Runtime unrolling will create a prologue to execute the extra
iterations which is can't divided by the unroll factor. It
generates an if-then-else sequence to jump into a factor -1
times unrolled loop body, like
extraiters = tripcount % loopfactor
if (extraiters == 0) jump Loop:
if (extraiters == loopfactor) jump L1
if (extraiters == loopfactor-1) jump L2
...
L1: LoopBody;
L2: LoopBody;
...
if tripcount < loopfactor jump End
Loop:
...
End:
It means if the unroll factor is 4, the loop body will be 7
times unrolled, 3 are in loop prologue, and 4 are in the loop.
This commit is to use a loop to execute the extra iterations
in prologue, like
extraiters = tripcount % loopfactor
if (extraiters == 0) jump Loop:
else jump Prol
Prol: LoopBody;
extraiters -= 1 // Omitted if unroll factor is 2.
if (extraiters != 0) jump Prol: // Omitted if unroll factor is 2.
if (tripcount < loopfactor) jump End
Loop:
...
End:
Then when unroll factor is 4, the loop body will be copied by
only 5 times, 1 in the prologue loop, 4 in the original loop.
And if the unroll factor is 2, new loop won't be created, just
as the original solution.
llvm-svn: 218604
nodes, and rely exclusively on its logic. This removes a ton of
duplication from the blend lowering and centralizes it in one place.
One downside is that it requires a bunch of hacks to make this work with
the current legalization framework. We have to manually speculate one
aspect of legalizing VSELECT nodes to get everything to work nicely
because the existing legalization framework isn't *actually* bottom-up.
The other grossness is that we somewhat duplicate the analysis of
constant blends. I'm on the fence here. If reviewers thing this would
look better with VSELECT when it has constant operands dumping over tho
VECTOR_SHUFFLE, we could go that way. But it would be a substantial
change because currently all of the actual blend instructions are
matched via patterns in the TD files based around VSELECT nodes (despite
them not being perfect fits for that). Suggestions welcome, but at least
this removes the rampant duplication in the backend.
llvm-svn: 218600
X86 target-specific DAG combining that tried to convert VSELECT nodes
into VECTOR_SHUFFLE nodes that it "knew" would lower into
immediate-controlled blend nodes.
Turns out, we have perfectly good lowering of all these VSELECT nodes,
and indeed that lowering already knows how to handle lowering through
BLENDI to immediate-controlled blend nodes. The code just wasn't getting
used much because this thing forced the world to go through the vector
shuffle lowering. Yuck.
This also exposes that I was too aggressive in avoiding domain crossing
in v218588 with that lowering -- when the other option is to expand into
two 128-bit vectors, it is worth domain crossing. Restore that behavior
now that we have nice tests covering it.
The test updates here fall into two camps. One is where previously we
ended up with an unsigned encoding of the blend operand and now we get
a signed encoding. In most of those places there were elaborate comments
explaining exactly what these operands really mean. Rather than that,
just switch these tests to use the nicely decoded comments that make it
obvious that the final shuffle matches.
The other updates are just removing pointless domain crossing by
blending integers with PBLENDW rather than BLENDPS.
llvm-svn: 218589
AVX-512.
There is no interesting logic yet. Everything ends up eventually
delegating to the generic code to split the vector and shuffle the
halves. Interestingly, that logic does a significantly better job of
lowering all of these types than the generic vector expansion code does.
Mostly, it lets most of the cases fall back to nice AVX2 code rather
than all the way back to SSE code paths.
Step 2 of basic AVX-512 support in the new vector shuffle lowering. Next
up will be to incrementally add direct support for the basic instruction
set to each type (adding tests first).
llvm-svn: 218585
vectors.
Someone will need to build the AVX512 lowering, which should follow
AVX1 and AVX2 *very* closely for AVX512F and AVX512BW resp. I've added
a dummy test which is a port of the v8f32 and v8i32 tests from AVX and
AVX2 to v8f64 and v8i64 tests for AVX512F and AVX512BW. Hopefully this
is enough information for someone to implement proper lowering here. If
not, I'll be happy to help, but right now the AVX-512 support isn't
a priority for me.
llvm-svn: 218583
lowerings.
This was hopelessly broken. First, the x86 backend wants '-1' to be the
element value representing true in a boolean vector, and second the
operand order for VSELECT is backwards from the actual x86 instructions.
To make matters worse, the backend is just using '-1' as the true value
to get the high bit to be set. It doesn't actually symbolically map the
'-1' to anything. But on x86 this isn't quite how it works: there *only*
the high bit is relevant. As a consequence weird non-'-1' values like
0x80 actually "work" once you flip the operands to be backwards.
Anyways, thanks to Hal for helping me sort out what these *should* be.
llvm-svn: 218582
new vector shuffle target DAG combines -- it helps to actually test for
the value you want rather than just using an integer in a boolean
context.
Have I mentioned that I loathe implicit conversions recently? :: sigh ::
llvm-svn: 218576
of widening masks.
We can't widen a zeroing mask unless both elements that would be merged
are either zeroed or undef. This is the only way to widen a mask if it
has a zeroed element.
Also clean up the code here by ordering the checks in a more logical way
and by using the symoblic values for undef and zero. I'm actually torn
on using the symbolic values because the existing code is littered with
the assumption that -1 is undef, and moreover that entries '< 0' are the
special entries. While that works with the values given to these
constants, using the symbolic constants actually makes it a bit more
opaque why this is the case.
llvm-svn: 218575
If there is a store followed by a store with the same value to the same location, then the store is dead/noop. It can be removed.
This problem is found in spec2006-197.parser.
For example,
stur w10, [x11, #-4]
stur w10, [x11, #-4]
Then one of the two stur instructions can be removed.
Patch by David Xu!
llvm-svn: 218569
and in the target shuffle combining when trying to widen vector
elements.
Previously only one of these was correct, and we didn't correctly
propagate zeroing target shuffle masks (which have a different sentinel
value from undef in non- target shuffle masks now). This isn't just
a missed optimization, this caused us to drop zeroing shuffles on the
floor and miscompile code. The added test case is one example of that.
There are other fixes to the test suite as a consequence of this as well
as restoring the undef elements in some of the masks that were lost when
I brought sanity to the actual *value* of the undef and zero sentinels.
I've also just cleaned up some of the PSHUFD and PSHUFLW and PSHUFHW
combining code, but that code really needs to go. It was a nice initial
attempt, but it isn't very principled and the recursive shuffle combiner
is much more powerful.
llvm-svn: 218562
to significantly more sane sentinels. Notably, everywhere else in the
backend's representation of shuffles uses '-1' to represent undef. The
target shuffle masks really shouldn't diverge from that, especially as
in a few places they are manipulated by shared code.
This causes us to lose some undef lanes in various test masks. I want to
get these back, but technically it isn't invalid and there are a *lot*
of bugs here so I want to try to establish a saner baseline for fixing
some of the bugs by aligning the specific senitnel values used.
llvm-svn: 218561
This is purely refactoring. No functional changes intended. PowerPC is the only target
that is currently using this interface.
The ultimate goal is to allow targets other than PowerPC (certainly X86 and Aarch64) to turn this:
z = y / sqrt(x)
into:
z = y * rsqrte(x)
And:
z = y / x
into:
z = y * rcpe(x)
using whatever HW magic they can use. See http://llvm.org/bugs/show_bug.cgi?id=20900 .
There is one hook in TargetLowering to get the target-specific opcode for an estimate instruction
along with the number of refinement steps needed to make the estimate usable.
Differential Revision: http://reviews.llvm.org/D5484
llvm-svn: 218553
Users of getSectionContents shouldn't try to pass in BSS or virtual
sections. In all instances, this is a bug in the code calling this
routine.
N.B. Some COFF implementations (like CL) will mark their BSS sections as
taking space on disk. This would confuse COFFObjectFile into thinking
the section is larger than the file.
llvm-svn: 218549
So in fully linked images when a call is made through a stub it now gets a
comment like the following in the disassembly:
callq 0x100000f6c ## symbol stub for: _printf
indicating the call is to a symbol stub and which symbol it is for. This is
done for branch reference types and seeing if the branch target is in a stub
section and if so using the indirect symbol table entry for that stub and
using that symbol table entries symbol name.
llvm-svn: 218546
that managed to elude all of my fuzz testing historically. =/
Something changed to allow this code path to actually be exercised and
it was doing bad things. It is especially heavily exercised by the
patterns that emerge when doing AVX shuffles that end up lowered through
the 128-bit code path.
llvm-svn: 218540
This has weird operand requirements so it's worthwhile
to have very strict checks for its operands.
Add different combinations of SGPR operands.
llvm-svn: 218535
Instead of moving the first SGPR that is different than the first,
legalize the operand that requires the fewest moves if one
SGPR is used for multiple operands.
This saves extra moves and is also required for some instructions
which require that the same operand be used for multiple operands.
llvm-svn: 218532
Disable the SGPR usage restriction parts of the DAG legalizeOperands.
It now should only be doing immediate folding until it can be replaced
later. The real legalization work is now done by the other
SIInstrInfo::legalizeOperands
llvm-svn: 218531
e.g. v_cndmask_b32 requires the condition operand be an SGPR.
If one of the source operands were an SGPR, that would be considered
the one SGPR use and the condition operand would be illegally moved.
llvm-svn: 218529
No test since the current SIISelLowering::legalizeOperands
effectively hides this, and the general uses seem to only fire
on SALU instructions which don't have modifiers between
the operands.
When trying to use legalizeOperands immediately after
instruction selection, it now sees a lot more patterns
it did not see before which break on this.
llvm-svn: 218527
The annotation instructions are dropped during codegen and have no
impact on size. In some cases, the annotations were preventing the
unroller from unrolling a loop because the annotation calls were
pushing the cost over the unrolling threshold.
Differential Revision: http://reviews.llvm.org/D5335
llvm-svn: 218525
layer of tie-breaking sorting, it really helps to check that you're in
a tie first. =] Otherwise the whole thing cycles infinitely. Test case
added, another one found through fuzz testing.
llvm-svn: 218523
AVX support.
New test cases included. Note that none of the existing test cases
covered these buggy code paths. =/ Also, it is clear from this that
SHUFPS and SHUFPD are the most bug prone shuffle instructions in x86. =[
These were all detected by fuzz-testing. (I <3 fuzz testing.)
llvm-svn: 218522
This patch makes the ARM backend transform 3 operand instructions such as
'adds/subs' to the 2 operand version of the same instruction if the first
two register operands are the same.
Example: 'adds r0, r0, #1' will is transformed to 'adds r0, #1'.
Currently for some instructions such as 'adds' if you try to assemble
'adds r0, r0, #8' for thumb v6m the assembler would throw an error message
because the immediate cannot be encoded using 3 bits.
The backend should be smart enough to transform the instruction to
'adds r0, #8', which allows for larger immediate constants.
Patch by Ranjeet Singh.
llvm-svn: 218521
based on the Function. This is currently used to implement
mips16 support in the mips backend via the existing module
pass resetting the subtarget.
Things to note:
a) This involved running resetTargetOptions before creating a
new subtarget so that code generation options like soft-float
could be recognized when creating the new subtarget. This is
to deal with initialization code in isel lowering that only
paid attention to the initial value.
b) Many of the existing testcases weren't using the soft-float
feature correctly. I've corrected these based on the check
values assuming that was the desired behavior.
c) The mips port now pays attention to the target-cpu and
target-features strings when generating code for a particular
function. I've removed these from one function where the
requested cpu and features didn't match the check lines in
the testcase.
llvm-svn: 218492
Machine Sink uses loop depth information to select between successors BBs to
sink machine instructions into, where BBs within smaller loop depths are
preferable. This patch adds support for choosing between successors by using
profile information from BlockFrequencyInfo instead, whenever the information
is available.
Tested it under SPEC2006 train (average of 30 runs for each program); ~1.5%
execution speedup in average on x86-64 darwin.
<rdar://problem/18021659>
llvm-svn: 218472
Summary:
I originally tried doing this specifically for X86 in the backend in D5091,
but it was rather brittle and generally running too late to be general.
Furthermore, other targets may want to implement similar optimizations.
So I reimplemented it at the IR-level, fitting it into AtomicExpandPass
as it interacts with that pass (which could not be cleanly done before
at the backend level).
This optimization relies on a new target hook, which is only used by X86
for now, as the correctness of the optimization on other targets remains
an open question. If it is found correct on other targets, it should be
trivial to enable for them.
Details of the optimization are discussed in D5091.
Test Plan: make check-all + a new test
Reviewers: jfb
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D5422
llvm-svn: 218455
These instructions do not indicate they are extendable or the
number of bits in the extendable operand. Rename to match
architected names. Add a testcase for the intrinsics.
llvm-svn: 218453
Summary:
The N32/N64 ABI's require that structs passed in registers are laid out
such that spilling the register with 'sd' places the struct at the lowest
address. For little endian this is trivial but for big-endian it requires
that structs are shifted into the upper bits of the register.
We also require that structs passed in registers have the 'inreg'
attribute for big-endian N32/N64 to work correctly. This is because the
tablegen-erated calling convention implementation only has access to the
lowered form of struct arguments (one or more integers of up to 64-bits
each) and is unable to determine the original type.
Reviewers: vmedic
Reviewed By: vmedic
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D5286
llvm-svn: 218451
On ARM NEON, VAND with immediate (16/32 bits) is an alias to VBIC ~imm with
the same type size. Adding that logic to the parser, and generating VBIC
instructions from VAND asm files.
This patch also fixes the validation routines for NEON splat immediates which
were wrong.
Fixes PR20702.
llvm-svn: 218450
v4f64 and v8f32 shuffles when they are lane-crossing. We have fully
general lane-crossing permutation functions in AVX2 that make this easy.
Part of this also changes exactly when and how these vectors are split
up when we don't have AVX2. This isn't always a win but it usually is
a win, so on the balance I think its better. The primary regressions are
all things that just need to be fixed anyways such as modeling when
a blend can be completely accomplished via VINSERTF128, etc.
Also, this highlights one of the few remaining big features: we do
a really poor job of inserting elements into AVX registers efficiently.
This completes almost all of the big tricks I have in mind for AVX2. The
only things left that I plan to add:
1) element insertion smarts
2) palignr and other fairly specialized lowerings when they happen to
apply
llvm-svn: 218449
256-bit vectors with lane-crossing.
Rather than immediately decomposing to 128-bit vectors, try flipping the
256-bit vector lanes, shuffling them and blending them together. This
reduces our worst case shuffle by a pretty significant margin across the
board.
llvm-svn: 218446
The Thumb2 BXJ instruction (Branch and Exchange Jazelle) is not
defined for v7M or v8A. It is defined for all other Thumb2-supporting
architectures (v6T2, v7A and v7R).
llvm-svn: 218445
lowering where it only used the mask of the low 128-bit lane rather than
the entire mask.
This allows the new lowering to correctly match the unpack patterns for
v8i32 vectors.
For reference, the reason that we check for the the entire mask rather
than checking the repeated mask is because the repeated masks don't
abide by all of the invariants of normal masks. As a consequence, it is
safer to use the full mask with functions like the generic equivalence
test.
llvm-svn: 218442
lowering.
This completes the basic AVX2 feature support, but there are still some
improvements I'd like to do to really get the last mile of performance
here.
llvm-svn: 218440
Kevin Qin