Since this method can apply to cmpxchg operations, make sure it's clear
what value we're actually retrieving. This will help ensure we don't
accidentally ignore the failure ordering of cmpxchg in the future.
We could potentially introduce a getOrdering() method on AtomicSDNode
that asserts the operation isn't cmpxchg, but not sure that's
worthwhile.
Differential Revision: https://reviews.llvm.org/D103338
This avoids new warnings due to D68912 adds -Wrange-loop-analysis to -Wall.
Also removed the top-level const as requested by Aaron Ballman in similar
patches.
Differential Revision: https://reviews.llvm.org/D71812
This has two main effects:
- Optimizes debug info size by saving 221.86 MB of obj file size in a
Windows optimized+debug build of 'all'. This is 3.03% of 7,332.7MB of
object file size.
- Incremental step towards decoupling target intrinsics.
The enums are still compact, so adding and removing a single
target-specific intrinsic will trigger a rebuild of all of LLVM.
Assigning distinct target id spaces is potential future work.
Part of PR34259
Reviewers: efriedma, echristo, MaskRay
Reviewed By: echristo, MaskRay
Differential Revision: https://reviews.llvm.org/D71320
* Implements scalable size queries for MVTs, split out from D53137.
* Contains a fix for FindMemType to avoid using scalable vector type
to contain non-scalable types.
* Explicit casts for several places where implicit integer sign
changes or promotion from 32 to 64 bits caused problems.
* CodeGenDAGPatterns will treat scalable and non-scalable vector types
as different.
Reviewers: greened, cameron.mcinally, sdesmalen, rovka
Reviewed By: rovka
Differential Revision: https://reviews.llvm.org/D66871
Summary:
Both the input Value pointer and the returned Value
pointers in GetUnderlyingObjects are now declared as
const.
It turned out that all current (in-tree) uses of
GetUnderlyingObjects were trivial to update, being
satisfied with have those Value pointers declared
as const. Actually, in the past several of the users
had to use const_cast, just because of ValueTracking
not providing a version of GetUnderlyingObjects with
"const" Value pointers. With this patch we get rid
of those const casts.
Reviewers: hfinkel, materi, jkorous
Reviewed By: jkorous
Subscribers: dexonsmith, jkorous, jholewinski, sdardis, eraman, hiraditya, jrtc27, atanasyan, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61038
llvm-svn: 359072
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
`MachineMemOperand` pointers attached to `MachineSDNodes` and instead
have the `SelectionDAG` fully manage the memory for this array.
Prior to this change, the memory management was deeply confusing here --
The way the MI was built relied on the `SelectionDAG` allocating memory
for these arrays of pointers using the `MachineFunction`'s allocator so
that the raw pointer to the array could be blindly copied into an
eventual `MachineInstr`. This creates a hard coupling between how
`MachineInstr`s allocate their array of `MachineMemOperand` pointers and
how the `MachineSDNode` does.
This change is motivated in large part by a change I am making to how
`MachineFunction` allocates these pointers, but it seems like a layering
improvement as well.
This would run the risk of increasing allocations overall, but I've
implemented an optimization that should avoid that by storing a single
`MachineMemOperand` pointer directly instead of allocating anything.
This is expected to be a net win because the vast majority of uses of
these only need a single pointer.
As a side-effect, this makes the API for updating a `MachineSDNode` and
a `MachineInstr` reasonably different which seems nice to avoid
unexpected coupling of these two layers. We can map between them, but we
shouldn't be *surprised* at where that occurs. =]
Differential Revision: https://reviews.llvm.org/D50680
llvm-svn: 339740
According to PTX ISA .volatile has the same memory synchronization
semantics as .relaxed.sys, so it can be used to implement monotonic
atomic loads and stores. This is important for OpenMP's atomic
construct where
- 'read's and 'write's are lowered to atomic loads and stores, and
- an update of float or double types are lowered into a cmpxchg loop.
(Note that PTX could do better because it has atom.add.f{32,64} but
LLVM's atomicrmw instruction only allows integer types.)
Higher levels of atomicity (like acquire and release) need additional
synchronization properties which were added with PTX ISA 6.0 / sm_70.
So using these instructions still results in an error.
Differential Revision: https://reviews.llvm.org/D50391
llvm-svn: 339316
Const/local/shared address spaces are all < 4GB and we can always use
32-bit pointers to access them. This has substantial performance impact
on kernels that uses shared memory for intermediary results.
The feature is disabled by default.
Differential Revision: https://reviews.llvm.org/D46147
llvm-svn: 331941
v2f16 is a special case in NVPTX. v4f16 may be loaded as a pair of v2f16
and that was not previously handled correctly by tryLDGLDU()
Differential Revision: https://reviews.llvm.org/D45339
llvm-svn: 329456
Now that patterns can handle intrinsics returning multiple results,
use tablegen'ed pattern matching instead of custom lowering.
Differential Revision: https://reviews.llvm.org/D43890
llvm-svn: 326457
Summary:
After D43914, loads from global variables in addrspace(1) happen with
ld.global. But since they're constants, even better would be to use
ld.global.nc, aka ldg.
Reviewers: tra
Subscribers: jholewinski, sanjoy, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D43915
llvm-svn: 326390
WMMA = "Warp Level Matrix Multiply-Accumulate".
These are the new instructions introduced in PTX6.0 and available
on sm_70 GPUs.
Differential Revision: https://reviews.llvm.org/D38645
llvm-svn: 315601
Make opcode selection code for the load instruction a bit easier
to read and maintain.
This patch also catches number of f16 load/store variants that were
not handled before.
Differential Revision: https://reviews.llvm.org/D30513
llvm-svn: 296785
This patch enables support for .f16x2 operations.
Added new register type Float16x2.
Added support for .f16x2 instructions.
Added handling of vectorized loads/stores of v2f16 values.
Differential Revision: https://reviews.llvm.org/D30057
Differential Revision: https://reviews.llvm.org/D30310
llvm-svn: 296032
Summary:
DADToDAG has access to TargetLowering, but not vice versa, so this is
the more general location for these functions.
NFC
Reviewers: tra
Subscribers: jholewinski, llvm-commits
Differential Revision: https://reviews.llvm.org/D28795
llvm-svn: 292693
Only scalar half-precision operations are supported at the moment.
- Adds general support for 'half' type in NVPTX.
- fp16 math operations are supported on sm_53+ GPUs only
(can be disabled with --nvptx-no-f16-math).
- Type conversions to/from fp16 are supported on all GPU variants.
- On GPU variants that do not have full fp16 support (or if it's disabled),
fp16 operations are promoted to fp32 and results are converted back
to fp16 for storage.
Differential Revision: https://reviews.llvm.org/D28540
llvm-svn: 291956
Previously we'd always lower @llvm.{sin,cos}.f32 to {sin.cos}.approx.f32
instruction even when unsafe FP math was not allowed.
Clang-generated IR is not affected by this as it uses precise sin/cos
from CUDA's libdevice when unsafe math is disabled.
Differential Revision: https://reviews.llvm.org/D28619
llvm-svn: 291936
Summary:
With this change (plus some changes to prevent !invariant from being
clobbered within llvm), clang will be able to model the __ldg CUDA
builtin as an invariant load, rather than as a target-specific llvm
intrinsic. This will let the optimizer play with these loads --
specifically, we should be able to vectorize them in the load-store
vectorizer.
Reviewers: tra
Subscribers: jholewinski, hfinkel, llvm-commits, chandlerc
Differential Revision: https://reviews.llvm.org/D23477
llvm-svn: 281152
Avoid unnecessary spills of byval arguments of device functions to
local space on SASS level and subsequent pointer conversion to generic
address space that follows. Instead, make a local copy in IR, provide
a way to access arguments directly, and let LLVM optimize the copy away
when possible.
Differential Review: https://reviews.llvm.org/D21421
llvm-svn: 276153
Avoid unnecessary spills of such vars to local space on SASS level and
pointer space conversion.
Instead, make a local copy with appropriate addrspacecasts and let
LLVM optimize them away when possible.
This allows loading value of the argument using [symbol+offset]
instead of converting argument to general space pointer and using it
for indexing (which also implicitly converts param space pointer to
local space one on SASS level and triggers copying of argument into
local space in the process).
This reduces call overhead, uses less registers and reduces overall
SASS size by 2-4%.
Differential Review: http://reviews.llvm.org/D21421
llvm-svn: 273313
- Where we were returning a node before, call ReplaceNode instead.
- Where we would return null to fall back to another selector, rename
the method to try* and return a bool for success.
Part of llvm.org/pr26808.
llvm-svn: 269483
This is a step towards removing the rampant undefined behaviour in
SelectionDAG, which is a part of llvm.org/PR26808.
We rename SelectionDAGISel::Select to SelectImpl and update targets to
match, and then change Select to return void and consolidate the
sketchy behaviour we're trying to get away from there.
Next, we'll update backends to implement `void Select(...)` instead of
SelectImpl and eventually drop the base Select implementation.
llvm-svn: 268693
Summary:
We don't have sign-/zero-extending ldg/ldu instructions defined,
so we need to emulate them with explicit CVTs. We were originally
handling the i8 case, but not any other cases.
Fixes PR26185
Reviewers: jingyue, jlebar
Subscribers: jholewinski
Differential Revision: http://reviews.llvm.org/D19615
llvm-svn: 268272
More specifically, make NVPTXISelDAGToDAG able to emit cached loads (LDG) for pointer induction variables.
Also fix latent bug where LDG was not restricted to kernel functions. I believe that this could not be triggered so far since we do not currently infer that a pointer is global outside a kernel function, and only loads of global pointers are considered for cached loads.
llvm-svn: 244166
Summary:
[NVPTX] make load on global readonly memory to use ldg
Summary:
As describe in [1], ld.global.nc may be used to load memory by nvcc when
__restrict__ is used and compiler can detect whether read-only data cache
is safe to use.
This patch will try to check whether ldg is safe to use and use them to
replace ld.global when possible. This change can improve the performance
by 18~29% on affected kernels (ratt*_kernel and rwdot*_kernel) in
S3D benchmark of shoc [2].
Patched by Xuetian Weng.
[1] http://docs.nvidia.com/cuda/kepler-tuning-guide/#read-only-data-cache
[2] https://github.com/vetter/shoc
Test Plan: test/CodeGen/NVPTX/load-with-non-coherent-cache.ll
Reviewers: jholewinski, jingyue
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D11314
llvm-svn: 242713
[DebugInfo] Add debug locations to constant SD nodes
This adds debug location to constant nodes of Selection DAG and updates
all places that create constants to pass debug locations
(see PR13269).
Can't guarantee that all locations are correct, but in a lot of cases choice
is obvious, so most of them should be. At least all tests pass.
Tests for these changes do not cover everything, instead just check it for
SDNodes, ARM and AArch64 where it's easy to get incorrect locations on
constants.
This is not complete fix as FastISel contains workaround for wrong debug
locations, which drops locations from instructions on processing constants,
but there isn't currently a way to use debug locations from constants there
as llvm::Constant doesn't cache it (yet). Although this is a bit different
issue, not directly related to these changes.
Differential Revision: http://reviews.llvm.org/D9084
llvm-svn: 235989
This adds debug location to constant nodes of Selection DAG and updates
all places that create constants to pass debug locations
(see PR13269).
Can't guarantee that all locations are correct, but in a lot of cases choice
is obvious, so most of them should be. At least all tests pass.
Tests for these changes do not cover everything, instead just check it for
SDNodes, ARM and AArch64 where it's easy to get incorrect locations on
constants.
This is not complete fix as FastISel contains workaround for wrong debug
locations, which drops locations from instructions on processing constants,
but there isn't currently a way to use debug locations from constants there
as llvm::Constant doesn't cache it (yet). Although this is a bit different
issue, not directly related to these changes.
Differential Revision: http://reviews.llvm.org/D9084
llvm-svn: 235977
The operand flag word for ISD::INLINEASM nodes now contains a 15-bit
memory constraint ID when the operand kind is Kind_Mem. This constraint
ID is a numeric equivalent to the constraint code string and is converted
with a target specific hook in TargetLowering.
This patch maps all memory constraints to InlineAsm::Constraint_m so there
is no functional change at this point. It just proves that using these
previously unused bits in the encoding of the flag word doesn't break
anything.
The next patch will make each target preserve the current mapping of
everything to Constraint_m for itself while changing the target independent
implementation of the hook to return Constraint_Unknown appropriately. Each
target will then be adapted in separate patches to use appropriate
Constraint_* values.
PR22883 was caused the matching operands copying the whole of the operand flags
for the matched operand. This included the constraint id which needed to be
replaced with the operand number. This has been fixed with a conversion
function. Following on from this, matching operands also used the operand
number as the constraint id. This has been fixed by looking up the matched
operand and taking it from there.
llvm-svn: 232165
Eli Friedman