Currently, the custom SGPR spill lowering pass spills
SGPRs into physical VGPR lanes and the remaining VGPRs
are used by regalloc for vector regclass allocation.
This imposes many restrictions that we ended up with
unsuccessful SGPR spilling when there won't be enough
VGPRs and we are forced to spill the leftover into
memory during PEI. The custom spill handling during PEI
has many edge cases and often breaks the compiler time
to time.
This patch implements spilling SGPRs into virtual VGPR
lanes. Since we now split the register allocation for
SGPRs and VGPRs, the virtual registers introduced for
the spill lanes would get allocated automatically in
the subsequent regalloc invocation for VGPRs.
Spill to virtual registers will always be successful,
even in the high-pressure situations, and hence it avoids
most of the edge cases during PEI. We are now left with
only the custom SGPR spills during PEI for special registers
like the frame pointer which isn an unproblematic case.
This patch also implements the whole wave spills which
might occur if RA spills any live range of virtual registers
involved in the whole wave operations. Earlier, we had
been hand-picking registers for such machine operands.
But now with SGPR spills into virtual VGPR lanes, we are
exposing them to the allocator.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D124196
value() has undesired exception checking semantics and calls
__throw_bad_optional_access in libc++. Moreover, the API is unavailable without
_LIBCPP_NO_EXCEPTIONS on older Mach-O platforms (see
_LIBCPP_AVAILABILITY_BAD_OPTIONAL_ACCESS).
This fixes clang.
C++17 allows us to call constructors pair and tuple instead of helper
functions make_pair and make_tuple.
Differential Revision: https://reviews.llvm.org/D139828
- [Clang] Declare AMDGPU target as supporting BF16 for storage-only purposes on amdgcn
- Add Sema & CodeGen tests cases.
- Also add cases that D138651 would have covered as this patch replaces it.
- [AMDGPU] Add BF16 storage-only support
- Support legalization/dealing with bf16 operations in DAGIsel.
- bf16 as a type remains illegal and is represented as i16 for storage purposes.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D139398
Renames the current lowering scheme to "module" and introduces two new
ones, "kernel" and "table", plus a "hybrid" that chooses between those three
on a per-variable basis.
Unit tests are set up to pass with the default lowering of "module" or "hybrid"
with this patch defaulting to "module", which will be a less dramatic codegen
change relative to the current. This reflects the sparsity of test coverage for
the table lowering method. Hybrid is better than module in every respect and
will be default in a subsequent patch.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D139433
Don't bother pre-shrinking "s_cmp_lg_u32 reg, 0" to s_cmpk_lg_u32
because 0 is already an inline constant so the s_cmpk form is no
smaller.
This is just for consistency with the surrounding code and to simplify a
downstream patch.
Differential Revision: https://reviews.llvm.org/D138993
Fixes a longstanding TODO in the codebase where we were using S_GETREG + shift to do something that could simply be done with an inline constant (register).
Patch based on D31874 by @kzhuravl
Depends on D137767
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D137542
Following up on the removal of BufferPSV in commit 43b86bf992 ("AMDGPU:
Remove BufferPseudoSourceValue")
It is unclear what exactly the right address space for images should be.
They seem morally closest to buffers, so that's what I went with. In
practical terms, address space 7 is better than address space 0 because
it can't alias with LDS.
Differential Revision: https://reviews.llvm.org/D138949
We were missing legality checks. The device library build was broken
for targets without f16 support. Technically the first pattern isn't
tested by this patch; it only triggers with the isBeforeLegalize check
in performAndCombine removed. I'm not sure how to trick this into
appearing post-legalization.
The use of a PSV for buffer intrinsics is misleading because it may be
misinterpreted as all buffer intrinsics accessing the same address in
memory, which is clearly not true.
Instead, build MachineMemOperands without a pointer value but with an
address space, so that address space-based alias analysis can still
work.
There is a lot of test churn because previously address space 4
(constant address space) was used as an address space for buffer
intrinsics. This doesn't make much sense and seems to have been an
accident -- see the change in
AMDGPUTargetMachine::getAddressSpaceForPseudoSourceKind.
Differential Revision: https://reviews.llvm.org/D138711
class support and introduce GlobalISel implementation for AMDGPU
Uses existing SelectionDAG lowering of the llvm.amdgcn.class intrinsic
for llvm.is.fpclass
A target can return if a misaligned access is 'fast' as defined
by the target or not. In reality there can be different levels
of 'fast' and 'slow'. This patch changes the boolean 'Fast'
argument of the allowsMisalignedMemoryAccesses family of functions
to an unsigned representing its speed.
A target can still define it as it wants and the direct translation
of the current code uses 0 and 1 for current false and true. This
makes the change an NFC.
Subsequent patch will start using an actual value of speed in
the load/store vectorizer to compare if a vectorized access going
to be not just fast, but not slower than before.
Differential Revision: https://reviews.llvm.org/D124217
The TableGen implementation was using a homegrown implementation of
FunctionModRefInfo. This switches it to use MemoryEffects instead.
This makes the code simpler, and will allow exposing the full
representational power of MemoryEffects in the future. Among other
things, this will allow us to map IntrHasSideEffects to an
inaccessiblemem readwrite, rather than just ignoring it entirely
in most cases.
To avoid layering issues, this moves the ModRef.h header from IR
to Support, so that it can be included in the TableGen layer.
Differential Revision: https://reviews.llvm.org/D137641
In a lot of places, we were just calling `getNamedOperandIdx` to check if the result was != or == to -1.
This is fine in itself, but it's verbose and doesn't make the intention clear, IMHO. I added a `hasNamedOperand` and replaced all cases I could find with regexes and manually.
Reviewed By: arsenm, foad
Differential Revision: https://reviews.llvm.org/D137540
The 32-bit floating-point atomic add instructions on AMDGPUs does not support a
"flat" or "generic" address space. So, if the address space cannot be determined
statically, the AMDGPU backend will fall back to a CAS loop (which does support
"flat" addressing). Instead, this patch emits runtime address-space checks to
allow native FP atomic add instructions for global and LDS memory (and non-atomic
FP add instructions for private/scratch memory).
In order to do that, this patch introduces a new interface function
`emitExpandAtomicRMW`. It is expected to be called when a common atomic expand
doesn't work for a specific target, such as the case we discussed here.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D129690
This switches everything to use the memory attribute proposed in
https://discourse.llvm.org/t/rfc-unify-memory-effect-attributes/65579.
The old argmemonly, inaccessiblememonly and inaccessiblemem_or_argmemonly
attributes are dropped. The readnone, readonly and writeonly attributes
are restricted to parameters only.
The old attributes are auto-upgraded both in bitcode and IR.
The bitcode upgrade is a policy requirement that has to be retained
indefinitely. The IR upgrade is mainly there so it's not necessary
to update all tests using memory attributes in this patch, which
is already large enough. We could drop that part after migrating
tests, or retain it longer term, to make it easier to import IR
from older LLVM versions.
High-level Function/CallBase APIs like doesNotAccessMemory() or
setDoesNotAccessMemory() are mapped transparently to the memory
attribute. Code that directly manipulates attributes (e.g. via
AttributeList) on the other hand needs to switch to working with
the memory attribute instead.
Differential Revision: https://reviews.llvm.org/D135780
The bitmask used to extract the bits assumed 16 bit elements and wasn't taking the size of the elements into account.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D135156
Recognize more opcodes in the function.
Fixes some regressions introduced in D134857 for fdiv.f16 too.
Depends on D134857
Reviewed By: arsenm, foad
Differential Revision: https://reviews.llvm.org/D134862
Surprisingly these were getting legalized to something
zero initialized.
This fixes an infinite loop when combining some vector types.
Also fixes zero initializing some undef values.
SimplifyDemandedVectorElts / SimplifyDemandedBits are not checking
for the legality of the output undefs they are replacing unused
operations with. This resulted in turning vectors into undefs
that were later re-legalized back into zero vectors.
A kernel may have an associated struct for laying out LDS variables.
This patch puts that instance, if present, at a deterministic address by
allocating it at the same time as the module scope instance.
This is relatively likely to be where the instance was allocated anyway (~NFC)
but will allow later patches to calculate where a given field can be found,
which means a function which is only reachable from a single kernel will be
able to access a LDS variable with zero overhead. That will be particularly
helpful for applications that instantiate a function template containing LDS
variables once per kernel.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D127052
Make MIMG NSA minimum addresses threshold an attribute that can
be set on a function or configured via command line.
This enables frontend tuning which allows increased NSA usage
where beneficial.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D134780
Remove manual selection for atomic fadd from global-isel.
Stop pre-isel translation to AtomicLoadFAdd/G_ATOMICRMW_FADD
which corresponds to llvm-ir's atomicrmw fadd instruction.
global and flat atomic fadd patterns changes:
Split rtn/no-rtn patterns
Add missing patterns or fix predicates
Remove atomicrmw patterns for v2f16 (atomic rmw doesn't support vectors).
Patterns now check addrspace of pointer, added patterns for flat intrinsic.
with global addrspace pointer that selects into global atomic instruction.
buffer atomic fadd patterns changes:
Rdit patterns to import into global-isel.
Remove gfx6/gfx7 _addr64 and _offset patterns.
Remove patterns that can't be reached (same pattern but different feature).
Differential Revision: https://reviews.llvm.org/D130579
Use same atomicrmw fadd expansion rules for gfx908, gfx940 and gfx11
as for gfx90a. Add missing globalisel legalizer support for flat
atomicrmw fadd f32 on gfx940 and gfx11.
Isel support for gfx11 will be added in D130579.
Differential Revision: https://reviews.llvm.org/D131560
This patch contains changes necessary to carry physical condition register (SCC) dependencies through the SDNode scheduler. It adds the edge in the SDNodeScheduler dependency graph instead of inserting the SCC copy between each definition and use. This approach lets the scheduler place instructions in an optimal way placing the copy only when the dependency cannot be resolved.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D133593
All in-tree targets pass pointer-sized ConstantSDNodes to the
method. This overload reduced amount of boilerplate code a bit. This
also makes getCALLSEQ_END consistent with getCALLSEQ_START, which
already takes uint64_ts.
Currently there isn't a generic way to get a smaller register class
that can be produced from a subregister of a larger class. Replaces a
manually implemented version for AMDGPU. This will be used to improve
subregister support in the allocator.
TragetLowering had two last InstructionCost related `getTypeLegalizationCost()`
and `getScalingFactorCost()` members, but all other costs are processed in TTI.
E.g. it is not comfortable to use other TTI members in these two functions
overrided in a target.
Minor refactoring: `getTypeLegalizationCost()` now doesn't need DataLayout
parameter - it was always passed from TTI.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D117723
Implement an intrinsic for use lowering LDS variables to different
addresses from different kernels. This will allow kernels that cannot
reach an LDS variable to avoid wasting space for it.
There are a number of implicit arguments accessed by intrinsic already
so this implementation closely follows the existing handling. It is slightly
novel in that this SGPR is written by the kernel prologue.
It is necessary in the general case to put variables at different addresses
such that they can be compactly allocated and thus necessary for an
indirect function call to have some means of determining where a
given variable was allocated. Claiming an arbitrary SGPR into which
an integer can be written by the kernel, in this implementation based
on metadata associated with that kernel, which is then passed on to
indirect call sites is sufficient to determine the variable address.
The intent is to emit a __const array of LDS addresses and index into it.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D125060
