A lot of CTS tests write a u8vec4 or an i8vec4 to an SSBO. This results
in a lot of shifts and MOVs. When that pattern can be recognized, the
individual 8-bit components can be packed much more efficiently.
v2: Rebase on b4369de27f ("nir/lower_packing: use
shader_instructions_pass")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9025>
Emitting the instructions one by one results in two MOV instructions
that won't be propagated. By handling both instructions at once, a
single MOV is emitted. For example, on Ice Lake this helps
dEQP-VK.spirv_assembly.type.vec3.i8.bitwise_xor_frag:
SIMD8 shader: 49 instructions. 1 loops. 4044 cycles. 0:0 spills:fills, 5 sends
SIMD8 shader: 41 instructions. 1 loops. 3804 cycles. 0:0 spills:fills, 5 sends
Without "intel/fs: Allow copy propagation between MOVs of mixed sizes,"
the improvement is still 8 instructions, but there are more instructions
to begin with:
SIMD8 shader: 52 instructions. 1 loops. 4164 cycles. 0:0 spills:fills, 5 sends
SIMD8 shader: 44 instructions. 1 loops. 3944 cycles. 0:0 spills:fills, 5 sends
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9025>
In the vec4 compiler, 8-bit types should never exist.
In the scalar compiler, 8-bit types should only ever be able to exist on
Gfx ver 8 and 9.
Some instructions are handled in non-obvious ways.
Hopefully this will save the next person some time.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9025>
The gfx6_gs_visitor overrides emit_urb_write_opcode but with a different
function signature. This causes warnings with -Woverloaded-virtual.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12308>
On Gfx4 and Gfx5, sel.l (for min) and sel.ge (for max) are implemented
using a separte cmpn and sel instruction. This lowering occurs in
fs_vistor::lower_minmax which is called very, very late... a long, long
time after the first calls to opt_cmod_propagation. As a result,
conditional modifiers can be incorrectly propagated across sel.cond on
those platforms.
No tests were affected by this change, and I find that quite shocking.
After just changing flags_written(), all of the atan tests started
failing on ILK. That required the change in cmod_propagatin (and the
addition of the prop_across_into_sel_gfx5 unit test).
Shader-db results for ILK and GM45 are below. I looked at a couple
before and after shaders... and every case that I looked at had
experienced incorrect cmod propagation. This affected a LOT of apps!
Euro Truck Simulator 2, The Talos Principle, Serious Sam 3, Sanctum 2,
Gang Beasts, and on and on... :(
I discovered this bug while working on a couple new optimization
passes. One of the passes attempts to remove condition modifiers that
are never used. The pass made no progress except on ILK and GM45.
After investigating a couple of the affected shaders, I noticed that
the code in those shaders looked wrong... investigation led to this
cause.
v2: Trivial changes in the unit tests.
v3: Fix type in comment in unit tests. Noticed by Jason and Priit.
v4: Tweak handling of BRW_OPCODE_SEL special case. Suggested by Jason.
Fixes: df1aec763e ("i965/fs: Define methods to calculate the flag subset read or written by an fs_inst.")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Tested-by: Dave Airlie <airlied@redhat.com>
Iron Lake
total instructions in shared programs: 8180493 -> 8181781 (0.02%)
instructions in affected programs: 541796 -> 543084 (0.24%)
helped: 28
HURT: 1158
helped stats (abs) min: 1 max: 1 x̄: 1.00 x̃: 1
helped stats (rel) min: 0.35% max: 0.86% x̄: 0.53% x̃: 0.50%
HURT stats (abs) min: 1 max: 3 x̄: 1.14 x̃: 1
HURT stats (rel) min: 0.12% max: 4.00% x̄: 0.37% x̃: 0.23%
95% mean confidence interval for instructions value: 1.06 1.11
95% mean confidence interval for instructions %-change: 0.31% 0.38%
Instructions are HURT.
total cycles in shared programs: 239420470 -> 239421690 (<.01%)
cycles in affected programs: 2925992 -> 2927212 (0.04%)
helped: 49
HURT: 157
helped stats (abs) min: 2 max: 284 x̄: 62.69 x̃: 70
helped stats (rel) min: 0.04% max: 6.20% x̄: 1.68% x̃: 1.96%
HURT stats (abs) min: 2 max: 48 x̄: 27.34 x̃: 24
HURT stats (rel) min: 0.02% max: 2.91% x̄: 0.31% x̃: 0.20%
95% mean confidence interval for cycles value: -0.80 12.64
95% mean confidence interval for cycles %-change: -0.31% <.01%
Inconclusive result (value mean confidence interval includes 0).
GM45
total instructions in shared programs: 4985517 -> 4986207 (0.01%)
instructions in affected programs: 306935 -> 307625 (0.22%)
helped: 14
HURT: 625
helped stats (abs) min: 1 max: 1 x̄: 1.00 x̃: 1
helped stats (rel) min: 0.35% max: 0.82% x̄: 0.52% x̃: 0.49%
HURT stats (abs) min: 1 max: 3 x̄: 1.13 x̃: 1
HURT stats (rel) min: 0.12% max: 3.90% x̄: 0.34% x̃: 0.22%
95% mean confidence interval for instructions value: 1.04 1.12
95% mean confidence interval for instructions %-change: 0.29% 0.36%
Instructions are HURT.
total cycles in shared programs: 153827268 -> 153828052 (<.01%)
cycles in affected programs: 1669290 -> 1670074 (0.05%)
helped: 24
HURT: 84
helped stats (abs) min: 2 max: 232 x̄: 64.33 x̃: 67
helped stats (rel) min: 0.04% max: 4.62% x̄: 1.60% x̃: 1.94%
HURT stats (abs) min: 2 max: 48 x̄: 27.71 x̃: 24
HURT stats (rel) min: 0.02% max: 2.66% x̄: 0.34% x̃: 0.14%
95% mean confidence interval for cycles value: -1.94 16.46
95% mean confidence interval for cycles %-change: -0.29% 0.11%
Inconclusive result (value mean confidence interval includes 0).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12191>
Changes:
- removal of attr_wa_state (it's passed directly)
- nir_metadata_preserve(..., nir_metadata_all) is called when pass doesn't
make progress
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12324>
Setting it to GPU can cause an L3$ flush in certain cases. That's not
what we want as we really only care about coherency within the GPU.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Sagar Ghuge <sagar@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12291>
Also switch to array of strings to show high/low dwords.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: daba2894ff ("intel/disasm: decode/describe more send messages")
Acked-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12183>
This is where it should be rather than having to pass it into the
optimisation pass every time.
It also allows us to call the loop analysis pass without having to
duplicate these options which we will do later in this series.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12064>
This fixes a hang in the following piglit test when GCM moves a
UBO load outside of the loop.
tests/shaders/ssa/fs-if-def-else-break.shader_test
The end NIR ends up looking like this:
vec2 32 ssa_3 = intrinsic load_ubo (ssa_2, ssa_0) (0, 1073741824, 0, 0, 8)
vec1 32 ssa_4 = mov ssa_3.x
vec1 32 ssa_5 = inot ssa_3.y
/* succs: block_1 */
loop {
...
if ssa_5 { }
}
Fixes: 1edf67fc3f ("intel/fs: Generate if instructions with inverted conditions")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12064>
There are two problems with the current architecture.
In OpenGL, the id is supposed to be a unique identifier for a particular
log source. This is done so that applications can (theoretically)
filter particular log messages. The debug callback infrastructure in
Mesa assigns a uniqe value when a value of 0 is passed in. This causes
the id to get set once to a unique value for each message.
By passing a stack variable that is initialized to 0 on every call,
every time the same message is logged, it will have a different id.
This isn't great, but it's also not catastrophic.
When threaded shader compiles are used, the id *pointer* is saved and
dereferenced at a possibly much later time on a possibly different
thread. This causes one thread to access the stack from a different
thread... and that stack frame might not be valid any more. :(
I have not observed any crashes related to this particular issue.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12136>
There are two problems with the current architecture.
In OpenGL, the id is supposed to be a unique identifier for a particular
log source. This is done so that applications can (theoretically)
filter particular log messages. The debug callback infrastructure in
Mesa assigns a uniqe value when a value of 0 is passed in. This causes
the id to get set once to a unique value for each message.
By passing a stack variable that is initialized to 0 on every call,
every time the same message is logged, it will have a different id.
This isn't great, but it's also not catastrophic.
When threaded shader compiles are used, the id *pointer* is saved and
dereferenced at a possibly much later time on a possibly different
thread. This causes one thread to access the stack from a different
thread... and that stack frame might not be valid any more. :(
This fixes shader-db crashes of various kinds on Iris with threaded
shader compiles enabled.
Fixes: 42c34e1ac8 ("iris: Enable threaded shader compilation")
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12136>
Only fragment and some compute shaders support implicit derivatives.
They're totally meaningless without helper invocations and some
understanding of the dispatch pattern. We've got code to lower
nir_texop_tex in these shader stages to use an explicit derivative of 0
but it was pretty badly broken:
1. It only handled nir_texop_tex, not nir_texop_txb or nir_texop_lod.
2. It didn't take min_lod into account
3. It was conflated with adding a missing LOD parameter to opcodes
which expect one such as nir_texop_txf. While not really a bug,
this does make it way harder to reason about the code.
4. Unless you set a flag (which most drivers don't), it left the
opcode nir_texop_tex instead of nir_texop_txl which it should have
been.
This reworks it to go through roughly the same path as other LOD
lowering only with a constant lod of 0 instead of calling out to
nir_texop_lod. We also get rid of the lower_tex_without_implicit_lod
flag because most drivers set it and those that don't are probably
subtly broken. If someone really wants to get nir_texop_tex in their
vertex shaders, they can write a new patch to add the flag back in.
Fixes: e382890e25 "nir: set default lod to texture opcodes that..."
Fixes: d5ac5d6e83 "nir: Add option to lower tex to txl when..."
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11775>
It's intel-specific, used to get at MSAA compression information.
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11775>
We need to lower the add3 instruction simd width otherwise in simd32
mode, we endup writing 4 register wide data which is not allowed.
Reported-by: Jordan Justen <jordan.l.justen@intel.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11985>
CoverityID: 1487496
Fixes: cde9ca616d "intel/compiler: Make decision based on source type instead of opcode"
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11985>
This patch restructure code a little bit to check if source can be
represented as immediate operand. This is a foundation for next patch
which add checks for integer operand as well.
Suggested-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11596>
This is distinct form max_cs_threads because it also encodes
restrictions about the way we use GPGPU/COMPUTE_WALKER. This gets rid
of the MIN2(64, devinfo->max_cs_threads) we have scattered all over the
driver and puts it in a central place.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11861>
No difference proven at 95.0% confidence (n=10) in
dEQP-VK.ssbo.phys.layout.random.16bit.scalar.13.
v2: Only update each block's IP data once instead of once per block.
Suggested by Emma.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11632>
Performance improvement in
dEQP-VK.ssbo.phys.layout.random.16bit.scalar.13 for n=30:
release build (w/Fedora build flags): -0.82% ± 0.23%
Meson -Dbuildtype=debugoptimized: -0.74% ± 0.27%
The difference in the debugoptimized build is the calls to
inst_is_in_block(block, this) still exist on each call to remove().
v2: Only update each block's IP data once instead of once per block.
Suggested by Emma.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11632>
Performance improvement in
dEQP-VK.ssbo.phys.layout.random.16bit.scalar.13 for n=30:
release build (w/Fedora build flags): -7.79% ± 0.25%
Meson -Dbuildtype=debugoptimized: -5.10% ± 0.40%
The difference in the debugoptimized build is the calls to
inst_is_in_block(block, this) still exist on each call to remove().
v2: Only update each block's IP data once instead of once per block.
Suggested by Emma.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11632>
We say that they're for debug only but we don't really have a good
policy around when to set them and when not to. In particular,
nir_lower_system_values and nir_lower_vars_to_ssa which are the chief
producers of SSA values which might reasonably have a name do not bother
to set one. We have some names set from things like BLORP and RADV's
meta shaders but AFAICT, they're setting a name more because it's there
than because they actually care.
Also, most things other than nir_clone and nir_serialize don't bother to
try and preserve them. You can see in the diffstat of this commit
exactly what passes attempt to preserve names. Notably missing from the
list is opt_algebraic which is the single largest source of SSA def
churn and it happily throws names away.
These observations lead me to question whether or not names are actually
useful at all or if they're just taking up space (8B per instruction)
and wasting CPU cycles (to ralloc_strdup on the off chance we do have
one). I don't think I can think of a single time in recent history
where I've been debugging a shader issue and a SSA value name has been
there and been useful. If anything, the few times they are there, they
just throw me off because they mess up the indentation in nir_print.
iris shader-db on my system gets runtime -2.07734% +/- 1.26933% (n=5)
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5439>
Qualcomm has a mode with a subgroup size of 128, so just emitting larger
integer operations and then lowering them later isn't an option. This
makes the pass able to handle the lowering itself, so that we don't have
to go down to 64-thread wavefronts when ballots are used.
(The GLSL and legacy SPIR-V extensions only support a maximum of 64
threads, but I guess we'll cross that bridge when we come to it...)
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6752>
We set the ex_desc to 0, since the address surface type is FLAT.
v2 (Sagar Ghuge):
- Fix message descriptor encoding
v2 (Jason Ekstrand):
- Drop support for block messages
Co-authored-by: Sagar Ghuge <sagar.ghuge@intel.com>
Co-authored-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11600>
