Insert waitcnts for loads and atomics before stores with system scope.
Scope is field in instruction encoding and corresponds to desired
coherence level in cache hierarchy.
Intrinsic stores can set scope in cache policy operand.
If volatile keyword is used on generic stores memory legalizer will set
scope to system. Generic stores, by default, get lowest scope level.
Waitcnts are not required if it is guaranteed that memory is cached.
For example vulkan shaders can guarantee this.
TODO: implement flag for frontends to give us a hint not to insert
waits.
Expecting vulkan flag to be implemented as vulkan:private MMRA.
At the moment, the emergency spill slot is a fixed object for entry
functions and chain functions, and a regular stack object otherwise.
This patch adopts the latter behaviour for entry/chain functions too. It
seems this was always the intention [1] and it will also save us a bit
of stack space in cases where the first stack object has a large
alignment.
[1]
34c8b835b1
Update SIMemoryLegalizer and SIInsertWaitcnts to use separate wait
instructions per counter (e.g. S_WAIT_LOADCNT) and split VMCNT into
separate LOADCNT, SAMPLECNT and BVHCNT counters.
https://github.com/llvm/llvm-project/pull/70634 has disabled use
of potentially negative scratch offsets, but we can use it on GFX12.
---------
Co-authored-by: Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
For scratch load/store, our hardware only accept non-negative value in
SGPR/VGPR. Besides the case that we can prove from known bits, we can
also prove that the value in `base` will be non-negative: 1.) When the
ADD for the address calculation has NonUnsignedWrap flag. 2.) When the
immediate offset is already negative.
The issue is uncovered by #47698: for IR files without a target triple,
-mtriple= specifies the full target triple while -march= merely sets the
architecture part of the default target triple, leaving a target triple which
may not make sense, e.g. riscv64-apple-darwin.
Therefore, -march= is error-prone and not recommended for tests without a target
triple. The issue has been benign as we recognize $unknown-apple-darwin as ELF instead
of rejecting it outrightly.
SIInsertWaitcnts inserts waitcnt instructions to resolve data
dependencies. The GFX10+ vscnt (VMEM store count) counter is never used
in this way. It is only used to resolve memory dependencies, and that is
handled by SIMemoryLegalizer. Hence there is no need to conservatively
wait for vscnt to be 0 on function entry and before returns.
Differential Revision: https://reviews.llvm.org/D153537
Frame index elimination runs backwards so we must use backwards
scavenging. Otherwise, when a scavenged register is spilled, the
scavenger will remember that the register is in use until the restore
point, but it will never reach that restore point. The result is that in
some cases it will keep scavenging different registers instead of
reusing the same one.
Differential Revision: https://reviews.llvm.org/D152394
This matches what scavengeRegisterBackwards does.
This is in preparation for converting most uses of scavengeRegister to
scavengeRegisterBackwards, to reduce test case churn when that lands and
to help with bisection if anything goes wrong.
Differential Revision: https://reviews.llvm.org/D150792
Values in SGPR and VGPR register are treated as unsigned by hardware.
When value in 32-bit SGPR or VGPR base can be negative calculate offset
using 32-bit add instructions, otherwise use
sgpr(unsigned) + vgpr(unsigned) + offset.
LoopStrengthReduce.cpp changes offsets to negative and in some
iterations value in SGPR or VGPR register could be negative.
Differential Revision: https://reviews.llvm.org/D144957
Values in VGPR register are treated as unsigned by hardware.
When value in 32-bit VGPR base can be negative calculate offset using
32-bit add instruction, otherwise use vgpr base(unsigned) + offset.
Does not affect case where whole offset comes from VGPR register
(immediate offset is 0).
LoopStrengthReduce.cpp changes offsets to negative and in some
iterations value in VGPR register could be negative.
Differential Revision: https://reviews.llvm.org/D144956
This reverts commit 122efef8ee9be57055d204d52c38700fe933c033.
- Patch fixed to not reuse definitions from predecessors in EH landing pads.
- Late review suggestions (by MaskRay) have been addressed.
- M68k/pipeline.ll test updated.
- Init captures added in processBlock() to avoid capturing structured bindings.
- RISCV has this disabled for now.
Original commit message:
A new pass MachineLateInstrsCleanup is added to be run after PEI.
This is a simple pass that removes redundant and identical instructions
whenever found by scanning the MF once while keeping track of register
definitions in a map. These instructions are typically immediate loads
resulting from rematerialization, and address loads emitted by target in
eliminateFrameInde().
This is enabled by default, but a target could easily disable it by means of
'disablePass(&MachineLateInstrsCleanupID);'.
This late cleanup is naturally not "optimal" in removing instructions as it
is done by looking at phys-regs, but still quite effective. It would be
desirable to improve other parts of CodeGen and avoid these redundant
instructions in the first place, but there are no ideas for this yet.
Differential Revision: https://reviews.llvm.org/D123394
Reviewed By: RKSimon, foad, craig.topper, arsenm, asb
Init captures added in processBlock() to avoid capturing structured bindings,
which caused the build problems (with clang).
RISCV has this disabled for now until problems relating to post RA pseudo
expansions are resolved.
A new pass MachineLateInstrsCleanup is added to be run after PEI.
This is a simple pass that removes redundant and identical instructions
whenever found by scanning the MF once while keeping track of register
definitions in a map. These instructions are typically immediate loads
resulting from rematerialization, and address loads emitted by target in
eliminateFrameInde().
This is enabled by default, but a target could easily disable it by means of
'disablePass(&MachineLateInstrsCleanupID);'.
This late cleanup is naturally not "optimal" in removing instructions as it
is done by looking at phys-regs, but still quite effective. It would be
desirable to improve other parts of CodeGen and avoid these redundant
instructions in the first place, but there are no ideas for this yet.
Differential Revision: https://reviews.llvm.org/D123394
Reviewed By: RKSimon, foad, craig.topper, arsenm, asb
This reverts commit e05ce03cfa0b36e9b99149e21afcb1fc039df813.
Caused asan use-after-poison to 4 DebugInfo/AMDGPU/ tests.
Triggered in PEI::replaceFrameIndicesBackward called llvm::MachineInstr::getNumOperands
memcpy has clamped dst stack alignment to NaturalStackAlignment if
hasStackRealignment is false. We should also clamp stack alignment
for memset and memmove. If we don't clamp, SelectionDAG may first
do tail call optimization which requires no stack realignment. Then
memmove, memset in same function may be lowered to load/store with
larger alignment leading to PEI emit stack realignment code which
is absolutely not correct.
Reviewed By: LuoYuanke
Differential Revision: https://reviews.llvm.org/D136456
We form VOPD instructions in the GCNCreateVOPD pass by combining
back-to-back component instructions. There are strict register
constraints for creating a legal VOPD, namely that the matching operands
(e.g. src0x and src0y, src1x and src1y) must be in different register
banks. We add a PostRA scheduler
mutation to put possible VOPD components back-to-back.
Depends on D128442, D128270
Reviewed By: #amdgpu, rampitec
Differential Revision: https://reviews.llvm.org/D128656
GFX11 has a new message type MSG_DEALLOC_VGPRS which can be used to
release a shader's VGPRs. Sending this at the end of a shader (just
before the s_endpgm) can help overall system performance in cases where
the s_endpgm would have to wait for outstanding VMEM stores to complete
before releasing the VGPRs.
Differential Revision: https://reviews.llvm.org/D128442
Fold immediates regardless of how many uses they have. This is expected
to increase overall code size, but decrease register usage.
Differential Revision: https://reviews.llvm.org/D114644
Previously SIFoldOperands::foldInstOperand would only fold a
non-inlinable immediate into a single user, so as not to increase code
size by adding the same 32-bit literal operand to many instructions.
This patch removes that restriction, so that a non-inlinable immediate
will be folded into any number of users. The rationale is:
- It reduces the number of registers used for holding constant values,
which might increase occupancy. (On the other hand, many of these
registers are SGPRs which no longer affect occupancy on GFX10+.)
- It reduces ALU stalls between the instruction that loads a constant
into a register, and the instruction that uses it.
- The above benefits are expected to outweigh any increase in code size.
Differential Revision: https://reviews.llvm.org/D114643
Use a subtarget feature instead of a command line argument to reduce
global state.
We want to enable flat scratch for graphics in some cases and this
doesn't work well with command line options.
Differential Revision: https://reviews.llvm.org/D119425
This was inheriting the mesa behavior, and as far as I know nobody is
using opencl kernels with amdpal. The isMesaKernel check was
irrelevant because this property needs to be held for all functions.
Change VOP_PAT_GEN to default to not generating an instruction selection
pattern for the VOP2 (e32) form of an instruction, only for the VOP3
(e64) form. This allows SIFoldOperands maximum freedom to fold copies
into the operands of an instruction, before SIShrinkInstructions tries
to shrink it back to the smaller encoding.
This affects the following VOP2 instructions:
v_min_i32
v_max_i32
v_min_u32
v_max_u32
v_and_b32
v_or_b32
v_xor_b32
v_lshr_b32
v_ashr_i32
v_lshl_b32
A further cleanup could simplify or remove VOP_PAT_GEN, since its
optional second argument is never used.
Differential Revision: https://reviews.llvm.org/D114252
AMDGPU is unusual in that the both stack is indexed in the same
direction as stack growth (up). We therefore always need the emergency
stack slots placed as low as possible to ensure they are in range of
load/store instruction immediate offsets. The existing logic is mostly
OK, but failed if we required stack realignment.
I don't understand what the existing control isFPCloseToIncomingSP is
supposed to mean, but can only be used to stop placing the scavenge
slots earlier. Make this explicit so that targets can opt-in rather
than opt-out only.
Use GCNHazardRecognizer in postra sched.
Updated tests for the new schedules.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D109536
Change-Id: Ia86ba2ae168f12fb34b4d8efdab491f84d936cde
Prevent SIFoldOperands from creating SALU instructions with a constant
and a frame index. Previously, only one operand was checked to be a
frame index, leading to too many constants when flat scratch is enabled
and stack offsets are large.
Differential Revision: https://reviews.llvm.org/D108368
This allows to convert the add instruction to s_addk_i32 and
v_add_nc_u32 instead of needing v_add_co_u32 when converting to a VALU
instruction.
Differential Revision: https://reviews.llvm.org/D103322
gfx9 does not work with negative offsets, gfx10 works only with
aligned negative offsets, but not with unaligned negative offsets.
This is slightly more conservative than needed, gfx9 does support
negative offsets when a VGPR address is used and gfx10 supports
negative, unaligned offsets when an SGPR address is used, but we
do not make use of that with this patch.
Differential Revision: https://reviews.llvm.org/D101292
Preexisting waitcnt may not update the scoreboard if the instruction
being examined needed to wait on fewer counters than what was encoded in
the old waitcnt instruction. Fixing this results in the elimination of
some redudnat waitcnt.
These changes also enable combining consecutive waitcnt into a single
S_WAITCNT or S_WAITCNT_VSCNT instruction.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D100281
Support for XNACK and SRAMECC is not static on some GPUs. We must be able
to differentiate between different scenarios for these dynamic subtarget
features.
The possible settings are:
- Unsupported: The GPU has no support for XNACK/SRAMECC.
- Any: Preference is unspecified. Use conservative settings that can run anywhere.
- Off: Request support for XNACK/SRAMECC Off
- On: Request support for XNACK/SRAMECC On
GCNSubtarget will track the four options based on the following criteria. If
the subtarget does not support XNACK/SRAMECC we say the setting is
"Unsupported". If no subtarget features for XNACK/SRAMECC are requested we
must support "Any" mode. If the subtarget features XNACK/SRAMECC exist in the
feature string when initializing the subtarget, the settings are "On/Off".
The defaults are updated to be conservatively correct, meaning if no setting
for XNACK or SRAMECC is explicitly requested, defaults will be used which
generate code that can be run anywhere. This corresponds to the "Any" setting.
Differential Revision: https://reviews.llvm.org/D85882
Treat a non-atomic volatile load and store as a relaxed atomic at
system scope for the address spaces accessed. This will ensure all
relevant caches will be bypassed.
A volatile atomic is not changed and still only bypasses caches upto
the level specified by the SyncScope operand.
Differential Revision: https://reviews.llvm.org/D94214
Extract the scratch offset from the scratch buffer descriptor that is
stored in the global table.
Differential Revision: https://reviews.llvm.org/D91701
