Make the "block after DO" more stable so that adding instructions after
a DO doesn't require repairing the CFG. Use a new SHADER_OPCODE_FLOW
instruction that is a placeholder representing "go to the next block"
and disappears at code generation.
For some context, there are a few facts about how CFG currently works
- Blocks are assumed to not be empty;
- DO is always by itself in a block, i.e. starts and ends a block;
- There are no empty blocks;
- Predicated WHILE and CONTINUE will link to the "block after DO";
- When nesting loops, it is possible that the "block after DO" is
another "DO".
Reasons and further explanations for those are in the brw_cfg.c comments.
What makes this new change useful is that a pass might want to add
instructions between two DO instructions. When that happens, a new
block must be created and any predicated WHILE and CONTINUE must be
repaired.
So, instead of requiring a repair (which has proven to be tricky in
the past), this change adds a block that can be "virtually" empty but
allow instructions to be added without further changes.
One alternative design would be allowing empty blocks, that would be
a deeper change since the blocks are currently assumed to be not empty
in various places. We'll save that for when other changes are made to
the CFG.
The problem described happens in brw_opt_combine_constants, and a
different patch will clean that up.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33536>
Memory fences do not refer to an element of a binding table. Rather,
the reason we had "BTI" in these opcodes was to distinguish what in
modern terms are called UGM (untyped memory data cache) vs. SLM
(cross-thread shared local memory) fences.
Icelake and older platforms used the "data cache" SFID for both
purposes, distinguishing them by having a special binding table
index, 254, meaning "this is actually SLM access". This is where
the notion that fences had BTIs came in. (In fact, prior to Icelake,
separate SLM fences were not a thing, so BTI wasn't used there either.)
To avoid confusion about BTI being involved, we choose a simpler lie: we
have Icelake SLM fences target GFX12_SFID_SLM (like modern platforms
would), even though it didn't really exist back then. Later lowering
code sets it back to the correct Data Cache SFID with magic SLM binding
table index. This eliminates BTI everywhere and an unnecessary source.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33297>
We can just specify this as a source to the logical FB read/write
opcodes. Notably FB reads had no sources before; now they have one.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33297>
Rather than using a bit in the generic fs_inst data structure, we can
simply set a source on our logical FB write messages. (We already do
so for many other cases.)
In the repclear shader, setting this wasn't actually having an effect,
as we were setting it on a SHADER_OPCODE_SEND message which ignored it.
(We had already correctly set the bit in the message descriptor.)
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33297>
This was used for legacy depth passthrough on older hardware. Gfx9+
doesn't actually have dst depth as part of the message, which is the
only hardware brw supports these days.
It sure looks like we were setting it though...
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33297>
We already have logical pixel interpolator messages that get lowered
to send messages. We can just add an extra boolean source to those
opcodes rather than sticking a opcode-specific boolean in the generic
fs_inst data structure.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33297>
Starting in Xe3, there's a variant of SEND that take the
register numbers from the ARF scalar register, and don't
require them to be contiguous. The new opcode added here
represents that kind of SEND.
To make the original sources still reachable, we keep them
around during the IR, just ignoring them at generator time.
This allow software scoreboard to properly reason the
dependencies without trying to decode the contents of ARF
scalar register being used.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Lionel Landwerlin <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32410>
We want to reuse the brw::nr field as a virtual address register
identifer. So we can't use brw::file=ARF brw::nr=ADDRESS.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28199>
This will translate to HDC Constant Cache loads or LSC UGM loads.
On LSC, MEMORY_MODE_UNTYPED would be fine, but for HDC we need to
distinguish between the regular and constant cache data ports.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32888>
And the SEND gather variant that uses a scalar register as its only
source.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32236>
Add opcodes for VOTE_ALL, VOTE_ANY and VOTE_EQUAL. The first two
are also used for the quad variants. Move their lowering from
NIR conversion to brw_lower_subgroup_ops.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31029>
We'll want to tune this setting based on other parameters.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Backport-to: 24.2
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31196>
These have now been replaced by the MEMORY_*_LOGICAL opcodes.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Acked-by: Rohan Garg <rohan.garg@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30828>
This is a new unified set of opcodes for memory access loosely patterned
after the new LSC-style data port messages introduced on Alchemist GPUs.
Rather than creating an opcode for every type of memory access, it has
only three opcodes: load, store, and atomic. It has various sources to
indicate the rest:
- Binding type (raw pointer, pointer to surface state, or BT index)
- Address size (A64, A32, A16)
- Data size (bit size, number of components)
- Opcode (atomic opcode, or LOAD/STORE vs. LOAD_CMASK/STORE_CMASK)
- Mode (typed vs. untyped vs. shared-local vs. scratch)
- Address (and its dimensionality)
- Data (0 for loads, 1 for stores, 2 for atomics)
- Whether we want block access
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Rohan Garg <rohan.garg@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30828>
The long names were originally meant to map to the HW encoding but
nowadays the actual encoding values depend on gfx version, whether
instruction is 3src, etc.
Suggested by Ken.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30704>
Create specific helper for register file encoding and handle it there.
Use ad-hoc structs to let the macro take optional named arguments.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30704>
SBID SET can only be used on SEND, SENDC, or DPAS instructions. The
existing code was handling SET for SEND/SENDC, but was using the wrong
encoding for DPAS. Add a new case to handle that and make it clear that
the existing code is only for SEND/SENDC.
While here, rewrite the encoder to use 2-bit binary immediates shifted
up into the mode [9:8] field, rather than pre-shifted hex values. This
matches the documentation better and is a little easier to follow.
On the decode side, we were incorrectly decoding MATH instructions.
Because they're marked is_unordered, we were hitting the SEND/SENDC
decoding, which is incorrect for MATH.
Fixes 22 cooperative matrix tests on Lunar Lake.
Huge thanks to Paulo Zanoni for bisecting failures to one of my commits,
then analyzing shaders and experimenting to discover that the failure
was really an unrelated bug, just being provoked by different choices of
registers. His work narrowing the problem down made it much easier to
discover and fix this bug.
Backport-to: 24.2
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30705>
We're going to need to handle encoding/decoding differently for DPAS vs.
SEND/SENDC vs. other instructions. Pass the opcode so we can figure out
the encodings for each type of instruction.
Backport-to: 24.2
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30705>
Our code to initialize gl_SubgroupInvocation uses multiple instructions
some of which are partial writes. This makes it difficult to analyze
expressions involving gl_SubgroupInvocation, which appear very
frequently in compute shaders.
To make this easier, we add a new virtual opcode which initializes
a full VGRF to the value of gl_SubgroupInvocation. (We also expand
it to UD for SIMD8 so there are not partial write issues.) We then
lower it to the original code later on in compilation, after we've
done the bulk of our optimizations.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28666>
These were removed with Icelake. While they technically still exist on
Skylake, which this compiler supports, we have never used these opcodes
in the 14 years we could have done so. So just scrap them.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29665>
In 2c65d90bc8 I forgot to add the new SHADER_OPCODE_READ_MASK_REG
opcode to the list of barrier instruction in the scheduler. Let's just
use a single opcode for all ARF registers that need special
scoreboarding and put the register as source (nicer for the debug
output).
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 2c65d90bc8 ("intel/brw: ensure find_live_channel don't access arch register without sync")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29446>
Another architecture register that requires some care before reading.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 49ee3ae9e8 ("intel/compiler: Lower FIND_[LAST_]LIVE_CHANNEL in IR on Gfx8+")
Tested-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29319>
We no longer support the old LINE+MAC lowering, and we already lower
this to MAD in NIR on Gfx11+, so the LINTERP virtual opcode always
corresponds the PLN. The only catch is that LINTERP's operands are
reversed from PLN, so we have to switch them.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28705>
We already have a logical opcode and lower to what is basically a send
instruction. We just weren't using SHADER_OPCODE_SEND, instead having
extra redundant infrastructure for no real gain.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28705>
We first generate the logical opcodes, and these days fully lower to
SHADER_OPCODE_SEND. In the past, we lowered to a non-logical variant
and handled that in the generator. These days, we were just using the
non-logical opcodes as an awkward intermediate opcode change during
the lowering...which isn't really necessary at all.
This patch eliminates them by using the original logical opcodes.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27908>
Thanks to Ken for suggesting this URB refactoring change and pointing
out that the LSC can operate on the byte offset granularity.
This should fix the geometry shader test cases where we have more than
32 vertices since previously we were failing to write the correct
control data bits because of incorrect write mask.
Shader-db results for Xe2:
total instructions in shared programs: 153475 -> 153437 (-0.02%)
instructions in affected programs: 1374 -> 1336 (-2.77%)
helped: 11
HURT: 0
helped stats (abs) min: 3 max: 5 x̄: 3.45 x̃: 3
helped stats (rel) min: 1.67% max: 4.92% x̄: 3.23% x̃: 2.70%
95% mean confidence interval for instructions value: -3.92 -2.99
95% mean confidence interval for instructions %-change: -4.10% -2.36%
Instructions are helped.
total loops in shared programs: 140 -> 140 (0.00%)
loops in affected programs: 0 -> 0
helped: 0
HURT: 0
total cycles in shared programs: 16002649 -> 16002329 (<.01%)
cycles in affected programs: 9174 -> 8854 (-3.49%)
helped: 11
HURT: 0
helped stats (abs) min: 22 max: 38 x̄: 29.09 x̃: 32
helped stats (rel) min: 2.62% max: 5.54% x̄: 3.78% x̃: 3.85%
95% mean confidence interval for cycles value: -33.56 -24.62
95% mean confidence interval for cycles %-change: -4.48% -3.08%
Cycles are helped.
total spills in shared programs: 52 -> 52 (0.00%)
spills in affected programs: 0 -> 0
helped: 0
HURT: 0
total fills in shared programs: 94 -> 94 (0.00%)
fills in affected programs: 0 -> 0
helped: 0
HURT: 0
total sends in shared programs: 4240 -> 4240 (0.00%)
sends in affected programs: 0 -> 0
helped: 0
HURT: 0
LOST: 0
GAINED: 0
Rework: (Sagar)
- Adjust offset/indirect offset calculation.
- Add shader-db results
- Always calculate dword index
- Drop changes for indirect writes
Signed-off-by: Rohan Garg <rohan.garg@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27602>
