This should only consider whether the pressure impact of the bundle at
the given point in the program will decrease the occupancy. High VGPR
pressure was incorrectly blocking the formation of scalar bundles, and
vice versa. This was also blocking bundling from high pressure
situations at other points in the program.
If the G_BR + G_BRCOND in this combine use the same MBB, then it will infinite
loop. Don't allow that to happen.
Differential Revision: https://reviews.llvm.org/D95895
Second land attempt. MachineVerifier DefRegState expensive check errors fixed.
Prologs and epilogs handle callee-save registers and tend to be irregular with
different immediate offsets that are not often handled by the MachineOutliner.
Commit D18619/a5335647d5e8 (combining stack operations) stretched irregularity
further.
This patch tries to emit homogeneous stores and loads with the same offset for
prologs and epilogs respectively. We have observed that this canonicalizes
(homogenizes) prologs and epilogs significantly and results in a greatly
increased chance of outlining, resulting in a code size reduction.
Despite the above results, there are still size wins to be had that the
MachineOutliner does not provide due to the special handling X30/LR. To handle
the LR case, his patch custom-outlines prologs and epilogs in place. It does
this by doing the following:
* Injects HOM_Prolog and HOM_Epilog pseudo instructions during a Prolog and
Epilog Injection Pass.
* Lowers and optimizes said pseudos in a AArchLowerHomogneousPrologEpilog Pass.
* Outlined helpers are created on demand. Identical helpers are merged by the linker.
* An opt-in flag is introduced to enable this feature. Another threshold flag
is also introduced to control the aggressiveness of outlining for application's need.
This reduced an average of 4% of code size on LLVM-TestSuite/CTMark targeting arm64/-Oz.
Differential Revision: https://reviews.llvm.org/D76570
Due to a clerical error, the sdiv operation was mapping to vdivu and
udiv to vdiv, when the opposite mapping is the correct one.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D95869
This allows the peephole optimizer to know that a MVE_VMOV_to_lane_32 is
the same as an insert subreg, allowing it to optimize some redundant
lane moves.
Differential Revision: https://reviews.llvm.org/D95433
The temporary register is only used to compute the frame pointer.
The frame pointer is overwritten and not used in between, so we
can reuse the frame pointer for the computation, saving one register.
Differential Revision: https://reviews.llvm.org/D95865
Saving callee-save registers happens in whole wave mode. Exec is saved
to a free register, which can be reused to save the frame pointer.
Therefore, saving the fp needs to happen after saving csrs.
Differential Revision: https://reviews.llvm.org/D95861
NOTE: This patch was originally written by Anil Mahmud. His code has been
rebased but otherwise left mostly unchanged.
A new instructon on Power 10 allows for the materialization of 34 bit
immediate values. This patch allows the compiler to take advantage of
the new instruction in this situation.
Reviewed By: amyk
Differential Revision: https://reviews.llvm.org/D92879
A v4i32 insert of an extract can become a simple lane move, as opposed
to round-tripping via a GPR. This adds a patterns that turns an v4i32
insert-extract pair into a EXTRACT_SUBREG/INSERT_SUBREG, with the
required COPY_TO_REGCLASS. These get better optimized into a simple lane
move by the rest of the backend.
Differential Revision: https://reviews.llvm.org/D95428
This patch adds tablegen patterns for pairs of i16/f16 insert/extracts.
If we are inserting into two adjacent vector lanes (0 and 1 for
example), we can use either a vmov;vins or vmovx;vins to insert the pair
together, avoiding a round-trip from GRP registers. This is quite a
large patterns with a number of EXTRACT_SUBREG/INSERT_SUBREG/
COPY_TO_REGCLASS nodes, but hopefully as most of those become copies all
that will be cleaned up by further optimizations.
The VINS pattern was also adjusted to allow it to represent that it is
inserting into the top half of an existing register.
Differential Revision: https://reviews.llvm.org/D95381
With predicate masks, AVX512 can efficiently perform variable-index vector insertion with 2 broadcasts + 1 comparison, avoiding a lot of aliased memory traffic.
Differential Revision: https://reviews.llvm.org/D95779
A DLS lr, lr instruction only moves lr to itself. It need not be emitted
on it's own to save a instruction in the loop preheader.
Differential Revision: https://reviews.llvm.org/D78916
I guess instructions were marked as frame-setup by accident, they are
restores as part of the epilog.
Differential Revision: https://reviews.llvm.org/D95783
Prologs and epilogs handle callee-save registers and tend to be irregular with
different immediate offsets that are not often handled by the MachineOutliner.
Commit D18619/a5335647d5e8 (combining stack operations) stretched irregularity
further.
This patch tries to emit homogeneous stores and loads with the same offset for
prologs and epilogs respectively. We have observed that this canonicalizes
(homogenizes) prologs and epilogs significantly and results in a greatly
increased chance of outlining, resulting in a code size reduction.
Despite the above results, there are still size wins to be had that the
MachineOutliner does not provide due to the special handling X30/LR. To handle
the LR case, his patch custom-outlines prologs and epilogs in place. It does
this by doing the following:
* Injects HOM_Prolog and HOM_Epilog pseudo instructions during a Prolog and
Epilog Injection Pass.
* Lowers and optimizes said pseudos in a AArchLowerHomogneousPrologEpilog Pass.
* Outlined helpers are created on demand. Identical helpers are merged by the linker.
* An opt-in flag is introduced to enable this feature. Another threshold flag
is also introduced to control the aggressiveness of outlining for application's need.
This reduced an average of 4% of code size on LLVM-TestSuite/CTMark targeting arm64/-Oz.
Differential Revision: https://reviews.llvm.org/D76570
DBG_VALUES placed between memory instructions would change
codegen. Skip over these and re-insert them after the bundle instead
of giving up on bundling.
This is directly analogous to the existing no_caller_saved_registers, but with the opposite intention. A function or call so marked shifts the responsibility of spilling the usual CSRs to it's caller.
An indirect call site and callee which don't agree on the attribute is ill defined.
The motivation for this change is that being able to prune callee saves (without modifying other details of the calling convention) is sometimes useful when generating stubs and adapters. There's no intention to expose this as a source language feature; this is expected to be used by frontends to implement adapters where warranted.
Some specific examples of use cases:
* GC compatible compiled code wants to call an externally defined library function without needing to track pointer values through CSRs.
* debug enabled code wants to call precompiled library which doesn't provide enough information to track CSRs while preserving debug quality in caller.
* adapter stub entering hand written assembler which doesn't follow normal calling conventions.
Given a shuffle(vqdmulh(shuffle, shuffle), we can flatter the shuffles
out if they become an identity mask. This can come up during lane
interleaving, when we do that better.
Differential Revision: https://reviews.llvm.org/D94034
When we have a zeroext parameter, emit G_ASSERT_ZEXT.
Add a check that we actually emit it.
This is a 0.1% code size win on CTMark/7zip and CTMark/consumer-typeset at -Os.
Differential Revision: https://reviews.llvm.org/D95567
A follow up patch will add support for commuting operands or
changing opcode to vfmacc and friends.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D95662
When replacing the dst reg with the src reg, we need to make sure that we
propagate the dst reg's register class through to the src.
Otherwise, we aren't meeting the requirements for G_ASSERT_ZEXT, and so the
verifier will fail.
Differential Revision: https://reviews.llvm.org/D95708
Rather than materializing the 0xffff immediate for the AND, use
a shift left to remove the upper bits and then shift in zeros
from the right.
This pattern occurs when type legalizing an i16 right shift.
I've implemented this with custom selection code for a number of
reasons. I've limited this to the AND having a single use. We need
to compensate for SimplifyDemandedBits altering the AND mask. I'm
using *W opcodes on RV64. We may want to generlize this in the
future. For all these reason it seemed easiest to do it this way.
Reviewed By: luismarques
Differential Revision: https://reviews.llvm.org/D95774
This would assert with amdgpu-spill-sgpr-to-vgpr disabled when trying to
spill the FP.
Fixes: SWDEV-262704
Reviewed By: RamNalamothu
Differential Revision: https://reviews.llvm.org/D95768
Under the softfp calling convention, we are often left with
VMOVRRD(extract(bitcast(build_vector(a, b, c, d)))) for the return value
of the function. These can be simplified to a,b or c,d directly,
depending on the value of the extract.
Big endian is a little different because the bitcast switches the lanes
around, meaning we end up with b,a or d,c.
Differential Revision: https://reviews.llvm.org/D94989
This adds a DAG combine for converting sext_inreg of VGetLaneu into
VGetLanes, providing the types match correctly.
Differential Revision: https://reviews.llvm.org/D95073
We can only legally extract from the lowest 128-bit subvector, so extract the correct subvector to allow us to handle 256/512-bit vector element extracts.
Under SoftFP calling conventions, we can be left with
extract(bitcast(BUILD_VECTOR(VMOVDRR(a, b), ..))) patterns that can
simplify to a or b, depending on the extract lane.
Differential Revision: https://reviews.llvm.org/D94990
If we're going to end up expanding anyway, we should do it early
so we don't create extra operations to handle the bytes added by
promotion.
This is helfpul on RISCV where we might have to promote i16 all
the way to i64.
Differential Revision: https://reviews.llvm.org/D95756
With a context instruction, this would produce a context
error. However, it would continue on and do an out of bounds access of
the empty allocation order array.
This demonstrates a missed optimization: the `vmv.x.s` instruction is
used to extract the element from the vector, and this instruction
already sign-extends the value to XLEN.
This is an optimized approach for D94155.
Previous code build the model that tile config register is the user of
each AMX instruction. There is a problem for the tile config register
spill. When across function, the ldtilecfg instruction may be inserted
on each AMX instruction which use tile config register. This cause all
tile data register clobber.
To fix this issue, we remove the model of tile config register. Instead,
we analyze the AMX instructions between one call to another. We will
insert ldtilecfg after the first call if we find any AMX instructions.
Reviewed By: LuoYuanke
Differential Revision: https://reviews.llvm.org/D95136
Source Drift happens when the sources are updated after profiling the binary
but before building the final optimized binary. If the source has changed since
the profiles were obtained, optimizing basic blocks might be sub-optimal. This
only applies to BasicBlockSection::List as it creates clusters of basic blocks
using basic block ids. Source drift can invalidate these groupings leading to
sub-optimal code generation with regards to performance.
PGO source drift for a particular function can be detected using function
metadata added in D95495.
When source drift is deected, disable basic block clusters by default
which can be re-enabled with -mllvm option
bbsections-detect-source-drift=false.
Differential Revision: https://reviews.llvm.org/D95593
