This hardware bug is the result of a control flow optimization present
in Gfx8-9 meant to prevent the ELSE instruction from disabling all
channels and update the control flow stack only to have them
re-enabled at the ENDIF instruction executed immediately after it.
Instead, on Gfx8-9 an ELSE instruction that would normally have ended
up with all channels disabled would pop off the last element of the
stack and jump directly to JIP+1 instead of to the ENDIF at JIP,
skipping over the ENDIF instruction. In simple cases this would work
okay (though it's actual performance benefit is questionable), but in
cases where a branch instruction within the IF block (e.g. BREAK or
CONTINUE) caused all active channels to jump outside the IF
conditional, the optimization would break the JIP chain of "join"
instructions by skipping the ENDIF, causing the block of instructions
immediately after the ENDIF to execute with all channels disabled
until execution reaches the reconvergence point.
This issue was observed on SKL in the
dEQP-VK.reconvergence.subgroup_uniform_control_flow_elect.compute.nesting4.0.38
test in combination with some Vulkan binding model changes Lionel is
working on. In such cases the execution with all channels disabled
was leading to corruption of an indirect message descriptor, causing a
hang.
Unfortunately the hardware bug doesn't provide a recommended
workaround. In order to fix the problem we point the JIP of an ELSE
instruction to the instruction immediately before the ENDIF -- However
that's not expected to work due to the restriction that JIP and UIP
must be equal if and only if BranchCtrl is disabled -- So this patch
also enables BranchCtrl, which is intended to support join
instructions within the "ELSE" block, which in turn disables the
optimization described above, which in turn causes us to execute the
instruction immediately *before* the ENDIF with all channels disabled
-- So in order to avoid further fallout from executing code with all
channels disabled we need to insert a NOP before ENDIF instructions
that have a matching ELSE instruction.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20921>
Setting default expected values as default in the xml.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21105>
On integrated products this makes almost no difference but on discrete
it's pretty important.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Tested-by: Chuansheng Liu <chuansheng.liu@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21131>
The decoder context needs to know what engine it's associated with.
Nowadays, we have render, compute, blitter, even video engines being
used from the same driver. Rather than trying to have a single decoder
and thwacking the engine field back and forth between calls, we make
one per queue family, and stash a pointer in anv_queue for easy access.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21149>
This will be useful for RADV since it hashes the state.
v3dv changes:
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20731>
This new helper, do_emit_fb_writes() does the actual walk over all the
render targets to emit each of the different FB writes. We want this in
a helper because we're about to go a bit crazy with coarse.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
This is required by code-gen since it generates a 1-wide OR and it'll
blow up if the register width > 1. It's also way better than the "your
register is the wrong size" assert you get from the more generic
validation check.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
This allows us to communicate to the back-end that we don't actually
know if the framebuffer is multisampled or not. No drivers set anything
but ALWAYS/NEVER and we still have a few ALWAYS/NEVER assumptions but
those should be asserted.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
This allows for the possibility that we may not know at compile time if
sample shading is enabled through the API. While we're here, also
document exactly what this bit means so we don't confuse ourselves.
v2: Fixup coarse pixel values (Lionel)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
Whenever one of them is BRW_SOMETIMES, we depend on dynamic flag pushed
in as a push constant. In this case, we have to often have to do the
calculation both ways and SEL the result. It's a bit more code but
decouples MSAA from the shader key.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
This is more similar to what we do for single-sample and it should be
more clear going forward once our lowering gets more complex.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21094>
Builds on the work of !15121. This gets to delete even more code
because many drivers shared a lot of code for i2b and f2b.
No shader-db or fossil-db changes on any Intel platform.
v2: Rebase on 1a35acd8d9.
v3: Update a comment in nir_opcodes_c.py. Suggested by Konstantin.
v4: Another rebase. Remove f2b stuff from Midgard.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20509>
Submitting a batch with the first command buffer with the simultaneous
bit set followed by a command buffer without the bit set gets past the
check and triggers this assert attempting to chain them:
../src/intel/vulkan/anv_batch_chain.c:1147: anv_cmd_buffer_chain_command_buffers: Assertion `num_cmd_buffers == 1' failed.
Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21056>
This file hasn't really been updated since 2016, apart from a single
search-replace two years ago.
That's an eternity in ANV-land, so let's just remove these.
While we're at it, also remove the duplicate in hasvk.
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21044>
We lower NIR's load_constant to load_global_constant, which uses A64
bindless messages. As such, we do the following math to produce the
address for each load:
base_lo@32 <- BRW_SHADER_RELOC_CONST_DATA_ADDR_LOW
base_hi@32 <- BRW_SHADER_RELOC_CONST_DATA_ADDR_HIGH
base@64 <- pack_64_2x32_split(base_lo, base_hi)
addr@64 <- iadd(base@64, u2u64(offset@32))
On platforms that emulate 64-bit math, we have to emit additional code
for the 64-bit iadd to handle the possibility of a carry happening and
affecting the top bits.
However, NIR constant data is always uploaded adjacent to the shader
assembly, in the same buffer. These buffers are required to live in a
4GB region of memory starting at Instruction State Base Address. We
always place the base address at a 4GB address. So the constant data
always lives in a buffer entirely contained within a 4GB region, which
means any offsets from the start of the buffer cannot possibly affect
the high bits.
So instead, we can simply do a 32-bit addition between the low bits of
the base and the offset, then pack that with the unchanged high bits.
On anv, INSTRUCTION_STATE_POOL_MIN_ADDRESS is 8GB, so the high bits are
always 0x2. We don't even need to patch that portion of the address and
can just use an immediate value. We do still need to pack, however.
fossil-db on Icelake indicates the following for affected shaders:
Instrs: 10830023 -> 10750080 (-0.74%)
Cycles: 1048521282 -> 1046770379 (-0.17%); split: -0.33%, +0.16%
Subgroup size: 103104 -> 103112 (+0.01%)
Send messages: 570886 -> 570760 (-0.02%)
Loop count: 14428 -> 14429 (+0.01%)
Spill count: 14246 -> 14244 (-0.01%); split: -0.06%, +0.04%
Fill count: 22802 -> 22794 (-0.04%); split: -0.04%, +0.01%
Scratch Memory Size: 654336 -> 662528 (+1.25%)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20999>
Be conservative in Gfx11+ and always stall in a fence. Since there are
two different fences, and shader might want to synchronize between them.
This change also brings back the original code block for the stall
between the fence and comment from the commit
b390ff3517.
v2: (Caio)
- Re-arrange code block.
- Adjust comment.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6958
Fixes: f7262462 ("intel/fs: Rework fence handling in brw_fs_nir.cpp")
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Tested-by: Mark Janes <markjanes@swizzler.org>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20996>
u_vector_add() don't keep the returned pointers valid.
After the initial size allocated in u_vector_init() is reached it will
allocate a bigger buffer and copy data from older buffer to the new
one and free the old buffer, making all the previous pointers returned
by u_vector_add() invalid and crashing the application when trying to
access it.
This is reproduced when running
dEQP-VK.synchronization.signal_order.timeline_semaphore.* in DG2 SKUs
that has 4 CCS engines, INTEL_COMPUTE_CLASS=1 is set and of course
perfetto build is enabled.
To fix this issue here I'm moving the storage/allocation of
struct intel_ds_queue to struct anv_queue/iris_batch and using
struct list_head to maintain a chain of intel_ds_queue of the
intel_ds_device.
This allows us to append or remove queues dynamically in future if
necessary.
Fixes: e760c5b37b ("anv: add perfetto source")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20977>
Patch changes comment to refer to the lineage 14014097488, this
workaround applies for ICL as well.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20952>
Allowing longer writes reduces the number of send messages needed
to support unaligned 4-component writes.
Note: nothing currently generates 8-component writes, so this change
makes "second_mask" code path in emit_urb_direct_writes and
emit_urb_indirect_writes_mod dead.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20858>
Our hardware requires that we write to URB using full vec4s at aligned
addresses. It gives us an ability to mask-off dwords within vec4 we don't
want to write, but we have to know their positions at compile time.
Let's assume that:
- V represents one dword we want to write
- ? is an unitinitialized value
- "|" is a vec4 boundary.
When we want to write 2-dword value at offset 0 we generate 1 write message:
| V1 V2 ? ? |
with mask:
| 1 1 0 0 |
When we want to write 4-dword value at offset 2 we generate 2 write messages:
| ? ? V1 V2 | V3 V4 ? ? |
with mask:
| 0 0 1 1 | 1 1 0 0 |
However if we don't know the offset within vec4 at *compile time* we
currently generate 4 write messages:
| V1 V1 V1 V1 |
| 0 0 1 0 |
| V2 V2 V2 V2 |
| 0 0 0 1 |
| V3 V3 V3 V3 |
| 1 0 0 0 |
| V4 V4 V4 V4 |
| 0 1 0 0 |
where masks are determined at *run time*.
This is quite wasteful and slow.
However, if we could determine the offset modulo 4 statically at compile time,
we could generate only 1 or 2 write messages (1 if modulo is 0) instead of 4.
This is what this patch does: it analyzes the addressing expression for
modulo 4 value and if it can determine it at compile time, we generate
1 or 2 writes, and if it can't we fallback to the old 4 writes method.
In mesh shader, the value of offset modulo 4 should be known for all outputs,
with an exception of primitive indices.
The modulo value should be known because of MUE layout restrictions, which
require that user per-primitive and per-vertex data start at address aligned
to 8 dwords and we should statically always know the offset from this base.
There can be some cases where the offset from the base is more dynamic
(e.g. indirect array access inside a per-vertex value), so we always do
the analysis.
Primitive indices are an exception, because they form vec3s (for triangles),
which means that the offset will not be easy to analyse.
When U888X index format lands, primitive indices will use only one dword
per triangle, which means that we'll always write them using one message.
Task shaders don't have any predetermined structure of output memory, so
always do the analysis.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20050>
When writing descriptor with a null buffer/image we expect that
writing 0 will point to the null surface. For that to work the null
surface has to be in the bindless surface heap.
This fixes some new failures in dEQP-VK.robustness.* tests once
rewritten from the NV_ray_tracing to KHR_ray_tracing extension.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 4ceaed7839 ("anv: split internal surface states from descriptors")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7762
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20953>
The anv_execbuf_add_bo() function already sets the offsets for the
exec_objects. Since we're always using softpin and never using
relocations all these objects should have non-changing offsets, all
set during anv_bo creation and never changed. Not only we should not
change these offsets, we definitely don't change them between
anv_execbuf_add_bo() and this loop we're removing.
Previously, we'd have the offset set as -1 for BOs that had never been
submitted when we were not using softpin.
Notice that with games we can have several hundreds of BOs in this
array.
This loop was added by:
c5f7e1f5b4 ("anv: Delete relocation support from batch submission")
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20885>
Commit 582bf4d9 turned on write-combining for most (all?) memory
allocations. This caused a fairly large performance drop in some of
our VMware tests (application traces, such as Windows Metro Paint).
This patch adds a third memory type configuration: DEVICE_LOCAL,
HOST_VISIBLE, HOST_COHERENT. This is uncached. Then, in
anv_AllocateMemory() we only use write-combining for this uncached
type. This memory type is found in the Intel Windows Vulkan driver.
And according to
https://asawicki.info/news_1740_vulkan_memory_types_on_pc_and_how_to_use_them
uncached memory correlates to write-combined memory.
This fixes our performance regression (and actually produced the
fastest ever results for our test suite).
Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20770>
We can lower FS_OPCODE_UNIFORM_PULL_CONSTANT_LOAD into other more
generic sends and drop this internal opcode.
The idea behind this change is to allow bindless surfaces to be used
for UBO pulls and why it's interesting to be able to reuse
setup_surface_descriptors(). But that will come in a later change.
No shader-db changes on TGL & DG2.
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/20416>
If you're only affecting one or a couple of drivers, it would be nice if
your pipeline buttons on the web UI weren't full of manual run buttons for
all the other drivers.
This is a bunch of duplicated lines, but less than it could have been now
that we have !references.
In some of these cases (i915g, nouveau, etnaviv), we have no non-manual
jobs for those drivers, so I could have just rewritten the original
"driver-rules" to "driver-manual-rules". I decided to keep things
consistent between drivers, though, because this is all esoteric enough to
readers already without making different drivers' rules look different.
Fixes: #4891
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17445>
The ACP entries created by copy propagation to track the implied
copies of LOAD_PAYLOAD instructions don't model the behavior of
LOAD_PAYLOAD correctly, since (as of 41868bb682) header
moves are implicitly retyped to UD and the destination of non-header
copies implicitly uses the same type as the corresponding source, even
though the ACP entries created for such copies could incorrectly
represent a type conversion, which can lead to mis-optimization of the
program.
According to Marcin, this fixes the func.mesh.ext.workgroup_id.task.q0
crucible test.
Fixes: 41868bb682 ("i965/fs: Rework the fs_visitor LOAD_PAYLOAD instruction")
Reported-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Tested-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/18980>
This value is also stored in vk_queue, so we can nuke from anv_queue.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20888>
