When vectorising for AArch64 targets if you specify the SVE attribute
we automatically then treat masked loads and stores as legal. Also,
since we have no cost model for masked memory ops we believe it's
cheap to use the masked load/store intrinsics even for fixed width
vectors. This can lead to poor code quality as the intrinsics will
currently be scalarised in the backend. This patch adds a basic
cost model that marks fixed-width masked memory ops as significantly
more expensive than for scalable vectors.
Tests for the cost model are added here:
Transforms/LoopVectorize/AArch64/masked-op-cost.ll
Differential Revision: https://reviews.llvm.org/D100745
`X86TTIImpl::getInterleavedMemoryOpCostAVX2()` currently contains data
only for a handful of tuples. For now, at least add tests for a few more.
I'm guessing that we care how well the patterns codegen since
we use their presumed cost for vectorization decisions,
so i've added codegen tests too.
There's one really easy caveat for these codegen tests:
for interleaved load tests, we really have to ensure that the
deinterleaved vectors are escaped separately. Similarly for stores.
In cases when ScalarizationCostPassed has no value, UINT_MAX is actually used
for cost estimation in `return ScalarCalls * ScalarCost + ScalarizationCost`.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D101099
We were missing some instruction costs when converting vectors of
floating point half types into integers, so I've added those here.
I also manually generated assembly code for each FP->int case and
looked at the number of instructions generated, which meant
adjusting some of the existing costs too.
I've updated an existing test to reflect the new costs:
Analysis/CostModel/AArch64/sve-fptoi.ll
Differential Revision: https://reviews.llvm.org/D99935
Introduced the cost of thre reverse shuffles for AArch64, currently just
copied the costs for PermuteSingleSrc.
Differential Revision: https://reviews.llvm.org/D100871
This is similar to the subvector extractions,
except that the 0'th subvector isn't free to insert,
because we generally don't know whether or not
the upper elements need to be preserved:
https://godbolt.org/z/rsxP5W4sW
This is needed to avoid regressions in D100684
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D100698
Test SSE41, since that added float/i64/i32/i8 inserts/extracts.
Don't forget to test vectors of pointers.
Do test byte-aligned loads/stores.
Fixup test coverage to be rather more exhaustive,
testing all reasonable element sizes vs element counts permutations
that fit up to witin ZMM.
Sometimes LV has to produce really wide vectors,
and sometimes they end up being not powers of two.
As it can be seen from the diff, the cost computation
is currently completely non-sensical in those cases.
Instead of just scalarizing everything, split/factorize the wide vector
into a number of subvectors, each one having a power-of-two elements,
recurse to get the cost of op on this subvector. Also, check how we'd
legalize this subvector, and if the legalized type is scalar,
also account for the scalarization cost.
Note that for sub-vector loads, we might be able to do better,
when the vectors are properly aligned.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D100099
At the moment, getMemoryOpCost returns 1 for all inputs if CostKind is
CodeSize or SizeAndLatency. This fools LoopUnroll into thinking memory
operations on large vectors have a cost of one, even if they will get
expanded to a large number of memory operations in the backend.
This patch updates getMemoryOpCost to return the cost for the type
legalization for both CodeSize and SizeAndLatency. This should more
accurately reflect the number of memory operations required.
I am not sure how latency should properly be included in SizeAndLatency
from the description, but returning the size cost should be clearly more
accurate.
This does not cause any binary changes when building
MultiSource/SPEC2000/SPEC2006 with -O3 -flto for AArch64, likely because
large vector memops are not really formed by code emitted from Clang.
But using the C/C++ matrix extension can easily result in code with very
large vector operations directly from Clang, e.g.
https://clang.godbolt.org/z/6xzxcTGvb
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D100291
Added cost estimation for switch instruction, updated costs of branches, fixed
phi cost.
Had to increase `-amdgpu-unroll-threshold-if` default value since conditional
branch cost (size) was corrected to higher value.
Test renamed to "control-flow.ll".
Removed redundant code in `X86TTIImpl::getCFInstrCost()` and
`PPCTTIImpl::getCFInstrCost()`.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D96805
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.
After rG47321c311bdbe0145b9bf45d822185c37b19fa50 we promote vXi8 reductions to vXi16 to create a much faster PMULLW mul reduction, followed by a (free) truncation. This avoids the high cost of repeated vXi8 multiplications (which extend+multiply+truncate to/from vXi16 types....).
Fixes the missing vXi8 mul reduction vectorization in PR42674 (Comment #20) 'mul16' test case.
This marks FSIN and other operations to EXPAND for scalable
vectors, so that they are not assumed to be legal by the cost-model.
Depends on D97470
Reviewed By: dmgreen, paulwalker-arm
Differential Revision: https://reviews.llvm.org/D97471
Let getIntrinsicInstrCost call getTypeBasedIntrinsicInstrCost for scalable vectors,
similar to how this is done for fixed-width vectors, instead of falling back
on BaseT::getIntrinsicInstrCost().
If the intrinsic cannot be costed (or is not overloaded by the target),
it will return InstructionCost::getInvalid() instead.
Depends on D97469
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D97470
We previously made a change to getUserCost to return a Invalid cost
when one of the TTI costs returned '-1' (meaning 'unknown' or
'infinitely expensive'). It makes no sense to say that:
shufflevector <2 x i8> %x, <2 x i8> %y, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
has an invalid cost. Perhaps the cost is not known, but the IR is valid
and can be code-generated. Invalid should only be used for IR that
cannot possibly be code-generated and where a cost is nonsensical.
With more passes now asserting that the cost must be valid, it is possible
that those assertions will fail for perfectly valid IR. An incomplete
cost-model probably shouldn't be a reason for the compiler to break.
It's better to consider these costs as 'very expensive' and ignore them
for other reasons. At some point, we should consider replacing -1 with
some other mechanism.
Reviewed By: paulwalker-arm, dmgreen
Differential Revision: https://reviews.llvm.org/D99502
The following operations have no associated cost for them
when applied to scalable vectors, and as a consequence
can trigger a crash when a call is made to
AArch64TTIImpl::getCastInstrCost():
- fptrunc
- trunc
- fpext
- fpto(u,s)i
This patch adds costs for these operations and
relevant regression tests.
Differential Revision: https://reviews.llvm.org/D98934
getMinRVVVectorSizeInBits() asserts if the V extension isn't
enabled. So check that gather/scatter is legal first since it
already contains a check for V extension being enabled. It
also already checks getMinRVVVectorSizeInBits for fixed length
vectors so we don't need a check in getGatherScatterOpCost.
This patch adds a new llvm.experimental.stepvector intrinsic,
which takes no arguments and returns a linear integer sequence of
values of the form <0, 1, ...>. It is primarily intended for
scalable vectors, although it will work for fixed width vectors
too. It is intended that later patches will make use of this
new intrinsic when vectorising induction variables, currently only
supported for fixed width. I've added a new CreateStepVector
method to the IRBuilder, which will generate a call to this
intrinsic for scalable vectors and fall back on creating a
ConstantVector for fixed width.
For scalable vectors this intrinsic is lowered to a new ISD node
called STEP_VECTOR, which takes a single constant integer argument
as the step. During lowering this argument is set to a value of 1.
The reason for this additional argument at the codegen level is
because in future patches we will introduce various generic DAG
combines such as
mul step_vector(1), 2 -> step_vector(2)
add step_vector(1), step_vector(1) -> step_vector(2)
shl step_vector(1), 1 -> step_vector(2)
etc.
that encourage a canonical format for all targets. This hopefully
means all other targets supporting scalable vectors can benefit
from this too.
I've added cost model tests for both fixed width and scalable
vectors:
llvm/test/Analysis/CostModel/AArch64/neon-stepvector.ll
llvm/test/Analysis/CostModel/AArch64/sve-stepvector.ll
as well as codegen lowering tests for fixed width and scalable
vectors:
llvm/test/CodeGen/AArch64/neon-stepvector.ll
llvm/test/CodeGen/AArch64/sve-stepvector.ll
See this thread for discussion of the intrinsic:
https://lists.llvm.org/pipermail/llvm-dev/2021-January/147943.html
The scalarization overhead was set deliberately high for MVE, whilst the
codegen was new. It helps protect us against the negative ramifications
of mixing scalar and vector instructions. This decreases that,
especially for floating point where the cost of extracting/inserting
lane elements can be low. For integer the cost is still fairly high due
to the cross-register-bank copy, but is no longer n^2 in the length of
the vector.
In general, this will decrease the cost of scalarizing floats and long
integer vectors. i64 increase in cost, having a high cost before and
after this patch. For floats this allows up to start doing things like
vectorizing fdiv instructions, even if they are scalarized.
Differential Revision: https://reviews.llvm.org/D98245
The generic cost of logical or/and reductions should be cost of bitcast
<ReduxWidth x i1> to iReduxWidth + cmp eq|ne iReduxWidth.
Differential Revision: https://reviews.llvm.org/D97961
This uses the shuffle mask cost from D98206 to give a better cost of MVE
VREV instructions. This helps especially in VectorCombine where the cost
of shuffles is used to reorder bitcasts, which this helps keep the phase
ordering test for fp16 reductions producing optimal code. The isVREVMask
has been moved to a header file to allow it to be used across target
transform and isel lowering.
Differential Revision: https://reviews.llvm.org/D98210
This patch adds the cost model for experimental.vector.reverse
with scalable vector types: nxv16i1, nxv8i1, nxv4i1 and nxv2i1.
These types are missing from the previous cost model patch D95603.
The cost model for experimental.vector.reverse with 1 bit mask is used by
loop vectorization in the patch D95363
Differential Revision: https://reviews.llvm.org/D97758
This is a patch that updates the cost of `select i1 a, b, false` to be equivalent to that of `and i1 a, b`
as well as the cost of `select i1 a, true, b` equivalent to `or i1 a, b`.
Until now, these selects were folded into and/or i1 by InstCombine, but the transformation is poison-unsafe.
This is a step towards removing the unsafe transformation. D93065 has relevant transformations linked.
These selects should be translated into the assemblies as and/or i1 do in the same manner. The cost should be equivalent.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D97360
Moved some of the `sve-getIntrinsicCost-<..>` into a single sve-intrinsics.ll
file, and simplified the tests a bit by bundling all the intrinsics in one
function (instead of testing one intrinsic per function). That makes it easier
to see the cost of the intrinsics.
As a followup to D95291, getOperandsScalarizationOverhead was still
using a VF as a vector factor if the arguments were scalar, and would
assert on certain matrix intrinsics with differently sized vector
arguments. This patch removes the VF arg, instead passing the Types
through directly. This should allow it to more accurately compute the
cost without having to guess at which operands will be vectorized,
something difficult with more complex intrinsics.
This adjusts one SVE test as it is now calling the wrong intrinsic vs
veccall. Without invalid InstructCosts the cost of the scalarized
intrinsic is too low. This should get fixed when the cost of
scalarization is accounted for with scalable types.
Differential Revision: https://reviews.llvm.org/D96287
It appears that pointer types were causing issues for the min/max cost
code in getIntrinsicInstrCost. This makes sure that when matching
icmp/select to a min/max, we only do that for normal int or float types.
A v8i32 compare will produce a v8i1 predicate, but during codegen the
v8i32 will be split into two v4i32, potentially requiring two v4i1
predicates to be merged into a single v8i1. Because this merging of two
v4i1's into a v8i1 is very expensive, we need to make the cost of the
compare equally high.
This patch adds the cost of that to ARMTTIImpl::getCmpSelInstrCost.
Because we don't know whether the user of the predicate can be split,
and the cost model is mostly pre-instruction, we may be pessimistic but
that should only be for larger and legal types. This also adds min/max
detection to the costmodel where it can be detected, to keep those in
line with the cost of simple min/max instructions. Otherwise for the
most part, costs that were already expensive have become more expensive.
Differential Revision: https://reviews.llvm.org/D96692
This adds basic MVE costs for SMIN/SMAX/UMIN/UMAX, as well as MINNUM and
MAXNUM representing fmin and fmax. It tightens up the costs, not using a
ICmp+Select cost.
Differential Revision: https://reviews.llvm.org/D96603
This patch uses the function getShuffleCost with SK_Reverse to compute the cost
for experimental.vector.reverse.
For scalable vector type, it adds a table will the legal types on
AArch64TTIImpl::getShuffleCost to not assert in BasicTTIImpl::getShuffleCost,
and for fixed vector, it relies on the existing cost model in BasicTTIImpl.
Depends on D94883
Differential Revision: https://reviews.llvm.org/D95603
This fixes an overly restrictive assumption that the vector is a FixedVectorType,
in code that tries to calculate the cost of a cast operation when splitting
a too-wide vector. The algorithm works the same for scalable vectors, so this
patch removes the cast<FixedVectorType>.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D96253
COST(zext (<4 x i32> load(...) to <4 x i64>)) != 0 when
<4 x i64> is an illegal result type that requires splitting
of the operation.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D96250
This adds the CostKind to getMVEVectorCostFactor, so that it can
automatically account for CodeSize costs, where it returns a cost of 1
not the MVEFactor used for Throughput/Latency. This helps simplify the
caller code and allows us to get the codesize cost more correct in more
cases.
This patch adds a cost model for SK_Broadcast in
AArch64TTIImpl::getShuffleCost with scalable vector.
Without this patch, the scalable vector type relies on BasicTTIImpl cost
implementation and assert.
Differential Revision: https://reviews.llvm.org/D95598
This adds sadd.sat, uadd.sat, ssub.sat and usub.sat costs for AArch64,
similar to how they were recently added for ARM.
Differential Revision: https://reviews.llvm.org/D95292
I have removed an unnecessary assert in LoopVectorizationCostModel::getInstructionCost
that prevented a cost being calculated for select instructions when using
scalable vectors. In addition, I have changed AArch64TTIImpl::getCmpSelInstrCost
to only do special cost calculations for fixed width vectors and fall
back to the base version for scalable vectors.
I have added a simple cost model test for cmps and selects:
test/Analysis/CostModel/sve-cmpsel.ll
and some simple tests that show we vectorize loops with cmp and select:
test/Transforms/LoopVectorize/AArch64/sve-basic-vec.ll
Differential Revision: https://reviews.llvm.org/D95039
Just like llvm.assume, there are a lot of cases where we can just ignore llvm.experimental.noalias.scope.decl.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93042
We have no lowering for VSELECT vXi1, vXi1, vXi1, so mark them as
expanded to turn them into a series of logical operations.
Differential Revision: https://reviews.llvm.org/D94946
This adds some basic MVE sadd_sat/ssub_sat/uadd_sat/usub_sat costs,
based on when the instruction is legal. With smaller than legal types
that are promoted we generate shr(qadd(shl, shl)), so the cost is 4
appropriately.
Differential Revision: https://reviews.llvm.org/D94958
This patch computes the cost for vector.reduce<operand> for scalable vectors.
The cost is split into two parts: the legalization cost and the horizontal
reduction.
Differential Revision: https://reviews.llvm.org/D93639
We did not have specific costs for larger than legal truncates that were
not otherwise cheap (where they were next to stores, for example). As
MVE does not have a dedicated instruction for them (and we do not use
loads/stores yet), they should be expensive as they get expanded to a
series of lane moves.
Differential Revision: https://reviews.llvm.org/D94260
This patch fixes a bug introduced in the patch:
https://reviews.llvm.org/D93030
This patch pulls the test for scalable vector to be the first instruction
to be checked. This avoids the Gather and Scatter cost model for AArch64 to
compute the number of vector elements for something that is not a vector and
therefore crashing.
A new TTI interface has been added 'Optional <unsigned>getMaxVScale' that
returns the maximum vscale for a given target.
When known getMaxVScale is used to compute the cost of masked gather scatter
for scalable vector.
Depends on D92094
Differential Revision: https://reviews.llvm.org/D93030
Adds cost model support for the new llvm.experimental.vector.{extract,insert}
intrinsics, using the existing getExtractSubvectorOverhead and
getInsertSubvectorOverhead functions for shuffles.
Previously this case would throw an assertion.
Differential Revision: https://reviews.llvm.org/D93043
This patch replaces FixedVectorType by VectorType in getIntrinsicInstrCost
in BasicTTIImpl.h. It re-arranges the scalable type test earlier return
and add tests for scalable types.
Depends on D91532
Differential Revision: https://reviews.llvm.org/D92094
This adds some basic MVE masked load/store costs, notably changing the
cost of legal loads/stores to the MVECostFactor and the cost of
scalarized instructions to 8*NumElts.
Differential Revision: https://reviews.llvm.org/D86538
Noticed while looking at D92701 - we only really handle TCK_RecipThroughput gather/scatter costs - for now drop back to the default implementation for non-legal gathers/scatters.
Without FMF, we lower these intrinsics into something like this:
vmaxsd %xmm0, %xmm1, %xmm2
vcmpunordsd %xmm0, %xmm0, %xmm0
vblendvpd %xmm0, %xmm1, %xmm2, %xmm0
But if we can ignore NANs, the single min/max instruction is enough
because there is no need to fix up the x86 logic that corresponds to
X > Y ? X : Y.
We probably want to make other adjustments for FP intrinsics with FMF
to account for specialized codegen (for example, FSQRT).
Differential Revision: https://reviews.llvm.org/D92337
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
If usubsat() is legal, this is likely to result in smaller codegen expansion than the default cmp+select codegen expansion.
Allows us to move the x86-specific lowering to the generic expansion code.
Differential Revision: https://reviews.llvm.org/D92183
The cost-model is not getting the cost right for a mul with <2 x i64>
operands, i.e. we don't have a MUL.2d, and this is precommitting some
tests before adjusting this.
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
Update costs now that D92095 and D92102 have tweaked the SSE2 implementation
The SSE42 BLENDVPD cost can actually be used on SSE41 as we don't attempt to generate PCMPGT anymore
Add scalar i16/i32/i64 costs as we can do this cheaply with CMOV
This might be a regression for some ARM targets, but that should
be changed in the target-specific overrides.
There is apparently still no default lowering for these nodes,
so I am assuming these intrinsics are not in common use.
X86, PowerPC, and RISC-V for example, just crash given the most
basic IR.
This is re-applying a combination of f7eac51b9b and 8ec7ea3ddc as one patch
to avoid regressions now that we have better testing in place.
Those were reverted with 32dd5870ee because of crashing in experimental intrinsics.
That bug should be fixed with 7ae346434.
Paraphrased original commit messages:
This is the last step in removing cost-kind as a consideration in the
basic class model for intrinsics.
See D89461 for the start of that.
Subsequent commits dealt with each of the special-case intrinsics that
had customization here in the basic class. This should remove a barrier
to retrying D87188 (canonicalization to the abs intrinsic).
The ARM and x86 cost diffs seen here may be wrong because the
target-specific overrides have their own bugs, but we hope this is
less wrong - if something has a significant throughput cost, then it
should have a significant size / blended cost too by default.
The only behavioral diff in current regression tests is shown in the
x86 scatter-gather test (which is misplaced or broken because it runs
the entire -O3 pipeline) - we unrolled less, and we assume that is
a improvement.
Exception: in general, we want the *size* cost for a scalar call to be
cheap even if the other costs are expensive - we expect it to just be
a branch with some optional stack manipulation.
It is likely that we will want to carve out some
exceptions/overrides to this rule as follow-up patches for
calls that have some general and/or target-specific difference
to the expected lowering.
This was noticed as a regression in unrolling, so we have a test
for that now along with a couple of direct cost model tests.
If the assumed scalarization costs for the oversized vector
calls are not realistic, that would be another follow-up
refinement of the cost models.
Differential Revision: https://reviews.llvm.org/D90554
The constrained intrinsics have metadata arguments, so the
tests here were crashing as noted in D90554 (and that was
reverted even though this bug exists independently of that
change).
This is a partial un-revert of 32dd5870ee (originally df09f82599 ).
I'm adding back the baseline tests first, so we don't have
to back-track as much in case there are still problems.
as it's causing crashes in the optimizer. A reduced testcase has been posted as a follow-up.
This reverts commit f7eac51b9b.
Temporarily Revert "[CostModel] make default size cost for libcalls small (again)" as it depends upon the primary revert.
This reverts commit 8ec7ea3ddc.
Temporarily Revert "[CostModel] add tests for math library calls; NFC" as it depends upon the primary revert.
This reverts commit df09f82599.
Temporarily Revert "[LoopUnroll] add test for full unroll that is sensitive to cost-model; NFC" as it depends upon the primary revert.
This reverts commit 618d555e8d.
We can use GF2P8AFFINEQB to reverse bits in a byte. Shuffles are needed to reverse the bytes in elements larger than i8. LegalizeVectorOps takes care of inserting the shuffle for the larger element size.
We already have Custom lowering for v16i8 with SSSE3, v32i8 with AVX, and v64i8 with AVX512BW.
I think we might be able to use this for scalars too by moving into a vector and back. But I'll save that for a follow up as its a little more involved.
Reviewed By: RKSimon, pengfei
Differential Revision: https://reviews.llvm.org/D91515
This patch fixes the function isWideningInstruction for scalable vectors.
Now the cost model can check the widening pattern for SVE.
Differential Revision: https://reviews.llvm.org/D91260
This was changed recently with D90554 / f7eac51b9b
...because we had a regression testing blindspot for intrinsics
that are expected to be lowered to libcalls.
In general, we want the *size* cost for a scalar call to be cheap
even if the other costs are expensive - we expect it to just be
a branch with some optional stack manipulation.
It is likely that we will want to carve out some
exceptions/overrides to this rule as follow-up patches for
calls that have some general and/or target-specific difference
to the expected lowering.
This was noticed as a regression in unrolling, so we have a test
for that now along with a couple of direct cost model tests.
If the assumed scalarization costs for the oversized vector
calls are not realistic, that would be another follow-up
refinement of the cost models.
This is the last step in removing cost-kind as a consideration in the basic class model for intrinsics.
See D89461 for the start of that.
Subsequent commits dealt with each of the special-case intrinsics that had customization here in the
basic class. This should remove a barrier to retrying
D87188 (canonicalization to the abs intrinsic).
The ARM and x86 cost diffs seen here may be wrong because the target-specific overrides have their own
bugs, but we hope this is less wrong - if something has a significant throughput cost, then it should
have a significant size / blended cost too by default.
The only behavioral diff in current regression tests is shown in the x86 scatter-gather test (which is
misplaced or broken because it runs the entire -O3 pipeline) - we unrolled less, and we assume that is
a improvement.
Differential Revision: https://reviews.llvm.org/D90554
This is the cmp/sel sibling to D90692.
Again, the reasoning is: the throughput cost is number of instructions/uops,
so size/blended costs are identical except in special cases (for example,
fdiv or other known-expensive machine instructions or things like MVE that
may require cracking into >1 uops).
We need to check for a valid (non-null) condition type parameter because
SimplifyCFG may pass nullptr for that (and so we will crash multiple
regression tests without that check). I'm not sure if passing nullptr makes
sense, but other code in the cost model does appear to check if that param
is set or not.
Differential Revision: https://reviews.llvm.org/D90781
This is based on the same idea that I am using for the basic model implementation
and what I have partly already done for x86: throughput cost is number of
instructions/uops, so size/blended costs are identical except in special cases
(for example, fdiv or other known-expensive machine instructions or things like
MVE that may require cracking into >1 uop)).
Differential Revision: https://reviews.llvm.org/D90692
As noted in D90554, there's an opcode typo in using an easily
misused cost model API: getCmpSelInstrCost(). Beyond that, the
assumed sequence of ops is questionable, but that would be
another patch.
My guess is that the x86 test diffs show that we are probably
wrong both before and after this change, so there will be no
practical difference.
As an example, I tried this test which shows a cost of '7'
either way:
define <4 x i32> @sadd(<4 x i32> %va, <4 x i32> %vb) {
%V4I32 = call {<4 x i32>, <4 x i1>} @llvm.sadd.with.overflow.v4i32(<4 x i32> %va, <4 x i32> %vb)
%ov = extractvalue {<4 x i32>, <4 x i1>} %V4I32, 1
%r = extractvalue {<4 x i32>, <4 x i1>} %V4I32, 0
%z = select <4 x i1> %ov, <4 x i32> <i32 42, i32 42, i32 42, i32 42>, <4 x i32> %r
ret <4 x i32> %z
}
$ llc -o - sadd.ll -mattr=avx
vpaddd %xmm1, %xmm0, %xmm2
vpcmpgtd %xmm2, %xmm0, %xmm0
vpxor %xmm0, %xmm1, %xmm0
vblendvps %xmm0, LCPI0_0(%rip), %xmm2, %xmm0a
Differential Revision: https://reviews.llvm.org/D90681
This patch changes the intrinsics cost model to assume that by default
target intrinsics are cheap. This didn't seem to be the case for all
intrinsics, and is potentially an MVE problem due to our scalarization
overheads. Cheap seems to be a good default in general though.
Differential Revision: https://reviews.llvm.org/D90597
As noticed in D90554 ,
the AVX2 costs for 256-bit vectors did not include FMAXNUM entries,
so we fell back to AVX1 which assumes those ops will be split into
128-bit halves or something close to that.
Differential Revision: https://reviews.llvm.org/D90613
This reverts the revert commit 408c4408fa.
This version of the patch includes a fix for a crash caused by
treating ICmp/FCmp constant expressions as instructions.
Original message:
On some targets, like AArch64, vector selects can be efficiently lowered
if the vector condition is a compare with a supported predicate.
This patch adds a new argument to getCmpSelInstrCost, to indicate the
predicate of the feeding select condition. Note that it is not
sufficient to use the context instruction when querying the cost of a
vector select starting from a scalar one, because the condition of the
vector select could be composed of compares with different predicates.
This change greatly improves modeling the costs of certain
compare/select patterns on AArch64.
I am also planning on putting up patches to make use of the new argument in
SLPVectorizer & LV.
If the elt size is unknown due to it being a pointer, a comparison
against 0 will cause an assert. Make sure the elt size is large enough
before comparing and for the moment just return the scalar cost.
I'm assuming the standard size integer instructions for this end up as something like:
mulq %rsi
seto %al
And the 'mul' generally has reciprocal throughput of 1 on typical implementations
(higher latency, but that's not handled here).
The default costs may end up much higher than that, and that's what we see in the test diffs.
Vector types are left as a 'TODO'.
Differential Revision: https://reviews.llvm.org/D90431
On some targets, like AArch64, vector selects can be efficiently lowered
if the vector condition is a compare with a supported predicate.
This patch adds a new argument to getCmpSelInstrCost, to indicate the
predicate of the feeding select condition. Note that it is not
sufficient to use the context instruction when querying the cost of a
vector select starting from a scalar one, because the condition of the
vector select could be composed of compares with different predicates.
This change greatly improves modeling the costs of certain
compare/select patterns on AArch64.
I am also planning on putting up patches to make use of the new argument in
SLPVectorizer & LV.
Reviewed By: dmgreen, RKSimon
Differential Revision: https://reviews.llvm.org/D90070
Completing the series of FIXME removals for special-case intrinsics:
50dfa19cc7f2c25c7079c963bde01501ea93d85d
This one looks quite different than the others. The size/blended
cost is still potentially very far off from the throughput cost,
but this is hopefully not worse on the whole. It looks like the
underlying costs for the expanded shift/logic have their own
cost-kind limitations. Also, we are not asking the target if
it has a legal funnel shift op, so we just assume that the
intrinsic gets expanded.
vnot (xor -1) should be equivalent to the AArch64 specific AArch64ISD::NOT
node, but allow more folding thanks to all the target independent
optimizations. Specifically this allows select(icmp ne, x, y) to
become "cmeq; bsl y, x" as opposed to needing to convert the predicate
with "cmeq; mvn; bsl x, y"
Unfortunately there is a regression in a cmtst test, but the code it
selected from was already non-canonical, with instcombine preferring to
use an eq predicate instead. Plus the more common case of icmp ne is
improved.
Differential Revision: https://reviews.llvm.org/D90126
This was originally part of:
f2c25c7079
but that was reverted because there was an underlying bug in
processing the vector type of these intrinsics. That was
fixed with:
74ffc823ed
This is similar in spirit to 01ea93d85d (memcpy) except that
here the underlying caller assumptions were created for vectorizer
use (throughput) rather than other passes.
That meant targets could have an enormous throughput cost with no
corresponding size, latency, or blended cost increase.
Paraphrasing from the previous commits:
This may not make sense for some callers, but at least now the
costs will be consistently wrong instead of mysteriously wrong.
Targets should provide better overrides if the current modeling
is not accurate.
In each 128-lane, if there is at least one index is demanded and not all
indices are demanded and this 128-lane is not the first 128-lane of the
legalized-vector, then this 128-lane needs a extracti128;
If in each 128-lane, there is at least one index is demanded, this 128-lane
needs a inserti128.
The following cases will help you build a better understanding:
Assume we insert several elements into a v8i32 vector in avx2,
Case#1: inserting into 1th index needs vpinsrd + inserti128
Case#2: inserting into 5th index needs extracti128 + vpinsrd +
inserti128
Case#3: inserting into 4,5,6,7 index needs 4*vpinsrd + inserti128.
Reviewed By: pengfei, RKSimon
Differential Revision: https://reviews.llvm.org/D89767
The warning would fire when calling getGEPCost for analyzing the cost of
a GEP instruction. This would result in the use of the now deprecated
implicit cast of TypeSize to uint64_t through the overloaded operator.
This patch fixes the issue by using getKnownMinSize instead of the
implicit cast. This is possible because the code is already
scalable-vector aware. The semantic behaviour of the code is unchanged
by this patch.
Reviewed By: sdesmalen, fpetrogalli
Differential Revision: https://reviews.llvm.org/D89872
This allows using annotation in a much more contexts than it currently has.
especially when annotation with template or constexpr.
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D88645
This is a modified 2nd try of 22d10b8ab4
(reverted by 1c8371692d because it managed
to expose an existing crashing bug that should be fixed by
74ffc823 ).
Original commit message:
This is similar in spirit to 01ea93d85d (memcpy) except that
here the underlying caller assumptions were created for vectorizer
use (throughput) rather than other passes.
That meant targets could have an enormous throughput cost with no
corresponding size, latency, or blended cost increase.
The ARM costs show a small difference between throughput and
size because there's an underlying difference in cmp/sel
costs that is also predicated on cost-kind.
Paraphrasing from the previous commits:
This may not make sense for some callers, but at least now the
costs will be consistently wrong instead of mysteriously wrong.
Targets should provide better overrides if the current modeling
is not accurate.
I'm not sure if/how this ever worked, but it must not be tested
currently because the basic tests added here were crashing as
noted in the post-review comments for 1c83716 (which reverted
another cost-model fix in 22d10b8ab4).
This is similar in spirit to 01ea93d85d (memcpy) except that
here the underlying caller assumptions were created for vectorizer
use (throughput) rather than other passes.
That meant targets could have an enormous throughput cost with no
corresponding size, latency, or blended cost increase.
The ARM costs show a small difference between throughput and
size because there's an underlying difference in cmp/sel
costs that is also predicated on cost-kind.
Paraphrasing from the previous commits:
This may not make sense for some callers, but at least now the
costs will be consistently wrong instead of mysteriously wrong.
Targets should provide better overrides if the current modeling
is not accurate.
1. Throughput and codesize costs estimations was separated and updated.
2. Updated fdiv cost estimation for different cases.
3. Added scalarization processing for types that are treated as !isSimple() to
improve codesize estimation in getArithmeticInstrCost() and
getArithmeticInstrCost(). The code was borrowed from TCK_RecipThroughput path
of base implementation.
Next step is unify scalarization part in base class that is currently works for
TCK_RecipThroughput path only.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D89973
This patch adds a specialized implementation of getIntrinsicInstrCost
and add initial cost-modeling for min/max vector intrinsics.
AArch64 NEON support umin/smin/umax/smax for vectors
<8 x i8>, <16 x i8>, <4 x i16>, <8 x i16>, <2 x i32> and <4 x i32>.
Notably, it does not support vectors with i64 elements.
This change by itself should have very little impact on codegen, but in
follow-up patches I plan to teach the vectorizers to consider using
those intrinsics on platforms where it is profitable, e.g. because there
is no general 'select'-like instruction.
The current cost returned should be better for throughput, latency and size.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D89953
This is similar in spirit to 01ea93d85d (memcpy) except that
here the underlying caller assumptions were created for vectorizer
use (throughput) rather than other passes.
That meant ARM could have an enormous throughput cost with no
corresponding size, latency, or blended cost increase. X86 has
the same throughput restriction as the basic implementation, so
it is still unchanged.
Paraphrasing from the previous commit:
This may not make sense for some callers, but at least now the
costs will be consistently wrong instead of mysteriously wrong.
Targets should provide better overrides if the current modeling
is not accurate.
The default implementation base returns TCC_Expensive (currently
set to '4'), so that explains the test diff. This probably does
not make sense for most callers, but at least now the costs will
be consistently wrong instead of mysteriously wrong.
The ARM target has an override that tries to model codegen expansion,
and that should likely be adapted for general usage.
This probably does not affect anything because the vectorizers are
the primary users of the throughput cost, but memcpy is not listed
as a trivially vectorizable intrinsic.
This adds some basic costs for MVE reductions - currently just costing
the simple legal add vectors as a single MVE instruction. More complex
costing can be added in the future when the framework more readily
allows it.
Differential Revision: https://reviews.llvm.org/D88980
This adds a very basic cost for active_lane_mask under MVE - making the
assumption that they will be free and then apologizing for that in a
comment.
In reality they may either be free (by being nicely folded into a tail
predicated loop), cost the same as a VCTP or be expanded into vdup's,
adds and cmp's. It is difficult to detect the difference from a single
getIntrinsicInstrCost call, so makes the assumption that the vectorizer
is adding them, and only added them where it makes sense.
We may need to change this in the future to better model predicate costs
in the vectorizer, especially at -Os or non-tail predicated loops. The
vectorizer currently does not query the cost of these instructions but
that will change in the future and a zero cost there probably makes the
most sense at the moment.
Differential Revision: https://reviews.llvm.org/D88989
The cost modeling for intrinsics is a patchwork based on different
expectations from the callers, so it's a mess. I'm hoping to untangle
this to allow canonicalization to the new min/max intrinsics in IR.
The general goal is to remove the cost-kind restriction here in the
basic implementation class. Ie, if some intrinsic has throughput cost
of 104, assume that it has the same size, latency, and blended costs.
Effectively, an intrinsic with cost N is composed of N simple
instructions. If that's not correct, the target should provide a more
accurate override.
The x86-64 SSE2 subtarget cost diffs require explanation:
1. The scalar ctlz/cttz are assuming "BSR+XOR+CMOV" or
"TEST+BSF+CMOV/BRANCH", so not cheap.
2. The 128-bit SSE vector width versions assume cost of 18 or 26
(no explanation provided in the tables, but this corresponds to a
bunch of shift/logic/compare).
3. The 512-bit vectors in the test file are scaled up by a factor of
4 from the legal vector width costs.
4. The plain latency cost-kind is not affected in this patch because
that calc is diverted before we get to getIntrinsicInstrCost().
Differential Revision: https://reviews.llvm.org/D89461
This is bigger/uglier than before, but it should allow fixing
all of the broken paths more easily. Test coverage added with
rGfab028b and other commits.
This is not NFC - the scalable vector test would crash
without this patch.
Testing for the various cost model "TargetCostKind" is limited,
and testing for scalable vectors is limited. The motivating
example of an intrinsic is not included here yet because that
just crashes.
This is my first LLVM patch, so please tell me if there are any process issues.
The main observation for this patch is that we can lower UMIN/UMAX with v8i16 by using unsigned saturated subtractions in a clever way. Previously this operation was lowered by turning the signbit of both inputs and the output which turns the unsigned minimum/maximum into a signed one.
We could use this trick in reverse for lowering SMIN/SMAX with v16i8 instead. In terms of latency/throughput this is the needs one large move instruction. It's just that the sign bit turning has an increased chance of being optimized further. This is particularly apparent in the "reduce" test cases. However due to the slight regression in the single use case, this patch no longer proposes this.
Unfortunately this argument also applies in reverse to the new lowering of UMIN/UMAX with v8i16 which regresses the "horizontal-reduce-umax", "horizontal-reduce-umin", "vector-reduce-umin" and "vector-reduce-umax" test cases a bit with this patch. Maybe some extra casework would be possible to avoid this. However independent of that I believe that the benefits in the common case of just 1 to 3 chained min/max instructions outweighs the downsides in that specific case.
Patch By: @TomHender (Tom Hender) ActuallyaDeviloper
Differential Revision: https://reviews.llvm.org/D87236
Better use isZero() and isIntN() in SystemZTargetTransformInfo rather than
calling getZExtValue() since the immediate operand may be wider than 64 bits,
which is not allowed with getZExtValue().
Fixes https://bugs.llvm.org/show_bug.cgi?id=47600
Review: Simon Pilgrim
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.
When optimising for size, make the cost of i1 logical operations
relatively expensive so that optimisations don't try to combine
predicates.
Differential Revision: https://reviews.llvm.org/D86525
MVE Gather scatter codegeneration is looking a lot better than it used
to, but still has some issues. The instructions we currently model as 1
cycle per element, which is a bit low for some cases. Increasing the
cost by the MVECostFactor brings them in-line with our other instruction
costs. This will have the effect of only generating then when the extra
benefit is more likely to overcome some of the issues. Notably in
running out of registers and vectorizing loops that could otherwise be
SLP vectorized.
In the short-term whilst we look at other ways of dealing with those
more directly, we can increase the costs of gathers to make them more
likely to be beneficial when created.
Differential Revision: https://reviews.llvm.org/D86444
In getCastInstrCost when the instruction is a truncate we were relying
upon the implicit TypeSize -> uint64_t cast when asking if a given type
has the same size as a legal integer. I've changed the code to only
ask the question if the type is fixed length.
I have also changed InstCombinerImpl::SimplifyDemandedUseBits to bail
out for now if the type is a scalable vector.
I've added the following new tests:
Analysis/CostModel/AArch64/sve-trunc.ll
Transforms/InstCombine/AArch64/sve-trunc.ll
for both of these fixes.
Differential revision: https://reviews.llvm.org/D86432
test referenced a relative path to a file, but the path was not correct
relative to the project the test is in
Differential Revision: https://reviews.llvm.org/D86368
Modify the ARM getCmpSelInstrCost implementation for the code size
costs of selects. Now consider the legalization cost and increase
the cost of i1 because those values wouldn't live in a general purpose
register. We also make selects +1 more expensive to account for the IT
instruction.
Differential Revision: https://reviews.llvm.org/D82091
Add cases of fused fmul+fadd/fsub with f16 and f64 operands to cost model.
Also added operations with contract attribute.
Fixed line endings in test.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D84995
As with other targets, set the throughput cost of control-flow
instructions to free so that we don't miss out of vectorization
opportunities.
Differential Revision: https://reviews.llvm.org/D85283
This patch uses the feature added in D79162 to fix the cost of a
sext/zext of a masked load, or a trunc for a masked store.
Previously, those were considered cheap or even free, but it's
not the case as we cannot split the load in the same way we would for
normal loads.
This updates the costs to better reflect reality, and adds a test for it
in test/Analysis/CostModel/ARM/cast.ll.
It also adds a vectorizer test that showcases the improvement: in some
cases, the vectorizer will now choose a smaller VF when
tail-predication is enabled, which results in better codegen. (Because
if it were to use a higher VF in those cases, the code we see above
would be generated, and the vmovs would block tail-predication later in
the process, resulting in very poor codegen overall)
Original Patch by Pierre van Houtryve
Differential Revision: https://reviews.llvm.org/D79163
Summary:
If result of fmul(b,c) has one use, in almost all cases (except denormals are
IEEE) the pair of operations will be fused in one fma/mad/mac/etc.
Reviewers: rampitec
Reviewed By: rampitec
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits, kerbowa
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83919
In getCastInstrCost() when comparing different sizes for src and
dst types we should be using the TypeSize comparison operators
instead of relying upon TypeSize being converted a uin64_t.
Previously this meant we were dropping the scalable property and
treating fixed and scalable vector types the same.
Differential Revision: https://reviews.llvm.org/D83461
Even though wide vectors are legal they still cost more as we
will have to eventually split them. Not all operations can
be uniformly done on vector types.
Conservatively add the cost of splitting at least to 8 dwords,
which is our widest possible load.
We are more or less lying to cost mode with this change but
this can prevent vectorizer from creation of wide vectors which
results in RA problems for us.
Differential Revision: https://reviews.llvm.org/D83078
This adjusts the MVE fp16 cost model, similar to how we already do for
integer casts. It uses the base cost of 1 per cvt for most fp extend /
truncates, but adjusts it for loads and stores where we know that a
extending load has been used to get the load into the correct lane, and
only an MVE VCVTB is then needed.
Differential Revision: https://reviews.llvm.org/D81813
This adds some default costs for fp extends and truncates, generally
costing them as 1 per lane. If the type is not legal then the cost will
include a call to an __aeabi_ function.
Some NEON code is also adjusted to make sure it applies to the expected
types, now that fp16 is a more common thing.
Differential Revision: https://reviews.llvm.org/D82458
This expands the existing extend costs with a few extras for larger
types than legal, which will usually be split under MVE. It also adds
trunk support for the same thing. These should not have a large effect
on many things, but makes the costs explicit and keeps a certain balance
between the trunks and extends.
Differential Revision: https://reviews.llvm.org/D82457
This alters getMemoryOpCost to use the Base TargetTransformInfo version
that includes some additional checks for whether extending loads are
legal. This will generally have the effect of making <2 x ..> and some
<4 x ..> loads/stores more expensive, which in turn should help favour
larger vector factors.
Notably it alters the cost of a <4 x half>, which with the current
codegen will be expensive if it is not extended.
Differential Revision: https://reviews.llvm.org/D82456
D79164/2596da31740f changed getCFInstrCost to return 1 per default.
AArch64 did not have its own implementation, hence the throughput cost
of CFI instructions is overestimated. On most cores, most branches should
be predicated and essentially free throughput wise.
This restores a 9% performance regression on a SPEC2006 benchmark on
AArch64 with -O3 LTO & PGO.
This patch effectively restores pre 2596da3174 behavior for AArch64
and undoes the AArch64 test changes of the patch.
Reviewers: samparker, dmgreen, anemet
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D82755
This file has grown quite large and could do with being split up. This
splits away the load/store + cast tests into a separate file. Some
masked load/store + cast tests have been added too, along with some
extra load/store + fpcast tests.
Adds aarch64-sve-vector-bits-{min,max} to allow the size of SVE
data registers (in bits) to be specified. This allows the code
generator to make assumptions it normally couldn't. As a starting
point this information is used to mark fixed length vector types
that can fit within the specified size as legal.
Reviewers: rengolin, efriedma
Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80384
Have BasicTTI call the base implementation so that both agree on the
default behaviour, which the default being a cost of '1'. This has
required an X86 specific implementation as it seems to be very
reliant on those instructions being free. Changes are also made to
AMDGPU so that their implementations distinguish between cost kinds,
so that the unrolling isn't affected. PowerPC also has its own
implementation to prevent changes to the reg-usage vectorizer test.
The cost model test changes now reflect that ret instructions are not
generally free.
Differential Revision: https://reviews.llvm.org/D79164
Recommitting part of "[CostModel] Unify Intrinsic Costs."
de71def3f5
Now that the 'free' intrinsic information has been sunk to the lowest
level, query the base implementation in BasicTTI before doing
anything else. I suspect this is the change that was causing the main
changes, particularly the large effects on debug builds.
Differential Revision: https://reviews.llvm.org/D80012
This has not been implemented by any backends which appear to cover
the functionality through getCastInstrCost. Sink what there is in the
default implementation into BasicTTI.
Differential Revision: https://reviews.llvm.org/D78922
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
We need to use it to handle <16 x double> indirect indexes
in the AMDGPU BE.
The only visible change from adding it is in ARM cost model.
To me it looks reasonable. With doubling a vector size it
quadruples the cost up to the size 8 and then it did only
double it. Now it also quadruples, which seems a logical
progression to me.
Actual AMDGPU code is to follow, this is a common part, plus
load/store legalization in the AMDGPU BE not to break what
works now.
Differential Revision: https://reviews.llvm.org/D79952
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
- Specifically check for sext/zext users which have 'long' form NEON
instructions.
- Add more entries to the table for sext/zexts so that we can report
more accurately the number of vmovls required for NEON.
- Pass the instruction to the pass implementation.
Differential Revision: https://reviews.llvm.org/D79561
If both the source and the destination need to be split then the two halves of the split operation are completely independent and don't need to be split or joined. So we don't need to assess a cost for the split or join.
Differential Revision: https://reviews.llvm.org/D79111
Also fix some cost tables for vXi1 types to match the costs entries for the types they will be promoted to.
Differential Revision: https://reviews.llvm.org/D79045
vpermw is 2 uops. vpermt2b/vpermt2w are two shuffle uops and a port 015 uop. Weirdly vpermb is a single uop.
This patch bumps the cost to 2 for these operations. Maybe should go to 3 for the vpermt2*, but I've started conservative.
I've also removed a few entries that were now the same as earlier subtargets or that I didn't think we really did. Like I don't think we extend v32i8 to v32i16, shuffle, and then truncate.
Differential Revision: https://reviews.llvm.org/D79148
We generate much better code these days than we used to. And we use the same sequence for AVX1 and AVX2 for these
For v4i64->v4i32 we generate:
vextractf128 xmm1, ymm0, 1
vshufps xmm0, xmm0, xmm1, 136 # xmm0 = xmm0[0,2],xmm1[0,2]
And for v8i64->v8i32 we generate:
vperm2f128 ymm2, ymm0, ymm1, 49 # ymm2 = ymm0[2,3],ymm1[2,3]
vinsertf128 ymm0, ymm0, xmm1, 1
vshufps ymm0, ymm0, ymm2, 136 # ymm0 = ymm0[0,2],ymm2[0,2],ymm0[4,6],ymm2[4,6]
Differential Revision: https://reviews.llvm.org/D79109
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
I've modified isTruncateFree to get an accurate cost for types that need to be split. I'm planning to look into fixing it for all vectors, but need more cost cleanups first.
Differential Revision: https://reviews.llvm.org/D78973
All avx512 truncate instructions except vXi64->vXi32 are 2 uops
on port 5. So raise their costs to 2. Except when we have an
earlier faster sequence like pshufb for 128 bit input vectors.
Add a lower cost of 3 v16i16->v16i8 with avx512f where we can
extend to v16i32 then truncate. And a cost of 2 for avx512bw with
and without avx512vl. There we can use vpmovwb with either a ymm
or zmm input. Both of these beat masking, splitting, and using
packuswb which is our avx/avx2 codegen.
We aren't properly costing extends from k-registers. I also added
command lines without avx512bw to be able to show all the different
extending strategies we have.
Previously, we just always returned 1. But that ignores that we have to do the operation for each subvector or scalar component.
Differential Revision: https://reviews.llvm.org/D78824
This moves v32i16/v64i8 to a model consistent with how we
treat integer types with avx1.
This does change the ABI for types vXi16/vXi8 vectors larger than
512 bits to pass in multiple zmms instead of multiple ymms. We'd
already hacked some code to make v64i8/v32i16 pass in zmm.
Cost model is still a bit of a mess. In some place I tried to
match existing behavior. But really we need to account for
splitting and concating costs. Cost model for shuffles is
especially pessimistic.
Differential Revision: https://reviews.llvm.org/D76212
If we're inserting into v2i8/v4i8/v8i8/v2i16/v4i16 style sub-128bit vectors ensure we don't use the SK_PermuteTwoSrc cost of the legalized value type - this is a followup to rG12c629ec6c59 which added equivalent sub-128bit shuffle costs
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
v2i8/v4i8/v8i8 + v2i16/v4i16 all show up in vectorizer code and by just using the legalized types (v16i8/v8i16) we're highly exaggerating the actual cost of the shuffle.
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
Previously we multiplied the cost for the table entries by the number of splits needed. But that implies that each split goes through a reduction to scalar independently. I think what really happens is that the we AND/OR the split pieces until we're down to a single value with a legal type and then do special reduction sequence on that.
So to model that this patch takes the number of splits minus one multiplied by the cost of a AND/OR at the legal element count and adds that on top of the table lookup.
Differential Revision: https://reviews.llvm.org/D76400
Refines the gather/scatter cost model, but also changes the TTI
function getIntrinsicInstrCost to accept an additional parameter
which is needed for the gather/scatter cost evaluation.
This did require trivial changes in some non-ARM backends to
adopt the new parameter.
Extending gathers and truncating scatters are now priced cheaper.
Differential Revision: https://reviews.llvm.org/D75525
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.
A cost query for a vector instruction should return a cost even without
target vector support, and not trigger an assert.
VectorCombine does this with an input containing source code vectors.
Review: Ulrich Weigand
We seem to be inheriting the cost from sse4.1. But if we have 256-bit registers we should be able to do this with just one extract to split the 16i16 and two v8i16->v8i32 operations so our cost should be 3 not 4.
Differential Revision: https://reviews.llvm.org/D73646
This is a very basic MVE gather/scatter cost model, based roughly on the
code that we will currently produce. It does not handle truncating
scatters or extending gathers correctly yet, as it is difficult to tell
that they are going to be correctly extended/truncated from the limited
information in the cost function.
This can be improved as we extend support for these in the future.
Based on code originally written by David Sherwood.
Differential Revision: https://reviews.llvm.org/D73021
This attempts to teach the cost model in Arm that code such as:
%s = shl i32 %a, 3
%a = and i32 %s, %b
Can under Arm or Thumb2 become:
and r0, r1, r2, lsl #3
So the cost of the shift can essentially be free. To do this without
trying to artificially adjust the cost of the "and" instruction, it
needs to get the users of the shl and check if they are a type of
instruction that the shift can be folded into. And so it needs to have
access to the actual instruction in getArithmeticInstrCost, which if
available is added as an extra parameter much like getCastInstrCost.
We otherwise limit it to shifts with a single user, which should
hopefully handle most of the cases. The list of instruction that the
shift can be folded into include ADC, ADD, AND, BIC, CMP, EOR, MVN, ORR,
ORN, RSB, SBC and SUB. This translates to Add, Sub, And, Or, Xor and
ICmp.
Differential Revision: https://reviews.llvm.org/D70966
This adds some extra cost model tests for shifts, and does some minor
adjustments to some Neon code to make it clear as to what it applies to.
Both NFC.
This is a follow-up to D70607 where we made any
extract element on SLM more costly than default. But that is
pessimistic for extract from element 0 because that corresponds
to x86 movd/movq instructions. These generally have >1 cycle
latency, but they are probably implemented as single uop
instructions.
Note that no vectorization tests are affected by this change.
Also, no targets besides SLM are affected because those are
falling through to the default cost of 1 anyway. But this will
become visible/important if we add more specializations via cost
tables.
Differential Revision: https://reviews.llvm.org/D71023
The Power 9 CPU has some features that are unlikely to be passed on to future
versions of the CPU. This patch separates this out so that future CPU does not
inherit them.
Differential Revision: https://reviews.llvm.org/D70466
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
This is no longer needed after widening legalization as we
custom legalize v8i8 ourselves.
Added entries to the cost model, but bumped the cost slightly
to account for the truncate shuffle that wasn't costed before.
ShuffleVectorInst::isExtractSubvectorMask, introduced in
[CostModel] Add SK_ExtractSubvector handling to getInstructionThroughput (PR39368)
erroneously thought that
%340 = shufflevector <4 x float> %339, <4 x float> undef, <3 x i32> <i32 2, i32 3, i32 undef>
is a subvector extract, even though it goes off the end of the parent
vector with the undef index. That then caused an assert in
BasicTTIImplBase::getExtractSubvectorOverhead.
This commit fixes that, by not considering the above a subvector
extract.
Differential Revision: https://reviews.llvm.org/D70005
Change-Id: I87b8b00b24bef19ffc9a1b82ef4eca3b8a246eaf
This better represents the kshift+binop we'd get for each stage
before the final extract. Its likely we'll do even better by
doing a kmov and a cmp with a GPR, but this is a good start.
The default handling was costing a worst case single source
permute shuffle of the vector before the binop. This worst
case assumes the shuffle might have to be emulated with
extracts and inserts. But since we know we're doing a reduction
we can assume we'll get kshift lowering.
There's still some room for improvement here, but this is
much better than it was.
Add specific scalar costs for CTLZ instructions, we can't discriminate between CTLZ and CTLZ_ZERO_UNDEF so we have to assume the worst. Given how BSR is often a microcoded nightmare on some older targets we might still be underestimating it.
For targets supporting LZCNT (Intel Haswell+ or AMD Fam10+), we provide overrides that assume 1cy costs.
llvm-svn: 374786
Add specific scalar costs for ctpop instructions, these are based on the llvm-mca's SLM throughput numbers (the oldest model we have).
For targets supporting POPCNT, we provide overrides that assume 1cy costs.
llvm-svn: 374775
I can't see any notable differences in costs between SSE2 and SSE42 arches for FADD/ADD reduction, so I've lowered the target to just SSE2.
I've also added vXi8 sum reduction costs in line with the PSADBW codegen and discussions on PR42674.
llvm-svn: 374655
SLM is 2 x slower for <2 x i64> comparison ops than other vector types, we should account for this like we do for SLM <2 x i64> add/sub/mul costs.
This should remove some of the SLM codegen diffs in D43582
llvm-svn: 372954
We are missing costs for a lot of truncation cases, I'm hoping to address all the 'zero cost' cases in trunc.ll
I thought this was a vector widening side effect, but even before this we had some interesting LV decisions (notably over indvars) being made due to these zero costs.
llvm-svn: 372498
The recently announced IBM z15 processor implements the architecture
already supported as "arch13" in LLVM. This patch adds support for
"z15" as an alternate architecture name for arch13.
The patch also uses z15 in a number of places where we used arch13
as long as the official name was not yet announced.
llvm-svn: 372435
I don't really understand the costs we're using for fp_to_sint,
but prior to widening legalization we used 20 as the cost for this
via the v2i64->v2f64 entry. That number seems better than the 40
we got with widening legalization. So now we need either a
v2i32->v2f64 entry or a v4i32->v2f64 entry depending on whether
AVX is enabled or not since we skip the first SSE2 table look up
under AVX.
llvm-svn: 369628
This adds some sext costs for MVE, taken from the length of assembly sequences
that we currently generate.
Differential Revision: https://reviews.llvm.org/D66010
llvm-svn: 369244
MVE also has some sext of loads, which will be free just as scalar
instructions are.
Differential Revision: https://reviews.llvm.org/D66008
llvm-svn: 369118
Now that we're using widening legalization. We need to improve our extract_subvector cost model for these types. This patch begins by modeling these as a subvector extract followed by a permute. I've left FIXMEs in the code for future improvements.
Differential Revision: https://reviews.llvm.org/D65892
llvm-svn: 369022
Now that we legalize by widening, the element types here won't change. Previously these were modeled as the elements being widened and then the instruction might become an AND or SHL/ASHR pair. But now they'll become something like a ZERO_EXTEND_VECTOR_INREG/SIGN_EXTEND_VECTOR_INREG.
For AVX2, when the destination type is legal its clear the cost should be 1 since we have extend instructions that can produce 256 bit vectors from less than 128 bit vectors. I'm a little less sure about AVX1 costs, but I think the ones I changed were definitely too high, but they might still be too high.
Differential Revision: https://reviews.llvm.org/D66169
llvm-svn: 368858
The MVE architecture has the idea of "beats", where a vector instruction can be
executed over several ticks of the architecture. This adds a similar system
into the Arm backend cost model, multiplying the cost of all vector
instructions by a factor.
This factor essentially becomes the expected difference between scalar code
and vector code, on average. MVE Vector instructions can also overlap so the a
true cost of them is often lower. But equally scalar instructions can in some
situations be dual issued, or have other optimisations such as unrolling or
make use of dsp instructions. The default is chosen as 2. This should not
prevent vectorisation is a most cases (as the vector instructions will still be
doing at least 4 times the work), but it will help prevent over vectorising in
cases where the benefits are less likely.
This adds things so far to the obvious places in ARMTargetTransformInfo, and
updates a few related costs like not treating float instructions as cost 2 just
because they are floats.
Differential Revision: https://reviews.llvm.org/D66005
llvm-svn: 368733
This teaches the cost model that the sext or zext of a load is going to be
free.
Differential Revision: https://reviews.llvm.org/D66006
llvm-svn: 368593
A VDUP will perform a vector broadcast in a single instruction. Update the cost
model for MVE accordingly.
Code originally by David Sherwood.
Differential Revision: https://reviews.llvm.org/D63448
llvm-svn: 368589
This puts some of the calls in ARMTargetTransformInfo.cpp behind hasNeon()
checks, now that we have MVE, and updates all the tests accordingly.
Differential Revision: https://reviews.llvm.org/D63447
llvm-svn: 368587
This adds a number of cost model tests for ARM, useful for MVE. It also re-jigs
some of the existing tests to make them easier to update and read.
llvm-svn: 368586
The assert that caused this to be reverted should be fixed now.
Original commit message:
This patch changes our defualt legalization behavior for 16, 32, and
64 bit vectors with i8/i16/i32/i64 scalar types from promotion to
widening. For example, v8i8 will now be widened to v16i8 instead of
promoted to v8i16. This keeps the elements widths the same and pads
with undef elements. We believe this is a better legalization strategy.
But it carries some issues due to the fragmented vector ISA. For
example, i8 shifts and multiplies get widened and then later have
to be promoted/split into vXi16 vectors.
This has the potential to cause regressions so we wanted to get
it in early in the 10.0 cycle so we have plenty of time to
address them.
Next steps will be to merge tests that explicitly test the command
line option. And then we can remove the option and its associated
code.
llvm-svn: 368183
This patch changes our defualt legalization behavior for 16, 32, and
64 bit vectors with i8/i16/i32/i64 scalar types from promotion to
widening. For example, v8i8 will now be widened to v16i8 instead of
promoted to v8i16. This keeps the elements widths the same and pads
with undef elements. We believe this is a better legalization strategy.
But it carries some issues due to the fragmented vector ISA. For
example, i8 shifts and multiplies get widened and then later have
to be promoted/split into vXi16 vectors.
This has the potential to cause regressions so we wanted to get
it in early in the 10.0 cycle so we have plenty of time to
address them.
Next steps will be to merge tests that explicitly test the command
line option. And then we can remove the option and its associated
code.
llvm-svn: 367901
This patch series adds support for the next-generation arch13
CPU architecture to the SystemZ backend.
This includes:
- Basic support for the new processor and its features.
- Assembler/disassembler support for new instructions.
- CodeGen for new instructions, including new LLVM intrinsics.
- Scheduler description for the new processor.
- Detection of arch13 as host processor.
Note: No currently available Z system supports the arch13
architecture. Once new systems become available, the
official system name will be added as supported -march name.
llvm-svn: 365932
This patch uses the mechanism from D62995 to strengthen the
definitions of the reduction intrinsics by letting the scalar
result/accumulator type be overloaded from the vector element type.
For example:
; The LLVM LangRef specifies that the scalar result must equal the
; vector element type, but this is not checked/enforced by LLVM.
declare i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> %a)
This patch changes that into:
declare i32 @llvm.experimental.vector.reduce.or.v4i32(<4 x i32> %a)
Which has the type-constraint more explicit and causes LLVM to check
the result type with the vector element type.
Reviewers: RKSimon, arsenm, rnk, greened, aemerson
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D62996
llvm-svn: 363240
Types such as float and i64's do not have legal loads in Thumb1, but will still
be loaded with a LDR (or potentially multiple LDR's). As such we can treat the
cost of addressing mode calculations the same as an i32 and get some optimisation
benefits.
Differential Revision: https://reviews.llvm.org/D62968
llvm-svn: 362874
Now with MVE being added, we can add the vector addressing mode costs for it.
These are generally imm7 multiplied by the size of the type being loaded /
stored.
Differential Revision: https://reviews.llvm.org/D62967
llvm-svn: 362873
The fp16 version of VLDR takes a imm8 multiplied by 2. This updates the costs
to account for those, and adds extra testing. It is dependant upon hasFPRegs16
as this is what the load/store instructions require.
Differential Revision: https://reviews.llvm.org/D62966
llvm-svn: 362872
For some reason multiple places need to do this, and the variant the
loop unroller and inliner use was not handling it.
Also, introduce a new wrapper to be slightly more precise, since on
AMDGPU some addrspacecasts are free, but not no-ops.
llvm-svn: 362436
Summary:
This reuses the getArithmeticInstrCost, but passes dummy values of the second
operand flags.
The X86 costs are wrong and can be improved in a follow up. I just wanted to
stop it from reporting an unknown cost first.
Reviewers: RKSimon, spatel, andrew.w.kaylor, cameron.mcinally
Reviewed By: spatel
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D62444
llvm-svn: 361788
getUserCost() currently returns TCC_Free for any extend of a compare (i1)
result. It seems this is only true in a limited number of cases where for
example two compares are chained. Even in those types of cases it seems
unlikely that they are generally free, while they may be in some cases.
This patch therefore removes this special handling of cast of i1. No tests
are failing because of this.
If some target want the old behavior, it could override getUserCost().
Review: Hal Finkel, Chandler Carruth, Evgeny Astigeevich, Simon Pilgrim,
Ulrich Weigand
https://reviews.llvm.org/D54742/new/
llvm-svn: 360970
The original costs stopped at SSE42, I've added conservative estimates for everything down to SSE1/SSE2 and moved some of the SSE42 costs to SSE41 (really only the addition of PCMPGT makes any difference).
I've also added missing vXi8 costs (we use PHMINPOSUW for i8/i16 for scarily quick results) and 256-bit vector costs for AVX1.
llvm-svn: 360528
This implements TargetTransformInfo method getMemcpyCost, which estimates the
number of instructions to which a memcpy instruction expands to.
Differential Revision: https://reviews.llvm.org/D59787
llvm-svn: 359547
The PPC vector cost model values for insert/extract element reflect older
processors that lacked vector insert/extract and move-to/move-from VSR
instructions. Update getVectorInstrCost to give appropriate values for when
the newer instructions are present.
Differential Revision: https://reviews.llvm.org/D60160
llvm-svn: 359313
This does two main things, firstly adding some at least basic addressing modes
for i64 types, and secondly treats floats and doubles sensibly when there is no
fpu. The floating point change can help codesize in some cases, especially with
D60294.
Most backends seems to not consider the exact VT in isLegalAddressingMode,
instead switching on type size. That is now what this does when the target does
not have an fpu (as the float data will be loaded using LDR's). i64's currently
use the address range of an LDRD (even though they may be legalised and loaded
with an LDR). This is at least better than marking them all as illegal
addressing modes.
I have not attempted to do much with vectors yet. That will need changing once
MVE is added.
Differential Revision: https://reviews.llvm.org/D60677
llvm-svn: 358845
On pre-AVX512 targets we can use MOVMSK to extract reduced boolean results. This is properly optimized, annoyingly AVX512 isn't and produces code that is almost as bad as the (unchanged) costs suggest......
Differential Revision: https://reviews.llvm.org/D60403
llvm-svn: 358574
This adds new function getMemcpyCost to TTI so that the cost of a memcpy can be
modeled and queried. The default implementation returns Expensive, but targets
can override this function to model the cost more accurately.
Differential Revision: https://reviews.llvm.org/D59252
llvm-svn: 356555
There are a few different issues, mostly stemming from using
generation based checks for anything instead of subtarget
features. Stop adding flat-address-space as a feature for HSA, as it
should only be a device property. This was incorrectly allowing flat
instructions to select for SI.
Increase the default generation for HSA to avoid the encoding error
when emitting objects. This has some other side effects from various
checks which probably should be separate subtarget features (in the
cost model and for dealing with the DS offset folding issue).
Partial fix for bug 41070. It should probably be an error to try using
amdhsa without flat support.
llvm-svn: 356347
AMDGPU would like to have MVTs for v3i32, v3f32, v5i32, v5f32. This
commit does not add them, but makes preparatory changes:
* Fixed assumptions of power-of-2 vector type in kernel arg handling,
and added v5 kernel arg tests and v3/v5 shader arg tests.
* Added v5 tests for cost analysis.
* Added vec3/vec5 arg test cases.
Some of this patch is from Matt Arsenault, also of AMD.
Differential Revision: https://reviews.llvm.org/D58928
Change-Id: I7279d6b4841464d2080eb255ef3c589e268eabcd
llvm-svn: 356342
Based on an IR equivalent of target lowering's generic expansion - target specific costs will typically be lower (IR doesn't have a good mull/mulh equivalent) but we need a baseline.
Differential Revision: https://reviews.llvm.org/D57925
llvm-svn: 354774
Followup to D56636, this time handling the UADDSAT case by expanding
uadd.sat(a, b) to umin(a, ~b) + b.
Differential Revision: https://reviews.llvm.org/D56869
llvm-svn: 352409
Add generic costs calculation for SADDSAT/SSUBSAT intrinsics, this uses generic costs for sadd_with_overflow/ssub_with_overflow, an extra sign comparison + a selects based on the sign/overflow.
This completes PR40316
Differential Revision: https://reviews.llvm.org/D57239
llvm-svn: 352315
For the power9 CPU, vector operations consume a pair of execution units rather
than one execution unit like a scalar operation. Update the target transform
cost functions to reflect the higher cost of vector operations when targeting
Power9.
Patch by RolandF.
Differential revision: https://reviews.llvm.org/D55461
llvm-svn: 352261
Add generic costs calculation for UADDSAT/USUBSAT intrinsics, this fallbacks to using generic costs for uadd_with_overflow/usub_with_overflow + a select.
Differential Revision: https://reviews.llvm.org/D56907
llvm-svn: 352044
This patch replaces the existing LLVMVectorSameWidth matcher with LLVMScalarOrSameVectorWidth.
The matching args must be either scalars or vectors with the same number of elements, but in either case the scalar/element type can differ, specified by LLVMScalarOrSameVectorWidth.
I've updated the _overflow intrinsics to demonstrate this - allowing it to return a i1 or <N x i1> overflow result, matching the scalar/vectorwidth of the other (add/sub/mul) result type.
The masked load/store/gather/scatter intrinsics have also been updated to use this, although as we specify the reference type to be llvm_anyvector_ty we guarantee the mask will be <N x i1> so no change in behaviour
Differential Revision: https://reviews.llvm.org/D57090
llvm-svn: 351957
First step towards PR40376, this patch adds support for getCmpSelInstrCost to use the (optional) Instruction CmpInst predicate to indicate the type of integer comparison we're performing and alter the costs accordingly.
Differential Revision: https://reviews.llvm.org/D57013
llvm-svn: 351810
Prior to SSE41 (and sometimes on AVX1), vector select has to be performed as a ((X & C)|(Y & ~C)) bit select.
Exposes a couple of issues with the min/max reduction costs (which only go down to SSE42 for some reason).
The increase pre-SSE41 selection costs also prevent a couple of tests from firing any longer, so I've either tweaked the target or added AVX tests as well to the existing SSE2 tests.
llvm-svn: 351685
Roman Lebedev