Replace insertelement instructions for splats with just single
insertelement + broadcast shuffle. Also, try to merge these instructions
if they come from the same/shuffled gather node.
Differential Revision: https://reviews.llvm.org/D107104
For the nodes with reused scalars the user may be not only of the size
of the final shuffle but also of the size of the scalars themselves,
need to check for this. It is safe to just modify the check here, since
the order of the scalars themselves is preserved, only indeces of the
reused scalars are changed. So, the users with the same size as the
number of scalars in the node, will not be affected, they still will get
the operands in the required order.
Reported by @mstorsjo in D105020.
Differential Revision: https://reviews.llvm.org/D107080
If the instruction was previously deleted, it should not be treated as
an external user. This fixes cost estimation and removes dead
extractelement instructions.
Differential Revision: https://reviews.llvm.org/D107106
Need to check that the minimum acceptable vector factor is at least 2,
not 0, to avoid compiler crash during gathered loads analysis.
Differential Revision: https://reviews.llvm.org/D107058
Reworked reordering algorithm. Originally, the compiler just tried to
detect the most common order in the reordarable nodes (loads, stores,
extractelements,extractvalues) and then fully rebuilding the graph in
the best order. This was not effecient, since it required an extra
memory and time for building/rebuilding tree, double the use of the
scheduling budget, which could lead to missing vectorization due to
exausted scheduling resources.
Patch provide 2-way approach for graph reodering problem. At first, all
reordering is done in-place, it doe not required tree
deleting/rebuilding, it just rotates the scalars/orders/reuses masks in
the graph node.
The first step (top-to bottom) rotates the whole graph, similarly to the previous
implementation. Compiler counts the number of the most used orders of
the graph nodes with the same vectorization factor and then rotates the
subgraph with the given vectorization factor to the most used order, if
it is not empty. Then repeats the same procedure for the subgraphs with
the smaller vectorization factor. We can do this because we still need
to reshuffle smaller subgraph when buildiong operands for the graph
nodes with lasrger vectorization factor, we can rotate just subgraph,
not the whole graph.
The second step (bottom-to-top) scans through the leaves and tries to
detect the users of the leaves which can be reordered. If the leaves can
be reorder in the best fashion, they are reordered and their user too.
It allows to remove double shuffles to the same ordering of the operands in
many cases and just reorder the user operations instead. Plus, it moves
the final shuffles closer to the top of the graph and in many cases
allows to remove extra shuffle because the same procedure is repeated
again and we can again merge some reordering masks and reorder user nodes
instead of the operands.
Also, patch improves cost model for gathering of loads, which improves
x264 benchmark in some cases.
Gives about +2% on AVX512 + LTO (more expected for AVX/AVX2) for {625,525}x264,
+3% for 508.namd, improves most of other benchmarks.
The compile and link time are almost the same, though in some cases it
should be better (we're not doing an extra instruction scheduling
anymore) + we may vectorize more code for the large basic blocks again
because of saving scheduling budget.
Differential Revision: https://reviews.llvm.org/D105020
Need to fix several cost-related problems. The final type may be defined
incorrectly because of to early definition (we may end up with the wider
type), the CommonCost should not be redefined in ExtractElements
cost related calculations and the shuffle of the final insertelements
vectors should be calculated as a cost of single vector permutations
+ costs of two vector permutations for other n-1 incoming vectors.
Differential Revision: https://reviews.llvm.org/D106578
Need to fix several cost-related problems. The final type may be defined
incorrectly because of to early definition (we may end up with the wider
type), the CommonCost should not be redefined in ExtractElements
cost related calculations and the shuffle of the final insertelements
vectors should be calculated as a cost of single vector permutations
+ costs of two vector permutations for other n-1 incoming vectors.
Differential Revision: https://reviews.llvm.org/D106578
This changes the cost to (LT.first-1) * cost(add) + 2, where the cost of
an add is assumed to be 1. This brings it inline with the other
reductions.
Differential Revision: https://reviews.llvm.org/D106240
The incoming values for PHI nodes may come from unreachable BasicBlocks,
need to handle this case.
Differential Revision: https://reviews.llvm.org/D106264
Part of D105020. Also, fixed FIXMEs that need to use wider vector type
when trying to calculate the cost of reused scalars. This may cause
regressions unless D100486 is landed to improve the cost estimations
for long vectors shuffling.
Differential Revision: https://reviews.llvm.org/D106060
The cost of the InsertSubvector shuffle kind cost is not complete and
may end up with just extracts + inserts costs in many cases. Added
a workaround to represent it as a generic PermuteSingleSrc, which is
still pessimistic but better than InsertSubvector.
Differential Revision: https://reviews.llvm.org/D105827
This bug was introduced with D105730 / 25ee55c0ba .
If we are not converting all of the operations of a reduction
into a vector op, we need to preserve the existing select form
of the remaining ops. Otherwise, we are potentially leaking
poison where it did not in the original code.
Alive2 agrees that the version that freezes some inputs
and then falls back to scalar is correct:
https://alive2.llvm.org/ce/z/erF4K2
The cost of the InsertSubvector shuffle kind cost is not complete and
may end up with just extracts + inserts costs in many cases. Added
a workaround to represent it as a generic PermuteSingleSrc, which is
still pessimistic but better than InsertSubvector.
Differential Revision: https://reviews.llvm.org/D105827
This has been a work-in-progress for a long time...we finally have all of
the pieces in place to handle vectorization of compare code as shown in:
https://llvm.org/PR41312
To do this (see PhaseOrdering tests), we converted SimplifyCFG and
InstCombine to the poison-safe (select) forms of the logic ops, so now we
need to have SLP recognize those patterns and insert a freeze op to make
a safe reduction:
https://alive2.llvm.org/ce/z/NH54Ah
We get the minimal patterns with this patch, but the PhaseOrdering tests
show that we still need adjustments to get the ideal IR in some or all of
the motivating cases.
Differential Revision: https://reviews.llvm.org/D105730
We know that "CVTTPS2SI" returns 0x80000000 for out of range inputs (and for FP_TO_UINT, negative float values are undefined). We can use this to make unsigned conversions from vXf32 to vXi32 more efficient, particularly on targets without blend using the following logic:
small := CVTTPS2SI(x);
fp_to_ui(x) := small | (CVTTPS2SI(x - 2^31) & ARITHMETIC_RIGHT_SHIFT(small, 31))
Even on targets where "PBLENDVPS"/"PBLENDVB" exists, it is often a latency 2, low throughput instruction so this logic is applied there too (in particular for AVX2 also). It furthermore gets rid of one high latency floating point comparison in the previous lowering.
@TomHender checked the correctness of this for all possible floats between -1 and 2^32 (both ends excluded).
Original Patch by @TomHender (Tom Hender)
Differential Revision: https://reviews.llvm.org/D89697
Update (mainly) vXf32/vXf64 -> vXi8/vXi16 fptosi/fptoui costs based on the worst case costs from the script in D103695.
Move to using legalized types wherever possible, which allows us to prune the cost tables.
Update truncation costs based on the worst case costs from the script in D103695.
Move to using legalized types wherever possible, which allows us to prune the cost tables.
Patch tries to improve the vectorization of stores. Originally, we just
check the type and the base pointer of the store.
Patch adds some extra checks to avoid non-profitable vectorization
cases. It includes analysis of the scalar values to be stored and
triggers the vectorization attempt only if the scalar values have
same/alt opcode and are from same basic block, i.e. we don't end up
immediately with the gather node, which is not profitable.
This also improves compile time by filtering out non-profitable cases.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D104122
Revived D101297 in its original form + added some changes in X86
legalization cehcking for masked gathers.
This solution is the most stable and the most correct one. We have to
check the legality before trying to build the masked gather in SLP.
Without this check we have incorrect cost (for SLP) in case if the masked gather
is not legal/slower than the gather. And we're missing some
vectorization opportunities.
This can be fixed in the cost model, but in this case we need to add
special checks for the cost of GEPs for ScatterVectorize node, add
special check for small trees, etc., i.e. there are a lot of corner
cases here and there, which insrease code base and make it harder to
maintain the code.
> Can't we rely on cost model to deal with this? This can be profitable for futher vectorization, when we can start from such gather loads as seed.
The question from D101297. Actually, no, it can't. Actually, simple
gather may give us better result, especially after we started
vectorization of insertelements. Plus, like I said before, the cost for
non-legal masked gathers leads to missed vectorization opportunities.
Differential Revision: https://reviews.llvm.org/D105042
Update costs based on the worst case costs from the script in D103695.
Move to using legalized types wherever possible, which allows us to prune the cost tables.
Update (mainly) vXi8/vXi16 -> vXf32/vXf64 sitofp/uitofp costs based on the worst case costs from the script in D103695.
Move to using legalized types wherever possible, which allows us to prune the cost tables.
Compare type IDs and DFS numbering for basic block instead of addresses
to fix non-determinism.
Differential Revision: https://reviews.llvm.org/D105031
Provide a generic fallback that extends sub-i32 scalars before using the existing sitofp instructions.
These numbers can be tweaked for specific sse levels, but we should get the default handling in place first.
We get the extension for free for non-vector loads.
The function vectorizeChainsInBlock does not support scalable vector,
because function like canReuseExtract and isCommutative in the code
path assert with scalable vectors.
This patch avoids vectorizing blocks that have extract instructions with scalable
vector..
Differential Revision: https://reviews.llvm.org/D104809
Loads of <4 x i8> vectors were modeled as extremely expensive. And while we
don't have a load instruction that supports this, it isn't that expensive to
create a vector of i8 elements. The codegen for this was fixed/optimised in
D105110. This now tweaks the cost model and enables SLP vectorisation of my
motivating case loadi8.ll.
Differential Revision: https://reviews.llvm.org/D103629
Update v4i64 -> v4f32/v4f64 uitofp costs based on the worst case costs from the script in D103695.
Fixes a few regressions before we start adding AVX costs for legalized types.
The compiler should not ignore UndefValue when gathering the scalars,
otherwise the resulting code may be less defined than the original one.
Also, grouped scalars to insert them at first to reduce the analysis in
further passes.
Differential Revision: https://reviews.llvm.org/D105275
Building on rG2a1ef8784ad9a, adjust the SSE cost tables to use the legalized types based on the worst case costs from the script in D103695.
To account for different numbers of src/dst legalized type registers we must scale the cost by maximum of the src/dst, not just use src
Based off the worse case numbers generated by D103695, the AVX1/2/512 sitofp/uitofp/fptosi/fptoui costs were higher than necessary (based off instruction counts instead of actual throughput).
The SSE costs still need further fixes, but I hit an issue with the order in which SSE costs are checked - we need to check CUSTOM costs (with non-legal types) first, and then fallback to LEGALIZED types. I'm looking at this now, and this should let us start thinning out a lot of the duplicates in the costs tables.
Then we can finally start work on vXi64 / vXi16 / vXi8 / vXi1 integers, which should let us look at sub-128-bit vectorization (D103925).
OR, XOR and AND entries are added to the cost table. An extra cost
is added when vector splitting occurs.
This is done to address the issue of a missed SLP vectorization
opportunity due to unreasonably high costs being attributed to the vector
Or reduction (see: https://bugs.llvm.org/show_bug.cgi?id=44593).
Differential Revision: https://reviews.llvm.org/D104538
Make getPointersDiff() and sortPtrAccesses() compatible with opaque
pointers by explicitly passing in the element type instead of
determining it from the pointer element type.
The SLPVectorizer result is slightly non-optimal in that unnecessary
pointer bitcasts are added.
Differential Revision: https://reviews.llvm.org/D104784
Perform better analysis when trying to vectorize PHIs.
1. Do not try to vectorize vector PHIs.
2. Do deeper analysis for more profitable nodes for the vectorization.
Before we just tried to vectorize the PHIs of the same type. Patch
improves this and tries to vectorize PHIs with incoming values which
come from the same basic block, have the same and/or alternative
opcodes.
It allows to save the compile time and provides better vectorization
results in general.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D103638
This can be seen as a follow up to commit 0ee439b705,
that changed the second argument of __powidf2, __powisf2 and
__powitf2 in compiler-rt from si_int to int. That was to align with
how those runtimes are defined in libgcc.
One thing that seem to have been missing in that patch was to make
sure that the rest of LLVM also handle that the argument now depends
on the size of int (not using the si_int machine mode for 32-bit).
When using __builtin_powi for a target with 16-bit int clang crashed.
And when emitting libcalls to those rtlib functions, typically when
lowering @llvm.powi), the backend would always prepare the exponent
argument as an i32 which caused miscompiles when the rtlib was
compiled with 16-bit int.
The solution used here is to use an overloaded type for the second
argument in @llvm.powi. This way clang can use the "correct" type
when lowering __builtin_powi, and then later when emitting the libcall
it is assumed that the type used in @llvm.powi matches the rtlib
function.
One thing that needed some extra attention was that when vectorizing
calls several passes did not support that several arguments could
be overloaded in the intrinsics. This patch allows overload of a
scalar operand by adding hasVectorInstrinsicOverloadedScalarOpd, with
an entry for powi.
Differential Revision: https://reviews.llvm.org/D99439
There is no need to schedule insertelement instructions. The compiler
did not schedule them before it started support their vectorization and
it should not do it after. We pre-schedule them manually when finding
a build vector sequence.
Disabling scheduling of insertelement instructions improves compile
time and vectorization of the very large basic blocks by saving
scheduling budget for other instructions.
Differential Revision: https://reviews.llvm.org/D104026
1. Better sorting of scalars to be gathered. Trying to insert
constants/arguments/instructions-out-of-loop at first and only then
the instructions which are inside the loop. It improves hoisting of
invariant insertelements instructions.
2. Better detection of shuffle candidates in gathering function.
3. The cost of insertelement for constants is 0.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D103458
No need to recalculate the cost of extractelements, just no need to
compensate the cost of all extractelements, need to check before if this
is actually going to be removed at the vectorization. Also, no need to
generate new extractelement instruction, we may just regenerate the
original one. It may improve the final vectorization.
Differential Revision: https://reviews.llvm.org/D102933
tryToVectorizeList function allows to reorder only 2 scalars. Patch
allows to reorder >2 scalars. Also, to avoid possible regressions, it
allows extra vectorization of the remaining parts of the scalars
elements if possible.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D103247
As noticed by NAKAMURA Takumi back in 2017, we cannot use
properlyDominates for std::stable_sort as properlyDominates only
partially orders blocks. That is, for blocks A, B, C, D, where A
dominates B and C dominates D, we have A == C, B == C, but A < B. This
is not a valid comparison function for std::stable_sort and causes
different results between libstdc++ and libc++. This change uses DFS
numbering to give deterministic results for all reachable blocks.
Unreachable blocks are ignored already, so do not need special
consideration.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D103441
As the existing test unreachable.ll shows, we should be doing more
work to avoid entering unreachable blocks: we should not stop
vectorization just because a PHI incoming value from an unreachable
block cannot be vectorized. We know that particular value will never
be used so we can just replace it with poison.
Implemented better scheme for perfect/shuffled matches of the gather
nodes which allows to fix the performance regressions introduced by
earlier patches. Starting detecting matches for broadcast nodes and
extractelement gathering.
Differential Revision: https://reviews.llvm.org/D102920
This is a patch that replaces shufflevector and insertelement's placeholder value with poison.
Underlying motivation is to fix the semantics of shufflevector with undef mask to return poison instead
(D93818)
The consensus has been made in the late 2020 via mailing list as well as the thread in https://bugs.llvm.org/show_bug.cgi?id=44185 .
This patch is a simple syntactic change to the existing code, hence directly pushed as a commit.
SLP vectorizer should not consider in sertelements with multiple uses as
a part of high level build vector, it must be considered as
a terminating insertelement in the vector build, otherwise it may
produce incorrect code.
Differential Revision: https://reviews.llvm.org/D103164
Determined from llvm-mca analysis, AVX1 capable targets have a higher throughput for VPBLENDVB and shuffle ops, making it cheaper to perform shift+shuffle/select shift patterns.
Based on worst case of sandybridge (vs btver2 + bdver2) llvm-mca analysis - which is a lot less than what we were predicting (I think based off total uop count).
BTVER2 has a weaker f64 multiplier that other AVX1-era targets, so we need to bump the worst case cost slightly - llvm-mca reports the new vectorization in simplebb is beneficial on btver2, bdver2 and sandybridge AVX1 targets
External insertelement users can be represented as a result of shuffle
of the vectorized element and noconsecutive insertlements too. Added
support for handling non-consecutive insertelements.
Differential Revision: https://reviews.llvm.org/D101555
If we gather extract elements and they actually are just shuffles, it
might be profitable to vectorize them even if the tree is tiny.
Differential Revision: https://reviews.llvm.org/D101460
Add new type of tree node for `InsertElementInst` chain forming vector.
These instructions could be either removed, or replaced by shuffles during
vectorization and we can add this node to cost model, so naturally estimating
their cost, getting rid of `CompensateCost` tricks and reducing further work
for InstCombine. This fixes PR40522 and PR35732 in a natural way. Also this
patch is the first step towards revectorization of partially vectorization
(to fix PR42022 completely). After adding inserts to tree the next step is
to add vector instructions there (for instance, to merge `store <2 x float>`
and `store <2 x float>` to `store <4 x float>`).
Fixes PR40522 and PR35732.
Differential Revision: https://reviews.llvm.org/D98714
The test example from https://llvm.org/PR50256 (and reduced here)
shows that we can match a load combine candidate even when there
are no "or" instructions. We can avoid that by confirming that we
do see an "or". This doesn't apply when matching an or-reduction
because that match begins from the operands of the reduction.
Differential Revision: https://reviews.llvm.org/D102074
Need to remove the old code for avoiding double counting of the gather
nodes with perfect diamond matches within the tree after we started
detecting perfect/shuffled matching in the previous patch D100495. We
may skip the cost for such nodes completely.
Differential Revision: https://reviews.llvm.org/D102023
Need to check if target allows/supports masked gathers before trying to
estimate its cost, otherwise we may fail to vectorize some of the
patterns because of too pessimistic cost model.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D101297
Need to check if target allows/supports masked gathers before trying to
estimate its cost, otherwise we may fail to vectorize some of the
patterns because of too pessimistic cost model.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D101297
If the extracts from the non-power-2 vectors are recognized as shuffles,
need some extra checks to not crash cost calculations if trying to gext
the ecost for subvector extracts. In this case need to check carefully
that we do not exit out of bounds of the original vector, otherwise the
TTI's cost model will crash on assert.
Differential Revision: https://reviews.llvm.org/D101477
Added an extra analysis for better choosing of shuffle kind in
getShuffleCost functions for better cost estimation if mask was
provided.
Differential Revision: https://reviews.llvm.org/D100865
Added an extra analysis for better choosing of shuffle kind in
getShuffleCost functions for better cost estimation if mask was
provided.
Differential Revision: https://reviews.llvm.org/D100865
If the first tree element is vectorize and the second is gather, it
still might be profitable to vectorize it if the gather node contains
less scalars to vectorize than the original tree node. It might be
profitable to use shuffles.
Differential Revision: https://reviews.llvm.org/D101397
We can skip check for undefs trying to find perfect/shuffled tree
entries matching, they can be ignored completely improving the final
cost/vectorization results.
Differential Revision: https://reviews.llvm.org/D101061
1. No need to call `areAllUsersVectorized` as later the cost is
calculated only if the instruction has one use and gets vectorized.
2. Need to calculate the cost of the dead extractelement more precisely,
taking the vector type of the vector operand, not the resulting
vector type.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D99980
SLP supports perfect diamond matching for the vectorized tree entries
but do not support it for gathered entries and does not support
non-perfect (shuffled) matching with 1 or 2 tree entries. Patch adds
support for this matching to improve cost of the vectorized tree.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D100495
SLP supports perfect diamond matching for the vectorized tree entries
but do not support it for gathered entries and does not support
non-perfect (shuffled) matching with 1 or 2 tree entries. Patch adds
support for this matching to improve cost of the vectorized tree.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D100495
SLP supports perfect diamond matching for the vectorized tree entries
but do not support it for gathered entries and does not support
non-perfect (shuffled) matching with 1 or 2 tree entries. Patch adds
support for this matching to improve cost of the vectorized tree.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D100495
SLP supports perfect diamond matching for the vectorized tree entries
but do not support it for gathered entries and does not support
non-perfect (shuffled) matching with 1 or 2 tree entries. Patch adds
support for this matching to improve cost of the vectorized tree.
Differential Revision: https://reviews.llvm.org/D100495
After D98856 these tests will by default break (fatal_error) if any of
the wrong interfaces are used, so there's no longer a need to have a
RUN line that checks for a warning message emitted by the compiler.
During vectorization better to postpone the vectorization of the CmpInst
instructions till the end of the basic block. Otherwise we may vectorize
it too early and may miss some vectorization patterns, like reductions.
Reworked part of D57059
Differential Revision: https://reviews.llvm.org/D99796
The ultimate reduction node may have multiple uses, but if the ultimate
reduction is min/max reduction and based on SelectInstruction, the
condition of this select instruction must have only single use.
Differential Revision: https://reviews.llvm.org/D99753
The motivation for this patch is to better estimate the cost of
extracelement instructions in cases were they are going to be free,
because the source vector can be used directly.
A simple example is
%v1.lane.0 = extractelement <2 x double> %v.1, i32 0
%v1.lane.1 = extractelement <2 x double> %v.1, i32 1
%a.lane.0 = fmul double %v1.lane.0, %x
%a.lane.1 = fmul double %v1.lane.1, %y
Currently we only consider the extracts free, if there are no other
users.
In this particular case, on AArch64 which can fit <2 x double> in a
vector register, the extracts should be free, independently of other
users, because the source vector of the extracts will be in a vector
register directly, so it should be free to use the vector directly.
The SLP vectorized version of noop_extracts_9_lanes is 30%-50% faster on
certain AArch64 CPUs.
It looks like this does not impact any code in
SPEC2000/SPEC2006/MultiSource both on X86 and AArch64 with -O3 -flto.
This originally regressed after D80773, so if there's a better
alternative to explore, I'd be more than happy to do that.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D99719
1. Need to cleanup InstrElementSize map for each new tree, otherwise might
use sizes from the previous run of the vectorization attempt.
2. No need to include into analysis the instructions from the different basic
blocks to save compile time.
Differential Revision: https://reviews.llvm.org/D99677
This is a 2nd try of:
3c8473ba53
which was reverted at:
a26312f9d4
because of crashing.
This version includes extra code and tests to avoid the known
crashing examples as discussed in PR49730.
Original commit message:
As noted in D98152, we need to patch SLP to avoid regressions when
we start canonicalizing to integer min/max intrinsics.
Most of the real work to make this possible was in:
7202f47508
Differential Revision: https://reviews.llvm.org/D98981
This was originally just an XFAIL test, but I modified it
to check output. To make that bot-friendly, I'm moving it
to the x86 dir since it specified an x86 target.
This reverts commit 3c8473ba53 and includes test diffs to
maintain testing status.
There's at least 1 place that was not updated with 7202f47508 ,
so we can crash mismatching select and intrinsics as shown in
PR49730.
The SCEV commit b46c085d2b [NFCI] SCEVExpander:
emit intrinsics for integral {u,s}{min,max} SCEV expressions
seems to reveal a new crash in SLPVectorizer.
SLP crashes expecting a SelectInst as an externally used value
but umin() call is found.
The patch relaxes the assumption to make the IR flag propagation safe.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D99328
We do not need to scan further if the upper end or lower end of the
basic block is reached already and the instruction is not found. It
means that the instruction is definitely in the lower part of basic
block or in the upper block relatively.
This should improve compile time for the very big basic blocks.
Differential Revision: https://reviews.llvm.org/D99266
Added getPointersDiff function to LoopAccessAnalysis and used it instead
direct calculatoin of the distance between pointers and/or
isConsecutiveAccess function in SLP vectorizer to improve compile time
and detection of stores consecutive chains.
Part of D57059
Differential Revision: https://reviews.llvm.org/D98967
Added getPointersDiff function to LoopAccessAnalysis and used it instead
direct calculatoin of the distance between pointers and/or
isConsecutiveAccess function in SLP vectorizer to improve compile time
and detection of stores consecutive chains.
Part of D57059
Differential Revision: https://reviews.llvm.org/D98967
As noted in D98152, we need to patch SLP to avoid regressions when
we start canonicalizing to integer min/max intrinsics.
Most of the real work to make this possible was in:
7202f47508
Differential Revision: https://reviews.llvm.org/D98981
Make sure we use PowerOf2Floor instead of PowerOf2Ceil when
calculating max number of elements that fits inside a vector
register (otherwise we could end up creating vectors larger
than the maximum vector register size).
Also make sure we honor the min/max VF (as given by TTI or
cmd line parameters) when doing vectorizeStores.
Reviewed By: anton-afanasyev
Differential Revision: https://reviews.llvm.org/D97691
If SLP vectorizer tries to extend the scheduling region and runs out of
the budget too early, but still extends the region to the new ending
instructions (i.e., it was able to extend the region for the first
instruction in the bundle, but not for the second), the compiler need to
recalculate dependecies in full, just like if the extending was
successfull. Without it, the schedule data chunks may end up with the
wrong number of (unscheduled) dependecies and it may end up with the
incorrect function, where the vectorized instruction does not dominate
on the extractelement instruction.
Differential Revision: https://reviews.llvm.org/D98531
Current SLP pass has this piece of code that inserts a trunc instruction
after the vectorized instruction. In the case that the vectorized instruction
is a phi node and not the last phi node in the BB, the trunc instruction
will be inserted between two phi nodes, which will trigger verify problem
in debug version or unpredictable error in another pass.
This patch changes the algorithm to 'if the last vectorized instruction
is a phi, insert it after the last phi node in current BB' to fix this problem.
Associative reduction matcher in SLP begins with select instruction but when
it reached call to llvm.umax (or alike) via def-use chain the latter also matched
as UMax kind. The routine's later code assumes matched instruction to be a select
and thus it merely died on the first encountered cast that did not fit.
Differential Revision: https://reviews.llvm.org/D98432
It is possible to merge reuse and reorder shuffles and reduce the total
cost of the vectorization tree/number of final instructions.
Differential Revision: https://reviews.llvm.org/D94992
getIntrinsicInstrCost takes a IntrinsicCostAttributes holding various
parameters of the intrinsic being costed. It can either be called with a
scalar intrinsic (RetTy==Scalar, VF==1), with a vector instruction
(RetTy==Vector, VF==1) or from the vectorizer with a scalar type and
vector width (RetTy==Scalar, VF>1). A RetTy==Vector, VF>1 is considered
an error. Both of the vector modes are expected to be treated the same,
but because this is confusing many backends end up getting it wrong.
Instead of trying work with those two values separately this removes the
VF parameter, widening the RetTy/ArgTys by VF used called from the
vectorizer. This keeps things simpler, but does require some other
modifications to keep things consistent.
Most backends look like this will be an improvement (or were not using
getIntrinsicInstrCost). AMDGPU needed the most changes to keep the code
from c230965ccf working. ARM removed the fix in
dfac521da1, webassembly happens to get a fixup for an SLP cost
issue and both X86 and AArch64 seem to now be using better costs from
the vectorizer.
Differential Revision: https://reviews.llvm.org/D95291
Pointer operand of scatter loads does not remain scalar in the tree (it
gest vectorized) and thus must not be marked as the scalar that remains
scalar in vectorized form.
Differential Revision: https://reviews.llvm.org/D96818
The vector reduction intrinsics started life as experimental ops, so backend support
was lacking. As part of promoting them to 1st-class intrinsics, however, codegen
support was added/improved:
D58015
D90247
So I think it is safe to now remove this complication from IR.
Note that we still have an IR-level codegen expansion pass for these as discussed
in D95690. Removing that is another step in simplifying the logic. Also note that
x86 was already unconditionally forming reductions in IR, so there should be no
difference for x86.
I spot checked a couple of the tests here by running them through opt+llc and did
not see any asm diffs.
If we do find functional differences for other targets, it should be possible
to (at least temporarily) restore the shuffle IR with the ExpandReductions IR
pass.
Differential Revision: https://reviews.llvm.org/D96552
This reverts commit 502a67dd7f.
This expose a failure in test-suite build on PowerPC,
revert to unblock buildbot first,
Dave will re-commit in https://reviews.llvm.org/D96287.
Thanks Dave.
getIntrinsicInstrCost takes a IntrinsicCostAttributes holding various
parameters of the intrinsic being costed. It can either be called with a
scalar intrinsic (RetTy==Scalar, VF==1), with a vector instruction
(RetTy==Vector, VF==1) or from the vectorizer with a scalar type and
vector width (RetTy==Scalar, VF>1). A RetTy==Vector, VF>1 is considered
an error. Both of the vector modes are expected to be treated the same,
but because this is confusing many backends end up getting it wrong.
Instead of trying work with those two values separately this removes the
VF parameter, widening the RetTy/ArgTys by VF used called from the
vectorizer. This keeps things simpler, but does require some other
modifications to keep things consistent.
Most backends look like this will be an improvement (or were not using
getIntrinsicInstrCost). AMDGPU needed the most changes to keep the code
from c230965ccf working. ARM removed the fix in
dfac521da1, webassembly happens to get a fixup for an SLP cost
issue and both X86 and AArch64 seem to now be using better costs from
the vectorizer.
Differential Revision: https://reviews.llvm.org/D95291
a6f0221276 enabled intersection of FMF on reduction instructions,
so it is safe to ease the check here.
There is still some room to improve here - it looks like we
have nearly duplicate flags propagation logic inside of the
LoopUtils helper but it is limited targets that do not form
reduction intrinsics (they form the shuffle expansion).
As shown in the test diffs, we could miscompile by
propagating flags that did not exist in the original
code.
The flags required for fmin/fmax reductions will be
fixed in a follow-up patch.
After much refactoring over the last 2 weeks to the reduction
matching code, I think this change is finally ready.
We effectively broke fmax/fmin vector reduction optimization
when we started canonicalizing to intrinsics in instcombine,
so this should restore that functionality for SLP.
There are still FMF problems here as noted in the code comments,
but we should be avoiding miscompiles on those for fmax/fmin by
restricting to full 'fast' ops (negative tests are included).
Fixing FMF propagation is a planned follow-up.
Differential Revision: https://reviews.llvm.org/D94913
In the spirit of commit fc783e91e0 (llvm-svn: 248943) we
shouldn't vectorize stores of non-packed types (i.e. types that
has padding between consecutive variables in a scalar layout,
but being packed in a vector layout).
The problem was detected as a miscompile in a downstream test case.
Reviewed By: anton-afanasyev
Differential Revision: https://reviews.llvm.org/D94446
We shouldn't vectorize stores of non-packed types (i.e. types that
has padding between consecutive variables in a scalar layout,
but being packed in a vector layout).
The problem was detected as a miscompile in a downstream test case.
This is a pre-commit of a test case for the fix in D94446.
After merging the shuffles, we cannot rely on the previous shuffle
anymore and need to shrink the final shuffle, if it is required.
Reported in D92668
Differential Revision: https://reviews.llvm.org/D93967
This patch makes SLP and LV emit operations with initial vectors set to poison constant instead of undef.
This is a part of efforts for using poison vector instead of undef to represent "doesn't care" vector.
The goal is to make nice shufflevector optimizations valid that is currently incorrect due to the tricky interaction between undef and poison (see https://bugs.llvm.org/show_bug.cgi?id=44185 ).
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D94061
Div/rem by zero is immediate undefined behavior and anything goes.
Currently we fold it to undef, this patch changes it to fold to
poison instead, which is slightly stronger.
Differential Revision: https://reviews.llvm.org/D93995
As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used
instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`.
Let's update them.
Actually, it would have been more natural if the patches were made in this order:
(1) let them use unary CreateShuffleVector first
(2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793)
The order is swapped, but in terms of correctness it is still fine.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D93923
This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93793
Currently undef is used as a don’t-care vector when constructing a vector using a series of insertelement.
However, this is problematic because undef isn’t undefined enough.
Especially, a sequence of insertelement can be optimized to shufflevector, but using undef as its placeholder makes shufflevector a poison-blocking instruction because undef cannot be optimized to poison.
This makes a few straightforward optimizations incorrect, such as:
```
; https://bugs.llvm.org/show_bug.cgi?id=44185
define <4 x float> @insert_not_undef_shuffle_translate_commute(float %x, <4 x float> %y, <4 x float> %q) {
%xv = insertelement <4 x float> %q, float %x, i32 2
%r = shufflevector <4 x float> %y, <4 x float> %xv, <4 x i32> { 0, 6, 2, undef }
ret <4 x float> %r ; %r[3] is undef
}
=>
define <4 x float> @insert_not_undef_shuffle_translate_commute(float %x, <4 x float> %y, <4 x float> %q) {
%r = insertelement <4 x float> %y, float %x, i32 1
ret <4 x float> %r ; %r[3] = %y[3], incorrect if %y[3] = poison
}
Transformation doesn't verify!
ERROR: Target is more poisonous than source
```
I’d like to suggest
1. Using poison as insertelement’s placeholder value (IRBuilder::CreateVectorSplat should be patched too)
2. Updating shufflevector’s semantics to return poison element if mask is undef
Note that poison is currently lowered into UNDEF in SelDag, so codegen part is okay.
m_Undef() matches PoisonValue as well, so existing optimizations will still fire.
The only concern is hidden miscompilations that will go incorrect when poison constant is given.
A conservative way is copying all tests having `insertelement undef` & replacing it with `insertelement poison` & run Alive2 on it, but it will create many tests and people won’t like it. :(
Instead, I’ll simply locally maintain the tests and run Alive2.
If there is any bug found, I’ll report it.
Relevant links: https://bugs.llvm.org/show_bug.cgi?id=43958 , http://lists.llvm.org/pipermail/llvm-dev/2019-November/137242.html
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93586
This commit copies existing tests at llvm/Transforms and replaces
'insertelement undef' in those files with 'insertelement poison'.
(see https://reviews.llvm.org/D93586)
Tests listed using this script:
grep -R -E '^[^;]*insertelement <.*> undef,' . | cut -d":" -f1 | uniq |
wc -l
Tests updated:
file_org=llvm/test/Transforms/$1
file=${file_org%.ll}-inseltpoison.ll
cp $file_org $file
sed -i -E 's/^([^;]*)insertelement <(.*)> undef/\1insertelement <\2> poison/g' $file
head -1 $file | grep "Assertions have been autogenerated by utils/update_test_checks.py" -q
if [ "$?" == 1 ]; then
echo "$file : should be manually updated"
# I manually updated the script
exit 1
fi
python3 ./llvm/utils/update_test_checks.py --opt-binary=./build-releaseassert/bin/opt $file
D82227 has added a proper check to limit PHI vectorization to the
maximum vector register size. That unfortunately resulted in at
least a couple of regressions on SystemZ and x86.
This change reverts PHI handling from D82227 and replaces it with
a more general check in SLPVectorizerPass::tryToVectorizeList().
Moved to tryToVectorizeList() it allows to restart vectorization
if initial chunk fails.
However, this function is more general and handles not only PHI
but everything which SLP handles. If vectorization factor would
be limited to maximum vector register size it would limit much
more vectorization than before leading to further regressions.
Therefore a new TTI callback getMaximumVF() is added with the
default 0 to preserve current behavior and limit nothing. Then
targets can decide what is better for them.
The callback gets ElementSize just like a similar getMinimumVF()
function and the main opcode of the chain. The latter is to avoid
regressions at least on the AMDGPU. We can have loads and stores
up to 128 bit wide, and <2 x 16> bit vector math on some
subtargets, where the rest shall not be vectorized. I.e. we need
to differentiate based on the element size and operation itself.
Differential Revision: https://reviews.llvm.org/D92059
Vector element size could be different for different store chains.
This patch prevents wrong computation of maximum number of elements
for that case.
Differential Revision: https://reviews.llvm.org/D93192
For stores chain vectorization we choose the size of vector
elements to ensure we fit to minimum and maximum vector register
size for the number of elements given. This patch corrects vector
element size choosing the width of value truncated just before
storing instead of the width of value stored.
Fixes PR46983
Differential Revision: https://reviews.llvm.org/D92824
It is possible to merge reuse and reorder shuffles and reduce the total
cost of the ivectorization tree/number of final instructions.
Differential Revision: https://reviews.llvm.org/D92668
This was modeled to have a cost of 1, but since we do not have a MUL.2d this is
scalarized into vector inserts/extracts and scalar muls.
Motivating precommitted test is test/Transforms/SLPVectorizer/AArch64/mul.ll,
which we don't want to SLP vectorize.
Test Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll
unfortunately needed changing, but the reason is documented in
LoopVectorize.cpp:6855:
// The cost of executing VF copies of the scalar instruction. This opcode
// is unknown. Assume that it is the same as 'mul'.
which I will address next as a follow up of this.
Differential Revision: https://reviews.llvm.org/D92208
Add a basic implementation of getGatherScatterOpCost to BasicTTIImpl.
The implementation estimates the cost of scalarizing the loads/stores,
the cost of packing/extracting the individual lanes and the cost of
only selecting enabled lanes.
This more accurately reflects the current cost on targets like AArch64.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D91984
For the scattered operands of load instructions it makes sense
to use gathering load intrinsic, which can lower to native instruction
for X86/AVX512 and ARM/SVE. This also enables building
vectorization tree with entries containing scattered operands.
The next step is to add scattered store.
Fixes PR47629 and PR47623
Differential Revision: https://reviews.llvm.org/D90445
Claim to not have any vector support to dissuade SLP, LV and friends
from generating SIMD IR for the VE target. We will take this back once
vector isel is stable.
Reviewed By: kaz7, fhahn
Differential Revision: https://reviews.llvm.org/D90462
This reverts the revert commit a1b53db324.
This patch includes a fix for a reported issue, caused by
matchSelectPattern returning UMIN for selects of pointers in
some cases by looking to some connected casts.
For now, ensure integer instrinsics are only returned for selects of
ints or int vectors.
This reverts commit 1922570489.
This appears to cause a crash in the following example
a, b, c;
l() {
int e = a, f = l, g, h, i, j;
float *d = c, *k = b;
for (;;)
for (; g < f; g++) {
k[h] = d[i];
k[h - 1] = d[j];
h += e << 1;
i += e;
}
}
clang -cc1 -triple i386-unknown-linux-gnu -emit-obj -target-cpu pentium-m -O1 -vectorize-loops -vectorize-slp reduced.c
llvm::Type *llvm::Type::getWithNewBitWidth(unsigned int) const: Assertion `isIntOrIntVectorTy() && "Original type expected to be a vector of integers or a scalar integer."' failed.
Use -0.0 instead of 0.0 as the start value. The previous use of 0.0
was fine for all existing uses of this function though, as it is
always generated with fast flags right now, and thus nsz.
Some architectures do not have general vector select instructions (e.g.
AArch64). But some cmp/select patterns can be vectorized using other
instructions/intrinsics.
One example is using min/max instructions for certain patterns.
This patch updates the cost calculations for selects in the SLP
vectorizer to consider using min/max intrinsics.
This patch does not change SLP vectorizer's codegen itself to actually
generate those intrinsics, but relies on the backends to lower the
vector cmps & selects. This keeps things simple on the SLP side and
works well in practice for AArch64.
This exposes additional SLP vectorization opportunities in some
benchmarks on AArch64 (-O3 -flto).
Metric: SLP.NumVectorInstructions
Program base slp diff
test-suite...ications/JM/ldecod/ldecod.test 502.00 697.00 38.8%
test-suite...ications/JM/lencod/lencod.test 1023.00 1414.00 38.2%
test-suite...-typeset/consumer-typeset.test 56.00 65.00 16.1%
test-suite...6/464.h264ref/464.h264ref.test 804.00 822.00 2.2%
test-suite...006/453.povray/453.povray.test 3335.00 3357.00 0.7%
test-suite...CFP2000/177.mesa/177.mesa.test 2110.00 2121.00 0.5%
test-suite...:: External/Povray/povray.test 2378.00 2382.00 0.2%
Reviewed By: RKSimon, samparker
Differential Revision: https://reviews.llvm.org/D89969
AArch64 does not have a flexible vector select instruction. In some
cases, the selects can be turned into min/max however, for which there
are dedicated vector instructions on AArch64.
This patch adds some tests for such cases.
D70365 allows us to make attributes default. This is a follow up to
actually make nosync, nofree and willreturn default. The approach we
chose, for now, is to opt-in to default attributes to avoid introducing
problems to target specific intrinsics. Intrinsics with default
attributes can be created using `DefaultAttrsIntrinsic` class.
If some leaves have the same instructions to be vectorized, we may
incorrectly evaluate the best order for the root node (it is built for the
vector of instructions without repeated instructions and, thus, has less
elements than the root node). In this case we just can not try to reorder
the tree + we may calculate the wrong number of nodes that requre the
same reordering.
For example, if the root node is \<a+b, a+c, a+d, f+e\>, then the leaves
are \<a, a, a, f\> and \<b, c, d, e\>. When we try to vectorize the first
leaf, it will be shrink to \<a, b\>. If instructions in this leaf should
be reordered, the best order will be \<1, 0\>. We need to extend this
order for the root node. For the root node this order should look like
\<3, 0, 1, 2\>. This patch allows extension of the orders of the nodes
with the reused instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D45263
If some leaves have the same instructions to be vectorized, we may
incorrectly evaluate the best order for the root node (it is built for the
vector of instructions without repeated instructions and, thus, has less
elements than the root node). In this case we just can not try to reorder
the tree + we may calculate the wrong number of nodes that requre the
same reordering.
For example, if the root node is \<a+b, a+c, a+d, f+e\>, then the leaves
are \<a, a, a, f\> and \<b, c, d, e\>. When we try to vectorize the first
leaf, it will be shrink to \<a, b\>. If instructions in this leaf should
be reordered, the best order will be \<1, 0\>. We need to extend this
order for the root node. For the root node this order should look like
\<3, 0, 1, 2\>. This patch allows extension of the orders of the nodes
with the reused instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D45263
This is one (small) part of improving PR41312:
https://llvm.org/PR41312
As shown there and in the smaller tests here, if we have some member of the
reduction values that does not match the others, we want to push it to the
end (bring the matching members forward and together).
In the regression tests, we have 5 candidates for the 4 slots of the reduction.
If the one "wrong" compare is grouped with the others, it prevents forming the
ideal v4i1 compare reduction.
Differential Revision: https://reviews.llvm.org/D87772
For scalable type, the aggregated size is unknown at compile-time.
Skip instructions with scalable type to ensure the list of instructions
for vectorizeSimpleInstructions does not contains any scalable-vector instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D87550
The test example based on PR47450 shows that we can
match non-byte-sized shifts, but those won't ever be
bswap opportunities. This isn't a full fix (we'd still
match if the shifts were by 8-bits for example), but
this should be enough until there's evidence that we
need to do more (this is a borderline case for
vectorization in the first place).
Previously we could match fcmp+select to a reduction if the fcmp had
the nonans fast math flag. But if the select had the nonans fast
math flag, InstCombine would turn it into a fminnum/fmaxnum intrinsic
before SLP gets to it. Seems fairly likely that if one of the
fcmp+select pair have the fast math flag, they both would.
My plan is to start vectorizing the fmaxnum/fminnum version soon,
but I wanted to get this code out as it had some of the strangest
fast math flag behaviors.
Other types can be handled in future patches but their uniform / non-uniform costs are more similar and don't appear to cause many vectorization issues.
The legacy SLPVectorizer has a dependency on InjectTLIMappingsLegacy.
That cannot be expressed in the new PM since they are both normal
passes. Explicitly add -inject-tli-mappings as a pass.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D86492
The 1st attempt (rG557b890) was reverted because it caused miscompiles.
That bug is avoided here by changing the order of folds and as verified
in the new tests.
Original commit message:
InstCombine currently has odd rules for folding insert-extract chains to shuffles,
so we miss collapsing seemingly simple cases as shown in the tests here.
But poison makes this not quite as easy as we might have guessed. Alive2 tests to
show the subtle difference (similar to the regression tests):
https://alive2.llvm.org/ce/z/hp4hv3 (this is ok)
https://alive2.llvm.org/ce/z/ehEWaN (poison leakage)
SLP tends to create these patterns (as shown in the SLP tests), and this could
help with solving PR16739.
Differential Revision: https://reviews.llvm.org/D86460
InstCombine currently has odd rules for folding insert-extract chains to shuffles,
so we miss collapsing seemingly simple cases as shown in the tests here.
But poison makes this not quite as easy as we might have guessed. Alive2 tests to
show the subtle difference (similar to the regression tests):
https://alive2.llvm.org/ce/z/hp4hv3 (this is ok)
https://alive2.llvm.org/ce/z/ehEWaN (poison leakage)
SLP tends to create these patterns (as shown in the SLP tests), and this could
help with solving PR16739.
Differential Revision: https://reviews.llvm.org/D86460
The legacy PM alias analysis pipeline by default includes basic-aa.
When running `opt -foo-pass` under the NPM and -disable-basic-aa is not
specified, use basic-aa.
This decreases the number of check-llvm failures under NPM from 913 to 752.
Reviewed By: ychen, asbirlea
Differential Revision: https://reviews.llvm.org/D86167
This reverts commit 52b71aa8b1.
The problem was a missing lit.local.cfg file, which was causing the
test to be incorrectly run on bots that had not built the WebAssembly
target.
8cc911fa5b refactored the `getIntrinsicInstrCost` function and was
meant to be a nonfunctional change, but it accidentally changed how
costs were calculated in the SLP vectorizer, which regressed
WebAssembly codegen and resulted in a downstream bug report at
https://github.com/emscripten-core/emscripten/issues/11449.
The fix for this regression is in D85759, and this patch just
pre-commits the test from that patch to demonstrate the regressed
behavior first.
The entries in VectorizableTree are not necessarily ordered by their
position in basic blocks. Collect them and order them by dominance so
later instructions are guaranteed to be visited first. For instructions
in different basic blocks, we only scan to the beginning of the block,
so their order does not matter, as long as all instructions in a basic
block are grouped together. Using dominance ensures a deterministic order.
The modified test case contains an example where we compute a wrong
spill cost (2) without this patch, even though there is no call between
any instruction in the bundle.
This seems to have limited practical impact, .e.g on X86 with a recent
Intel Xeon CPU with -O3 -march=native -flto on MultiSource,SPEC2000,SPEC2006
there are no binary changes.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D82444
Summary:
This patch takes the indices operands of `insertelement`/`insertvalue`
into account while generation of seed elements for `findBuildAggregate()`.
This function has kept the original order of `insert`s before.
Also this patch optimizes `findBuildAggregate()` preventing it from
redundant temporary vector allocations and its multiple reversing.
Fixes llvm.org/pr44067
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83779
In vectorizeChainsInBlock we try to collect chains of PHI nodes
that have the same element type, but the code is relying upon
the implicit conversion from TypeSize -> uint64_t. For now, I have
modified the code to ignore PHI nodes with scalable types.
Differential Revision: https://reviews.llvm.org/D83542
Teaches the SLPVectorizer to use vectorized library functions for
non-intrinsic calls.
This already worked for intrinsics that have vectorized library
functions, thanks to D75878, but schedules with library functions with a
vector variant were being rejected early.
- assume that there are no load/store dependencies between lib
functions with a vector variant; this would otherwise prevent the
bundle from becoming "ready"
- check during legalization that the vector variant can be used
- fix-up where we previously assumed that a call would be an intrinsic
Differential Revision: https://reviews.llvm.org/D82550
At the moment this place does not check maximum size set
by TTI and just creates a maximum possible vectors.
Differential Revision: https://reviews.llvm.org/D82227
The entries in VectorizableTree are not necessarily ordered by their
position in basic blocks. Collect them and order them by dominance so
later instructions are guaranteed to be visited first. For instructions
in different basic blocks, we only scan to the beginning of the block,
so their order does not matter, as long as all instructions in a basic
block are grouped together. Using dominance ensures a deterministic order.
The modified test case contains an example where we compute a wrong
spill cost (2) without this patch, even though there is no call between
any instruction in the bundle.
This seems to have limited practical impact, .e.g on X86 with a recent
Intel Xeon CPU with -O3 -march=native -flto on MultiSource,SPEC2000,SPEC2006
there are no binary changes.
Reviewers: craig.topper, RKSimon, xbolva00, ABataev, spatel
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D82444
D68667 introduced a tighter limit to the number of GEPs to simplify
together. The limit was based on the vector element size of the pointer,
but the pointers themselves are not actually put in vectors.
IIUC we try to vectorize the index computations here, so we should base
the limit on the vector element size of the computation of the index.
This restores the test regression on AArch64 and also restores the
vectorization for a important pattern in SPEC2006/464.h264ref on
AArch64 (@test_i16_extend). We get a large benefit from doing a single
load up front and then processing the index computations in vectors.
Note that we could probably even further improve the AArch64 codegen, if
we would do zexts to i32 instead of i64 for the sub operands and then do
a single vector sext on the result of the subtractions. AArch64 provides
dedicated vector instructions to do so. Sketch of proof in Alive:
https://alive2.llvm.org/ce/z/A4xYAB
Reviewers: craig.topper, RKSimon, xbolva00, ABataev, spatel
Reviewed By: ABataev, spatel
Differential Revision: https://reviews.llvm.org/D82418
Motivating examples are seen in the PhaseOrdering tests based on:
https://bugs.llvm.org/show_bug.cgi?id=43953#c2 - if we have
intrinsics there, some pass can fold them.
The intrinsics are still named "experimental" at this point, but
if there is no fallout from this patch, that will be a good
indicator that it is safe to finalize them.
Differential Revision: https://reviews.llvm.org/D80867
relevant aggregate build instructions only (UserCost).
Users are detected with findBuildAggregate routine and the trick is
that following SLP vectorization may end up vectorizing entire list
with smaller chunks. Cost adjustment then is applied for individual
chunks and these adjustments obviously have to be smaller than the
entire aggregate build cost.
Differential Revision: https://reviews.llvm.org/D80773
The bug is related to aggregate build cost model adjustment
that adds a bias to cost triggering vectorization of actually
unprofitable to vectorize tree.
Differential Revision: https://reviews.llvm.org/D80682
This test was failing verification because the
metadata is ill-formed. This commit is split
from D80401 because it is an independent fix
(although the test would break with that change).
Summary:
Replace any extant metadata uses of a dying instruction with undef to
preserve debug info accuracy. Some alternatives include:
- Treat Instruction like any other Value, and point its extant metadata
uses to an empty ValueAsMetadata node. This makes extant dbg.value uses
trivially dead (i.e. fair game for deletion in many passes), leading to
stale dbg.values being in effect for too long.
- Call salvageDebugInfoOrMarkUndef. Not needed to make instruction removal
correct. OTOH results in wasted work in some common cases (e.g. when all
instructions in a BasicBlock are deleted).
This came up while discussing some basic cases in
https://reviews.llvm.org/D80052.
Reviewers: jmorse, TWeaver, aprantl, dexonsmith, jdoerfert
Subscribers: jholewinski, qcolombet, hiraditya, jfb, sstefan1, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80264
This is D77454, except for stores. All the infrastructure work was done
for loads, so the remaining changes necessary are relatively small.
Differential Revision: https://reviews.llvm.org/D79968
For IR generated by a compiler, this is really simple: you just take the
datalayout from the beginning of the file, and apply it to all the IR
later in the file. For optimization testcases that don't care about the
datalayout, this is also really simple: we just use the default
datalayout.
The complexity here comes from the fact that some LLVM tools allow
overriding the datalayout: some tools have an explicit flag for this,
some tools will infer a datalayout based on the code generation target.
Supporting this properly required plumbing through a bunch of new
machinery: we want to allow overriding the datalayout after the
datalayout is parsed from the file, but before we use any information
from it. Therefore, IR/bitcode parsing now has a callback to allow tools
to compute the datalayout at the appropriate time.
Not sure if I covered all the LLVM tools that want to use the callback.
(clang? lli? Misc IR manipulation tools like llvm-link?). But this is at
least enough for all the LLVM regression tests, and IR without a
datalayout is not something frontends should generate.
This change had some sort of weird effects for certain CodeGen
regression tests: if the datalayout is overridden with a datalayout with
a different program or stack address space, we now parse IR based on the
overridden datalayout, instead of the one written in the file (or the
default one, if none is specified). This broke a few AVR tests, and one
AMDGPU test.
Outside the CodeGen tests I mentioned, the test changes are all just
fixing CHECK lines and moving around datalayout lines in weird places.
Differential Revision: https://reviews.llvm.org/D78403
Fix the assumption that all bitcasts of the same type sizes are free.
We now only assume that bitcasts between ints and ptrs of the same
size are free. This allows TTImpl to just call the concrete
implementation of getCastInstrCost.
Differential Revision: https://reviews.llvm.org/D78918
The original patch (rG86dfbc676ebe) exposed an existing bug:
we could wrongly cast a constant expression to BinaryOperator
because the pattern matching allows that. This adds a check
for that case, and there's a reduced test case to verify no
crashing.
Original commit message:
This builds on the or-reduction bailout that was added with D67841.
We still do not have IR-level load combining, although that could
be a target-specific enhancement for -vector-combiner.
The heuristic is narrowly defined to catch the motivating case from
PR39538:
https://bugs.llvm.org/show_bug.cgi?id=39538
...while preserving existing functionality.
That is, there's an unmodified test of pure load/zext/store that is
not seen in this patch at llvm/test/Transforms/SLPVectorizer/X86/cast.ll.
That's the reason for the logic difference to require the 'or'
instructions. The chances that vectorization would actually help a
memory-bound sequence like that seem small, but it looks nicer with:
vpmovzxwd (%rsi), %xmm0
vmovdqu %xmm0, (%rdi)
rather than:
movzwl (%rsi), %eax
movl %eax, (%rdi)
...
In the motivating test, we avoid creating a vector mess that is
unrecoverable in the backend, and SDAG forms the expected bswap
instructions after load combining:
movzbl (%rdi), %eax
vmovd %eax, %xmm0
movzbl 1(%rdi), %eax
vmovd %eax, %xmm1
movzbl 2(%rdi), %eax
vpinsrb $4, 4(%rdi), %xmm0, %xmm0
vpinsrb $8, 8(%rdi), %xmm0, %xmm0
vpinsrb $12, 12(%rdi), %xmm0, %xmm0
vmovd %eax, %xmm2
movzbl 3(%rdi), %eax
vpinsrb $1, 5(%rdi), %xmm1, %xmm1
vpinsrb $2, 9(%rdi), %xmm1, %xmm1
vpinsrb $3, 13(%rdi), %xmm1, %xmm1
vpslld $24, %xmm0, %xmm0
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpslld $16, %xmm1, %xmm1
vpor %xmm0, %xmm1, %xmm0
vpinsrb $1, 6(%rdi), %xmm2, %xmm1
vmovd %eax, %xmm2
vpinsrb $2, 10(%rdi), %xmm1, %xmm1
vpinsrb $3, 14(%rdi), %xmm1, %xmm1
vpinsrb $1, 7(%rdi), %xmm2, %xmm2
vpinsrb $2, 11(%rdi), %xmm2, %xmm2
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpinsrb $3, 15(%rdi), %xmm2, %xmm2
vpslld $8, %xmm1, %xmm1
vpmovzxbd %xmm2, %xmm2 # xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
vpor %xmm2, %xmm1, %xmm1
vpor %xmm1, %xmm0, %xmm0
vmovdqu %xmm0, (%rsi)
movl (%rdi), %eax
movl 4(%rdi), %ecx
movl 8(%rdi), %edx
movbel %eax, (%rsi)
movbel %ecx, 4(%rsi)
movl 12(%rdi), %ecx
movbel %edx, 8(%rsi)
movbel %ecx, 12(%rsi)
Differential Revision: https://reviews.llvm.org/D78997
This builds on the or-reduction bailout that was added with D67841.
We still do not have IR-level load combining, although that could
be a target-specific enhancement for -vector-combiner.
The heuristic is narrowly defined to catch the motivating case from
PR39538:
https://bugs.llvm.org/show_bug.cgi?id=39538
...while preserving existing functionality.
That is, there's an unmodified test of pure load/zext/store that is
not seen in this patch at llvm/test/Transforms/SLPVectorizer/X86/cast.ll.
That's the reason for the logic difference to require the 'or'
instructions. The chances that vectorization would actually help a
memory-bound sequence like that seem small, but it looks nicer with:
vpmovzxwd (%rsi), %xmm0
vmovdqu %xmm0, (%rdi)
rather than:
movzwl (%rsi), %eax
movl %eax, (%rdi)
...
In the motivating test, we avoid creating a vector mess that is
unrecoverable in the backend, and SDAG forms the expected bswap
instructions after load combining:
movzbl (%rdi), %eax
vmovd %eax, %xmm0
movzbl 1(%rdi), %eax
vmovd %eax, %xmm1
movzbl 2(%rdi), %eax
vpinsrb $4, 4(%rdi), %xmm0, %xmm0
vpinsrb $8, 8(%rdi), %xmm0, %xmm0
vpinsrb $12, 12(%rdi), %xmm0, %xmm0
vmovd %eax, %xmm2
movzbl 3(%rdi), %eax
vpinsrb $1, 5(%rdi), %xmm1, %xmm1
vpinsrb $2, 9(%rdi), %xmm1, %xmm1
vpinsrb $3, 13(%rdi), %xmm1, %xmm1
vpslld $24, %xmm0, %xmm0
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpslld $16, %xmm1, %xmm1
vpor %xmm0, %xmm1, %xmm0
vpinsrb $1, 6(%rdi), %xmm2, %xmm1
vmovd %eax, %xmm2
vpinsrb $2, 10(%rdi), %xmm1, %xmm1
vpinsrb $3, 14(%rdi), %xmm1, %xmm1
vpinsrb $1, 7(%rdi), %xmm2, %xmm2
vpinsrb $2, 11(%rdi), %xmm2, %xmm2
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpinsrb $3, 15(%rdi), %xmm2, %xmm2
vpslld $8, %xmm1, %xmm1
vpmovzxbd %xmm2, %xmm2 # xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
vpor %xmm2, %xmm1, %xmm1
vpor %xmm1, %xmm0, %xmm0
vmovdqu %xmm0, (%rsi)
movl (%rdi), %eax
movl 4(%rdi), %ecx
movl 8(%rdi), %edx
movbel %eax, (%rsi)
movbel %ecx, 4(%rsi)
movl 12(%rdi), %ecx
movbel %edx, 8(%rsi)
movbel %ecx, 12(%rsi)
Differential Revision: https://reviews.llvm.org/D78997
The improvements to the x86 vector insert/extract element costs in D74976 resulted in the estimated costs for vector initialization and scalarization increasing higher than should be expected. This is particularly noticeable on pre-SSE4 targets where the available of legal INSERT_VECTOR_ELT ops is more limited.
This patch does 2 things:
1 - it implements X86TTIImpl::getScalarizationOverhead to more accurately represent the typical costs of a ISD::BUILD_VECTOR pattern.
2 - it adds a DemandedElts mask to getScalarizationOverhead to permit the SLP's BoUpSLP::getGatherCost to be rewritten to use it directly instead of accumulating raw vector insertion costs.
This fixes PR45418 where a v4i8 (zext'd to v4i32) was no longer vectorizing.
A future patch should extend X86TTIImpl::getScalarizationOverhead to tweak the EXTRACT_VECTOR_ELT scalarization costs as well.
Reviewed By: @craig.topper
Differential Revision: https://reviews.llvm.org/D78216
Summary:
Currently, the internal options -vectorize-loops, -vectorize-slp, and
-interleave-loops do not have much practical effect. This is because
they are used to initialize the corresponding flags in the pass
managers, and those flags are then unconditionally overwritten when
compiling via clang or via LTO from the linkers. The only exception was
-vectorize-loops via opt because of some special hackery there.
While vectorization could still be disabled when compiling via clang,
using -fno-[slp-]vectorize, this meant that there was no way to disable
it when compiling in LTO mode via the linkers. This only affected
ThinLTO, since for regular LTO vectorization is done during the compile
step for scalability reasons. For ThinLTO it is invoked in the LTO
backends. See also the discussion on PR45434.
This patch makes it so the internal options can actually be used to
disable these optimizations. Ultimately, the best long term solution is
to mark the loops with metadata (similar to the approach used to fix
-fno-unroll-loops in D77058), but this enables a shorter term
workaround, and actually makes these internal options useful.
I constant propagated the initial values of these internal flags into
the pass manager flags (for some reasons vectorize-loops and
interleave-loops were initialized to true, while vectorize-slp was
initialized to false). As mentioned above, they are overwritten
unconditionally so this doesn't have any real impact, and these initial
values aren't particularly meaningful.
I then changed the passes to check the internl values and return without
performing the associated optimization when false (I changed the default
of -vectorize-slp to true so the options behave similarly). I was able
to remove the hackery in opt used to get -vectorize-loops=false to work,
as well as a special option there used to disable SLP vectorization.
Finally, I changed thinlto-slp-vectorize-pm.c to:
a) Only test SLP (moved the loop vectorization checking to a new test).
b) Use code that is slp vectorized when it is enabled, and check that
instead of whether the pass is enabled.
c) Test the new behavior of -vectorize-slp.
d) Test both pass managers.
The loop vectorization (and associated interleaving) testing I moved to
a new thinlto-loop-vectorize-pm.c test, with several changes:
a) Changed the flags on the interleaving testing so that it will
actually interleave, and check that.
b) Test the new behavior of -vectorize-loops and -interleave-loops.
c) Test both pass managers.
Reviewers: fhahn, wmi
Subscribers: hiraditya, steven_wu, dexonsmith, cfe-commits, davezarzycki, llvm-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77989
This is similar to what I recently did for getArithmeticReductionCost.
I'm trying to account for the narrowing from 512->256->128 as we go.
I've also added a new helper method getMinMaxCost that tries to
handle the cases where we have native min/max instructions and
fall back to cmp+select when we don't.
Differential Revision: https://reviews.llvm.org/D76634
There seems to be a small benefit to the legalized sequence for v2f16
round with packed instructions, so allow vectorizing it by reducing
the cost.
An unintended side effect is vectorization of f32 round also
happens. The current FMA logic seems off to me, and isn't checking for
packed instructions.
This patch attempts to more accurately model the reduction of
power of 2 vectors of types we natively support. This takes into
account the narrowing of vectors that occur as we go from 512
bits to 256 bits, to 128 bits. It also takes into account the use
of wider elements in the shuffles for the first 2 steps of a
reduction from 128 bits. And uses a v8i16 shift for the final step
of vXi8 reduction.
The default implementation uses the legalized type for the arithmetic
for all levels. And uses the single source permute cost of the
legalized type for all levels. This penalizes things like
lack of v16i8 pshufb on pre-sse3 targets and the splitting and
joining that needs to be done for integer types on AVX1. We never
need v16i8 shuffle for a reduction and we only need split AVX1 ops
when type the type wide and needs to be split. I think we're still
over costing splits and joins for AVX1, but we're closer now.
I've also removed all pairwise special casing because I don't
think we ever want to generate that on X86. I've also adjusted
the add handling to more accurately account for any type splitting
that occurs before we reach a legal type.
Differential Revision: https://reviews.llvm.org/D76478
Summary:
SLPVectorizer try to vectorize list of scalar instructions of the same type,
instructions already vectorized are rejected through isValidElementType().
Without this patch, tryToVectorizeList() will first try to determine vectorization
factor of a list of Instructions before checking whether each instruction has unsupported
type or not. For instructions already vectorized for SVE, it will crash at getVectorElementSize(),
where it try to return a fixed size.
This patch make sure invalid element types are rejected before trying to get vectorization
factor. This make sure we are not trying to vectorize instructions already vectorized.
Reviewers: sdesmalen, efriedma, spatel, RKSimon, ABataev, apazos, rengolin
Reviewed By: efriedma
Subscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D76017
It seems like the SLPVectorizer is currently not aware of vector
versions of functions provided by libraries like Accelerate [1].
This patch updates SLPVectorizer to use the same infrastructure
the LoopVectorizer uses to detect vectorizable library functions.
For calls, it computes the cost of an intrinsic call (existing behavior)
and the cost of a vector function library call, if available. Like
LoopVectorizer, it assumes the cost of the vector function is simply the
cost of a call to a vector function.
[1] https://developer.apple.com/documentation/accelerate
Reviewers: ABataev, RKSimon, spatel
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D75878
This tries to improve the accuracy of extract/insert element costs by accounting for subvector extraction/insertion for >128-bit vectors and the shuffling of elements to/from the 0'th index.
It also adds INSERTPS for f32 types and PINSR/PEXTR costs for integer types (at the moment we assume the same cost as MOVD/MOVQ - which isn't always true).
Differential Revision: https://reviews.llvm.org/D74976
D74976 will handle larger vector types, but since SLM doesn't support AVX+ then we will always be extracting from 128-bit vectors so don't need to scale the cost.
The index of an ExtractElementInst is not guaranteed to be a
ConstantInt. It can be any integer value. Check explicitly for
ConstantInts.
The new test cases illustrate scenarios where we crash without
this patch. I've also added another test case to check the matching
of extractelement vector ops works.
Reviewers: RKSimon, ABataev, dtemirbulatov, vporpo
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D74758
We may calculate reassociable math ops in arbitrary order when creating a shuffle reduction,
so there's no guarantee that things like 'nsw' hold on those intermediate values. Drop all
poison-generating flags for safety.
This change is limited to shuffle reductions because I don't think we have a problem in the
general case (where we intersect flags of each scalar op that goes into a vector op), but if
there's evidence of other cases being wrong, we can extend this fix to cover those cases.
https://bugs.llvm.org/show_bug.cgi?id=44536
Differential Revision: https://reviews.llvm.org/D73727
Summary:
We don't have control/verify what will be the RHS of the division, so it might
happen to be zero, causing UB.
Reviewers: Vasilis, RKSimon, ABataev
Reviewed By: ABataev
Subscribers: vporpo, ABataev, hiraditya, llvm-commits, vdmitrie
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72740
Summary:
Make SLPVectorize to recognize homogeneous aggregates like
`{<2 x float>, <2 x float>}`, `{{float, float}, {float, float}}`,
`[2 x {float, float}]` and so on.
It's a follow-up of https://reviews.llvm.org/D70068.
Merged `findBuildVector()` and `findBuildAggregate()` to
one `findBuildAggregate()` function making it recursive
to recognize multidimensional aggregates. Aggregates required
to be homogeneous.
Reviewers: RKSimon, ABataev, dtemirbulatov, spatel, vporpo
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70587
I'm not sure what the effect of this change will be on all of the affected
tests or a larger benchmark, but it fixes the horizontal add/sub problems
noted here:
https://reviews.llvm.org/D59710?vs=227972&id=228095&whitespace=ignore-most#toc
The costs are based on reciprocal throughput numbers in Agner's tables for
PEXTR*; these appear to be very slow ops on Silvermont.
This is a small step towards the larger motivation discussed in PR43605:
https://bugs.llvm.org/show_bug.cgi?id=43605
Also, it seems likely that insert/extract is the source of perf regressions on
other CPUs (up to 30%) that were cited as part of the reason to revert D59710,
so maybe we'll extend the table-based approach to other subtargets.
Differential Revision: https://reviews.llvm.org/D70607
Summary:
Vector aggregate is homogeneous aggregate of vectors like `{ <2 x float>, <2 x float> }`.
This patch allows `findBuildAggregate()` to consider vector aggregates as
well as scalar ones. For instance, `{ <2 x float>, <2 x float> }` maps to `<4 x float>`.
Fixes vector part of llvm.org/PR42022
Reviewers: RKSimon
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70068
After speaking with Sanjay - seeing a number of miscompiles and working
on tracking down a testcase. None of the follow on patches seem to
have helped so far.
This reverts commit 8a0aa5310b.
Alexey Bataev