Summary:
vectorizer-maximize-bandwidth is generally useful in terms of performance. I've tested the impact of changing this to default on speccpu benchmarks on sandybridge machines. The result shows non-negative impact:
spec/2006/fp/C++/444.namd 26.84 -0.31%
spec/2006/fp/C++/447.dealII 46.19 +0.89%
spec/2006/fp/C++/450.soplex 42.92 -0.44%
spec/2006/fp/C++/453.povray 38.57 -2.25%
spec/2006/fp/C/433.milc 24.54 -0.76%
spec/2006/fp/C/470.lbm 41.08 +0.26%
spec/2006/fp/C/482.sphinx3 47.58 -0.99%
spec/2006/int/C++/471.omnetpp 22.06 +1.87%
spec/2006/int/C++/473.astar 22.65 -0.12%
spec/2006/int/C++/483.xalancbmk 33.69 +4.97%
spec/2006/int/C/400.perlbench 33.43 +1.70%
spec/2006/int/C/401.bzip2 23.02 -0.19%
spec/2006/int/C/403.gcc 32.57 -0.43%
spec/2006/int/C/429.mcf 40.35 +0.27%
spec/2006/int/C/445.gobmk 26.96 +0.06%
spec/2006/int/C/456.hmmer 24.4 +0.19%
spec/2006/int/C/458.sjeng 27.91 -0.08%
spec/2006/int/C/462.libquantum 57.47 -0.20%
spec/2006/int/C/464.h264ref 46.52 +1.35%
geometric mean +0.29%
The regression on 453.povray seems real, but is due to secondary effects as all hot functions are bit-identical with and without the flag.
I started this patch to consult upstream opinions on this. It will be greatly appreciated if the community can help test the performance impact of this change on other architectures so that we can decided if this should be target-dependent.
Reviewers: hfinkel, mkuper, davidxl, chandlerc
Reviewed By: chandlerc
Subscribers: rengolin, sanjoy, javed.absar, bjope, dorit, magabari, RKSimon, llvm-commits, mzolotukhin
Differential Revision: https://reviews.llvm.org/D33341
llvm-svn: 306336
Fixes bug 33597.
Use of substituteRegister in the tied operand case messes
up the register use iterator, causing some uses to be left
unprocessed.
llvm-svn: 306333
Instead of providing access to the internal MapStorage holding all Values
associated with a given Key, used for setting or resetting them all together,
ValueMap keeps its MapStorage internal; its new interface allows getting,
setting or resetting a single Value, per part or per part-and-lane.
Follows the discussion in https://reviews.llvm.org/D32871.
Differential Revision: https://reviews.llvm.org/D34473
llvm-svn: 306331
This patch fills holes in executable sections with 0xd4 (ARM) or
0xef (MIPS). These trap instructions were suggested by Theo de Raadt.
llvm-svn: 306322
modules to preprocessing of nested .pcm files.
Making those module files available results in loading more .pcm files than
necessary, and potentially in misbehavior if a module makes itself visible
during its own compilation (as parts of that module that have not yet been
processed would then become visible).
llvm-svn: 306320
Summary: On BSDs, there is no `libdl.so`, and functions like `dlopen`
are implemented in the main C library instead. Use the `CMAKE_DL_LIBS`
variable instead of hardcoding a dependency on the `dl` library.
Reviewers: grokos, joerg, emaste
Reviewed By: emaste
Subscribers: jlpeyton, mgorny, openmp-commits
Differential Revision: https://reviews.llvm.org/D34632
llvm-svn: 306319
When we forward a stored value to a load and eliminate it entirely we need to
make sure the liveness of the register is maintained all the way to its use.
Previously we only cleared liveness on the store doing the forwarding, but
there could be other killing uses in between.
We already do the right thing when the load has to be converted into something
else, it was just this one path that skipped it.
llvm-svn: 306318
definition or non-reference class type.
The crash occurs when there is a template parameter list in a class that
is missing the closing angle bracket followed by a definition of a
struct. For example:
class C0 {
public:
template<typename T, typename T1 = T // missing closing angle bracket
struct S0 {};
C0() : m(new S0<int>) {}
S0<int> *m;
};
This happens because the parsed struct is added to the scope of the
enclosing class without having its access specifier set, which results
in an assertion failure in SemaAccess.cpp later.
This commit fixes the crash by adding the parsed struct to the enclosing
file scope and marking structs as invalid if they are defined in
template parameter lists.
rdar://problem/31783961
rdar://problem/19570630
Differential Revision: https://reviews.llvm.org/D33606
llvm-svn: 306317
Some forms have sizes that depend on the DWARF version, DWARF format
(32/64-bit), or the size of an address. Collect these into a struct
to simplify passing them around. Require callers to provide one when
they query a form's size.
Differential Revision: http://reviews.llvm.org/D34570
llvm-svn: 306315
The recommit fixes three bugs: The first one is to use CurrentBlock instead of
PREInstr's Parent as param of performScalarPREInsertion because the Parent
of a clone instruction may be uninitialized. The second one is stop PRE when
CurrentBlock to its predecessor is a backedge and an operand of CurInst is
defined inside of CurrentBlock. The same value defined inside of loop in last
iteration can not be regarded as available. The third one is an out-of-bound
array access in a flipped if guard.
Right now scalarpre doesn't have phi-translate support, so it will miss some
simple pre opportunities. Like the following testcase, current scalarpre cannot
recognize the last "a * b" is fully redundent because a and b used by the last
"a * b" expr are both defined by phis.
long a[100], b[100], g1, g2, g3;
__attribute__((pure)) long goo();
void foo(long a, long b, long c, long d) {
g1 = a * b;
if (__builtin_expect(g2 > 3, 0)) {
a = c;
b = d;
g2 = a * b;
}
g3 = a * b; // fully redundant.
}
The patch adds phi-translate support in scalarpre. This is only a temporary
solution before the newpre based on newgvn is available.
llvm-svn: 306313
This patch removes the dependency on the external rc.exe tool by writing
a simple .res file using our own library. In this patch I also added an
explicit definition for the .res file magic. Furthermore, I added a
unittest for embeded manifests and fixed a bug exposed by the test.
llvm-svn: 306311
attribute.
This is needed because older versions of libc++ do not have these
operators. If users target an older deployment target and try to compile
programs in which these operators are explicitly called, the compiler
will complain.
The following is the list of minimum deployment targets for the four
OSes:
macosx: 10.13
ios: 11.0
tvos: 11.0
watchos: 4.0
rdar://problem/32664169
Differential Revision: https://reviews.llvm.org/D34556
llvm-svn: 306310
Summary: Minor changes that allow opt-stats.py to support both Python 2 and 3.
Reviewers: anemet, davidxl
Reviewed By: anemet
Subscribers: llvm-commits, fhahn
Differential Revision: https://reviews.llvm.org/D34564
llvm-svn: 306306
We sometimes need emergency spill slots for the register scavenger.
This may be the case when code needs to access a stack slot that
has an offset of 4096 or more relative to the stack pointer.
To make that determination, processFunctionBeforeFrameFinalized
currently simply checks the total stack frame size of the current
function. But this is not enough, since code may need to access
stack slots in the caller's stack frame as well, in particular
incoming arguments stored on the stack.
This commit fixes the problem by taking argument slots into account.
llvm-svn: 306305
Summary:
They do the obvious thing: provide the section index of .bss and the
offset of the symbol in .bss.
Reviewers: ruiu
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D34628
llvm-svn: 306304
Over time we've started to add inputs and test cases using the .yaml
extension, which seems to be preferred over the .objtxt extension that
we were using initially. One nice thing about using .yaml is that it
triggers existing editor highlighting and formatting support.
Fix two pdb*.yaml test cases that I added that weren't being run as part
of check-lld.
llvm-svn: 306303
The non-AVX-512 behavior was changed in r248266 to match N1778
(C bindings for IEEE-754 (2008)), which defined the four functions
to not raise the inexact exception ("rint" is still defined as raising
it).
Update the AVX-512 lowering of these functions to match that: it should
not be different.
llvm-svn: 306299
Summary:
The main change is that we can have SECREL and SECTION relocations
against ___safe_se_handler_table, which is important for handling the
debug info in the MSVCRT.
Previously we were using DefinedRelative for __safe_se_handler_table and
__ImageBase, and after we implement CFGuard, we plan to extend it to
handle __guard_fids_table, __guard_longjmp_table, and more. However,
DefinedRelative is really only suitable for implementing __ImageBase,
because it lacks a Chunk, which you need in order to figure out the
output section index and output section offset when resolving SECREl and
SECTION relocations.
This change renames DefinedRelative to DefinedSynthetic and gives it a
Chunk. One wart is that __ImageBase doesn't have a chunk. It points to
the PE header, effectively. We could split DefinedRelative and
DefinedSynthetic if we think that's cleaner and creates fewer special
cases.
I also added safeseh.s, which checks that we don't emit a safe seh table
entries pointing to garbage collected handlers and that we don't emit a
table at all when there are no handlers.
Reviewers: ruiu
Reviewed By: ruiu
Subscribers: inglorion, pcc, llvm-commits, aprantl
Differential Revision: https://reviews.llvm.org/D34577
llvm-svn: 306293
Most "reserved" symbols are in ElfSym and it looks like there's no
reason to not do the same thing for _GLOBAL_OFFSET_TABLE_. This should
help https://reviews.llvm.org/D34618 too.
llvm-svn: 306292
Convert vector increment or decrement to sub/add with an all-ones constant:
add X, <1, 1...> --> sub X, <-1, -1...>
sub X, <1, 1...> --> add X, <-1, -1...>
The all-ones vector constant can be materialized using a pcmpeq instruction that is
commonly recognized as an idiom (has no register dependency), so that's better than
loading a splat 1 constant.
AVX512 uses 'vpternlogd' for 512-bit vectors because there is apparently no better
way to produce 512 one-bits.
The general advantages of this lowering are:
1. pcmpeq has lower latency than a memop on every uarch I looked at in Agner's tables,
so in theory, this could be better for perf, but...
2. That seems unlikely to affect any OOO implementation, and I can't measure any real
perf difference from this transform on Haswell or Jaguar, but...
3. It doesn't look like it from the diffs, but this is an overall size win because we
eliminate 16 - 64 constant bytes in the case of a vector load. If we're broadcasting
a scalar load (which might itself be a bug), then we're replacing a scalar constant
load + broadcast with a single cheap op, so that should always be smaller/better too.
4. This makes the DAG/isel output more consistent - we use pcmpeq already for padd x, -1
and psub x, -1, so we should use that form for +1 too because we can. If there's some
reason to favor a constant load on some CPU, let's make the reverse transform for all
of these cases (either here in the DAG or in a later machine pass).
This should fix:
https://bugs.llvm.org/show_bug.cgi?id=33483
Differential Revision: https://reviews.llvm.org/D34336
llvm-svn: 306289
Alex Shlyapnikov