Summary:
Without this patch, the jump threading pass ignores profiling data
whenever we invoke the pass with the new pass manager.
Specifically, JumpThreadingPass::run calls runImpl with class variable
HasProfileData always set to false. In turn, runImpl sets
HasProfileData to false again:
HasProfileData = HasProfileData_;
In the end, we don't use profiling data at all with the new pass
manager.
This patch fixes the problem by passing F.hasProfileData() to runImpl.
The bug appears to have been introduced at:
https://reviews.llvm.org/D41461
which removed local variable HasProfileData in JumpThreadingPass::run
even though there was one more use left in the same function. As a
result, the remaining use ended referring to the class variable
instead.
Note that F.hasProfileData is an extremely lightweight function, so I
don't see the need to cache its result. Once this patch is approved,
I'm planning to stop caching the result of F.hasProfileData in
runOnFunction.
Reviewers: wmi, eli.friedman
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70509
The verification inside loop passes should be done under the
VerifyMemorySSA flag (enabled by EXPESIVE_CHECKS or explicitly with
opt), in order to not add to compile time during regular builds.
We may end up with a case where we have a widenable branch above the loop, but not all widenable branches within the loop have been removed. Since a widenable branch inhibit SCEVs ability to reason about exit counts (by design), we have a tradeoff between effectiveness of this optimization and allowing future widening of the branches within the loop. LoopPred is thought to be one of the most important optimizations for range check elimination, so let's pay the cost.
Bit-Tracking Dead Code Elimination (bdce) do not mark dbg.value as undef after
deleting instruction. which shows invalid state of variable in debugger. This
patches fixes this by marking the dbg.value as undef which depends on dead
instruction.
This fixes https://bugs.llvm.org/show_bug.cgi?id=41925
Patch by kamlesh kumar!
Differential Revision: https://reviews.llvm.org/D70040
Summary:
This patch moves various checks from ThreadEdge to new function
TryThreadEdge The rational behind this is that I'd like to use
ThreadEdge without its checks in my upcoming patch.
This patch preserves lightweight checks as assertions in ThreadEdge.
ThreadEdge does not repeat the cost check, however.
Reviewers: wmi
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70338
Summary:
Most libraries are defined in the lib/ directory but there are also a
few libraries defined in tools/ e.g. libLLVM, libLTO. I'm defining
"Component Libraries" as libraries defined in lib/ that may be included in
libLLVM.so. Explicitly marking the libraries in lib/ as component
libraries allows us to remove some fragile checks that attempt to
differentiate between lib/ libraries and tools/ libraires:
1. In tools/llvm-shlib, because
llvm_map_components_to_libnames(LIB_NAMES "all") returned a list of
all libraries defined in the whole project, there was custom code
needed to filter out libraries defined in tools/, none of which should
be included in libLLVM.so. This code assumed that any library
defined as static was from lib/ and everything else should be
excluded.
With this change, llvm_map_components_to_libnames(LIB_NAMES, "all")
only returns libraries that have been added to the LLVM_COMPONENT_LIBS
global cmake property, so this custom filtering logic can be removed.
Doing this also fixes the build with BUILD_SHARED_LIBS=ON
and LLVM_BUILD_LLVM_DYLIB=ON.
2. There was some code in llvm_add_library that assumed that
libraries defined in lib/ would not have LLVM_LINK_COMPONENTS or
ARG_LINK_COMPONENTS set. This is only true because libraries
defined lib lib/ use LLVMBuild.txt and don't set these values.
This code has been fixed now to check if the library has been
explicitly marked as a component library, which should now make it
easier to remove LLVMBuild at some point in the future.
I have tested this patch on Windows, MacOS and Linux with release builds
and the following combinations of CMake options:
- "" (No options)
- -DLLVM_BUILD_LLVM_DYLIB=ON
- -DLLVM_LINK_LLVM_DYLIB=ON
- -DBUILD_SHARED_LIBS=ON
- -DBUILD_SHARED_LIBS=ON -DLLVM_BUILD_LLVM_DYLIB=ON
- -DBUILD_SHARED_LIBS=ON -DLLVM_LINK_LLVM_DYLIB=ON
Reviewers: beanz, smeenai, compnerd, phosek
Reviewed By: beanz
Subscribers: wuzish, jholewinski, arsenm, dschuff, jyknight, dylanmckay, sdardis, nemanjai, jvesely, nhaehnle, mgorny, mehdi_amini, sbc100, jgravelle-google, hiraditya, aheejin, fedor.sergeev, asb, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, jrtc27, MaskRay, zzheng, edward-jones, atanasyan, steven_wu, rogfer01, MartinMosbeck, brucehoult, the_o, dexonsmith, PkmX, jocewei, jsji, dang, Jim, lenary, s.egerton, pzheng, sameer.abuasal, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70179
As a reminder, a "widenable branch" is the pattern "br i1 (and i1 X, WC()), label %taken, label %untaken" where "WC" is the widenable condition intrinsics. The semantics of such a branch (derived from the semantics of WC) is that a new condition can be added into the condition arbitrarily without violating legality.
Broaden the definition in two ways:
Allow swapped operands to the br (and X, WC()) form
Allow widenable branch w/trivial condition (i.e. true) which takes form of br i1 WC()
The former is just general robustness (e.g. for X = non-instruction this is what instcombine produces). The later is specifically important as partial unswitching of a widenable range check produces exactly this form above the loop.
Differential Revision: https://reviews.llvm.org/D70502
Moving accesses in MemorySSA at InsertionPlace::End, when an instruction is
moved into a block, almost always means insert at the end of the block, but
before the block terminator. This matters when the block terminator is a
MemoryAccess itself (an invoke), and the insertion must be done before
the terminator for the update to be correct.
Insert an additional position: InsertionPlace:BeforeTerminator and update
current usages where this applies.
Resolves PR44027.
This is mostly NFC, but I removed the setting of the guard's calling convention onto the WC call. Why? Because it was untested, and was producing an ill defined output as the declaration's convention wasn't been changed leaving a mismatch which is UB.
This code has never been enabled. While it is tested, it's complicating some refactoring. If we decide to re-implement this, doing it in SimplifyCFG would probably make more sense anyways.
With the widenable condition construct, we have the ability to reason about branches which can be 'widened' (i.e. made to fail more often). We've got a couple o transforms which leverage this. This patch just cleans up the API a bit.
This is prep work for generalizing our definition of a widenable branch slightly. At the moment "br i1 (and A, wc()), ..." is considered widenable, but oddly, neither "br i1 (and wc(), B), ..." or "br i1 wc(), ..." is. That clearly needs addressed, so first, let's centralize the code in one place.
Unswitch (and other loop transforms) like to generate loop exit blocks with unconditional successors, and phi nodes (LCSSA, or simple multiple exiting blocks sharing an exit). Generalize the "likely very rare exit" check slightly to handle this form.
This implements a version of the predicateLoopExits transform from IndVarSimplify extended to exploit widenable conditions - and thus be much wider in scope of legality. The code structure ends up being almost entirely different, so I chose to duplicate this into the LoopPredication pass instead of trying to reuse the code in the IndVars.
The core notions of the transform are as follows:
If we have a widenable condition which controls entry into the loop, we're allowed to widen it arbitrarily. Given that, it's simply a *profitability* question as to what conditions to fold into the widenable branch.
To avoid pass ordering issues, we want to avoid widening cases that would otherwise be dischargeable. Or... widen in a form which can still be discharged. Thus, we phrase the transform as selecting one analyzeable exit from the set of analyzeable exits to keep. This avoids creating pass ordering complexities.
Since none of the above proves that we actually exit through our analyzeable exits - we might exit through something else entirely - we limit ourselves to cases where a) the latch is analyzeable and b) the latch is predicted taken, and c) the exit being removed is statically cold.
Differential Revision: https://reviews.llvm.org/D69830
It was added in 2014 in 732e0aa9fb with one use in Scalarizer.cpp.
That one use was then removed when porting to the new pass manager in
2018 in b6f76002d9.
While the RFC and the desire to get off of static initializers for
cl::opt all still stand, this code is now dead, and I think we should
delete this code until someone is ready to do the migration.
There were many clients of CommandLine.h that were it transitively
through LLVMContext.h, so I cleaned that up in 4c1a1d3cf9.
Reviewers: beanz
Differential Revision: https://reviews.llvm.org/D70280
Summary:
When scalarizing PHI nodes we might try to examine/rewrite
InsertElement nodes in predecessors. If those predecessors
are unreachable from entry, then the IR in those blocks could
have unexpected properties resulting in infinite loops in
Scatterer::operator[].
By simply treating values originating from instructions in
unreachable blocks as undef we do not need to analyse them
further.
This fixes PR41723.
Reviewers: bjope
Reviewed By: bjope
Subscribers: bjope, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70171
This file lists every pass in LLVM, and is included by Pass.h, which is
very popular. Every time we add, remove, or rename a pass in LLVM, it
caused lots of recompilation.
I found this fact by looking at this table, which is sorted by the
number of times a file was changed over the last 100,000 git commits
multiplied by the number of object files that depend on it in the
current checkout:
recompiles touches affected_files header
342380 95 3604 llvm/include/llvm/ADT/STLExtras.h
314730 234 1345 llvm/include/llvm/InitializePasses.h
307036 118 2602 llvm/include/llvm/ADT/APInt.h
213049 59 3611 llvm/include/llvm/Support/MathExtras.h
170422 47 3626 llvm/include/llvm/Support/Compiler.h
162225 45 3605 llvm/include/llvm/ADT/Optional.h
158319 63 2513 llvm/include/llvm/ADT/Triple.h
140322 39 3598 llvm/include/llvm/ADT/StringRef.h
137647 59 2333 llvm/include/llvm/Support/Error.h
131619 73 1803 llvm/include/llvm/Support/FileSystem.h
Before this change, touching InitializePasses.h would cause 1345 files
to recompile. After this change, touching it only causes 550 compiles in
an incremental rebuild.
Reviewers: bkramer, asbirlea, bollu, jdoerfert
Differential Revision: https://reviews.llvm.org/D70211
Currently we have limited support for outer loops with multiple basic
blocks after the inner loop exit. But the current checks for creating
PHIs for loop exit values only assumes the header and latches of the
outer loop. It is better to just skip incoming values defined in the
original inner loops. Those are handled earlier.
Reviewers: efriedma, mcrosier
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D70059
This patch implements a correct, but not terribly useful, transform. In particular, if we have a dynamic alloca in a loop which is guaranteed to execute, and provably not captured, we hoist the alloca out of the loop. The capture tracking is needed so that we can prove that each previous stack region dies before the next one is allocated. The transform decreases the amount of stack allocation needed by a linear factor (e.g. the iteration count of the loop).
Now, I really hope no one is actually using dynamic allocas. As such, why this patch?
Well, the actual problem I'm hoping to make progress on is allocation hoisting. There's a large draft patch out for review (https://reviews.llvm.org/D60056), and this patch was the smallest chunk of testable functionality I could come up with which takes a step vaguely in that direction.
Once this is in, it makes motivating the changes to capture tracking mentioned in TODOs testable. After that, I hope to extend this to trivial malloc free regions (i.e. free dominating all loop exits) and allocation functions for GCed languages.
Differential Revision: https://reviews.llvm.org/D69227
The change itself is straight forward and obvious, but ... there's an existing test checking for exactly the opposite. Both I and Artur think this is simply conservatism in the initial implementation. If anyone bisects a problem to this, a counter example will be very interesting.
Differential Revision: https://reviews.llvm.org/D69907
The basic idea of the transform is to convert variant loop exit conditions into invariant exit conditions by changing the iteration on which the exit is taken when we know that the trip count is unobservable. See the original patch which introduced the code for a more complete explanation.
The individual parts of this have been reviewed, the result has been fuzzed, and then further analyzed by hand, but despite all of that, I will not be suprised to see breakage here. If you see problems, please don't hesitate to revert - though please do provide a test case. The most likely class of issues are latent SCEV bugs and without a reduced test case, I'll be essentially stuck on reducing them.
(Note: A bunch of tests were opted out of the new transform to preserve coverage. That landed in a previous commit to simplify revert cycles if they turn out to be needed.)
Summary:
This patch factors out code to clone instructions -- partly for
readability and partly to facilitate an upcoming patch of my own.
Reviewers: wmi
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69861
We had a subtle, but nasty bug in our definition of a widenable branch, and thus in the transforms which used that utility. Specifically, we returned true for any branch which included a widenable condition within it's condition, regardless of whether that widenable condition also had other uses.
The problem is that the result of the WC() call is defined to be one particular value. As such, all users must agree as to what that value is. If we widen a branch without also updating *all other users* of the WC in the same way, we have broken the required semantics.
Most of the textual diff is updating existing transforms not to leave dead uses hanging around. They're largely NFC as the dead instructions would be immediately deleted by other passes. The reason to make these changes is so that the transforms preserve the widenable branch form.
In practice, we don't get bitten by this only because it isn't profitable to CSE WC() calls and the lowering pass from guards uses distinct WC calls per branch.
Differential Revision: https://reviews.llvm.org/D69916
This patch fixes two issues noticed by inspection when going to enable the loop predication code in IndVarSimplify.
Issue 1 - Both the LoopPredication transform, and the already on by default optimizeLoopExits transform, modify the exit count of the exits they modify. (either to 0 or Infinity) Looking at the code more closely, this was not reflected into SCEV and we were instead running later transforms with incorrect SCEVs. Fixing this requires forgetting the loop, weakening a too strong assert, and updating SCEV to not pessimize results when a loop is provable untaken. I haven't been able to find a test case to demonstrate the miscompile.
Issue 2 - For modules without a data layout, we can end up with unsized pointer typed exit counts. Just bail out of this case.
I think these are the last two issues which need addressed before we enable this by default. The code has already survived a decent amount of fuzzing without revealing either of the above.
Differential Revision: https://reviews.llvm.org/D69695
Summary:
This patch factors out code to merge a basic block with its sole
successor -- partly for readability and partly to facilitate an
upcoming patch of my own.
Reviewers: wmi
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69852
Summary:
This patch factors out common code to update the SSA form in
JumpThreading.cpp -- partly for readability and partly to facilitate
an coming patch of my own.
Reviewers: wmi
Subscribers: hiraditya, jfb, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69811
As discussed in https://bugs.llvm.org/show_bug.cgi?id=43870,
this transform is missing a crucial legality check:
the old (non-countable) loop would early-return upon first mismatch,
but there is no such guarantee for bcmp/memcmp.
We'd need to ensure that [PtrA, PtrA+NBytes) and [PtrB, PtrB+NBytes)
are fully dereferenceable memory regions. But that would limit
the transform to constant loop trip counts and would further
cripple it because dereferenceability analysis is *very* partial.
Furthermore, even if all that is done, every single test
would need to be rewritten from scratch.
So let's just give up.
Since SCEV can cache information about location of an instruction, it should be invalidated when the instruction is moved.
There should be similar bug in code sinking part of LICM, it will be fixed in a follow-up change.
Patch Author: Daniil Suchkov
Reviewers: asbirlea, mkazantsev, reames
Reviewed By: asbirlea
Subscribers: hiraditya, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D69370
This phi simplification transform was added with:
D45448
However as shown in PR43802:
https://bugs.llvm.org/show_bug.cgi?id=43802
...we must be careful not to propagate poison when we do the substitution.
There might be some more complicated analysis possible to retain the overflow flag,
but it should always be safe and easy to drop flags (we have similar behavior in
instcombine and other passes).
Differential Revision: https://reviews.llvm.org/D69442
We were already going to all of the trouble of computing maximum constant exit counts for each loop exit, we might as well expose them through the API. The change in IndVars is mostly to demonstrate that the wired up code works, but it als very slightly strengthens the transform. The strengthened case is rather narrow though: it requires one exactly analyzeable exit, one imprecisely analyzeable exit (with the upper bound less than the precise one), and one unanalyzeable exit. I coudn't construct a reasonably stable test case.
This does increase the memory usage of the BackedgeTakenCount by a factor of 2 in the worst case.
I also noticed the loop in IndVars is O(#Exits ^ 2). This doesn't change with this patch. A future patch will cache this result inside of SCEV to avoid requering.
The static analyzer is warning about a potential null dereference, but we should be able to use cast<> directly and if not assert will fire for us.
llvm-svn: 375429
The static analyzer is warning about a potential null dereference, but we should be able to use cast<> directly and if not assert will fire for us.
llvm-svn: 375426
Summary:
When MemCpyOpt is handling aggregate type values, if an instruction (let's call it P) between the targeting load (L) and store (S) clobbers the source pointer of L, it will try to hoist S before P. This process will also hoist S's data dependency instructions.
However, the current implementation has a bug that if one of S's dependency instructions is //also// a user of P, MemCpyOpt will not prevent it from being hoisted above P and cause a use-before-define error. For example, in the newly added test file (i.e. `aggregate-type-crash.ll`), it will try to hoist both `store %my_struct %1, %my_struct* %3` and its dependent, `%3 = bitcast i8* %2 to %my_struct*`, above `%2 = call i8* @my_malloc(%my_struct* %0)`. Creating the following BB:
```
entry:
%1 = bitcast i8* %4 to %my_struct*
%2 = bitcast %my_struct* %1 to i8*
%3 = bitcast %my_struct* %0 to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %2, i8* align 4 %3, i64 8, i1 false)
%4 = call i8* @my_malloc(%my_struct* %0)
ret void
```
Where there is a use-before-define error between `%1` and `%4`.
Update: The compiler for the Pony Programming Language [also encounter the same bug](https://github.com/ponylang/ponyc/issues/3140)
Patch by Min-Yih Hsu (myhsu)
Reviewers: eugenis, pcc, dblaikie, dneilson, t.p.northover, lattner
Reviewed By: eugenis
Subscribers: lenary, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D66060
llvm-svn: 375403
We can end up with two loop exits whose exit counts are equivalent, but whose textual representation is different and non-obvious. For the sub-case where we have a series of exits which dominate one another (common), eliminate any exits which would iterate *after* a previous exit on the exiting iteration.
As noted in the TODO being removed, I'd always thought this was a good idea, but I've now seen this in a real workload as well.
Interestingly, in review, Nikita pointed out there's let another oppurtunity to leverage SCEV's reasoning. If we kept track of the min of dominanting exits so far, we could discharge exits with EC >= MDE. This is less powerful than the existing transform (since later exits aren't considered), but potentially more powerful for any case where SCEV can prove a >= b, but neither a == b or a > b. I don't have an example to illustrate that oppurtunity, but won't be suprised if we find one and return to handle that case as well.
Differential Revision: https://reviews.llvm.org/D69009
llvm-svn: 375379
Summary:
CVP, unlike InstCombine, does not run till exaustion.
It only does a single pass.
When dealing with those special binops, if we prove that they can
safely be demoted into their usual binop form,
we do set the no-wrap we deduced. But when dealing with usual binops,
we try to deduce both no-wraps.
So if we convert e.g. @llvm.uadd.with.overflow() to `add nuw`,
we won't attempt to check whether it can be `add nuw nsw`.
This patch proposes to call `processBinOp()` on newly-created binop,
which is identical to what we do for div/rem already.
Reviewers: nikic, spatel, reames
Reviewed By: nikic
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69183
llvm-svn: 375273
Summary:
It looks like this is the only missing statistic in the CVP pass.
Since we prove NSW and NUW separately i'd think we should count them separately too.
Reviewers: nikic, spatel, reames
Reviewed By: spatel
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68740
llvm-svn: 375230
In the process of writing D69009, I realized we have two distinct sets of invariants within this single function, and basically no shared logic. The optimize loop exit transforms (including the new one in D69009) only care about *analyzeable* exits. Loop predication, on the other hand, has to reason about *all* exits. At the moment, we have the property (due to the requirement for an exact btc) that all exits are analyzeable, but that will likely change in the future as we add widenable condition support.
llvm-svn: 375138
The static analyzer is warning about a potential null dereference, but we should be able to use cast<> directly and if not assert will fire for us.
llvm-svn: 375103
We can't normally stumble into that assertion because a tautological
*conditional* `br` in loop body is required, one that always
branches to loop latch. But that should have been always folded
to an unconditional branch before we get it.
But that is not guaranteed if the pass is run standalone.
So let's just promote the assertion into a proper check.
Fixes https://bugs.llvm.org/show_bug.cgi?id=43687
llvm-svn: 375100
Summary:
There are two cases where a block is merged into its predecessor and the
MergeBlockIntoPredecessor API is not used. Update the API so it can be
reused in the other cases, in order to avoid code duplication.
Cleanup motivated by D68659.
Reviewers: chandlerc, sanjoy.google, george.burgess.iv
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68670
llvm-svn: 375050
The problem is that we can have two loop exits, 'a' and 'b', where 'a' and 'b' would exit at the same iteration, 'a' precedes 'b' along some path, and 'b' is predicated while 'a' is not. In this case (see the previously submitted test case), we causing the loop to exit through 'b' whereas it should have exited through 'a'.
This only applies to loop exits where the exit counts are not provably inequal, but that isn't as much of a restriction as it appears. If we could order the exit counts, we'd have already removed one of the two exits. In theory, we might be able to prove inequality w/o ordering, but I didn't really explore that piece. Instead, I went for the obvious restriction and ensured we didn't predicate exits following non-predicateable exits.
Credit goes to Evgeny Brevnov for figuring out the problematic case. Fuzzing probably also found it (failures seen), but due to some silly infrastructure problems I hadn't gotten to the results before Evgeny hand reduced it from a benchmark (he manually enabled the transform). Once this is fixed, I'll try to filter through the fuzzer failures to see if there's anything additional lurking.
Differential Revision https://reviews.llvm.org/D68956
llvm-svn: 375038
Check that a call has an attached MemoryAccess before calling
getClobbering on the instruction.
If no access is attached, the instruction does not access memory.
Resolves PR43441.
llvm-svn: 374920
As reported by Joerg Sonnenberger in IRC, for 32-bit systems,
where pointer and size_t are 32-bit, if you use 64-bit-wide variable
in the loop, you could end up with loop exit count being of the type
wider than the size_t. Now, i'm not sure if we can produce `bcmp`
from that (just truncate?), but we certainly should not assert/miscompile.
llvm-svn: 374811
Add a pass to lower is.constant and objectsize intrinsics
This pass lowers is.constant and objectsize intrinsics not simplified by
earlier constant folding, i.e. if the object given is not constant or if
not using the optimized pass chain. The result is recursively simplified
and constant conditionals are pruned, so that dead blocks are removed
even for -O0. This allows inline asm blocks with operand constraints to
work all the time.
The new pass replaces the existing lowering in the codegen-prepare pass
and fallbacks in SDAG/GlobalISEL and FastISel. The latter now assert
on the intrinsics.
Differential Revision: https://reviews.llvm.org/D65280
llvm-svn: 374784
This pass lowers is.constant and objectsize intrinsics not simplified by
earlier constant folding, i.e. if the object given is not constant or if
not using the optimized pass chain. The result is recursively simplified
and constant conditionals are pruned, so that dead blocks are removed
even for -O0. This allows inline asm blocks with operand constraints to
work all the time.
The new pass replaces the existing lowering in the codegen-prepare pass
and fallbacks in SDAG/GlobalISEL and FastISel. The latter now assert
on the intrinsics.
Differential Revision: https://reviews.llvm.org/D65280
llvm-svn: 374743
Summary:
If the underlying alloca did not change, we do not necessarily need new
lifetime markers. This patch adds a check and reuses the old ones if
possible.
Reviewers: reames, ssarda, t.p.northover, hfinkel
Subscribers: hiraditya, bollu, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68900
llvm-svn: 374692
Summary:
This is a recommit, this originally landed in rL370454 but was
subsequently reverted in rL370788 due to
https://bugs.llvm.org/show_bug.cgi?id=43206
The reduced testcase was added to bcmp-negative-tests.ll
as @pr43206_different_loops - we must ensure that the SCEV's
we got are both for the same loop we are currently investigating.
Original commit message:
@mclow.lists brought up this issue up in IRC.
It is a reasonably common problem to compare some two values for equality.
Those may be just some integers, strings or arrays of integers.
In C, there is `memcmp()`, `bcmp()` functions.
In C++, there exists `std::equal()` algorithm.
One can also write that function manually.
libstdc++'s `std::equal()` is specialized to directly call `memcmp()` for
various types, but not `std::byte` from C++2a. https://godbolt.org/z/mx2ejJ
libc++ does not do anything like that, it simply relies on simple C++'s
`operator==()`. https://godbolt.org/z/er0Zwf (GOOD!)
So likely, there exists a certain performance opportunities.
Let's compare performance of naive `std::equal()` (no `memcmp()`) with one that
is using `memcmp()` (in this case, compiled with modified compiler). {F8768213}
```
#include <algorithm>
#include <cmath>
#include <cstdint>
#include <iterator>
#include <limits>
#include <random>
#include <type_traits>
#include <utility>
#include <vector>
#include "benchmark/benchmark.h"
template <class T>
bool equal(T* a, T* a_end, T* b) noexcept {
for (; a != a_end; ++a, ++b) {
if (*a != *b) return false;
}
return true;
}
template <typename T>
std::vector<T> getVectorOfRandomNumbers(size_t count) {
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<T> dis(std::numeric_limits<T>::min(),
std::numeric_limits<T>::max());
std::vector<T> v;
v.reserve(count);
std::generate_n(std::back_inserter(v), count,
[&dis, &gen]() { return dis(gen); });
assert(v.size() == count);
return v;
}
struct Identical {
template <typename T>
static std::pair<std::vector<T>, std::vector<T>> Gen(size_t count) {
auto Tmp = getVectorOfRandomNumbers<T>(count);
return std::make_pair(Tmp, std::move(Tmp));
}
};
struct InequalHalfway {
template <typename T>
static std::pair<std::vector<T>, std::vector<T>> Gen(size_t count) {
auto V0 = getVectorOfRandomNumbers<T>(count);
auto V1 = V0;
V1[V1.size() / size_t(2)]++; // just change the value.
return std::make_pair(std::move(V0), std::move(V1));
}
};
template <class T, class Gen>
void BM_bcmp(benchmark::State& state) {
const size_t Length = state.range(0);
const std::pair<std::vector<T>, std::vector<T>> Data =
Gen::template Gen<T>(Length);
const std::vector<T>& a = Data.first;
const std::vector<T>& b = Data.second;
assert(a.size() == Length && b.size() == a.size());
benchmark::ClobberMemory();
benchmark::DoNotOptimize(a);
benchmark::DoNotOptimize(a.data());
benchmark::DoNotOptimize(b);
benchmark::DoNotOptimize(b.data());
for (auto _ : state) {
const bool is_equal = equal(a.data(), a.data() + a.size(), b.data());
benchmark::DoNotOptimize(is_equal);
}
state.SetComplexityN(Length);
state.counters["eltcnt"] =
benchmark::Counter(Length, benchmark::Counter::kIsIterationInvariant);
state.counters["eltcnt/sec"] =
benchmark::Counter(Length, benchmark::Counter::kIsIterationInvariantRate);
const size_t BytesRead = 2 * sizeof(T) * Length;
state.counters["bytes_read/iteration"] =
benchmark::Counter(BytesRead, benchmark::Counter::kDefaults,
benchmark::Counter::OneK::kIs1024);
state.counters["bytes_read/sec"] = benchmark::Counter(
BytesRead, benchmark::Counter::kIsIterationInvariantRate,
benchmark::Counter::OneK::kIs1024);
}
template <typename T>
static void CustomArguments(benchmark::internal::Benchmark* b) {
const size_t L2SizeBytes = []() {
for (const benchmark::CPUInfo::CacheInfo& I :
benchmark::CPUInfo::Get().caches) {
if (I.level == 2) return I.size;
}
return 0;
}();
// What is the largest range we can check to always fit within given L2 cache?
const size_t MaxLen = L2SizeBytes / /*total bufs*/ 2 /
/*maximal elt size*/ sizeof(T) / /*safety margin*/ 2;
b->RangeMultiplier(2)->Range(1, MaxLen)->Complexity(benchmark::oN);
}
BENCHMARK_TEMPLATE(BM_bcmp, uint8_t, Identical)
->Apply(CustomArguments<uint8_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint16_t, Identical)
->Apply(CustomArguments<uint16_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint32_t, Identical)
->Apply(CustomArguments<uint32_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint64_t, Identical)
->Apply(CustomArguments<uint64_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint8_t, InequalHalfway)
->Apply(CustomArguments<uint8_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint16_t, InequalHalfway)
->Apply(CustomArguments<uint16_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint32_t, InequalHalfway)
->Apply(CustomArguments<uint32_t>);
BENCHMARK_TEMPLATE(BM_bcmp, uint64_t, InequalHalfway)
->Apply(CustomArguments<uint64_t>);
```
{F8768210}
```
$ ~/src/googlebenchmark/tools/compare.py --no-utest benchmarks build-{old,new}/test/llvm-bcmp-bench
RUNNING: build-old/test/llvm-bcmp-bench --benchmark_out=/tmp/tmpb6PEUx
2019-04-25 21:17:11
Running build-old/test/llvm-bcmp-bench
Run on (8 X 4000 MHz CPU s)
CPU Caches:
L1 Data 16K (x8)
L1 Instruction 64K (x4)
L2 Unified 2048K (x4)
L3 Unified 8192K (x1)
Load Average: 0.65, 3.90, 4.14
---------------------------------------------------------------------------------------------------
Benchmark Time CPU Iterations UserCounters...
---------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000 432131 ns 432101 ns 1613 bytes_read/iteration=1000k bytes_read/sec=2.20706G/s eltcnt=825.856M eltcnt/sec=1.18491G/s
BM_bcmp<uint8_t, Identical>_BigO 0.86 N 0.86 N
BM_bcmp<uint8_t, Identical>_RMS 8 % 8 %
<...>
BM_bcmp<uint16_t, Identical>/256000 161408 ns 161409 ns 4027 bytes_read/iteration=1000k bytes_read/sec=5.90843G/s eltcnt=1030.91M eltcnt/sec=1.58603G/s
BM_bcmp<uint16_t, Identical>_BigO 0.67 N 0.67 N
BM_bcmp<uint16_t, Identical>_RMS 25 % 25 %
<...>
BM_bcmp<uint32_t, Identical>/128000 81497 ns 81488 ns 8415 bytes_read/iteration=1000k bytes_read/sec=11.7032G/s eltcnt=1077.12M eltcnt/sec=1.57078G/s
BM_bcmp<uint32_t, Identical>_BigO 0.71 N 0.71 N
BM_bcmp<uint32_t, Identical>_RMS 42 % 42 %
<...>
BM_bcmp<uint64_t, Identical>/64000 50138 ns 50138 ns 10909 bytes_read/iteration=1000k bytes_read/sec=19.0209G/s eltcnt=698.176M eltcnt/sec=1.27647G/s
BM_bcmp<uint64_t, Identical>_BigO 0.84 N 0.84 N
BM_bcmp<uint64_t, Identical>_RMS 27 % 27 %
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000 192405 ns 192392 ns 3638 bytes_read/iteration=1000k bytes_read/sec=4.95694G/s eltcnt=1.86266G eltcnt/sec=2.66124G/s
BM_bcmp<uint8_t, InequalHalfway>_BigO 0.38 N 0.38 N
BM_bcmp<uint8_t, InequalHalfway>_RMS 3 % 3 %
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000 127858 ns 127860 ns 5477 bytes_read/iteration=1000k bytes_read/sec=7.45873G/s eltcnt=1.40211G eltcnt/sec=2.00219G/s
BM_bcmp<uint16_t, InequalHalfway>_BigO 0.50 N 0.50 N
BM_bcmp<uint16_t, InequalHalfway>_RMS 0 % 0 %
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000 49140 ns 49140 ns 14281 bytes_read/iteration=1000k bytes_read/sec=19.4072G/s eltcnt=1.82797G eltcnt/sec=2.60478G/s
BM_bcmp<uint32_t, InequalHalfway>_BigO 0.40 N 0.40 N
BM_bcmp<uint32_t, InequalHalfway>_RMS 18 % 18 %
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000 32101 ns 32099 ns 21786 bytes_read/iteration=1000k bytes_read/sec=29.7101G/s eltcnt=1.3943G eltcnt/sec=1.99381G/s
BM_bcmp<uint64_t, InequalHalfway>_BigO 0.50 N 0.50 N
BM_bcmp<uint64_t, InequalHalfway>_RMS 1 % 1 %
RUNNING: build-new/test/llvm-bcmp-bench --benchmark_out=/tmp/tmpQ46PP0
2019-04-25 21:19:29
Running build-new/test/llvm-bcmp-bench
Run on (8 X 4000 MHz CPU s)
CPU Caches:
L1 Data 16K (x8)
L1 Instruction 64K (x4)
L2 Unified 2048K (x4)
L3 Unified 8192K (x1)
Load Average: 1.01, 2.85, 3.71
---------------------------------------------------------------------------------------------------
Benchmark Time CPU Iterations UserCounters...
---------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000 18593 ns 18590 ns 37565 bytes_read/iteration=1000k bytes_read/sec=51.2991G/s eltcnt=19.2333G eltcnt/sec=27.541G/s
BM_bcmp<uint8_t, Identical>_BigO 0.04 N 0.04 N
BM_bcmp<uint8_t, Identical>_RMS 37 % 37 %
<...>
BM_bcmp<uint16_t, Identical>/256000 18950 ns 18948 ns 37223 bytes_read/iteration=1000k bytes_read/sec=50.3324G/s eltcnt=9.52909G eltcnt/sec=13.511G/s
BM_bcmp<uint16_t, Identical>_BigO 0.08 N 0.08 N
BM_bcmp<uint16_t, Identical>_RMS 34 % 34 %
<...>
BM_bcmp<uint32_t, Identical>/128000 18627 ns 18627 ns 37895 bytes_read/iteration=1000k bytes_read/sec=51.198G/s eltcnt=4.85056G eltcnt/sec=6.87168G/s
BM_bcmp<uint32_t, Identical>_BigO 0.16 N 0.16 N
BM_bcmp<uint32_t, Identical>_RMS 35 % 35 %
<...>
BM_bcmp<uint64_t, Identical>/64000 18855 ns 18855 ns 37458 bytes_read/iteration=1000k bytes_read/sec=50.5791G/s eltcnt=2.39731G eltcnt/sec=3.3943G/s
BM_bcmp<uint64_t, Identical>_BigO 0.32 N 0.32 N
BM_bcmp<uint64_t, Identical>_RMS 33 % 33 %
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000 9570 ns 9569 ns 73500 bytes_read/iteration=1000k bytes_read/sec=99.6601G/s eltcnt=37.632G eltcnt/sec=53.5046G/s
BM_bcmp<uint8_t, InequalHalfway>_BigO 0.02 N 0.02 N
BM_bcmp<uint8_t, InequalHalfway>_RMS 29 % 29 %
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000 9547 ns 9547 ns 74343 bytes_read/iteration=1000k bytes_read/sec=99.8971G/s eltcnt=19.0318G eltcnt/sec=26.8159G/s
BM_bcmp<uint16_t, InequalHalfway>_BigO 0.04 N 0.04 N
BM_bcmp<uint16_t, InequalHalfway>_RMS 29 % 29 %
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000 9396 ns 9394 ns 73521 bytes_read/iteration=1000k bytes_read/sec=101.518G/s eltcnt=9.41069G eltcnt/sec=13.6255G/s
BM_bcmp<uint32_t, InequalHalfway>_BigO 0.08 N 0.08 N
BM_bcmp<uint32_t, InequalHalfway>_RMS 30 % 30 %
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000 9499 ns 9498 ns 73802 bytes_read/iteration=1000k bytes_read/sec=100.405G/s eltcnt=4.72333G eltcnt/sec=6.73808G/s
BM_bcmp<uint64_t, InequalHalfway>_BigO 0.16 N 0.16 N
BM_bcmp<uint64_t, InequalHalfway>_RMS 28 % 28 %
Comparing build-old/test/llvm-bcmp-bench to build-new/test/llvm-bcmp-bench
Benchmark Time CPU Time Old Time New CPU Old CPU New
---------------------------------------------------------------------------------------------------------------------------------------
<...>
BM_bcmp<uint8_t, Identical>/512000 -0.9570 -0.9570 432131 18593 432101 18590
<...>
BM_bcmp<uint16_t, Identical>/256000 -0.8826 -0.8826 161408 18950 161409 18948
<...>
BM_bcmp<uint32_t, Identical>/128000 -0.7714 -0.7714 81497 18627 81488 18627
<...>
BM_bcmp<uint64_t, Identical>/64000 -0.6239 -0.6239 50138 18855 50138 18855
<...>
BM_bcmp<uint8_t, InequalHalfway>/512000 -0.9503 -0.9503 192405 9570 192392 9569
<...>
BM_bcmp<uint16_t, InequalHalfway>/256000 -0.9253 -0.9253 127858 9547 127860 9547
<...>
BM_bcmp<uint32_t, InequalHalfway>/128000 -0.8088 -0.8088 49140 9396 49140 9394
<...>
BM_bcmp<uint64_t, InequalHalfway>/64000 -0.7041 -0.7041 32101 9499 32099 9498
```
What can we tell from the benchmark?
* Performance of naive equality check somewhat improves with element size,
maxing out at eltcnt/sec=1.58603G/s for uint16_t, or bytes_read/sec=19.0209G/s
for uint64_t. I think, that instability implies performance problems.
* Performance of `memcmp()`-aware benchmark always maxes out at around
bytes_read/sec=51.2991G/s for every type. That is 2.6x the throughput of the
naive variant!
* eltcnt/sec metric for the `memcmp()`-aware benchmark maxes out at
eltcnt/sec=27.541G/s for uint8_t (was: eltcnt/sec=1.18491G/s, so 24x) and
linearly decreases with element size.
For uint64_t, it's ~4x+ the elements/second.
* The call obvious is more pricey than the loop, with small element count.
As it can be seen from the full output {F8768210}, the `memcmp()` is almost
universally worse, independent of the element size (and thus buffer size) when
element count is less than 8.
So all in all, bcmp idiom does indeed pose untapped performance headroom.
This diff does implement said idiom recognition. I think a reasonable test
coverage is present, but do tell if there is anything obvious missing.
Now, quality. This does succeed to build and pass the test-suite, at least
without any non-bundled elements. {F8768216} {F8768217}
This transform fires 91 times:
```
$ /build/test-suite/utils/compare.py -m loop-idiom.NumBCmp result-new.json
Tests: 1149
Metric: loop-idiom.NumBCmp
Program result-new
MultiSourc...Benchmarks/7zip/7zip-benchmark 79.00
MultiSource/Applications/d/make_dparser 3.00
SingleSource/UnitTests/vla 2.00
MultiSource/Applications/Burg/burg 1.00
MultiSourc.../Applications/JM/lencod/lencod 1.00
MultiSource/Applications/lemon/lemon 1.00
MultiSource/Benchmarks/Bullet/bullet 1.00
MultiSourc...e/Benchmarks/MallocBench/gs/gs 1.00
MultiSourc...gs-C/TimberWolfMC/timberwolfmc 1.00
MultiSourc...Prolangs-C/simulator/simulator 1.00
```
The size changes are:
I'm not sure what's going on with SingleSource/UnitTests/vla.test yet, did not look.
```
$ /build/test-suite/utils/compare.py -m size..text result-{old,new}.json --filter-hash
Tests: 1149
Same hash: 907 (filtered out)
Remaining: 242
Metric: size..text
Program result-old result-new diff
test-suite...ingleSource/UnitTests/vla.test 753.00 833.00 10.6%
test-suite...marks/7zip/7zip-benchmark.test 1001697.00 966657.00 -3.5%
test-suite...ngs-C/simulator/simulator.test 32369.00 32321.00 -0.1%
test-suite...plications/d/make_dparser.test 89585.00 89505.00 -0.1%
test-suite...ce/Applications/Burg/burg.test 40817.00 40785.00 -0.1%
test-suite.../Applications/lemon/lemon.test 47281.00 47249.00 -0.1%
test-suite...TimberWolfMC/timberwolfmc.test 250065.00 250113.00 0.0%
test-suite...chmarks/MallocBench/gs/gs.test 149889.00 149873.00 -0.0%
test-suite...ications/JM/lencod/lencod.test 769585.00 769569.00 -0.0%
test-suite.../Benchmarks/Bullet/bullet.test 770049.00 770049.00 0.0%
test-suite...HMARK_ANISTROPIC_DIFFUSION/128 NaN NaN nan%
test-suite...HMARK_ANISTROPIC_DIFFUSION/256 NaN NaN nan%
test-suite...CHMARK_ANISTROPIC_DIFFUSION/64 NaN NaN nan%
test-suite...CHMARK_ANISTROPIC_DIFFUSION/32 NaN NaN nan%
test-suite...ENCHMARK_BILATERAL_FILTER/64/4 NaN NaN nan%
Geomean difference nan%
result-old result-new diff
count 1.000000e+01 10.00000 10.000000
mean 3.152090e+05 311695.40000 0.006749
std 3.790398e+05 372091.42232 0.036605
min 7.530000e+02 833.00000 -0.034981
25% 4.243300e+04 42401.00000 -0.000866
50% 1.197370e+05 119689.00000 -0.000392
75% 6.397050e+05 639705.00000 -0.000005
max 1.001697e+06 966657.00000 0.106242
```
I don't have timings though.
And now to the code. The basic idea is to completely replace the whole loop.
If we can't fully kill it, don't transform.
I have left one or two comments in the code, so hopefully it can be understood.
Also, there is a few TODO's that i have left for follow-ups:
* widening of `memcmp()`/`bcmp()`
* step smaller than the comparison size
* Metadata propagation
* more than two blocks as long as there is still a single backedge?
* ???
Reviewers: reames, fhahn, mkazantsev, chandlerc, craig.topper, courbet
Reviewed By: courbet
Subscribers: miyuki, hiraditya, xbolva00, nikic, jfb, gchatelet, courbet, llvm-commits, mclow.lists
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61144
llvm-svn: 374662
In loop-vectorize, interleave count and vector factor depend on target register number. Currently, it does not
estimate different register pressure for different register class separately(especially for scalar type,
float type should not be on the same position with int type), so it's not accurate. Specifically,
it causes too many times interleaving/unrolling, result in too many register spills in loop body and hurting performance.
So we need classify the register classes in IR level, and importantly these are abstract register classes,
and are not the target register class of backend provided in td file. It's used to establish the mapping between
the types of IR values and the number of simultaneous live ranges to which we'd like to limit for some set of those types.
For example, POWER target, register num is special when VSX is enabled. When VSX is enabled, the number of int scalar register is 32(GPR),
float is 64(VSR), but for int and float vector register both are 64(VSR). So there should be 2 kinds of register class when vsx is enabled,
and 3 kinds of register class when VSX is NOT enabled.
It runs on POWER target, it makes big(+~30%) performance improvement in one specific bmk(503.bwaves_r) of spec2017 and no other obvious degressions.
Differential revision: https://reviews.llvm.org/D67148
llvm-svn: 374634
This is really a known bits style transformation, but known bits isn't context sensitive. The particular case which comes up happens to involve a range which allows range based reasoning to eliminate the mask pattern, so handle that case specifically in CVP.
InstCombine likes to generate the mask-by-low-bits pattern when widening an arithmetic expression which includes a zext in the middle.
Differential Revision: https://reviews.llvm.org/D68811
llvm-svn: 374506
Summary:
The rule for the moveAllAfterMergeBlocks API si for all instructions
from `From` to have been moved to `To`, while keeping the CFG edges (and
block terminators) unchanged.
Update all the callsites for moveAllAfterMergeBlocks to follow this.
Pending follow-up: since the same behavior is needed everytime, merge
all callsites into one. The common denominator may be the call to
`MergeBlockIntoPredecessor`.
Resolves PR43569.
Reviewers: george.burgess.iv
Subscribers: Prazek, sanjoy.google, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68659
llvm-svn: 374177
Also Revert "[LoopVectorize] Fix non-debug builds after rL374017"
This reverts commit 9f41deccc0.
This reverts commit 18b6fe07bc.
The patch is breaking PowerPC internal build, checked with author, reverting
on behalf of him for now due to timezone.
llvm-svn: 374091
* Adds a TypeSize struct to represent the known minimum size of a type
along with a flag to indicate that the runtime size is a integer multiple
of that size
* Converts existing size query functions from Type.h and DataLayout.h to
return a TypeSize result
* Adds convenience methods (including a transparent conversion operator
to uint64_t) so that most existing code 'just works' as if the return
values were still scalars.
* Uses the new size queries along with ElementCount to ensure that all
supported instructions used with scalable vectors can be constructed
in IR.
Reviewers: hfinkel, lattner, rkruppe, greened, rovka, rengolin, sdesmalen
Reviewed By: rovka, sdesmalen
Differential Revision: https://reviews.llvm.org/D53137
llvm-svn: 374042
Summary: LoopRotate is a loop pass and SE should always be available.
Reviewers: anemet, asbirlea
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D68573
llvm-svn: 374026
In loop-vectorize, interleave count and vector factor depend on target register number. Currently, it does not
estimate different register pressure for different register class separately(especially for scalar type,
float type should not be on the same position with int type), so it's not accurate. Specifically,
it causes too many times interleaving/unrolling, result in too many register spills in loop body and hurting performance.
So we need classify the register classes in IR level, and importantly these are abstract register classes,
and are not the target register class of backend provided in td file. It's used to establish the mapping between
the types of IR values and the number of simultaneous live ranges to which we'd like to limit for some set of those types.
For example, POWER target, register num is special when VSX is enabled. When VSX is enabled, the number of int scalar register is 32(GPR),
float is 64(VSR), but for int and float vector register both are 64(VSR). So there should be 2 kinds of register class when vsx is enabled,
and 3 kinds of register class when VSX is NOT enabled.
It runs on POWER target, it makes big(+~30%) performance improvement in one specific bmk(503.bwaves_r) of spec2017 and no other obvious degressions.
Differential revision: https://reviews.llvm.org/D67148
llvm-svn: 374017
Doing this makes MSVC complain that `empty(someRange)` could refer to
either C++17's std::empty or LLVM's llvm::empty, which previously we
avoided via SFINAE because std::empty is defined in terms of an empty
member rather than begin and end. So, switch callers over to the new
method as it is added.
https://reviews.llvm.org/D68439
llvm-svn: 373935
There are no users that pass in LazyValueInfo, so we can simplify the
function a bit.
Reviewers: brzycki, asbirlea, davide
Reviewed By: davide
Differential Revision: https://reviews.llvm.org/D68297
llvm-svn: 373488
The static analyzer is warning about a potential null dereference, but we should be able to use cast<PHINode> directly and if not assert will fire for us.
llvm-svn: 373481
This patch implements a variation of a well known techniques for JIT compilers - we have an implementation in tree as LoopPredication - but with an interesting twist. This version does not assume the ability to execute a path which wasn't taken in the original program (such as a guard or widenable.condition intrinsic). The benefit is that this works for arbitrary IR from any frontend (including C/C++/Fortran). The tradeoff is that it's restricted to read only loops without implicit exits.
This builds on SCEV, and can thus eliminate the loop varying portion of the any early exit where all exits are understandable by SCEV. A key advantage is that fixing deficiency exposed in SCEV - already found one while writing test cases - will also benefit all of full redundancy elimination (and most other loop transforms).
I haven't seen anything in the literature which quite matches this. Given that, I'm not entirely sure that keeping the name "loop predication" is helpful. Anyone have suggestions for a better name? This is analogous to partial redundancy elimination - since we remove the condition flowing around the backedge - and has some parallels to our existing transforms which try to make conditions invariant in loops.
Factoring wise, I chose to put this in IndVarSimplify since it's a generally applicable to all workloads. I could split this off into it's own pass, but we'd then probably want to add that new pass every place we use IndVars. One solid argument for splitting it off into it's own pass is that this transform is "too good". It breaks a huge number of existing IndVars test cases as they tend to be simple read only loops. At the moment, I've opted it off by default, but if we add this to IndVars and enable, we'll have to update around 20 test files to add side effects or disable this transform.
Near term plan is to fuzz this extensively while off by default, reflect and discuss on the factoring issue mentioned just above, and then enable by default. I also need to give some though to supporting widenable conditions in this framing.
Differential Revision: https://reviews.llvm.org/D67408
llvm-svn: 373351
Kazu Hirata