If a fp scalar is loaded and then used as both a scalar and a vector broadcast, perform the load as a broadcast and then extract the scalar for 'free' from the 0th element.
This involved switching the order of the X86ISD::BROADCAST combines so we only convert to X86ISD::BROADCAST_LOAD once all other canonicalizations have been attempted.
Adds a DAGCombinerInfo::recursivelyDeleteUnusedNodes wrapper.
Fixes PR43217
Differential Revision: https://reviews.llvm.org/D68544
llvm-svn: 373871
This is patch aims to group together the `CRNotPat` multi class instantiations
within the `PPCInstrInfo.td` file.
Integer instantiations of the multi class are grouped together into a section,
and the floating point patterns are separated into its own section.
Differential Revision: https://reviews.llvm.org/D67975
llvm-svn: 373869
Replaces setTargetShuffleZeroElements with getTargetShuffleAndZeroables which reports the Zeroable elements but doesn't merge them into the decoded target shuffle mask (the merging has been moved up into getTargetShuffleInputs until we can get rid of it entirely).
This is part of the work to fix PR43024 and allow us to use SimplifyDemandedElts to simplify shuffle chains - we need to get to a point where the target shuffle mask isn't adjusted by its source inputs but instead we cache them in a parallel Zeroable mask.
llvm-svn: 373867
Summary:
The default legalization for v16i64->v16i8 tries to create a multiple stage truncate concatenating after each stage and truncating again. But avx512 implements truncates with multiple uops. So it should be better to truncate all the way to the desired element size and then concatenate the pieces using unpckl instructions. This minimizes the number of 2 uop truncates. The unpcks are all single uop instructions.
I tried to handle this by just custom splitting the v16i64->v16i8 shuffle. And hoped that the DAG combiner would leave the two halves in the state needed to make D68374 do the job for each half. This worked for the first half, but the second half got messed up. So I've implemented custom handling for v8i64->v8i8 when v8i64 needs to be split to produce the VTRUNCs directly.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68428
llvm-svn: 373864
Summary: The VSELECT splitting code tries to split a setcc input as well. But on avx512 where mask registers are well supported it should be better to just split the mask and use a single compare.
Reviewers: RKSimon, spatel, efriedma
Reviewed By: spatel
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68359
llvm-svn: 373863
Summary: The assertion in getLoopGuardBranch can be a 'return nullptr'
under if condition.
Authored By: DTharun
Reviewer: Whitney, fhahn
Reviewed By: Whitney, fhahn
Subscribers: fhahn, llvm-commits
Tag: LLVM
Differential Revision: https://reviews.llvm.org/D66084
llvm-svn: 373857
Summary: It ensures that the bswap is generated even when a part of the subtree already matches a bswap transform.
Reviewers: craig.topper, efriedma, RKSimon, lebedev.ri
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68250
llvm-svn: 373850
We can make use of the Zeroable mask to indicate which elements we can safely set to zero instead of creating a target shuffle mask on the fly.
This allows us to remove createTargetShuffleMask.
This is part of the work to fix PR43024 and allow us to use SimplifyDemandedElts to simplify shuffle chains - we need to get to a point where the target shuffle masks isn't adjusted by its source inputs in setTargetShuffleZeroElements but instead we cache them in a parallel Zeroable mask.
llvm-svn: 373846
As discussed on PR42025, with more complex boolean math we can end up with many truncations/extensions of the comparison results through each bitop.
This patch handles the cases introduced in combineBitcastvxi1 by pushing the sign extension through the AND/OR/XOR ops so its just the original SETCC ops that gets extended.
Differential Revision: https://reviews.llvm.org/D68226
llvm-svn: 373834
I don't see an ideal solution to these 2 related, potentially large, perf regressions:
https://bugs.llvm.org/show_bug.cgi?id=42708https://bugs.llvm.org/show_bug.cgi?id=43146
We decided that load combining was unsuitable for IR because it could obscure other
optimizations in IR. So we removed the LoadCombiner pass and deferred to the backend.
Therefore, preventing SLP from destroying load combine opportunities requires that it
recognizes patterns that could be combined later, but not do the optimization itself (
it's not a vector combine anyway, so it's probably out-of-scope for SLP).
Here, we add a scalar cost model adjustment with a conservative pattern match and cost
summation for a multi-instruction sequence that can probably be reduced later.
This should prevent SLP from creating a vector reduction unless that sequence is
extremely cheap.
In the x86 tests shown (and discussed in more detail in the bug reports), SDAG combining
will produce a single instruction on these tests like:
movbe rax, qword ptr [rdi]
or:
mov rax, qword ptr [rdi]
Not some (half) vector monstrosity as we currently do using SLP:
vpmovzxbq ymm0, dword ptr [rdi + 1] # ymm0 = mem[0],zero,zero,..
vpsllvq ymm0, ymm0, ymmword ptr [rip + .LCPI0_0]
movzx eax, byte ptr [rdi]
movzx ecx, byte ptr [rdi + 5]
shl rcx, 40
movzx edx, byte ptr [rdi + 6]
shl rdx, 48
or rdx, rcx
movzx ecx, byte ptr [rdi + 7]
shl rcx, 56
or rcx, rdx
or rcx, rax
vextracti128 xmm1, ymm0, 1
vpor xmm0, xmm0, xmm1
vpshufd xmm1, xmm0, 78 # xmm1 = xmm0[2,3,0,1]
vpor xmm0, xmm0, xmm1
vmovq rax, xmm0
or rax, rcx
vzeroupper
ret
Differential Revision: https://reviews.llvm.org/D67841
llvm-svn: 373833
Rename some variables to match lowerShuffleAsRepeatedMaskAndLanePermute - prep work toward adding some equivalent sublane functionality.
llvm-svn: 373832
The motivation is to reuse the key value parsing logic here to
parse instance specific pass options within the context of MLIR.
The primary functionality exposed is the "," splitting for
arrays and the logic for properly handling duplicate definitions
of a single flag.
Patch by: Parker Schuh <parkers@google.com>
Differential Revision: https://reviews.llvm.org/D68294
llvm-svn: 373815
This is an omission in rL371441. Loads which happened to be unordered weren't being added to the PendingLoad set, and thus weren't be ordered w/respect to side effects which followed before the end of the block.
Included test case is how I spotted this. We had an atomic load being folded into a using instruction after a fence that load was supposed to be ordered with. I'm sure it showed up a bunch of other ways as well.
Spotted via manual inspecting of assembly differences in a corpus w/and w/o the new experimental mode. Finding this with testing would have been "unpleasant".
llvm-svn: 373814
This reverts r371177 (git commit f879c68755)
It caused PR43566 by removing empty, address-taken MachineBasicBlocks.
Such blocks may have references from blockaddress or other operands, and
need more consideration to be removed.
See the PR for a test case to use when relanding.
llvm-svn: 373805
We do indeed already get it right in some cases, but only transitively,
with one-use restrictions. Since we only need to produce a single
comparison, it makes sense to match the pattern directly:
https://rise4fun.com/Alive/kPg
llvm-svn: 373802
Initially (D65380) i believed that if we have rightshift-trunc-rightshift,
we can't do any folding. But as it usually happens, i was wrong.
https://rise4fun.com/Alive/GEwhttps://rise4fun.com/Alive/gN2O
In https://bugs.llvm.org/show_bug.cgi?id=43564 we happen to have
this very sequence, of two right shifts separated by trunc.
And "just" so that happens, we apparently can fold the pattern
if the total shift amount is either 0, or it's equal to the bitwidth
of the innermost widest shift - i.e. if we are left with only the
original sign bit. Which is exactly what is wanted there.
llvm-svn: 373801
Outlining from noreturn functions doesn't do the correct thing right now. The
outliner should respect that the caller is marked noreturn. In the event that
we have a noreturn function, and the outlined code is in tail position, the
outliner will not see that the outlined function should be tail called. As a
result, you end up with a regular call containing a return.
Fixing this requires that we check that all candidates live inside noreturn
functions. So, for the sake of correctness, don't outline from noreturn
functions right now.
Add machine-outliner-noreturn.mir to test this.
llvm-svn: 373791
InstrEmitter's virtual register handling assumes that clones are emitted
after the cloned node. Make sure this assumption actually holds.
Fixes a "Node emitted out of order - early" assertion on the testcase.
This is probably a very rare case to actually hit in practice; even
without the explicit edge, the scheduler will usually end up scheduling
the nodes in the expected order due to other constraints.
Differential Revision: https://reviews.llvm.org/D68068
llvm-svn: 373782
The immediate form of VPCMP can represent these completely. The
vpcmpgt/eq are just shorter encodings.
This patch removes the isel patterns and just swaps the opcodes
and removes the immediate in MCInstLower. This matches where we do
some other encodings tricks.
Removes over 10K bytes from the isel table.
Differential Revision: https://reviews.llvm.org/D68446
llvm-svn: 373766
We already do this for ISD::TRUNCATE, but we can do the same for X86ISD::VTRUNC
Differential Revision: https://reviews.llvm.org/D68432
llvm-svn: 373765
This is a trivial point fix. Terminator instructions aren't scheduled, so
we shouldn't expect to be able to remap them.
This doesn't affect Hexagon and PPC because their terminators are always
hardware loop backbranches that have no register operands.
llvm-svn: 373762
Darwin platforms need the frame register to always point at a valid record even
if it's not updated in a leaf function. Backtraces are more important than one
extra GPR.
llvm-svn: 373738
Rather than having a mixture of location-state shared between DBG_VALUEs
and VarLoc objects in LiveDebugValues, this patch makes VarLoc the
master record of variable locations. The refactoring means that the
transfer of locations from one place to another is always a performed by
an operation on an existing VarLoc, that produces another transferred
VarLoc. DBG_VALUEs are only created at the end of LiveDebugValues, once
all locations are known. As a plus, there is now only one method where
DBG_VALUEs can be created.
The test case added covers a circumstance that is now impossible to
express in LiveDebugValues: if an already-indirect DBG_VALUE is spilt,
previously it would have been restored-from-spill as a direct DBG_VALUE.
We now don't lose this information along the way, as VarLocs always
refer back to the "original" non-transfer DBG_VALUE, and we can always
work out whether a location was "originally" indirect.
Differential Revision: https://reviews.llvm.org/D67398
llvm-svn: 373727
Simon Pilgrim