This patch makes JumpThreading fold br(freeze(undef)) if the freeze instruction
is only used by the branch.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D84818
This is to avoid the need to update a bunch of test files when the PGO
instrumentation function hashing changes.
Split off of D84782.
Differential Revision: https://reviews.llvm.org/D84865
I've been looking at missed vectorizations in one codebase.
One particular thing that stands out is that some of the loops
reach vectorizer in a rather mangled form, with weird PHI's,
and some of the loops aren't even in a rotated form.
After taking a more detailed look, that happened because
the loop's headers were too big by then. It is evident that
SimplifyCFG's common code hoisting transform is at fault there,
because the pattern it handles is precisely the unrotated
loop basic block structure.
Surprizingly, `SimplifyCFGOpt::HoistThenElseCodeToIf()` is enabled
by default, and is always run, unlike it's friend, common code sinking
transform, `SinkCommonCodeFromPredecessors()`, which is not enabled
by default and is only run once very late in the pipeline.
I'm proposing to harmonize this, and disable common code hoisting
until //late// in pipeline. Definition of //late// may vary,
here currently i've picked the same one as for code sinking,
but i suppose we could enable it as soon as right after
loop rotation happens.
Experimentation shows that this does indeed unsurprizingly help,
more loops got rotated, although other issues remain elsewhere.
Now, this undoubtedly seriously shakes phase ordering.
This will undoubtedly be a mixed bag in terms of both compile- and
run- time performance, codesize. Since we no longer aggressively
hoist+deduplicate common code, we don't pay the price of said hoisting
(which wasn't big). That may allow more loops to be rotated,
so we pay that price. That, in turn, that may enable all the transforms
that require canonical (rotated) loop form, including but not limited to
vectorization, so we pay that too. And in general, no deduplication means
more [duplicate] instructions going through the optimizations. But there's still
late hoisting, some of them will be caught late.
As per benchmarks i've run {F12360204}, this is mostly within the noise,
there are some small improvements, some small regressions.
One big regression i saw i fixed in rG8d487668d09fb0e4e54f36207f07c1480ffabbfd, but i'm sure
this will expose many more pre-existing missed optimizations, as usual :S
llvm-compile-time-tracker.com thoughts on this:
http://llvm-compile-time-tracker.com/compare.php?from=e40315d2b4ed1e38962a8f33ff151693ed4ada63&to=c8289c0ecbf235da9fb0e3bc052e3c0d6bff5cf9&stat=instructions
* this does regress compile-time by +0.5% geomean (unsurprizingly)
* size impact varies; for ThinLTO it's actually an improvement
The largest fallout appears to be in GVN's load partial redundancy
elimination, it spends *much* more time in
`MemoryDependenceResults::getNonLocalPointerDependency()`.
Non-local `MemoryDependenceResults` is widely-known to be, uh, costly.
There does not appear to be a proper solution to this issue,
other than silencing the compile-time performance regression
by tuning cut-off thresholds in `MemoryDependenceResults`,
at the cost of potentially regressing run-time performance.
D84609 attempts to move in that direction, but the path is unclear
and is going to take some time.
If we look at stats before/after diffs, some excerpts:
* RawSpeed (the target) {F12360200}
* -14 (-73.68%) loops not rotated due to the header size (yay)
* -272 (-0.67%) `"Number of live out of a loop variables"` - good for vectorizer
* -3937 (-64.19%) common instructions hoisted
* +561 (+0.06%) x86 asm instructions
* -2 basic blocks
* +2418 (+0.11%) IR instructions
* vanilla test-suite + RawSpeed + darktable {F12360201}
* -36396 (-65.29%) common instructions hoisted
* +1676 (+0.02%) x86 asm instructions
* +662 (+0.06%) basic blocks
* +4395 (+0.04%) IR instructions
It is likely to be sub-optimal for when optimizing for code size,
so one might want to change tune pipeline by enabling sinking/hoisting
when optimizing for size.
Reviewed By: mkazantsev
Differential Revision: https://reviews.llvm.org/D84108
All these tests already explicitly test against both legacy PM and NPM.
$ sed -i 's/ -attributor / -attributor -enable-new-pm=0 /g' $(rg --path-separator // -l -- -passes=)
$ sed -i 's/ -attributor-cgscc / -attributor-cgscc -enable-new-pm=0 /g' $(rg --path-separator // -l -- -passes=)
Now all tests in Transforms/Attributor/ pass under NPM.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D84813
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
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
In addition to removing phi nodes this patch removes any
landing pad that the dead exit block might have. Without
this fix Verifier complains about a new switch instruction
jumps to a block with a landing pad.
Differential Revision: https://reviews.llvm.org/D84320
This is the main icmp simplification shortcoming seen in D84655.
Alive2 agrees that the basic examples are correct at least:
define <2 x i1> @src(<2 x i8> %x) {
%0:
%r = icmp sle <2 x i8> { undef, 128 }, %x
ret <2 x i1> %r
}
=>
define <2 x i1> @tgt(<2 x i8> %x) {
%0:
ret <2 x i1> { 1, 1 }
}
Transformation seems to be correct!
define <2 x i1> @src(<2 x i32> %X) {
%0:
%A = or <2 x i32> %X, { 63, 63 }
%B = icmp ult <2 x i32> %A, { undef, 50 }
ret <2 x i1> %B
}
=>
define <2 x i1> @tgt(<2 x i32> %X) {
%0:
ret <2 x i1> { 0, 0 }
}
Transformation seems to be correct!
https://alive2.llvm.org/ce/z/omt2eehttps://alive2.llvm.org/ce/z/GW4nP_
Differential Revision: https://reviews.llvm.org/D84762
This patch adds a basic support for freeze instruction to JumpThreading
by making ComputeValueKnownInPredecessorsImpl look into its operand.
Reviewed By: efriedma, nikic
Differential Revision: https://reviews.llvm.org/D84598
To match NewPM pass name, and also for readability.
Also rename rpo-functionattrs -> rpo-function-attrs while we're here.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D84694
This patch added dependency graph to the attributor so that we can dump the dependencies between AAs more easily. We can also apply general graph algorithms to the graph, making it easier for us to create deep wrappers.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D78861
While this doesn't appear to help with the perf issue being exposed by
D84108, the function as-is is very weird, convoluted, and what's worse,
recursive.
There was no need for `SpeculativelyAvaliableAndUsedForSpeculation`,
tri-state choice is enough. We don't even ever check for that state.
The basic idea here is that we need to perform a depth-first traversal
of the predecessors of the basic block in question, either finding a
preexisting state for the block in a map, or inserting a "placeholder"
`SpeculativelyAvaliable`,
If we encounter an `Unavaliable` block, then we need to give up search,
and back-propagate the `Unavaliable` state to the each successor of
said block, more specifically to the each `SpeculativelyAvaliable`
we've just created.
However, if we have traversed entirety of the predecessors and have not
encountered an `Unavaliable` block, then it must mean the value is fully
available. We could update each inserted `SpeculativelyAvaliable` into
a `Avaliable`, but we don't need to, as assertion excersizes,
because we can assume that if we see an `SpeculativelyAvaliable` entry,
it is actually `Avaliable`, because during the time we've produced it,
if we would have found that it has an `Unavaliable` predecessor,
we would have updated it's successors, including this block,
into `Unavaliable`
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D84181
PGO profile is usually more precise than sample profile. However, PGO profile
needs to be collected from loadtest and loadtest may not be representative
enough to the production workload. Sample profile collected from production
can be used as a supplement -- for functions cold in loadtest but warm/hot
in production, we can scale up the related function in PGO profile if the
function is warm or hot in sample profile.
The implementation contains changes in compiler side and llvm-profdata side.
Given an instr profile and a sample profile, for a function cold in PGO
profile but warm/hot in sample profile, llvm-profdata will either mark
all the counters in the profile to be -1 or scale up the max count in the
function to be above hot threshold, depending on the zero counter ratio in
the profile. The assumption is if there are too many counters being zero
in the function profile, the profile is more likely to cause harm than good,
then llvm-profdata will mark all the counters to be -1 indicating the
function is hot but the profile is unaccountable. In compiler side, if a
function profile with all -1 counters is seen, the function entry count will
be set to be above hot threshold but its internal profile will be dropped.
In the long run, it may be useful to let compiler support using PGO profile
and sample profile at the same time, but that requires more careful design
and more substantial changes to make two profiles work seamlessly. The patch
here serves as a simple intermediate solution.
Differential Revision: https://reviews.llvm.org/D81981
SplitBlockPredecessors() can not split blocks that have such terminators,
and in two other places we already ensure that we don't end up calling
SplitBlockPredecessors() on such blocks. Do so in one more place.
Fixes https://bugs.llvm.org/show_bug.cgi?id=46857
This patch adds folding freeze into phi if it has only one operand to target.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D84601
This is a simple patch that makes canCreateUndefOrPoison use
Instruction::isBinaryOp because BinaryOperator inherits Instruction.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D84596
Reapply with DTU update moved after CFG update, which is a
requirement of the API.
-----
Non-feasible control-flow edges are currently removed by replacing
the branch condition with a constant and then calling
ConstantFoldTerminator. This happens in a rather roundabout manner,
by inspecting the users (effectively: predecessors) of unreachable
blocks, and further complicated by the need to explicitly materialize
the condition for "forced" edges. I would like to extend SCCP to
discard switch conditions that are non-feasible based on range
information, but this is incompatible with the current approach
(as there is no single constant we could use.)
Instead, this patch explicitly removes non-feasible edges. It
currently only needs to handle the case where there is a single
feasible edge. The llvm_unreachable() branch will need to be
implemented for the aforementioned switch improvement.
Differential Revision: https://reviews.llvm.org/D84264
This patch updates IPSCCP to drop argmemonly and
inaccessiblemem_or_argmemonly if it replaces a pointer argument.
Fixes PR46717.
Reviewers: efriedma, davide, nikic, jdoerfert
Reviewed By: efriedma, jdoerfert
Differential Revision: https://reviews.llvm.org/D84432
Function entry count might be zero after the profile counts reset and
before reentry to the function.
Zero profile entry count is very bad as the profile count from BFI will
be wrong.
A simple fix is to set the profile entry count to 1 if there are
non-zero profile counts in this function.
Differential Revision: https://reviews.llvm.org/D84378
Skip profile count promotion if any of the ExitBlocks contains a ret
instruction. This is to prevent dumping of incomplete profile -- if the
the loop is a long running loop and dump is called in the middle
of the loop, the result profile is incomplete.
ExitBlocks containing a ret instruction is an indication of a long running
loop -- early exit to error handling code.
Differential Revision: https://reviews.llvm.org/D84379
This is the second of two patches to address PR46753. We basically allow
SROA to promote allocas that are used in doppable instructions, for
now that means `llvm.assume`. The (transitive) uses are replaced by
`undef` in the droppable instructions.
See also D83976.
Reviewed By: Tyker
Differential Revision: https://reviews.llvm.org/D83978
This is the first of two patches to address PR46753. We basically allow
mem2reg to promote allocas that are used in doppable instructions, for
now that means `llvm.assume`. The uses of the alloca (or a bitcast or
zero offset GEP from there) are replaced by `undef` in the droppable
instructions.
Reviewed By: Tyker
Differential Revision: https://reviews.llvm.org/D83976
SROA knows that it can look through addrspacecast but
PromoteMemoryToRegister did not handle them. This caused an assertion
error for the test case, exposed while running
`Transforms/PhaseOrdering/inlining-alignment-assumptions.ll` with D83978
applied.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D84085
Summary: To match NewPM name. Also the new name is clearer and more consistent.
Subscribers: jvesely, nhaehnle, hiraditya, asbirlea, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D84542
Make sure we do not call
constainsConstantExpression/containsUndefElement on ConstantExpression,
which is not supported.
In particular, containsUndefElement/constainsConstantExpression are only
supported on constants which are supported by getAggregateElement.
Unfortunately there's no convenient way to check if a constant supports
getAggregateElement, so just check for non-constantexpressions with
vector type. Other users of those functions do so too.
Reviewers: spatel, nikic, craig.topper, lebedev.ri, jdoerfert, aqjune
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D84512
It breaks stage-2 build. Clang crashed when compiling
llvm/lib/Target/Hexagon/HexagonFrameLowering.cpp
llvm/Support/GenericDomTree.h eraseNode: Node is not a leaf node
This patch adds the ability to peel off iterations of the first loop in loop
fusion. This can allow for both loops to have the same trip count, making it
legal for them to be fused together.
Here is a simple scenario peeling can be used in loop fusion:
for (i = 0; i < 10; ++i)
a[i] = a[i] + 3;
for (j = 1; j < 10; ++j)
b[j] = b[j] + 5;
Here is we can make use of peeling, and then fuse the two loops together. We
can peel off the 0th iteration of the loop i, and then combine loop i and j for
i = 1 to 10.
a[0] = a[0] +3;
for (i = 1; i < 10; ++i) {
a[i] = a[i] + 3;
b[i] = b[i] + 5;
}
Currently peeling with loop fusion is only supported for loops with constant
trip counts and a single exit point. Both unguarded and guarded loops are
supported.
Reviewed By: bmahjour (Bardia Mahjour), MaskRay (Fangrui Song)
Differential Revision: https://reviews.llvm.org/D82927
Non-feasible control-flow edges are currently removed by replacing
the branch condition with a constant and then calling
ConstantFoldTerminator. This happens in a rather roundabout manner,
by inspecting the users (effectively: predecessors) of unreachable
blocks, and further complicated by the need to explicitly materialize
the condition for "forced" edges. I would like to extend SCCP to
discard switch conditions that are non-feasible based on range
information, but this is incompatible with the current approach
(as there is no single constant we could use.)
Instead, this patch explicitly removes non-feasible edges. It
currently only needs to handle the case where there is a single
feasible edge. The llvm_unreachable() branch will need to be
implemented for the aforementioned switch improvement.
Differential Revision: https://reviews.llvm.org/D84264
As long as RenamedOp is not guaranteed to be accurate, we cannot
assert here and should just return false. This was already done
for the other conditions in this function.
Fixes https://bugs.llvm.org/show_bug.cgi?id=46814.
The test is quite frafile, as the check lines match IR numbers and it is
not obvious why only a very small subset is checked.
Re-generate check lines, so further changes are more obvious.
This reverts commit 4a539faf74.
There is a __llvm_profile_instrument_range related crash in PGO-instrumented clang:
```
(gdb) bt
llvm::ConstantRange const&, llvm::APInt const&, unsigned int, bool) ()
llvm::ScalarEvolution::getRangeForAffineAR(llvm::SCEV const*, llvm::SCEV
const*, llvm::SCEV const*, unsigned int) ()
```
(The body of __llvm_profile_instrument_range is inlined, so we can only find__llvm_profile_instrument_target in the trace)
```
23│ 0x000055555dba0961 <+65>: nopw %cs:0x0(%rax,%rax,1)
24│ 0x000055555dba096b <+75>: nopl 0x0(%rax,%rax,1)
25│ 0x000055555dba0970 <+80>: mov %rsi,%rbx
26│ 0x000055555dba0973 <+83>: mov 0x8(%rsi),%rsi # %rsi=-1 -> SIGSEGV
27│ 0x000055555dba0977 <+87>: cmp %r15,(%rbx)
28│ 0x000055555dba097a <+90>: je 0x55555dba0a76 <__llvm_profile_instrument_target+342>
```
This patch includes the supporting code that enables always
instrumenting the function entry block by default.
This patch will NOT the default behavior.
It adds a variant bit in the profile version, adds new directives in
text profile format, and changes llvm-profdata tool accordingly.
This patch is a split of D83024 (https://reviews.llvm.org/D83024)
Many test changes from D83024 are also included.
Differential Revision: https://reviews.llvm.org/D84261
And adjust the indbrtest4 test to actually test what it's supposed
to. BB1 is supposed to be eliminated here, but isn't, because
BB0 still branches to it. This was lost due to the incomplete CHECK
lines.
For a long time, the InstCombine pass handled target specific
intrinsics. Having target specific code in general passes was noted as
an area for improvement for a long time.
D81728 moves most target specific code out of the InstCombine pass.
Applying the target specific combinations in an extra pass would
probably result in inferior optimizations compared to the current
fixed-point iteration, therefore the InstCombine pass resorts to newly
introduced functions in the TargetTransformInfo when it encounters
unknown intrinsics.
The patch should not have any effect on generated code (under the
assumption that code never uses intrinsics from a foreign target).
This introduces three new functions:
TargetTransformInfo::instCombineIntrinsic
TargetTransformInfo::simplifyDemandedUseBitsIntrinsic
TargetTransformInfo::simplifyDemandedVectorEltsIntrinsic
A few target specific parts are left in the InstCombine folder, where
it makes sense to share code. The largest left-over part in
InstCombineCalls.cpp is the code shared between arm and aarch64.
This allows to move about 3000 lines out from InstCombine to the targets.
Differential Revision: https://reviews.llvm.org/D81728
v3i16 and v3f16 currently cannot be legalized and lowered so they should
not be emitted by inst combining.
Moved the check down to still allow extracting 1 or 2 elements via the dmask.
Fixes image intrinsics being combined to return v3x16.
Differential Revision: https://reviews.llvm.org/D84223
We do not thread blocks with convergent calls, but this check was missing
when we decide to insert PR Phis into it (which we only do for threading).
Differential Revision: https://reviews.llvm.org/D83936
Reviewed By: nikic
store (load float*) can be optimized to store(load i32*) in InstCombine pass.
Add store (load float*) to isProfitableToHoist to make sure we don't break
the opt in InstCombine pass.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D82341
.. in isGuaranteedNotToBeUndefOrPoison.
This caused early exit of isGuaranteedNotToBeUndefOrPoison, making it return
imprecise result.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D84251
This reverts commit bb8850d34d.
It broke 3 check-llvm-transforms-loopfusion tests in an ASAN build.
LoopFuse.cpp `for (BasicBlock *Pred : predecessors(BB)) {` may operate on a deleted BB.
Summary:
This patch adds the ability to peel off iterations of the first loop in loop
fusion. This can allow for both loops to have the same trip count, making it
legal for them to be fused together.
Here is a simple scenario peeling can be used in loop fusion:
for (i = 0; i < 10; ++i)
a[i] = a[i] + 3;
for (j = 1; j < 10; ++j)
b[j] = b[j] + 5;
Here is we can make use of peeling, and then fuse the two loops together. We can
peel off the 0th iteration of the loop i, and then combine loop i and j for
i = 1 to 10.
a[0] = a[0] +3;
for (i = 1; i < 10; ++i) {
a[i] = a[i] + 3;
b[i] = b[i] + 5;
}
Currently peeling with loop fusion is only supported for loops with constant
trip counts and a single exit point. Both unguarded and guarded loops are
supported.
Author: sidbav (Sidharth Baveja)
Reviewers: kbarton, Meinersbur, bkramer, Whitney, skatkov, ashlykov, fhahn, bmahjour
Reviewed By: bmahjour
Subscribers: bmahjour, mgorny, hiraditya, zzheng
Tags: LLVM
Differential Revision: https://reviews.llvm.org/D82927
The new implementation makes it clear that there are exactly two
conditional stores (after the initial no-op optimization). By contrast
the old implementation had seven conditionals, some hidden inside other
functions.
This commit can change the order of operands in operand lists, hence the
tweak to one test case.
Differential Revision: https://reviews.llvm.org/D80116
If we inferred a range for the function return value, we can add !range
at all call-sites of the function, if the range does not include undef.
Reviewers: efriedma, davide, nikic
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D83952
This patch uses the TileInfo introduced in D77550 to generate a loop
nest for tiled matrix multiplication, instead of generating the
unrolled code for the whole multiplication. This makes code-generation
more scalable for larger matrixes.
Initially loops are only used if both the number of rows and columns are
divisible by the tile size. Other cases will be added as follow-up.
Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor, LuoYuanke, nicolasvasilache
Reviewed By: anemet
Differential Revision: https://reviews.llvm.org/D81308
Summary:
This is the InlineAdvisor used in 'development' mode. It enables two
scenarios:
- loading models via a command-line parameter, thus allowing for rapid
training iteration, where models can be used for the next exploration
phase without requiring recompiling the compiler. This trades off some
compilation speed for the added flexibility.
- collecting training logs, in the form of tensorflow.SequenceExample
protobufs. We generate these as textual protobufs, which simplifies
generation and testing. The protobufs may then be readily consumed by a
tensorflow-based training algorithm.
To speed up training, training logs may also be collected from the
'default' training policy. In that case, this InlineAdvisor does not
use a model.
RFC: http://lists.llvm.org/pipermail/llvm-dev/2020-April/140763.html
Reviewers: jdoerfert, davidxl
Subscribers: mgorny, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83733
This allows tracking the in-memory type of a pointer argument to a
function for ABI purposes. This is essentially a stripped down version
of byval to remove some of the stack-copy implications in its
definition.
This includes the base IR changes, and some tests for places where it
should be treated similarly to byval. Codegen support will be in a
future patch.
My original attempt at solving some of these problems was to repurpose
byval with a different address space from the stack. However, it is
technically permitted for the callee to introduce a write to the
argument, although nothing does this in reality. There is also talk of
removing and replacing the byval attribute, so a new attribute would
need to take its place anyway.
This is intended avoid some optimization issues with the current
handling of aggregate arguments, as well as fixes inflexibilty in how
frontends can specify the kernel ABI. The most honest representation
of the amdgpu_kernel convention is to expose all kernel arguments as
loads from constant memory. Today, these are raw, SSA Argument values
and codegen is responsible for turning these into loads.
Background:
There currently isn't a satisfactory way to represent how arguments
for the amdgpu_kernel calling convention are passed. In reality,
arguments are passed in a single, flat, constant memory buffer
implicitly passed to the function. It is also illegal to call this
function in the IR, and this is only ever invoked by a driver of some
kind.
It does not make sense to have a stack passed parameter in this
context as is implied by byval. It is never valid to write to the
kernel arguments, as this would corrupt the inputs seen by other
dispatches of the kernel. These argumets are also not in the same
address space as the stack, so a copy is needed to an alloca. From a
source C-like language, the kernel parameters are invisible.
Semantically, a copy is always required from the constant argument
memory to a mutable variable.
The current clang calling convention lowering emits raw values,
including aggregates into the function argument list, since using
byval would not make sense. This has some unfortunate consequences for
the optimizer. In the aggregate case, we end up with an aggregate
store to alloca, which both SROA and instcombine turn into a store of
each aggregate field. The optimizer never pieces this back together to
see that this is really just a copy from constant memory, so we end up
stuck with expensive stack usage.
This also means the backend dictates the alignment of arguments, and
arbitrarily picks the LLVM IR ABI type alignment. By allowing an
explicit alignment, frontends can make better decisions. For example,
there's real no advantage to an aligment higher than 4, so a frontend
could choose to compact the argument layout. Similarly, there is a
high penalty to using an alignment lower than 4, so a frontend could
opt into more padding for small arguments.
Another design consideration is when it is appropriate to expose the
fact that these arguments are all really passed in adjacent
memory. Currently we have a late IR optimization pass in codegen to
rewrite the kernel argument values into explicit loads to enable
vectorization. In most programs, unrelated argument loads can be
merged together. However, exposing this property directly from the
frontend has some disadvantages. We still need a way to track the
original argument sizes and alignments to report to the driver. I find
using some side-channel, metadata mechanism to track this
unappealing. If the kernel arguments were exposed as a single buffer
to begin with, alias analysis would be unaware that the padding bits
betewen arguments are meaningless. Another family of problems is there
are still some gaps in replacing all of the available parameter
attributes with metadata equivalents once lowered to loads.
The immediate plan is to start using this new attribute to handle all
aggregate argumets for kernels. Long term, it makes sense to migrate
all kernel arguments, including scalars, to be passed indirectly in
the same manner.
Additional context is in D79744.
This patch adds a new variant of the matrix lowering pass that only does
a minimal lowering and only depends on TTI. The main purpose of this pass
is to have a pass with minimal dependencies to run as part of the backend
pipeline.
At the moment, the only difference to the regular lowering pass is that it
does not support remarks. But in subsequent patches add support for tiling
to the lowering pass which will require more analysis, which we do not want
to run in the backend, as the lowering should happen in the middle-end in
practice and running it in the backend is mostly for convenience when
running llc.
Reviewers: anemet, Gerolf, efriedma, hfinkel
Reviewed By: anemet
Differential Revision: https://reviews.llvm.org/D76867
Common code sinking is already guarded with a (with default-off!) flag,
so add a flag for hoisting, too.
D84108 will hopefully make hoisting off-by-default too.
Summary:
All tests are updated, except wrapper.ll since it is not working nicely
with newly created functions.
Reviewers: jdoerfert, uenoku, baziotis, homerdin
Subscribers: arphaman, jfb, kuter, bbn, okura, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D84130
Summary:
This change added a new inline advisor that takes optimization remarks from previous inlining as input, and provides the decision as advice so current inlining can replay inline decisions of a different compilation. Dwarf inline stack with line and discriminator is used as anchor for call sites including call context. The change can be useful for Inliner tuning as it provides a channel to allow external input for tweaking inline decisions. Existing alternatives like alwaysinline attribute is per-function, not per-callsite. Per-callsite inline intrinsic can be another solution (not yet existing), but it's intrusive to implement and also does not differentiate call context.
A switch -sample-profile-inline-replay=<inline_remarks_file> is added to hook up the new inline advisor with SampleProfileLoader's inline decision for replay. Since SampleProfileLoader does top-down inlining, inline decision can be specialized for each call context, hence we should be able to replay inlining accurately. However with a bottom-up inliner like CGSCC inlining, the replay can be limited due to lack of specialization for different call context. Apart from that limitation, the new inline advisor can still be used by regular CGSCC inliner later if needed for tuning purpose.
Subscribers: mgorny, aprantl, hiraditya, llvm-commits
Tags: #llvm
Resubmit for https://reviews.llvm.org/D84086
Both users of predicteinfo (NewGVN and SCCP) are interested in
getting a cmp constraint on the predicated value. They currently
implement separate logic for this. This patch adds a common method
for this in PredicateBase.
This enables a missing bit of PredicateInfo handling in SCCP: Now
the predicate on the condition itself is also used. For switches
it means we know that the switched-on value is the same as the case
value. For assumes/branches we know that the condition is true or
false.
Differential Revision: https://reviews.llvm.org/D83640
This is a step towards trying to remove unnecessary FP compares
with infinity when compiling with -ffinite-math-only or similar.
I'm intentionally not checking FMF on the fcmp itself because
I'm assuming that will go away eventually.
The analysis part of this was added with rGcd481136 for use with
isKnownNeverNaN. Similarly, that could be an enhancement here to
get predicates like 'one' and 'ueq'.
Differential Revision: https://reviews.llvm.org/D84035
Fixes https://bugs.llvm.org/show_bug.cgi?id=46680.
Just like insertions through IRBuilder, InsertNewInstBefore()
should be using the deferred worklist mechanism, so that processing
of newly added instructions is prioritized.
There's one side-effect of the worklist order change which could be
classified as a regression. An add op gets pushed through a select
that at the time is not a umax. We could add a reverse transform
that tries to push adds in the reverse direction to restore a min/max,
but that seems like a sure way of getting infinite loops... Seems
like something that should best wait on min/max intrinsics.
Differential Revision: https://reviews.llvm.org/D84109
Summary:
This patch adds more function attribute information to the runtime function definitions in OMPKinds.def. The goal is to provide sufficient information about OpenMP runtime functions to perform more optimizations on OpenMP code.
Reviewers: jdoerfert
Subscribers: aaron.ballman cfe-commits yaxunl guansong sstefan1 llvm-commits
Tags: #OpenMP #clang #LLVM
Differential Revision: https://reviews.llvm.org/D81031
Yes, if operands are non-positive this comes at the extra cost
of two extra negations. But a. division is already just
ridiculously costly, two more subtractions can't hurt much :)
and b. we have better/more analyzes/folds for an unsigned division,
we could end up narrowing it's bitwidth, converting it to lshr, etc.
This is essentially a take two on 0fdcca07ad,
which didn't fix the potential regression i was seeing,
because ValueTracking's computeKnownBits() doesn't make use
of dominating conditions in it's analysis.
While i could teach it that, this seems like the more general fix.
This big hammer actually does catch said potential regression.
Over vanilla test-suite + RawSpeed + darktable
(10M IR instrs, 1M IR BB, 1M X86 ASM instrs), this fires/converts 5 more
(+2%) SDiv's, the total instruction count at the end of middle-end pipeline
is only +6, so out of +10 extra negations, ~half are folded away,
and asm instr count is only +1, so practically speaking all extra
negations are folded away and are therefore free.
Sadly, all these new UDiv's remained, none folded away.
But there are two less basic blocks.
https://rise4fun.com/Alive/VS6
Name: v0
Pre: C0 >= 0 && C1 >= 0
%r = sdiv i8 C0, C1
=>
%r = udiv i8 C0, C1
Name: v1
Pre: C0 <= 0 && C1 >= 0
%r = sdiv i8 C0, C1
=>
%t0 = udiv i8 -C0, C1
%r = sub i8 0, %t0
Name: v2
Pre: C0 >= 0 && C1 <= 0
%r = sdiv i8 C0, C1
=>
%t0 = udiv i8 C0, -C1
%r = sub i8 0, %t0
Name: v3
Pre: C0 <= 0 && C1 <= 0
%r = sdiv i8 C0, C1
=>
%r = udiv i8 -C0, -C1
Currently that fold requires both operands to be non-negative,
but the only real requirement for the fold is that we must know
the domains of the operands.
as it's causing a few unused variable warnings via the macro instantiation:
sources/llvm-project/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def:649:17: error: unused variable 'InaccessibleOnlyAttrs' [-Werror,-Wunused-variable]
__OMP_ATTRS_SET(InaccessibleOnlyAttrs,
^
This reverts commit 09fe0c5ab9.
Summary:
This patch adds more function attribute information to the runtime function definitions in OMPKinds.def. The goal is to provide sufficient information about OpenMP runtime functions to perform more optimizations on OpenMP code.
Reviewers: jdoerfert
Subscribers: aaron.ballman cfe-commits yaxunl guansong sstefan1 llvm-commits
Tags: #OpenMP #clang #llvm
Differential Revision: https://reviews.llvm.org/D81031
Summary:
This change added a new inline advisor that takes optimization remarks for previous inlining as input, and provide the decision as advice so current inlining can replay inline decision of a different compilation. Dwarf inline stack with line and discriminator is used as anchor for call sites. The change can be useful for Inliner tuning.
A switch -sample-profile-inline-replay=<inline_remarks_file> is added to hook up the new inliner advisor with SampleProfileLoader's inline decision for replay. The new inline advisor can also be used by regular CGSCC inliner later if needed.
Reviewers: davidxl, mtrofin, wmi, hoy
Subscribers: aprantl, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83743
This is the one i'm seeing as missed optimization,
although there are likely other possibilities, as usual.
There are 4 variants of a general sdiv->udiv fold:
https://rise4fun.com/Alive/VS6
Name: v0
Pre: C0 >= 0 && C1 >= 0
%r = sdiv i8 C0, C1
=>
%r = udiv i8 C0, C1
Name: v1
Pre: C0 <= 0 && C1 >= 0
%r = sdiv i8 C0, C1
=>
%t0 = udiv i8 -C0, C1
%r = sub i8 0, %t0
Name: v2
Pre: C0 >= 0 && C1 <= 0
%r = sdiv i8 C0, C1
=>
%t0 = udiv i8 C0, -C1
%r = sub i8 0, %t0
Name: v3
Pre: C0 <= 0 && C1 <= 0
%r = sdiv i8 C0, C1
=>
%r = udiv i8 -C0, -C1
If we really don't like sdiv (more than udiv that is),
and are okay with increasing instruction count (2 new negations),
and we ensure that we don't undo the fold,
then we could just implement these..
It was getting difficult to see which test was in which file, so this
reorganises the test files so that now all filenames start with tail-folding-*
followed by a more descriptive name what that group of tests check.
Summary:
This patch resolves an issue where the metadata of a loop is not added to the
new loop latch, and not removed from the old loop latch. This issue occurs in
the SplitBlockPredecessors function, which adds a new block in a loop, and
in the case that the block passed into this function is the header of the loop,
the loop can be modified such that the latch of the loop is replaced.
This patch applies to the Loop Simplify pass since it ensures that each loop
has exit blocks which only have predecessors that are inside of the loop. In
the case that this is not true, the pass will create a new exit block for the
loop. This guarantees that the loop preheader/header will dominate the exit blocks.
Author: sidbav (Sidharth Baveja)
Reviewers: asbirlea (Alina Sbirlea), chandlerc (Chandler Carruth), Whitney (Whitney Tsang), bmahjour (Bardia Mahjour)
Reviewed By: asbirlea (Alina Sbirlea)
Subscribers: hiraditya (Aditya Kumar), llvm-commits
Tag: LLVM
Differential Revision: https://reviews.llvm.org/D83869
This patch enables the LoopVectorizer to build a phi of pointer
type and provide the vector loads and stores with vector type
getelementptrs built from the pointer induction variable, which
produces much less instructions than the previous approach of
creating scalar getelementpointers and glue them together to a
vector.
Differential Revision: https://reviews.llvm.org/D81267
An additional test that allows to check the correctness of handling the case of the same
branch labels in the dominator when trying to replace select with phi-node.
Patch By: Kirill Polushin
Differential Revision: https://reviews.llvm.org/D84006
Reviewed By: mkazantsev
SimplifyCFG was incorrectly reporting to the pass manager that it had not made
changes after folding away a PHI. This is detected in the EXPENSIVE_CHECKS
build when the function's hash changes.
Differential Revision: https://reviews.llvm.org/D83985
The IR doesn't have a proper concept of invalid pointers, and "null"
constants are just all zeros (though it really needs one).
I think it's not possible to break this for AMDGPU due to the copy
semantics of byval. If you have an original stack object at 0, the
byval copy will be placed above it so I don't think it's really
possible to hit a 0 address.
This patch adds simplification for pattern:
```
if (cond)
/ \
... ...
\ /
p = phi [true] [false]
...
br p, succ_1, succ_2
```
If we can prove that top block's branches dominate respective
inputs of a block that has a Phi with constant inputs, we can
use the branch condition (maybe inverted) instead of Phi.
This will make proofs of implication for further jump threading
more transparent.
Differential Revision: https://reviews.llvm.org/D81375
Reviewed By: xbolva00
This reverts most of the following patches due to reports of miscompiles.
I've left the added test cases with comments updated to be FIXMEs.
1cf6f210a2 [IR] Disable select ? C : undef -> C fold in ConstantFoldSelectInstruction unless we know C isn't poison.
469da663f2 [InstSimplify] Re-enable select ?, undef, X -> X transform when X is provably not poison
122b0640fc [InstSimplify] Don't fold vectors of partial undef in SimplifySelectInst if the non-undef element value might produce poison
ac0af12ed2 [InstSimplify] Add test cases for opportunities to fold select ?, X, undef -> X when we can prove X isn't poison
9b1e95329a [InstSimplify] Remove select ?, undef, X -> X and select ?, X, undef -> X transforms
Unrolling a loop with compile-time unknown trip count results in a remainder loop. The remainder loop executes the remaining iterations of the original loop when the original trip count is not a multiple of the unroll factor. For better profile counts maintenance throughout the optimization pipeline, I'm assigning an artificial weight to the latch branch of the remainder loop.
A remainder loop runs up to as many times as the unroll factor subtracted by 1. Therefore I'm assigning the maximum possible trip count as the back edge weight. This should be more accurate than the default non-profile weight, which assumes the back edge runs much more frequently than the exit edge.
Differential Revision: https://reviews.llvm.org/D83187
I'm not sure if the test is truly minimal, but we need to
induce a situation where a value becomes a constant but is
not immediately folded before getting to the 'or' transform.
We need to specify legal integer widths to trigger PR46712,
so add those here. This doesn't appear to affect any existing
tests, and it's not clear why a datalayout would not include
any legal integer widths.
While here, change some variable names that include 'tmp' to
avoid warnings from the auto-generating script for CHECK lines.
Extend the memop value profile buckets to be more flexible (could accommodate a
mix of individual values and ranges) and to cover more value ranges (from 11 to
22 buckets).
Disabled behind a flag (to be enabled separately) and the existing code to be
removed later.
CodeGenPrepare keeps fairly close track of various instructions it's
seen, particularly GEPs, in maps and vectors. However, sometimes those
instructions become dead and get removed while it's still executing.
This triggers AssertingVH references to them in an asserts build and
could lead to miscompiles in a release build (I've only seen a later
segfault though).
So this patch adds a callback to
RecursivelyDeleteTriviallyDeadInstructions which can make sure the
instruction about to be deleted is removed from CodeGenPrepare's data
structures.
This fixes an instance where MemorySSA-using Dead Store Elimination is failing
to do a transformation that the non-MemorySSA-using version does.
Differential Revision: https://reviews.llvm.org/D83783
Some of the system registers readable on AArch64 and ARM platforms
return different values with each read (for example a timer counter),
these shouldn't be hoisted outside loops or otherwise interfered with,
but the normal @llvm.read_register intrinsic is only considered to read
memory.
This introduces a separate @llvm.read_volatile_register intrinsic and
maps all system-registers on ARM platforms to use it for the
__builtin_arm_rsr calls. Registers declared with asm("r9") or similar
are unaffected.
Since D82572, we keep "reference" edges for callback call sites. While
not strictly necessary they can improve the traversal order. However, we
did not update them properly in case a pass removed the callback call
site which caused a verification error (PR46687). With this patch we
update these reference edges properly during the invocation of
`CallGraphSCCPass::RefreshCallGraph` in non-checking mode.
Reviewed By: sdmitriev
Differential Revision: https://reviews.llvm.org/D83718
Summary: This patch added dependency graph to the attributor so that we can dump the dependencies between AAs more easily. We can also apply general graph algorithms to the graph, making it easier for us to create deep wrappers.
Reviewers: jdoerfert, sstefan1, uenoku, homerdin, baziotis
Reviewed By: jdoerfert
Subscribers: jfb, okura, mgrang, kuter, lebedev.ri, hiraditya, uenoku, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78861
Summary: This patch added dependency graph to the attributor so that we can dump the dependencies between AAs more easily. We can also apply general graph algorithms to the graph, making it easier for us to create deep wrappers.
Reviewers: jdoerfert, sstefan1, uenoku, homerdin, baziotis
Reviewed By: jdoerfert
Subscribers: jfb, okura, mgrang, kuter, lebedev.ri, hiraditya, uenoku, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D78861
Summary:
Add debug counter and stats counter to assume queries and assume builder
here is the collected stats on a build of check-llvm + check-clang.
"assume-builder.NumAssumeBuilt": 2720879,
"assume-builder.NumAssumesMerged": 761396,
"assume-builder.NumAssumesRemoved": 1576212,
"assume-builder.NumBundlesInAssumes": 6518809,
"assume-queries.NumAssumeQueries": 85566380,
"assume-queries.NumUsefullAssumeQueries": 2727360,
the NumUsefullAssumeQueries stat is actually pessimistic because in a few places queries
ask to keep providing information to try to get better information. and this isn't counted
as a usefull query evem tho it can be usefull
Reviewers: jdoerfert
Reviewed By: jdoerfert
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83506
This restores commit 80d0a137a5, and the
follow on fix in 873c0d0786, with a new
fix for test failures after a 2-stage clang bootstrap, and a more robust
fix for the Chromium build failure that an earlier version partially
fixed. See also discussion on D75201.
Reviewers: evgeny777
Subscribers: mehdi_amini, Prazek, hiraditya, steven_wu, dexonsmith, arphaman, davidxl, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D73242
This changes the matrix load/store intrinsic definitions to load/store from/to
a pointer, and not from/to a pointer to a vector, as discussed in D83477.
This also includes the recommit of "[Matrix] Tighten LangRef definitions and
Verifier checks" which adds improved language reference descriptions of the
matrix intrinsics and verifier checks.
Differential Revision: https://reviews.llvm.org/D83785
If a vector body has live-out values, it is probably a reduction, which needs a
final reduction step after the loop. MVE has a VADDV instruction to reduce
integer vectors, but doesn't have an equivalent one for float vectors. A
live-out value that is not recognised as reduction later in the optimisation
pipeline will result in the tail-predicated loop to be reverted to a
non-predicated loop and this is very expensive, i.e. it has a significant
performance impact, which is what we hope to avoid with fine tuning the ARM TTI
hook preferPredicateOverEpilogue implementation.
Differential Revision: https://reviews.llvm.org/D82953
Here we teach the ConstantFolding analysis pass that it is not legal to
replace a load of a bitcast constant (having a non-integral addrspace)
with a bitcast of the value of that constant (with a different
non-integral addrspace).
But also teach it that certain bit patterns are always known and
convertable (a fact it already uses elsewhere). This required us to also
fix a globalopt test, since, after this change, LLVM is able to realize
that the test actually is a valid transform (NULL is always a known
bit-pattern) and so it doesn't need to emit the failure remarks for it.
Also simplify some of the negative tests for transforms by avoiding a
type change in their bitcast, and add positive versions of the same
tests, to show that they otherwise should work.
Differential Revision: https://reviews.llvm.org/D59730
Summary:
NOTE: There is a mailing list discussion on this: http://lists.llvm.org/pipermail/llvm-dev/2019-December/137632.html
Complemantary to the assumption outliner prototype in D71692, this patch
shows how we could simplify the code emitted for an alignemnt
assumption. The generated code is smaller, less fragile, and it makes it
easier to recognize the additional use as a "assumption use".
As mentioned in D71692 and on the mailing list, we could adopt this
scheme, and similar schemes for other patterns, without adopting the
assumption outlining.
Reviewers: hfinkel, xbolva00, lebedev.ri, nikic, rjmccall, spatel, jdoerfert, sstefan1
Reviewed By: jdoerfert
Subscribers: thopre, yamauchi, kuter, fhahn, merge_guards_bot, hiraditya, bollu, rkruppe, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D71739
Currently, a transformation like pow(2.0, x) -> exp2(x) copies the pow
attribute list verbatim and applies it to exp2. This works out fine
when the attribute list is empty, but when it isn't clang may error due
due to the mismatch.
The source function and destination don't necessarily have anything
to do with one another, attribute-wise. So it makes sense to remove
the attribute lists (this is similar to what IPO does in this
situation).
This was discovered after implementing the `noundef` param attribute.
Differential Revision: https://reviews.llvm.org/D82820
This reverts commit 9649c2095f. See
discussion on the llvm-commits thread: if it's OK to preserve the
location when sinking a call, it's probably OK to always preserve the
location.
Place the ssa.copy instructions for assumes after the assume,
instead of before it. Both options are valid, but placing them
afterwards prevents assumes from being replaced with assume(true).
This fixes https://bugs.llvm.org/show_bug.cgi?id=37541 in NewGVN
and will avoid a similar issue in SCCP when we handle more
predicate infos.
Differential Revision: https://reviews.llvm.org/D83631
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
This refactors option -disable-mve-tail-predication to take different arguments
so that we have 1 option to control tail-predication rather than several
different ones.
This is also a prep step for D82953, in which we want to reject reductions
unless that is requested with this option.
Differential Revision: https://reviews.llvm.org/D83133
We can try to replace select with a Phi not in its parent block alone,
but also in blocks of its arguments. We benefit from it when select's
argument is a Phi.
Differential Revision: https://reviews.llvm.org/D83284
Reviewed By: nikic
Similar to rG40fcc42:
The base case only worked because we were relying on a
poison-unsafe select transform; if that is fixed, we
would regress on patterns like this.
The extra use tests show that the select transform can't
be applied consistently. So it may be a regression to have
an extra instruction on 1 test, but that result was not
created safely and does not happen reliably.
This patch fixes D81345 and PR46652.
If a loop with a small trip count is compiled w/o -Os/-Oz, Loop Access Analysis
still generates runtime checks for unit strides that will version the loop.
In such cases, the loop vectorizer should either re-run the analysis or bail-out
from vectorizing the loop, as done prior to D81345. The latter is applied for
now as the former requires refactoring.
Differential Revision: https://reviews.llvm.org/D83470
Summary:
- Skip unreachable predecessors during header detection in SCC. Those
unreachable blocks would be generated in the switch lowering pass in
the corner cases or other frontends. Even though they could be removed
through the CFG simplification, we should skip them during header
detection.
Reviewers: sameerds
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83562
The current implementation of Tail Recursion Elimination has a very restricted
pre-requisite: AllCallsAreTailCalls. i.e. it requires that no function
call receives a pointer to local stack. Generally, function calls that
receive a pointer to local stack but do not capture it - should not
break TRE. This fix allows us to do TRE if it is proved that no pointer
to the local stack is escaped.
Reviewed by: efriedma
Differential Revision: https://reviews.llvm.org/D82085
In non-SPMD mode we create a state machine like code to identify the
parallel region the GPU worker threads should execute next. The
identification uses the parallel region function pointer as that allows
it to work even if the kernel (=target region) and the parallel region
are in separate TUs. However, taking the address of a function comes
with various downsides. With this patch we will identify the most common
situation and replace the function pointer use with a dummy global
symbol (for identification purposes only). That means, if the parallel
region is only called from a single target region (or kernel), we do not
use the function pointer of the parallel region to identify it but a new
global symbol.
Fixes PR46450.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D83271
We now identify GPU kernels, that is entry points into the GPU code.
These kernels (can) correspond to OpenMP target regions. With this patch
we identify and on request print them via remarks.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D83269
This reverts commit 1d542f0ca8.
`recollectUses()` is added to prevent looking at dead uses after
Attributor run.
This is the first and most basic ICV Tracking implementation. For this
first version, we only support deduplication within the same BB.
Reviewers: jdoerfert, JonChesterfield, hamax97, jhuber6, uenoku,
baziotis, lebedev.ri
Differential Revision: https://reviews.llvm.org/D81788
Florian Hahn