In theory this could be generalized to move anything where
we prove the operands are available, but that would require
rewriting PRE. As NewGVN will hopefully come soon, and we're
trying to rewrite PRE in terms of NewGVN+MemorySSA, it's probably
not worth spending too much time on it. Fix provided by
Daniel Berlin!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284311 91177308-0d34-0410-b5e6-96231b3b80d8
computeKnownBits only returns the common bits of each vector element instead of only the elements that are actually used
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284308 91177308-0d34-0410-b5e6-96231b3b80d8
BasicBlock::size is O(insts), making this loop O(blocks*insts), which
can be really slow on generated code. getPrevNode already checks if
we're at the beginning of the block and returns nullptr if so, just use
that instead. No functionality change intended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284303 91177308-0d34-0410-b5e6-96231b3b80d8
Instead of annotating (most of) the APInt API, we can just annotate
the type directly. This is less code and it will warn in more cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284297 91177308-0d34-0410-b5e6-96231b3b80d8
- Removed unused class members.
- Made class internal data private.
- Made class scoped data function scoped where it's possible.
- Replace naked new/delete with unique_ptr.
- Made resources guaranteed to be freed.
Differential Revision: https://reviews.llvm.org/D25464
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284290 91177308-0d34-0410-b5e6-96231b3b80d8
The previous names were both misleading (the MachineLegalizer actually
contained the info tables) and inconsistent with the selector & translator (in
having a "Machine") prefix. This should make everything sensible again.
The only functional change is the name of a couple of command-line options.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284287 91177308-0d34-0410-b5e6-96231b3b80d8
This is essentially a more powerful version of our current
LLVM_ATTRIBUTE_UNUSED_RESULT, in that it can also be applied to types
and generate warnings whenever an object of that type is returned by
value and the value is discarded.
I'll replace uses of LLVM_ATTRIBUTE_UNUSED_RESULT and remove that
macro in follow up commits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284286 91177308-0d34-0410-b5e6-96231b3b80d8
This is a patch to implement pr30640.
When a 64bit constant has the same hi/lo words, we can use rldimi to copy the low word into high word of the same register.
This optimization caused failure of test case bperm.ll because of not optimal heuristic in function SelectAndParts64. It chooses AND or ROTATE to extract bit groups from a register, and OR them together. This optimization lowers the cost of loading 64bit constant mask used in AND method, and causes different code sequence. But actually ROTATE method is better in this test case. The reason is in ROTATE method the final OR operation can be avoided since rldimi can insert the rotated bits into target register directly. So this patch also enhances SelectAndParts64 to prefer ROTATE method when the two methods have same cost and there are multiple bit groups need to be ORed together.
Differential Revision: https://reviews.llvm.org/D25521
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284276 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
We are using this helper for our 24-bit arithmetic combines, so we are now able to eliminate multi-use operations that mask the high-bits of 24-bit inputs (e.g. and x, 0xffffff)
Reviewers: arsenm, nhaehnle
Subscribers: tony-tye, arsenm, kzhuravl, wdng, nhaehnle, llvm-commits, yaxunl
Differential Revision: https://reviews.llvm.org/D24672
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284267 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
The main purpose of this new helper is to enable simplifying operations that
have multiple uses. SimplifyDemandedBits does not handle multiple uses
currently, and this new function makes it possible to optimize:
and v1, v0, 0xffffff
mul24 v2, v1, v1 ; Multiply ignoring high 8-bits.
To:
mul24 v2, v0, v0
Where before this would not be optimized, because v1 has multiple uses.
Reviewers: bogner, arsenm
Subscribers: nhaehnle, wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D24964
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284266 91177308-0d34-0410-b5e6-96231b3b80d8
X86. The pass optimizes as a unit the entire wide load + shuffles pattern
produced by interleaved vectorization. This initial patch optimizes one pattern
(64-bit elements interleaved by a factor of 4). Future patches will generalize
to additional patterns.
Patch by Farhana Aleen
Differential revision: http://reviews.llvm.org/D24681
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284260 91177308-0d34-0410-b5e6-96231b3b80d8
Use PackedRegisterRef to store the register information in the graph nodes.
This commit also removes support for virtual registers. It has never been
tested or used. It will be possible to add it back if there is a need.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284255 91177308-0d34-0410-b5e6-96231b3b80d8
TargetMachine," as it's causing sanitizer/memory issues until I
can track down this set.
This reverts commit r284203
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284252 91177308-0d34-0410-b5e6-96231b3b80d8
Add support for loading multiple coverage readers into a single
CoverageMapping instance. This should make it easier to prepare a
unified coverage report for multiple binaries.
Differential Revision: https://reviews.llvm.org/D25535
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284251 91177308-0d34-0410-b5e6-96231b3b80d8
This is an improvement when compiling with llvm. llvm doesn't inline
the call to insert, so the align is always executed and shows up in
the profile.
With gcc the call to insert is inlined and the align computation moved
and done only if needed.
With this patch we explicitly only compute it if it is needed.
In the two tests with debug info, the speedup was
scylla
master 3.008959365
patch 2.932080942 1.02621974786x faster
firefox
master 6.709823604
patch 6.592387227 1.01781393795x faster
In all others the difference was in the noise.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284249 91177308-0d34-0410-b5e6-96231b3b80d8
This change adds transformations such as:
zext(or(setcc(eq, (cmp x, 0)), setcc(eq, (cmp y, 0))))
To:
srl(or(ctlz(x), ctlz(y)), log2(bitsize(x))
This optimisation is beneficial on Jaguar architecture only, where lzcnt has a good reciprocal throughput.
Other architectures such as Intel's Haswell/Broadwell or AMD's Bulldozer/PileDriver do not benefit from it.
For this reason the change also adds a "HasFastLZCNT" feature which gets enabled for Jaguar.
Differential Revision: https://reviews.llvm.org/D23446
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284248 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
* Describe new (3.3) parameter attribute group encoding, leaving old encoding there with a note about legacy
* Bring TYPE_BLOCK docs up to date
* Remove docs about obsolete (pre 3.0) TYPE_SYMTAB_BLOCK, TST_CODE_ENTRY
* Fix a couple of incorrect comments and remove one unused enum definition along the way
This addresses https://llvm.org/bugs/show_bug.cgi?id=28941.
Patch by: Ismail Badawi <ibadawi@cisco.com>
Differential Revision: https://reviews.llvm.org/D25623
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284246 91177308-0d34-0410-b5e6-96231b3b80d8
This test was apparently checking for 2 independent folds, but we have
plenty of tests for those individual folds already. We are lacking
vector tests, however, because we don't have the shift folds for vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@284243 91177308-0d34-0410-b5e6-96231b3b80d8
