300)]
anv: remove support for gfx7/8
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Fri, 2 Sep 2022 08:53:12 +0000 (11:53 +0300)]
anv: silence fedora build warning
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Mon, 22 Aug 2022 07:09:18 +0000 (10:09 +0300)]
hasvk: remove mesh code
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Mon, 22 Aug 2022 07:01:10 +0000 (10:01 +0300)]
hasvk: remove ray tracing code
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Mon, 22 Aug 2022 06:54:24 +0000 (09:54 +0300)]
hasvk: remove acceleration structure code
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Mon, 8 Aug 2022 19:11:19 +0000 (22:11 +0300)]
hasvk: stop advertising Vk 1.3 on non-softpin
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Wed, 3 Aug 2022 08:59:26 +0000 (11:59 +0300)]
hasvk: remove entrypoints for gfx9+
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Lionel Landwerlin [Wed, 3 Aug 2022 08:43:36 +0000 (11:43 +0300)]
intel: add a hasvk vulkan driver
This new driver is a copy of the current Anv code, it will only load
on gfx7/8 platforms though.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason.ekstrand@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18208>
Frank Binns [Fri, 26 Aug 2022 13:57:36 +0000 (14:57 +0100)]
pvr: remove image pointer from image view struct
A pointer is also stored in the base vk_image_view struct, so we can use this
one instead.
Signed-off-by: Frank Binns <frank.binns@imgtec.com>
Reviewed-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18373>
Jordan Justen [Fri, 2 Sep 2022 06:30:00 +0000 (23:30 -0700)]
intel/pci_ids: Add dg2 0x5698 pci-id
This motherboard-down pci-id was added in kernel commit
8618b8489ba6
("drm/i915: DG2 and ATS-M device ID updates").
Ref: bspec 44477
Ref: https://patchwork.freedesktop.org/patch/msgid/
20220701152231.529511-2-matthew.d.roper@intel.com
Fixes:
ad565f6b70d ("intel/dev: Enable first set of DG2 PCI IDs")
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18385>
Samuel Pitoiset [Tue, 30 Aug 2022 09:49:45 +0000 (11:49 +0200)]
radeonsi/ci: trigger radeonsi-raven-skqp for any RADV changes
Otherwise SKQP VK won't be run.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7120
Co-authored-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18316>
Jordan Justen [Mon, 29 Aug 2022 17:39:27 +0000 (10:39 -0700)]
intel/pci_ids: Update DG2 device names
Ref: bspec 44477
Ref: https://www.intel.com/content/www/us/en/products/details/discrete-gpus/arc/arc-a-series.html
Cc: mesa-stable
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18303>
Jordan Justen [Mon, 29 Aug 2022 17:39:27 +0000 (10:39 -0700)]
intel/pci_ids: Update ATS-M device names
Ref: bspec 44477
Ref: https://www.intel.com/content/www/us/en/products/details/discrete-gpus/data-center-gpu/flex-series/products.html
Cc: mesa-stable
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18303>
pal1000 [Thu, 21 Jul 2022 10:02:07 +0000 (13:02 +0300)]
meson: Only draw with llvm depends on native directly
Tests, softpipe or AMD drivers don't depend on it directly
Fixes:
3955dd07 ("meson/gallium: Add an option to not use LLVM for gallium draw module")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6817
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Tested-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17675>
Yonggang Luo [Mon, 22 Aug 2022 15:50:09 +0000 (23:50 +0800)]
ci/fedora: Re-enable vulkan-layers=device-select,overlay
Fixes:
ace13203f30 ("ci: remove broken device-select-layer from build")
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18044>
Yonggang Luo [Sun, 7 Aug 2022 16:40:31 +0000 (00:40 +0800)]
ci: Add debian-clang-release build pipeline
As the default build type is debug according to
.gitlab-ci/meson/build.sh, the specified line is:
-D buildtype=${BUILDTYPE:-debug}
So we use release for building optimized version for testing
if the compiler optimization are fine.
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18044>
Yonggang Luo [Sun, 7 Aug 2022 16:40:31 +0000 (00:40 +0800)]
ci: Enable all possible meson build options for clang
By doing this to makes more code be compiled with clang and guarded by bot
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18044>
Yonggang Luo [Sun, 7 Aug 2022 16:40:31 +0000 (00:40 +0800)]
ci: Enable GALLIUM_DUMP_CPU=true only in the clang job
This is used to show the cpu caps when running unit tests.
As clang have optimization bug about util_get_cpu_caps before, so
we monitor it by using GALLIUM_DUMP_CPU=true
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18044>
Yonggang Luo [Sun, 7 Aug 2022 08:37:24 +0000 (16:37 +0800)]
ci: EXTRA_OPTIONS -> EXTRA_OPTION for debian-clang
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18044>
Jose Fonseca [Tue, 30 Aug 2022 15:13:58 +0000 (23:13 +0800)]
util: Add macro util_get_cpu_caps in u_cpu_detect.c for avoid accidentally call it in u_cpu_detect.c
Signed-off-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 06:44:49 +0000 (14:44 +0800)]
docs: document LP_NATIVE_VECTOR_WIDTH GALLIUM_OVERRIDE_CPU_CAPS GALLIUM_NOSSE LP_FORCE_SSE2
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 06:49:00 +0000 (14:49 +0800)]
llvmpipe: Enable avx512 support in llvm
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Tue, 23 Aug 2022 17:20:43 +0000 (01:20 +0800)]
llvmpipe: Always using util_get_cpu_caps to get cpu caps for llvm on x86
As we can override cpu features with environment variables, so always using util_get_cpu_caps
instead llvm::sys::getHostCPUFeatures to get cpu features is a better option
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Tue, 30 Aug 2022 06:57:26 +0000 (14:57 +0800)]
util: add support for detecting avx512 vector bit size
Default to 256 until we're confident llvmpipe with 512 is
as correct and not slower than 256
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 09:11:35 +0000 (17:11 +0800)]
util: Move the detect of max_vector_bits into u_cpu_detect.c
has_avx2 implies has_avx, so have no need check then both
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 06:32:48 +0000 (14:32 +0800)]
util: Add ssse3 and avx cpu caps override option
llvmpipe can use these options to testing ssse3 and avx on cpus that support for avx512
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 17:10:29 +0000 (01:10 +0800)]
util: Refactoring check_cpu_caps_override out
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 27 Aug 2022 06:13:28 +0000 (14:13 +0800)]
meson: bump the minimal required llvm version to 5.0.0
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Yonggang Luo [Sat, 30 Jul 2022 04:17:27 +0000 (12:17 +0800)]
llvmpipe: Remove llvm version guard LLVM_VERSION_MAJOR <=> 4
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17813>
Igor Torrente [Fri, 26 Aug 2022 14:22:55 +0000 (11:22 -0300)]
venus: Fix dEQP-VK.pipeline.timestamp.calibrated.host_domain_test failure
The current implementation is getting its clock value from the host
and this value is not guaranteed to be the same as the VM guest.
This commit implements the CLOCK_MONOTONIC[_RAW] natively.
Signed-off-by: Igor Torrente <igor.torrente@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18281>
Igor Torrente [Mon, 29 Aug 2022 17:29:31 +0000 (14:29 -0300)]
radv: anv: Use the new vk_clock_gettime and vk_time_max_deviation
functions
Removes the duplicated code and start using the new common
code.
v2: split anv/radv parts to separate commits
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> (v1)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> (v1)
Signed-off-by: Igor Torrente <igor.torrente@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18281>
Igor Torrente [Mon, 29 Aug 2022 17:23:27 +0000 (14:23 -0300)]
vulkan: Add vk_clock_gettime and vk_time_max_deviation
These two new function are based on {anv,radv}_clock_gettime
and some other common code between radv and anv.
These new functions allow these drivers to share code and
more drivers can use it in the future.
v2: Drop the anv/radv changes in this commit (Yiwei Zhang)
v3: Add #ifndef _WIN32
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> (v1)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> (v1)
Signed-off-by: Igor Torrente <igor.torrente@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18281>
Chia-I Wu [Thu, 1 Sep 2022 20:48:10 +0000 (13:48 -0700)]
freedreno: get cpu timestamp again after FD_TIMESTAMP
Same as commit
91a0411d2ab ("turnip: improve perfetto sync_timestamp"),
but for freedreno.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18379>
Mike Blumenkrantz [Wed, 17 Aug 2022 22:09:22 +0000 (18:09 -0400)]
vulkan: initialize a variable
../src/vulkan/runtime/vk_fence.c: In function ‘vk_common_CreateFence’:
../src/vulkan/runtime/vk_fence.c:160:12: warning: ‘fence’ may be used uninitialized [-Wmaybe-uninitialized]
160 | *pFence = vk_fence_to_handle(fence);
| ~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/vulkan/runtime/vk_fence.c:154:21: note: ‘fence’ was declared here
154 | struct vk_fence *fence;
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18119>
Mike Blumenkrantz [Wed, 31 Aug 2022 19:37:31 +0000 (15:37 -0400)]
zink: don't emit illegal interpolation
this is not valid for vertex inputs or fragment outputs
cc: mesa-stable
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18365>
Mike Blumenkrantz [Wed, 31 Aug 2022 23:32:42 +0000 (19:32 -0400)]
zink: don't set MUTABLE by default
this is only used by texture views, so use the same mechanism as STORAGE
to add the flag only as-needed
the mutable flag must be set before create_ivci is called for the imageview,
so zink_resource_object_init_mutable() is moved out to the callers of
zink_get_surface instead of being conveniently located in that function
fixes #7174
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18358>
Mike Blumenkrantz [Thu, 1 Sep 2022 02:11:21 +0000 (22:11 -0400)]
zink: avoid needing mutable for z/s sampling
the aspect flags should be sufficient to determine which component the
imageview is created for, which means this can avoid changing the format
here to retain the immutable format
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18358>
Mike Blumenkrantz [Thu, 1 Sep 2022 00:16:35 +0000 (20:16 -0400)]
zink: only add srgb mutable for images with modifiers
these are supposed to be for dmabuf handling, so checking for mutable
swapchain is both pointless and wrong
Fixes:
28ee911ad67 ("zink: handle mutable swapchain images with dmabuf")
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18358>
Mike Blumenkrantz [Thu, 1 Sep 2022 00:37:21 +0000 (20:37 -0400)]
zink: store VkImageViewUsageCreateInfo for surface creation
this otherwise breaks surface rebinds if used since the pointer
will be a garbage stack value
cc: mesa-stable
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18358>
Mike Blumenkrantz [Tue, 14 Jun 2022 16:12:31 +0000 (12:12 -0400)]
zink: explicitly use unsigned types for bit shifts
fixes some ubsan errors
cc: mesa-stable
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18358>
Emma Anholt [Thu, 25 Aug 2022 21:34:20 +0000 (14:34 -0700)]
turnip: Enable lowering of mediump temps/CS shared to 16-bit.
In Aztec Ruins, we end up storing some big shared-mem arrays as 16-bit,
cutting shared mem size in half across many shaders while also reducing
conversions. gfxbench vk-5-normal perf +0.364983% +/- 0.189764% (n=4).
fossil-db:
Totals from 448 (2.99% of 14988) affected shaders:
MaxWaves: 6154 -> 6390 (+3.83%); split: +3.96%, -0.13%
Instrs: 174554 -> 165045 (-5.45%); split: -6.45%, +1.01%
CodeSize: 364224 -> 345558 (-5.12%); split: -6.03%, +0.90%
NOPs: 48224 -> 48024 (-0.41%); split: -3.33%, +2.91%
MOVs: 6985 -> 6104 (-12.61%); split: -19.11%, +6.50%
Full: 4577 -> 4101 (-10.40%); split: -11.08%, +0.68%
(ss): 3428 -> 3335 (-2.71%); split: -4.17%, +1.46%
(sy): 1250 -> 1205 (-3.60%); split: -4.72%, +1.12%
(ss)-stall: 14695 -> 14528 (-1.14%); split: -2.25%, +1.12%
(sy)-stall: 19565 -> 17998 (-8.01%); split: -9.55%, +1.54%
STPs: 1086 -> 870 (-19.89%)
LDPs: 162 -> 108 (-33.33%)
Cat0: 51400 -> 51120 (-0.54%); split: -3.31%, +2.76%
Cat1: 16861 -> 14688 (-12.89%); split: -18.18%, +5.30%
Cat2: 71161 -> 68454 (-3.80%); split: -4.52%, +0.72%
Cat3: 29572 -> 25306 (-14.43%); split: -14.49%, +0.06%
Cat4: 3128 -> 3131 (+0.10%)
Cat5: 1502 -> 1506 (+0.27%)
Cat6: 840 -> 750 (-10.71%)
aztec ruins is a big winner with the ldp/stp reductions. summoners_war
racks up an astounding 41% reduction in instructions and +15% max_waves.
Most affected apps show a minor win in instrs, with
fallout_shelter_online, and aztec ruins on ANGLE taking minor hits.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18259>
Emma Anholt [Thu, 25 Aug 2022 21:33:38 +0000 (14:33 -0700)]
spirv: Mark phis as mediump instead of directly lowering them to 16 bit.
This reverts commit
6f25d45877a1e1a7ac6250a7d051d33485e0cba7, replacing it
with GLSL_PRECISION_MEDIUM. The previous commit ended up not being the
right approach, as it affected only nir vars for spirv phis and not other
nir vars, and we want a tool that does both. The new
nir_lower_mediump_vars pass can do that for you.
No fossil-db change for my angle fossils run on radv.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18259>
Emma Anholt [Tue, 23 Aug 2022 21:54:37 +0000 (14:54 -0700)]
nir: Add a pass to lower mediump temps and shared mem.
SPIRV and GLSL are reasonable at converting ALU ops to mediump, but
variable storage would be wrapped in a 2f32/2mp on store/load, and if
nir_vars_to_ssa doesn't make that storage go away then you'd have extra
conversions. For compute shader shared mem, you'd waste memory too.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18259>
Alyssa Rosenzweig [Sun, 21 Aug 2022 01:47:57 +0000 (21:47 -0400)]
ail: Add unit tests for miptree layouts
The equations for calculating miptree offsets are complicated,
nonobvious, and full of subtle footguns. Worse, the driver doesn't
control the offsets -- it must simply agree with the offsets
implicitly calculated in the hardware. The CTS doesn't adequately
exercise all the corner cases. Make sure we have unit tests that do.
The tests themselves are generated by instrumenting agxdecode to scan
GPU memory after uploading test patterns in a variety of layout with a
Metal application.
Thank you to Asahi Lina and Dougall Johnson for the reverse-engineering
that led to this. The tests selected here are a subset of those used for
the reverse-engineering. The full set may be found in Lina's tilecalc
repo:
https://github.com/asahilina/tilecalc
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Sat, 20 Aug 2022 17:13:36 +0000 (13:13 -0400)]
asahi: Rename our fake twiddled DRM modifier
To account for non-64x64 tile sizes.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Sat, 20 Aug 2022 16:35:53 +0000 (12:35 -0400)]
asahi: Advertise ASTC formats
Now that we have layout and tiling code that can handle block-compressed
formats, including the non-square blocks found with some ASTC formats,
we can advertise ASTC formats. Passes dEQP-GLES3.*astc* which
exercises everything here. (These tests passed before by decompressing
the textures to RGBA8 UNORM in the frontend, but it's much more
efficient to use real ASTC textures as done here.)
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Thu, 18 Aug 2022 23:06:18 +0000 (19:06 -0400)]
asahi: Advertise ETC1
ETC2 is backwards compatible with ETC1, we just need to declare the
format mapping.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Fri, 19 Aug 2022 02:35:58 +0000 (22:35 -0400)]
asahi: Fix is_format_supported returns
Gallium can test multiple binds at once, we need to return the logical
AND, not the OR.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Fri, 19 Aug 2022 02:22:21 +0000 (22:22 -0400)]
asahi: Handle 2-channel sRGB textures
I'm not sure why we need to set this magic bit, but this fixes the
non-depth_component portion of
dEQP-GLES3.functional.texture.format.sized.*, e.g
dEQP-GLES3.functional.texture.format.sized.cube.srgb_rg8_pot
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Fri, 19 Aug 2022 00:13:50 +0000 (20:13 -0400)]
ail: Rewrite tiled memcpy for correctness
Move tiling.c into ail, using ail data structures and helpers to manage
the tiling. This fixes a staggering number of issues with the tiling
routines:
* NPOT block sizes defeatured. The hardware only supports POT block
sizes. There's no need to handle anything else.
* Use ail to determine tile sizes, instead of the broken
agx_select_tile_shift routine that didn't work for non-square tile
sizes (for instance).
* Handle up to 128x128 tiles, as required by 8bpp textures.
* Handle non-square tiles. If the block size is not a multiple of 4, the
tile size will be of the form 2n x n. This is easy with the ail_tile
data structure, but not possible architecturally with
agx_select_tile_shift. This is required for 16bpp and 64bpp textures.
* Express in terms of elements instead of pixels, using unit
suffixes to make the dimensional analysis obvious. In particular this
handles tiling of block-compressed textures by tiling the blocks
themselves. This is required for block-compressed textures (internally handled
like smaller 64bpp textures).
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Sun, 12 Jun 2022 18:14:58 +0000 (14:14 -0400)]
ail: Introduce image layout module
Introduce ail, a small library for working with the image (and buffer)
layouts encountered with AGX hardware. Its design is inspired by isl. In
particular, ail strives to use isl unit suffixes and to represent
quantities in a canonical, API-agnostic fashion [1].
ail replaces the old miptree code (based on some ad hoc heuristics that
passed a few dEQP tests). It is based on a thorough reverse-engineering
of AGX's twiddled format, courtesy of Asahi Lina, Dougall Johnson, and
me. This corrects our handling of many common cases that were totally
wrong in the old code, leading to GPU faults.
Unlike the code, ail differentiates between pixels and elements
consistently, allowing block-compressed formats like ETC2 to be
supported correctly. These formats will be enabled later in the series.
This commit fixes Inochi2D, glmark2 -brefract and -bterrain, and who
knows what else.
ail stands for { Asahi, AGX } Image { Layout, Library } at your
convenience. ail is best served warm.
Liberal use of ail is recommended. Yum!
[1] https://docs.mesa3d.org/isl/units.html
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Sat, 20 Aug 2022 16:26:58 +0000 (12:26 -0400)]
asahi: Fix ASTC enums
The ASTC enum only encodes the block width/height. By contrast the
LDR/HDR/sRGB distinction is encoded as UNORM/Float and via the sRGB bit.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Wed, 10 Aug 2022 00:44:17 +0000 (20:44 -0400)]
asahi: Fix bind sizes
Otherwise we get a nonsensical count of 129 in one Metal sample. Exact
size still not known but this bounds somewhat.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>
Alyssa Rosenzweig [Sun, 21 Aug 2022 01:15:07 +0000 (21:15 -0400)]
asahi: Add XML for multisampled textures
Metal supports MSAA 2x and MSAA 4x, neither at the same time as
mipmapping. That matches the GL spec requirement.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18167>