RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 11:42:57 +01:00
Code Issues Projects Releases Wiki Activity
145,248 Commits 4 Branches 5 Tags 704 MiB
130730ae28
Commit Graph

145248 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Craig Topper
01c6dd61f1 [AVX-512] Add more VPTERNLOG patterns to enable folding of broadcast loads that aren't in operand 2. 
llvm-svn: 295634
 2017-02-20 02:47:42 +00:00
Craig Topper
d5f42ccd0e [X86] Use memory form of shift right by 1 when the rotl immediate is one less than the operation size. 
An earlier commit already did this for the register form.

llvm-svn: 295626
 2017-02-20 00:37:23 +00:00
Craig Topper
9974d0dc36 [X86] Add test cases showing missed opportunities to use rotate right by 1 instructions when operation reads/writes memory. 
llvm-svn: 295625
 2017-02-20 00:37:20 +00:00
Daniel Jasper
ad0ca020f4 s/REQUIRES: Asserts/REQUIRES: asserts/ 
Other than this, we consistently use lower case.

llvm-svn: 295623
 2017-02-19 23:26:00 +00:00
Craig Topper
befbc1c4c5 [AVX-512] Remove AddedComplexity from masked operations. The size of the patterns already increases their priority. 
llvm-svn: 295619
 2017-02-19 21:44:35 +00:00
Simon Pilgrim
660cf776ec [X86] Use peekThroughOneUseBitcasts helper. NFCI. 
llvm-svn: 295618
 2017-02-19 21:40:51 +00:00
Davide Italiano
deb9c1307c [X86] Prefer static_cast<> to C-style cast. NFCI. 
llvm-svn: 295617
 2017-02-19 21:35:41 +00:00
Craig Topper
4f611ef630 [AVX-512] Disable peephole optimizations on the VPTERNLOG commute test. Add new patterns to enable isel to fold the loads on it own. 
llvm-svn: 295616
 2017-02-19 21:32:15 +00:00
Davide Italiano
61ab7700e5 [AArch64] Prefer static_cast<> to C-style cast. NFCI. 
llvm-svn: 295615
 2017-02-19 21:31:14 +00:00
Simon Pilgrim
006db24698 [X86][SSE] Use getTargetConstantBitsFromNode to find zeroable shuffle elements. 
Replaces existing approach that could only search BUILD_VECTOR nodes.

Requires getTargetConstantBitsFromNode to discriminate cases with all/partial UNDEF bits in each element - this should also be useful when we get around to supporting getTargetShuffleMaskIndices with UNDEF elements. 

llvm-svn: 295613
 2017-02-19 19:40:31 +00:00
Craig Topper
ebdbe16d9a [AVX-512] Add patterns to recognize masked vpternlog when the passthrough operand is not operand 0. 
This uses a SDNodeXForm to swizzle the appropriate immediate bits to allow this to be matched.

llvm-svn: 295612
 2017-02-19 19:36:58 +00:00
Craig Topper
060dd27aa0 [AVX-512] Add test cases that show failure to select masked VPTERNLOG when a select is used to force the passthru operand to be not operand 0. 
llvm-svn: 295611
 2017-02-19 19:36:54 +00:00
Simon Pilgrim
c0eb9d9be8 [X86][SSE] Enable initial support for domain crossing at high shuffle combine depths. 
As discussed on D27692, this permits another domain to be used to combine a shuffle at high depths.

We currently set the required depth at 4 or more combined shuffles, this is probably too high for most targets but is a good starting point and already helps avoid a number of costly variable shuffles.

llvm-svn: 295608
 2017-02-19 17:19:38 +00:00
Artyom Skrobov
ddf6bfae71 Remove redundant call to GluedNodes.back() [NFC] 
llvm-svn: 295607
 2017-02-19 16:56:18 +00:00
Simon Pilgrim
0424401efa [X86][SSE] Generalize INSERTPS/SHUFPS/SHUFPD combines across domains. 
Relax the INSERTPS/SHUFPS/SHUFPD combines to support integer inputs if permitted.

llvm-svn: 295606
 2017-02-19 15:15:40 +00:00
Igor Kudrin
94f3f93b77 [llvm-cov] Respect Windows line endings when parsing demangled symbols. 
Differential Revision: https://reviews.llvm.org/D30096

llvm-svn: 295605
 2017-02-19 14:26:52 +00:00
Simon Pilgrim
cf30c7a3b3 [X86][SSE] Add domain crossing support for target shuffle combines. 
Add the infrastructure to flag whether float and/or int domains are permitable.

A future patch will enable domain crossing based off shuffle depth and the value types of the source vectors.

llvm-svn: 295604
 2017-02-19 14:12:25 +00:00
Simon Pilgrim
e00f7b33b7 Removed extra ';' 
llvm-svn: 295603
 2017-02-19 12:32:44 +00:00
Craig Topper
bbf06838c3 [AVX-512] Add broadcast VPTERNLOG instructions to special case commuting switch. 
The instructions are marked commutable, but without special handling we don't get the immediate correct.

While here also remove the masked memory forms that aren't commutable.

llvm-svn: 295602
 2017-02-19 08:03:26 +00:00
Craig Topper
2084df9ff5 [AVX-512] Add patterns to show missed opportunities for folding vpternlog with broadcast loads. Also demonstrates a bug in the commuting of broadcast vpternlog instructions when we are able to select them. 
llvm-svn: 295601
 2017-02-19 08:03:23 +00:00
NAKAMURA Takumi
9eb0aec05d Untabify. 
llvm-svn: 295599
 2017-02-19 06:51:46 +00:00
Daniel Berlin
2b6a9d5992 Re-add debugcounter.ll with Requires: Asserts so that it only triggers when asserts are on 
llvm-svn: 295598
 2017-02-19 06:45:02 +00:00
Daniel Berlin
8ae939c514 Which, in turn, causes build bots to fail that have it unexpectedly passing. So remove debugcounter.ll for now 
llvm-svn: 295597
 2017-02-19 04:56:07 +00:00
Daniel Berlin
a5a34963cc XFAIL this test until we figure out what to do here, since it will fail if NDEBUG defined 
llvm-svn: 295596
 2017-02-19 04:55:02 +00:00
Daniel Berlin
1abf6f2aae Add two files lost in rebase, causing build break 
llvm-svn: 295595
 2017-02-19 04:29:50 +00:00
Daniel Berlin
79b0f25a92 Add a DebugCounter for PredicateInfo renaming, and an associated test 
llvm-svn: 295594
 2017-02-19 04:29:01 +00:00
Daniel Berlin
3a776413d0 Add initial support for debug counting 
Summary:

We have support for bisection, and bugpoint can reduce testcases
often to a single pass. But that doesn't help reduce it to a single
transform by a single pass.  Which debug counting lets us do.

Debug counting lets you instrument a pass so that it only executes a
certain thing (rwhatever you want) after skipping it a certain time of
times, and then only does a certain number of executions before saying
"skip" again.

To make it concrete, for predicateinfo, if i instrument use renaming,
i can make it so it skips renaming the first N uses, renames the next
N, and then skips the rest.

This lets you narrow down a miscompilation to, often, a single
transformation, and then also debug it (by using the same command line
parameters).

Reviewers: chandlerc, davide, mehdi_amini

Subscribers: mgorny, llvm-commits

Differential Revision: https://reviews.llvm.org/D29998

llvm-svn: 295593
 2017-02-19 04:28:56 +00:00
NAKAMURA Takumi
a8741d2ac3 llvm/test/CodeGen/AMDGPU/r600.alu-limits.ll should require +Asserts. This would run into infinite loop anyways with -Asserts. 
llvm-svn: 295591
 2017-02-19 02:31:06 +00:00
Craig Topper
90784d3126 [X86] Remove patterns for MOVSD with v4i32 types. We don't appear to really need them and if we do we should just use a bitcast to a 64-bit element type. 
llvm-svn: 295589
 2017-02-19 02:08:48 +00:00
Craig Topper
b0c5272c72 [X86] Tighten up some of the SDNode type constraints. 
llvm-svn: 295588
 2017-02-19 01:54:47 +00:00
Simon Pilgrim
cbed204090 Fix unused variable warning when assertions are disabled. 
llvm-svn: 295587
 2017-02-19 00:33:37 +00:00
Simon Pilgrim
b5311b43f0 [X86] Fix enumeral/non-enumeral conditional expression warning. 
gcc only allows you to mix enums / ints if they have the same signedness.

llvm-svn: 295586
 2017-02-19 00:04:30 +00:00
Simon Pilgrim
1c1550439f Fix 'variable set but not used' warning when assertions are disabled. 
llvm-svn: 295585
 2017-02-19 00:03:46 +00:00
Daniel Berlin
8b2af5b529 NewGVN: Start making use of predicateinfo pass. 
Summary: This begins using the predicateinfo pass in NewGVN.

Reviewers: davide

Subscribers: llvm-commits, Prazek

Differential Revision: https://reviews.llvm.org/D29682

llvm-svn: 295583
 2017-02-18 23:06:50 +00:00
Daniel Berlin
ca0c7e5a21 NewGVN: Make ranking prefer undef to constants. Fix direction of 
shouldSwapOperands to be correct.

llvm-svn: 295582
 2017-02-18 23:06:47 +00:00
Daniel Berlin
7ac0e46e8b PredicateInfo: Clean up predicate info a little, using insertion 
helpers, and fixing support for the renaming the comparison.

llvm-svn: 295581
 2017-02-18 23:06:38 +00:00
Simon Pilgrim
305f3a0dc5 Fix signed/unsigned comparison warning. 
llvm-svn: 295580
 2017-02-18 22:56:17 +00:00
Craig Topper
fc82e6add2 [X86][XOP] Reduce the size of a multiclass by moving more stuff to parameters instead of doing 128-bit and 256-bit simultaneously. 
This requires some instructions to be renamed to move the Y earlier in the instruction name. The new names are more consistent with other instructions.

llvm-svn: 295579
 2017-02-18 22:53:43 +00:00
Craig Topper
93ddfca815 [TableGen] Make sure EnforceSameSize populates the type sets if necessary. 
This was found by another commit I'm working on.

llvm-svn: 295578
 2017-02-18 22:53:38 +00:00
Simon Pilgrim
ac7a882516 [AArch64] Fix enumeral/non-enumeral conditional expression warning. 
gcc only allows you to mix enums / ints if they have the same signedness.

llvm-svn: 295577
 2017-02-18 22:50:28 +00:00
Simon Pilgrim
3d41169b7b [X86] Fix enumeral/non-enumeral comparison warning. 
gcc only allows you to mix enums / ints if they have the same signedness.

llvm-svn: 295576
 2017-02-18 22:40:58 +00:00
Simon Pilgrim
bd82f6c338 [X86][SSE] Avoid repeated calls to SDValue::getValueType. 
Added assertion to check input type of X86ISD::VZEXT during target known bits calculation.

llvm-svn: 295575
 2017-02-18 22:25:27 +00:00
Sanjay Patel
ee4ec69fbc [InstCombine] add nsw/nuw X, signbit --> or X, signbit 
Changing to 'or' (rather than 'xor' when no wrapping flags are set)
allows icmp simplifies to happen as expected.

Differential Revision: https://reviews.llvm.org/D29729

llvm-svn: 295574
 2017-02-18 22:20:09 +00:00
Sanjay Patel
ab7638b75f [InstSimplify] add nsw/nuw (xor X, signbit), signbit --> X 
The change to InstCombine in:
https://reviews.llvm.org/D29729
...exposes this missing fold in InstSimplify, so adding this
first to avoid a regression.

llvm-svn: 295573
 2017-02-18 21:59:09 +00:00
Sanjay Patel
dbc786ab1d [InstSimplify] add tests for add nsw/nuw (xor X, signbit), signbit --> X; NFC 
llvm-svn: 295572
 2017-02-18 21:51:14 +00:00
Craig Topper
109be6a74d Recommit "[X86] Remove XOP VPCMOV intrinsics and autoupgrade them to native IR." 
Clang has now been fixed to not use these intrinsics.

llvm-svn: 295571
 2017-02-18 21:50:58 +00:00
Sanjay Patel
e664e4e1d9 [x86] remove stale comments from tests; NFC 
llvm-svn: 295569
 2017-02-18 21:07:37 +00:00
Sanjay Patel
c9789bbb4f [x86] fold sext (xor Bool, -1) --> sub (zext Bool), 1 
This is the same transform that is current used for:
select Bool, 0, -1

llvm-svn: 295568
 2017-02-18 21:03:28 +00:00
Piotr Padlewski
e85556f918 [MemorySSA] NFC small fixes 
Summary:
2 small fixes extracted from
https://reviews.llvm.org/D29064

Reviewers: kuhar, davide, dberlin, george.burgess.iv

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D30109

llvm-svn: 295566
 2017-02-18 20:34:36 +00:00
Craig Topper
72006e7fa3 Revert "[X86] Remove XOP VPCMOV intrinsics and autoupgrade them to native IR." 
This reverts r295564. I missed that clang was still using the intrinsics despite our half implemented autoupgrade support.

llvm-svn: 295565
 2017-02-18 20:14:20 +00:00