llvm/test/Transforms/LoopVectorize
Renato Golin a447ca5c09 [ARM] Adding IEEE-754 SIMD detection to loop vectorizer
Some SIMD implementations are not IEEE-754 compliant, for example ARM's NEON.

This patch teaches the loop vectorizer to only allow transformations of loops
that either contain no floating-point operations or have enough allowance
flags supporting lack of precision (ex. -ffast-math, Darwin).

For that, the target description now has a method which tells us if the
vectorizer is allowed to handle FP math without falling into unsafe
representations, plus a check on every FP instruction in the candidate loop
to check for the safety flags.

This commit makes LLVM behave like GCC with respect to ARM NEON support, but
it stops short of fixing the underlying problem: sub-normals. Neither GCC
nor LLVM have a flag for allowing sub-normal operations. Before this patch,
GCC only allows it using unsafe-math flags and LLVM allows it by default with
no way to turn it off (short of not using NEON at all).

As a first step, we push this change to make it safe and in sync with GCC.
The second step is to discuss a new sub-normal's flag on both communitues
and come up with a common solution. The third step is to improve the FastMath
flags in LLVM to encode sub-normals and use those flags to restrict NEON FP.

Fixes PR16275.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@266363 91177308-0d34-0410-b5e6-96231b3b80d8
2016-04-14 20:42:18 +00:00
..
AArch64 [LoopUtils, LV] Fix PR27246 (first-order recurrences) 2016-04-11 19:48:18 +00:00
ARM [ARM] Adding IEEE-754 SIMD detection to loop vectorizer 2016-04-14 20:42:18 +00:00
PowerPC [LoopVectorize] Don't vectorize loops when everything will be scalarized 2016-03-30 19:37:08 +00:00
X86 Revert "Support arbitrary addrspace pointers in masked load/store intrinsics" 2016-04-14 08:47:17 +00:00
XCore [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
12-12-11-if-conv.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
2012-10-20-infloop.ll Rename getMaximumUnrollFactor -> getMaxInterleaveFactor; also rename option names controlling this variable. 2014-09-10 17:58:16 +00:00
2012-10-22-isconsec.ll [opaque pointer type] Add textual IR support for explicit type parameter to gep operator 2015-03-13 18:20:45 +00:00
align.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
bsd_regex.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
bzip_reverse_loops.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
calloc.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
cast-induction.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
conditional-assignment.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
control-flow.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
cpp-new-array.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
dbg.value.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
debugloc.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
duplicated-metadata.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
ee-crash.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
exact.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
flags.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
float-reduction.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
funcall.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
gcc-examples.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
gep_with_bitcast.ll LoopVectorizer - skip 'bitcast' between GEP and load. 2015-11-03 10:29:34 +00:00
global_alias.ll [opaque pointer type] Add textual IR support for explicit type parameter to gep operator 2015-03-13 18:20:45 +00:00
hoist-loads.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
i8-induction.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
if-conv-crash.ll Rename getMaximumUnrollFactor -> getMaxInterleaveFactor; also rename option names controlling this variable. 2014-09-10 17:58:16 +00:00
if-conversion-edgemasks.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
if-conversion-nest.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
if-conversion-reduction.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
if-conversion.ll [opaque pointer type] Add textual IR support for explicit type parameter to gep operator 2015-03-13 18:20:45 +00:00
if-pred-stores.ll [LV] Preserve LoopInfo when store predication is used 2016-03-15 18:06:20 +00:00
incorrect-dom-info.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
increment.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
induction_plus.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
induction.ll [LV] Don't bail to MiddleBlock if a runtime check fails, bail to ScalarPH instead 2015-09-02 10:15:39 +00:00
infiniteloop.ll Fix missing CHECK-LABELs 2013-10-02 20:29:00 +00:00
interleaved-accesses.ll [LV] Fix PR26600: avoid out of bounds loads for interleaved access vectorization 2016-02-19 15:46:10 +00:00
intrinsic.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
lcssa-crash.ll Rename getMaximumUnrollFactor -> getMaxInterleaveFactor; also rename option names controlling this variable. 2014-09-10 17:58:16 +00:00
lifetime.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
loop-form.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
loop-vect-memdep.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
memdep.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
metadata-unroll.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
metadata-width.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
metadata.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
miniters.ll The patch replace the overflow check in loop vectorization with the minimum loop iterations check. 2015-08-25 16:43:47 +00:00
minmax_reduction.ll [LoopUtils,LV] Propagate fast-math flags on generated FCmp instructions 2015-09-21 19:41:19 +00:00
multi-use-reduction-bug.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
multiple-address-spaces.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
no_array_bounds.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
no_idiv_reduction.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
no_int_induction.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
no_outside_user.ll Print vectorization analysis when loop hint is specified. 2015-08-11 01:09:15 +00:00
no_switch.ll testcase gardening: update the emissionKind enum to the new syntax. (NFC) 2016-04-01 00:16:49 +00:00
noalias-md-licm.ll [LoopVectorize] Annotate versioned loop with noalias metadata 2016-03-17 20:32:37 +00:00
noalias-md.ll [LoopVectorize] Annotate versioned loop with noalias metadata 2016-03-17 20:32:37 +00:00
nofloat.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
non-const-n.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
nontemporal.ll [LoopVectorize] Propagate 'nontemporal' attribute into vectorized instructions. 2015-08-20 22:27:38 +00:00
nsw-crash.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
opt.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
optsize.ll fix minsize detection: minsize attribute implies optimizing for size 2015-08-11 15:56:31 +00:00
phi-hang.ll [Verifier] Reject PHIs using defs from own block. 2016-03-26 23:32:57 +00:00
ptr_loops.ll [opaque pointer type] Add textual IR support for explicit type parameter to gep operator 2015-03-13 18:20:45 +00:00
ptr-induction.ll Add newline to test. NFC. 2015-09-02 14:06:16 +00:00
read-only.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
reduction.ll [LV] Don't bail to MiddleBlock if a runtime check fails, bail to ScalarPH instead 2015-09-02 10:15:39 +00:00
reverse_induction.ll [LV] Switch to using canonical induction variables. 2015-09-02 10:14:54 +00:00
reverse_iter.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
runtime-check-address-space.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
runtime-check-readonly-address-space.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
runtime-check-readonly.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
runtime-check.ll [DebugInfo] Fix tests so that each subprogram belongs to a CU. 2016-04-05 23:37:08 +00:00
runtime-limit.ll Fix test introduced in r246187 that failed on some systems. 2015-08-27 20:43:29 +00:00
safegep.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
same-base-access.ll [SCEV][LAA] Re-commit r260085 and r260086, this time with a fix for the memory 2016-02-08 17:02:45 +00:00
scalar-select.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
scev-exitlim-crash.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
simple-unroll.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
small-loop.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
start-non-zero.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
store-shuffle-bug.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
struct_access.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
tbaa-nodep.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
undef-inst-bug.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
unroll_novec.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
unroll.ll [X86] Disable loop unrolling in loop vectorization pass when VF is 1. 2015-05-06 17:12:25 +00:00
unsized-pointee-crash.ll [opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction 2015-02-27 19:29:02 +00:00
value-ptr-bug.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
vect.omp.persistence.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
vect.stats.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
vectorize-once.ll Introduce runtime unrolling disable matadata and use it to mark the scalar loop from vectorization. 2015-03-09 06:14:18 +00:00
version-mem-access.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
write-only.ll [opaque pointer type] Add textual IR support for explicit type parameter to load instruction 2015-02-27 21:17:42 +00:00
zero-sized-pointee-crash.ll [LoopVectorize] Don't crash on zero-sized types in isInductionPHI 2015-06-05 10:52:40 +00:00