Generalized the cost model estimation. Improved cost model estimation
for repeated scalars (no need to count their cost anymore), improved
cost model for extractelement instructions.
cpu2017
511.povray_r 0.57
520.omnetpp_r -0.98
521.wrf_r -0.01
525.x264_r 3.59 <+
526.blender_r -0.12
531.deepsjeng_r -0.07
538.imagick_r -1.42
Geometric mean: 0.21
Differential Revision: https://reviews.llvm.org/D115757
Generalized the cost model estimation. Improved cost model estimation
for repeated scalars (no need to count their cost anymore), improved
cost model for extractelement instructions.
cpu2017
511.povray_r 0.57
520.omnetpp_r -0.98
521.wrf_r -0.01
525.x264_r 3.59 <+
526.blender_r -0.12
531.deepsjeng_r -0.07
538.imagick_r -1.42
Geometric mean: 0.21
Differential Revision: https://reviews.llvm.org/D115757
Freeze instruction in some cases makes codegen worse, so need to be very
careful when emitting it. Instead improve analysis in isUndefVector
function to generate mask of unused elements and use it in the analysis.
Differential Revision: https://reviews.llvm.org/D135382
Revert rGef89409a59f3b79ae143b33b7d8e6ee6285aa42f "Fix 'unused-lambda-capture' gcc warning. NFCI."
Revert rG926ccfef032d206dcbcdf74ca1e3a9ebf4d1be45 "[SLP] ScalarizationOverheadBuilder - demand all elements for scalarization if the extraction index is unknown / out of bounds"
Revert ScalarizationOverheadBuilder sequence from D134605 - when accumulating extraction costs by Type (instead of specific Value), we are not distinguishing enough when they are coming from the same source or not, and we always just count the cost once. This needs addressing before we can use getScalarizationOverhead properly.
Need either follow the original order of the operands for bool logical
ops, or emit freeze instruction to avoid poison propagation.
Differential Revision: https://reviews.llvm.org/D126877
Add test cases for AArch64 that show over-eager SLP vectorization on
AArch64, where keeping the things scalar allows efficient lowering using
scalar fmas.
2xi64 is the legalized type for wide reductions (like 16xi64) and setting the
cost to 2 makes `load-reduce` and `load-zext-reduce` patterns profitable.
The few performance measurments that I did on an aarch64 machine confirm that
these patterns are actually faster when vectorized.
Differential Revision: https://reviews.llvm.org/D130740
This patch slightly extends the limit on the RecursionMaxDepth inside
the SLP vectorizer. It does it only when it hits a load (or zext/sext of
a load), which allows it to peek through in the places where it will be
the most valuable, without ballooning out the O(..) by any 2^n factors.
Differential Revision: https://reviews.llvm.org/D122148
Improved/fixed cost modeling for shuffles by providing masks, improved
cost model for non-identity insertelements.
Differential Revision: https://reviews.llvm.org/D115462
This reverts commit cac60940b771a0685d058a5b471c84cea05fdc46.
Caused -Os -fsanitize=memory -march=haswell miscompile to pytorch/cpuinfo.
See my latest comment (may update) on D115462.
This reverts commit f1ee2738b3d70fea803ac1f3401c2fc9f61e514a.
Revert due to the revert of a dependent commit `[SLP]Improve shuffles cost estimation where possible.`
If the OffsetBeg + InsertVecSz is greater than VecSz, need to estimate
the cost as shuffle of 2 vector, not as insert of subvector. Otherwise,
the inserted subvector is out of range and compiler may crash.
Differential Revision: https://reviews.llvm.org/D128071
Improved/fixed cost modeling for shuffles by providing masks, improved
cost model for non-identity insertelements.
Differential Revision: https://reviews.llvm.org/D115462
Improved/fixed cost modeling for shuffles by providing masks, improved
cost model for non-identity insertelements.
Differential Revision: https://reviews.llvm.org/D115462
Improved/fixed cost modeling for shuffles by providing masks, improved
cost model for non-identity insertelements.
Differential Revision: https://reviews.llvm.org/D115462
Need to handle a corner case correctly, if all elements are Undefs/Poisons,
need to emit actual values, not just poisons.
Differential Revision: https://reviews.llvm.org/D126298
Need to use all ReductionOps when propagating flags for the reduction
ops, otherwise transformation is not correct. Plus, need to drop nuw/nsw
flags.
Differential Revision: https://reviews.llvm.org/D126371
SLP vectorizer emits extracts for externally used vectorized scalars and
estimates the cost for each such extract. But in many cases these
scalars are input for insertelement instructions, forming buildvector,
and instead of extractelement/insertelement pair we can emit/cost
estimate shuffle(s) cost and generate series of shuffles, which can be
further optimized.
Tested using test-suite (+SPEC2017), the tests passed, SLP was able to
generate/vectorize more instructions in many cases and it allowed to reduce
number of re-vectorization attempts (where we could try to vectorize
buildector insertelements again and again).
Differential Revision: https://reviews.llvm.org/D107966
This reverts commit fc9c59c355cb255446e571b4515b5e41a76503c4.
The patch triggers an assertion when building SPEC on X86. Reduced
reproducer shared at D107966.
Also reverts follow-up commit 11a09af76d11ad5a9f1f95b561112af17ff81f80.
SLP vectorizer emits extracts for externally used vectorized scalars and
estimates the cost for each such extract. But in many cases these
scalars are input for insertelement instructions, forming buildvector,
and instead of extractelement/insertelement pair we can emit/cost
estimate shuffle(s) cost and generate series of shuffles, which can be
further optimized.
Tested using test-suite (+SPEC2017), the tests passed, SLP was able to
generate/vectorize more instructions in many cases and it allowed to reduce
number of re-vectorization attempts (where we could try to vectorize
buildector insertelements again and again).
Differential Revision: https://reviews.llvm.org/D107966
Given a load without a better order, this patch partially sorts the
elements to form clusters of adjacent elements in memory. These clusters
can potentially be loaded in fewer loads, meaning less overall shuffling
(for example loading v4i8 clusters of a v16i8 as a single f32 loads, as
opposed to multiple independent bytes loads and inserts).
Differential Revision: https://reviews.llvm.org/D122145
This adds fptosi_sat and fptoui_sat to the list of trivially
vectorizable functions, mainly so that the loop vectorizer can vectorize
the instruction. Marking them as trivially vectorizable also allows them
to be SLP vectorized, and Scalarized.
The signature of a fptosi_sat requires two type overrides
(@llvm.fptosi.sat.v2i32.v2f32), unlike other intrinsics that often only
take a single. This patch alters hasVectorInstrinsicOverloadedScalarOpd
to isVectorIntrinsicWithOverloadTypeAtArg, so that it can mark the first
operand of the intrinsic as a overloaded (but not scalar) operand.
Differential Revision: https://reviews.llvm.org/D124358
Currently SLP vectorizer walks through the instructions and selects
3 main classes of values: 1) reduction operations - instructions with same
reduction opcode (add, mul, min/max, etc.), which build the reduction,
2) reduced values - instructions with the same opcodes, but different
from the reduction opcode, 3) extra arguments - all other values,
instructions from the different basic block rather than the root node,
instructions with to many/less uses.
This scheme is not very efficient. It excludes some instructions and all
non-instruction values from the reductions (constants, proficient
gathers), to many possibly reduced values are marked as extra arguments.
Patch improves this process by introducing a bit extended analysis
stage. During this stage, we still try to select 3 classes of the
values: 1) reduction operations - same as before, 2) possibly reduced
values - all instructions from the current block/non-instructions, which
may build a vectorization tree, 3) extra arguments - instructions from
the different basic blocks. Additionally, an extra sorting of the
possibly reduced values occurs to build the scalar sequences which
highly likely will bed vectorized, e.g. loads are grouped by the
distance between them, constants are grouped together, cmp instructions
are sorted by their compare types and predicates, extractelement
instructions are sorted by the vector operand, etc. Also, these groups
are reordered by their length so the longest group is the first in the
list of the possibly reduced values.
The vectorization process tries to emit the reductions for all these
groups. These reductions, remaining non-vectorized possible reduced
values and extra arguments are then combined into the final expression
just like it was before.
Differential Revision: https://reviews.llvm.org/D114171
tryToVectorize() method implements one of searching paths for vectorizable tree roots in SLP vectorizer,
specifically for binary and comparison operations. Order of making probes for various scalar pairs
was defined by its implementation: the instruction operands, then climb over one operand if
the instruction is its sole user and then perform same actions for another operand if previous
attempts failed. Problem with this approach is that among these options we can have more than a
single vectorizable tree candidate and it is not necessarily the one that encountered first.
Trying to build vectorizable tree for each possible combination for just evaluation is expensive.
But we already have lookahead heuristics mechanism which we use for finding best pick among
operands of commutative instructions. It calculates cumulative score for candidates in two
consecutive lanes. This patch introduces use of the heuristics for choosing the best pair among
several combinations. We only try one that looks as most promising for vectorization.
Additional benefit is that we reduce total number of vectorization trees built for probes
because we skip those looking non-profitable early.
Reviewed By: Alexey Bataev (ABataev), Vasileios Porpodas (vporpo)
Differential Revision: https://reviews.llvm.org/D124309
The original patch (https://reviews.llvm.org/D121354) targets x86 and adjusts
the lookahead score of splat loads ad they can be done by the `movddup`
instruction that combines the load and the broadcast and is cheap to execute.
A similar issue shows up on AArch64. The `ld1r` instruction performs a broadcast
load and is cheap to execute.
This patch implements the TargetTransformInfo hooks for AArch64.
Differential Revision: https://reviews.llvm.org/D123638
SLP uses the distance between pointers to optimize
the getShallowScore. However the current code misses
the case where we are trying to vectorize for VF=4, and the distance
between pointers is 2. In that case the returned score
reflects the case of contiguous loads, when it's not actually
contiguous.
The attached unit tests have 5 loads, where the program order
is not the same as the offset order in the GEPs. So, the choice
of which 4 loads to bundle together matters. If we pick the
first 4, then we can vectorize with VF=4. If we pick the
last 4, then we can only vectorize with VF=2.
This patch makes a more conservative choice, to consider
all distances>1 to not be a case of contiguous load, and
give those cases a lower score.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D123516
Currently SLP vectorizer walks through the instructions and selects
3 main classes of values: 1) reduction operations - instructions with same
reduction opcode (add, mul, min/max, etc.), which build the reduction,
2) reduced values - instructions with the same opcodes, but different
from the reduction opcode, 3) extra arguments - all other values,
instructions from the different basic block rather than the root node,
instructions with to many/less uses.
This scheme is not very efficient. It excludes some instructions and all
non-instruction values from the reductions (constants, proficient
gathers), to many possibly reduced values are marked as extra arguments.
Patch improves this process by introducing a bit extended analysis
stage. During this stage, we still try to select 3 classes of the
values: 1) reduction operations - same as before, 2) possibly reduced
values - all instructions from the current block/non-instructions, which
may build a vectorization tree, 3) extra arguments - instructions from
the different basic blocks. Additionally, an extra sorting of the
possibly reduced values occurs to build the scalar sequences which
highly likely will bed vectorized, e.g. loads are grouped by the
distance between them, constants are grouped together, cmp instructions
are sorted by their compare types and predicates, extractelement
instructions are sorted by the vector operand, etc. Also, these groups
are reordered by their length so the longest group is the first in the
list of the possibly reduced values.
The vectorization process tries to emit the reductions for all these
groups. These reductions, remaining non-vectorized possible reduced
values and extra arguments are then combined into the final expression
just like it was before.
Differential Revision: https://reviews.llvm.org/D114171
If we vectorize a e.g. store, we leave around a bunch of getelementptrs for the individual scalar stores which we removed. We can go ahead and delete them as well.
This is purely for test output quality and readability. It should have no effect in any sane pipeline.
Differential Revision: https://reviews.llvm.org/D122493
The original commit exposed several missing dependencies (e.g. latent bugs in SLP scheduling). Most of these were fixed over the weekend and have had several days to bake. The last was fixed this morning after being noticed in manual review of test changes yesterday. See the review thread for links to each change.
Original commit message follows:
SLP currently schedules all instructions within a scheduling window which stretches from the first instruction potentially vectorized to the last. This window can include a very large number of unrelated instructions which are not being considered for vectorization. This change switches the code to only schedule the sub-graph consisting of the instructions being vectorized and their transitive users.
This has the effect of greatly reducing the amount of work performed in large basic blocks, and thus greatly improves compile time on degenerate examples. To understand the effects, I added some statistics (not planned for upstream contribution). Here's an illustration from my motivating example:
Before this patch:
704357 SLP - Number of calcDeps actions
699021 SLP - Number of schedule calls
5598 SLP - Number of ReSchedule actions
59 SLP - Number of ReScheduleOnFail actions
10084 SLP - Number of schedule resets
8523 SLP - Number of vector instructions generated
After this patch:
102895 SLP - Number of calcDeps actions
161916 SLP - Number of schedule calls
5637 SLP - Number of ReSchedule actions
55 SLP - Number of ReScheduleOnFail actions
10083 SLP - Number of schedule resets
8403 SLP - Number of vector instructions generated
I do want to highlight that there is a small difference in number of generated vector instructions. This example is hitting the bailout due to maximum window size, and the change in scheduling is slightly perturbing when and how we hit it. This can be seen in the RescheduleOnFail counter change. Given that, I think we can safely ignore.
The downside of this change can be seen in the large test diff. We group all vectorizable instructions together at the bottom of the scheduling region. This means that vector instructions can move quite far from their original point in code. While maybe undesirable, I don't see this as being a major problem as this pass is not intended to be a general scheduling pass.
For context, it's worth noting that the pre-scheduling that SLP does while building the vector tree is exactly the sub-graph scheduling implemented by this patch.
Differential Revision: https://reviews.llvm.org/D118538
SLP is currently assuming that control dependence in these cases is irrelevant. This is only valid if none of the lib-funcs involved can throw or infinite loop in the scalar forms. This appears to be true (or at least we infer the respective attributes) for the libfuncs I spot checked. This change is mostly for shrunking the diff on an upcoming patch.
No need to schedule entry nodes where all instructions are not memory
read/write instructions and their operands are either constants, or
arguments, or phis, or instructions from others blocks, or their users
are phis or from the other blocks.
The resulting vector instructions can be placed at
the beginning of the basic block without scheduling (if operands does
not need to be scheduled) or at the end of the block (if users are
outside of the block).
It may save some compile time and scheduling resources.
Differential Revision: https://reviews.llvm.org/D121121
No need to schedule entry nodes where all instructions are not memory
read/write instructions and their operands are either constants, or
arguments, or phis, or instructions from others blocks, or their users
are phis or from the other blocks.
The resulting vector instructions can be placed at
the beginning of the basic block without scheduling (if operands does
not need to be scheduled) or at the end of the block (if users are
outside of the block).
It may save some compile time and scheduling resources.
Differential Revision: https://reviews.llvm.org/D121121
