This is trying to clear a bit in the control register. However, it's
executing with whatever channel mask happens to be active. Typically
this is the one at the start of the program, so at least some channels
will be active. Typically the first channel will be active due to
packed dispatch, but that's not always guaranteed. Without NoMask,
the float controls writes may randomly not happen.
Recent GPUs also seem to have a hang issue when the first instruction in
the shader doesn't have any active channels. Having an instruction with
NoMask at the start of the program works around the issue. See HSD bug
14017989577. In our case, the float controls preamble was breaking that
restriction every time, causing us to run into this problem frequently.
Thanks to Tapani Pälli for finding this hang issue, and Francisco
Jerez and Lionel Landwerlin for helping pinpoint this issue during
review of a workaround patch in !20194.
Fixes GPU hangs in Elder Scrolls Online, Witcher 3, and likely more.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7639
Fixes: 9da56ffc52 ("i965/fs: add emit_shader_float_controls_execution_mode() and aux functions")
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20214>
In a tesselation control shader where an input array is accessed using
the index gl_InvocationID, we can end up accessing elements beyond the
number of input vertices specified in the shader key.
This happens because of the lowering in nir_lower_indirect_derefs().
This lowering will affect compact variables which happens in this
case :
in gl_PerVertex {
vec4 gl_Position;
float gl_ClipDistance[1];
} gl_in[gl_MaxPatchVertices];
The lowered code produced by NIR is somewhat ineffecient (implements a
binary seach) :
if (gl_InvocationID < 16) {
if (gl_InvocationID < 8) {
if (gl_InvocationID < 4) {
vec4 vals = load_at_offset(0);
value = bcsel(vals, gl_InvocationID);
} else {
vec4 vals = load_at_offset(4);
value = bcsel(vals, gl_InvocationID - 4);
}
} else {
if (gl_InvocationID < 12) {
vec4 vals = load_at_offset(8);
value = bcsel(vals, gl_InvocationID - 8);
} else {
vec4 vals = load_at_offset(12);
value = bcsel(vals, gl_InvocationID - 12);
}
}
} else {
if (gl_InvocationID < 24) {
...
} else {
...
}
}
By default the gl_MaxPatchVertices must be set at 32 items and that's
what the lowering code will use to divide the access into chunks of 4.
But when running with 3 input vertices, this means we'll pull one more
item than what was delivered in the shader payload.
This triggers issues further down the register scheduling where the
g5UD (register for the 4th item) is overwritten by a previous SEND,
leading the URB read to use an invalid handle.
This pass clamps any access load_per_vertex_input intrinsic vertex
indice to (input_vertices - 1).
Fixes issues with tests like :
dEQP-VK.clipping.user_defined.clip_distance.vert_tess.*
Also fixes a hang with zink/anv on :
KHR-GL46.draw_elements_base_vertex_tests.AEP_shader_stages
v2: Don't replace source register
v3: Implement in NIR
v4: Clamp per vertex array sizes in NIR (Jason)
v5: Move the clamping on the intel compiler
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9749>
For Tiger Lake and onward, we generally don't need to ambiguate the CCS
before accessing it. This is safe for two reasons:
- Tiger Lake and onward treat all CCS values as legal.
- We enable compression on all writable image layouts. The CCS will
receive all writes and will therefore always be valid.
When dealing with modifiers, we continue to allow ambiguates in some
instances.
Before this patch, I found ~19.5k ambiguates in Wolfenstein:
Youngblood's Riverside benchmark (note that this includes manually
entering the benchmark and exiting the app). With this patch, the number
of ambiguates goes down to zero.
Improves performance of Fallout 4 at 1080p/High settings on Arc A380 by
around 22%.
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20118>
Whether or not CCS can be used without initialization depends on the
platform:
- On gfx7-8, each CCS element is 1-bit and encodes "fast-cleared" or
"pass-through". So, those platforms have no illegal values.
- On gfx9-11, each CCS element is 2-bits and some bit combinations
are invalid.
- On gfx12+, each CCS element is 4-bits but they have no truly illegal
values. Unused encodings are interpreted as "pass-through".
Refer to the "MCS/CCS Buffers for Render Target(s)" sections of the
PRMs for more info.
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20118>
When INTEL_DEBUG=pc is set and a CCS operation is being performed, the
driver reports that flushes are happing before and after the operation.
It also reports that the operation is a fast clear, but that's not
always the case. We could be resolving for example.
Reporting the specific operation can help avoid confusion.
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20118>
Oh, for the days of Broadwell and earlier where compression was called
fast-clear. That was a simpler time. The birds sang in the trees, the
oceans weren't brown from oil spills, and Intel surface compression was
actually comprehendable by humans. To help the reviewer, keep the
following in mind:
1. CCS_E is SKL+
2. Implicit CCS is TGL+
3. The AUX TT (AKA aux map) is TGL+
4. HIZ+CCS, stencil CCS, and CCS for storage images are all TGL+
4. CCS_D surfaces only ever get full resolves and MCS surfaces only
ever get partial resolves
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19852>
This option didn't exist half a decade ago when I first implemented base
workgroup support in ANV. It's cleaner to just have split system values
like all the other zero_base+base things do.
We currently only do this for COMPUTE and not KERNEL because it lets us
avoid changing intel_clc for now. We can add KERNEL later if needed.
We also don't do this lowering for task/mesh.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20068>
Oleksii Bozhenko