The original version didn't properly account for the base register
being modified before the final jump, so caused miscompilations in
Chromium and LLVM. I've fixed this and tested with an LLVM self-host
(I don't have the means to build & test Chromium).
The general idea remains the same: in pathological cases jump tables
can be too far away from the instructions referencing them (like other
constants) so they need to be movable.
Should fix PR23627.
llvm-svn: 238680
best approach of each.
For vNi16, we use SHL + ADD + SRL pattern that seem easily the best.
For vNi32, we use the PUNPCK + PSADBW + PACKUSWB pattern. In some cases
there is a huge improvement with this in IACA's estimated throughput --
over 2x higher throughput!!!! -- but the measurements are too good to be
true. In one narrow case, the SHL + ADD + SHL + ADD + SRL pattern looks
slightly faster, but I'm not sure I believe any of the measurements at
this point. Both are the exact same uops though. Hard to be confident of
anything past that.
If anyone wants to collect very detailed (Agner-level) timings with the
result of this patch, or with the i32 case replaced with SHL + ADD + SHl
+ ADD + SRL, I'd be very interested. Note that you'll need to test it on
both Ivybridge and Haswell, with both SSE3, SSSE3, and AVX selected as
I saw unique behavior in each of these buckets with IACA all of which
should be checked against measured performance.
But this patch is still a useful improvement by dropping duplicate work
and getting the much nicer PSADBW lowering for v2i64.
I'd still like to rephrase this in terms of generic horizontal sum. It's
a bit lame to have a special case of that just for popcount.
llvm-svn: 238652
helper that skips creating a cast when it isn't necessary.
It's really somewhat concerning that this was caused by the the presence
of a no-op bitcast, but...
llvm-svn: 238642
shifting vectors of bytes as x86 doesn't have direct support for that.
This removes a bunch of redundant masking in the generated code for SSE2
and SSE3.
In order to avoid the really significant code size growth this would
have triggered, I also factored the completely repeatative logic for
shifting and masking into two lambdas which in turn makes all of this
much easier to read IMO.
llvm-svn: 238637
in-register LUT technique.
Summary:
A description of this technique can be found here:
http://wm.ite.pl/articles/sse-popcount.html
The core of the idea is to use an in-register lookup table and the
PSHUFB instruction to compute the population count for the low and high
nibbles of each byte, and then to use horizontal sums to aggregate these
into vector population counts with wider element types.
On x86 there is an instruction that will directly compute the horizontal
sum for the low 8 and high 8 bytes, giving vNi64 popcount very easily.
Various tricks are used to get vNi32 and vNi16 from the vNi8 that the
LUT computes.
The base implemantion of this, and most of the work, was done by Bruno
in a follow up to D6531. See Bruno's detailed post there for lots of
timing information about these changes.
I have extended Bruno's patch in the following ways:
0) I committed the new tests with baseline sequences so this shows
a diff, and regenerated the tests using the update scripts.
1) Bruno had noticed and mentioned in IRC a redundant mask that
I removed.
2) I introduced a particular optimization for the i32 vector cases where
we use PSHL + PSADBW to compute the the low i32 popcounts, and PSHUFD
+ PSADBW to compute doubled high i32 popcounts. This takes advantage
of the fact that to line up the high i32 popcounts we have to shift
them anyways, and we can shift them by one fewer bit to effectively
divide the count by two. While the PSHUFD based horizontal add is no
faster, it doesn't require registers or load traffic the way a mask
would, and provides more ILP as it happens on different ports with
high throughput.
3) I did some code cleanups throughout to simplify the implementation
logic.
4) I refactored it to continue to use the parallel bitmath lowering when
SSSE3 is not available to preserve the performance of that version on
SSE2 targets where it is still much better than scalarizing as we'll
still do a bitmath implementation of popcount even in scalar code
there.
With #1 and #2 above, I analyzed the result in IACA for sandybridge,
ivybridge, and haswell. In every case I measured, the throughput is the
same or better using the LUT lowering, even v2i64 and v4i64, and even
compared with using the native popcnt instruction! The latency of the
LUT lowering is often higher than the latency of the scalarized popcnt
instruction sequence, but I think those latency measurements are deeply
misleading. Keeping the operation fully in the vector unit and having
many chances for increased throughput seems much more likely to win.
With this, we can lower every integer vector popcount implementation
using the LUT strategy if we have SSSE3 or better (and thus have
PSHUFB). I've updated the operation lowering to reflect this. This also
fixes an issue where we were scalarizing horribly some AVX lowerings.
Finally, there are some remaining cleanups. There is duplication between
the two techniques in how they perform the horizontal sum once the byte
population count is computed. I'm going to factor and merge those two in
a separate follow-up commit.
Differential Revision: http://reviews.llvm.org/D10084
llvm-svn: 238636
a separate routine, generalize it to work for all the integer vector
sizes, and do general code cleanups.
This dramatically improves lowerings of byte and short element vector
popcount, but more importantly it will make the introduction of the
LUT-approach much cleaner.
The biggest cleanup I've done is to just force the legalizer to do the
bitcasting we need. We run these iteratively now and it makes the code
much simpler IMO. Other changes were minor, and mostly naming and
splitting things up in a way that makes it more clear what is going on.
The other significant change is to use a different final horizontal sum
approach. This is the same number of instructions as the old method, but
shifts left instead of right so that we can clear everything but the
final sum with a single shift right. This seems likely better than
a mask which will usually have to read the mask from memory. It is
certaily fewer u-ops. Also, this will be temporary. This and the LUT
approach share the need of horizontal adds to finish the computation,
and we have more clever approaches than this one that I'll switch over
to.
llvm-svn: 238635
It turns out that _except_handler3 and _except_handler4 really use the
same stack allocation layout, at least today. They just make different
choices about encoding the LSDA.
This is in preparation for lowering the llvm.eh.exceptioninfo().
llvm-svn: 238627
For some history here see the commit messages of r199797 and r169060.
The original intent was to fix cases like:
%EAX<def> = COPY %ECX<kill>, %RAX<imp-def>
%RCX<def> = COPY %RAX<kill>
where simply removing the copies would have RCX undefined as in terms of
machine operands only the ECX part of it is defined. The machine
verifier would complain about this so 169060 changed such COPY
instructions into KILL instructions so some super-register imp-defs
would be preserved. In r199797 it was finally decided to always do this
regardless of super-register defs.
But this is wrong, consider:
R1 = COPY R0
...
R0 = COPY R1
getting changed to:
R1 = KILL R0
...
R0 = KILL R1
It now looks like R0 dies at the first KILL and won't be alive until the
second KILL, while in reality R0 is alive and must not change in this
part of the program.
As this only happens after register allocation there is not much code
still performing liveness queries so the issue was not noticed. In fact
I didn't manage to create a testcase for this, without unrelated changes
I am working on at the moment.
The fix is simple: As of r223896 the MachineVerifier allows reads from
partially defined registers, so the whole transforming COPY->KILL thing
is not necessary anymore. This patch also changes a similar (but more
benign case as the def and src are the same register) case in the
VirtRegRewriter.
Differential Revision: http://reviews.llvm.org/D10117
llvm-svn: 238588
This patch corresponds to review:
http://reviews.llvm.org/D9941
It adds the various FMA instructions introduced in the version 2.07 of
the ISA along with the testing for them. These are operations on single
precision scalar values in VSX registers.
llvm-svn: 238578
This commit translates the line and column numbers for LLVM IR
errors from the numbers in the YAML block scalar to the numbers
in the MIR file so that the MIRParser users can report LLVM IR
errors with the correct line and column numbers.
Reviewers: Duncan P. N. Exon Smith
Differential Revision: http://reviews.llvm.org/D10108
llvm-svn: 238576
Small (really small!) C++ exception handling examples work on 32-bit x86
now.
This change disables the use of .seh_* directives in WinException when
CFI is not in use. It also uses absolute symbol references in the tables
instead of imagerel32 relocations.
Also fixes a cache invalidation bug in MMI personality classification.
llvm-svn: 238575
organize them by the width of vector.
This makes it a lot easier to see that we're covering all of the vector
types but not doing so excessively. This also adds tests across the
spectrum of SSE versions in addition to the AVX versions.
If you're really tired of seeing the *massive* sprawl of scalarized code
for this, don't worry, I'm just about to land Bruno's patch that
dramatically improve the situation for SSSE3 and newer.
llvm-svn: 238520
This commit introduces a serializable structure called
'llvm::yaml::MachineFunction' that stores the machine
function's name. This structure will mirror the machine
function's state in the future.
This commit prints machine functions as YAML documents
containing a YAML mapping that stores the state of a machine
function. This commit also parses the YAML documents
that contain the machine functions.
Reviewers: Duncan P. N. Exon Smith
Differential Revision: http://reviews.llvm.org/D9841
llvm-svn: 238519
This moves all the state numbering code for C++ EH to WinEHPrepare so
that we can call it from the X86 state numbering IR pass that runs
before isel.
Now we just call the same state numbering machinery and insert a bunch
of stores. It also populates MachineModuleInfo with information about
the current function.
llvm-svn: 238514
For x86 targets, do not do sibling call optimization when materializing
the callee's address would require a GOT relocation. We can still do
tail calls to internal functions, hidden functions, and protected
functions, because they do not require this kind of relocation. It is
still possible to get GOT relocations when the user explicitly asks for
it with musttail or -tailcallopt, both of which are supposed to
guarantee TCO.
Based on a patch by Chih-hung Hsieh.
Reviewers: srhines, timmurray, danalbert, enh, void, nadav, rnk
Subscribers: joerg, davidxl, llvm-commits
Differential Revision: http://reviews.llvm.org/D9799
llvm-svn: 238487
It caused a smaller number of failures than the previous attempt at committing but still caused a couple on the llvm-linux-mips builder. Reverting while I investigate the remainder.
llvm-svn: 238483
We were previously codegen'ing these as regular load/store operations and
hoping that the register allocator would allocate registers in ascending order
so that we could apply an LDM/STM combine after register allocation. According
to the commit that first introduced this code (r37179), we planned to teach
the register allocator to allocate the registers in ascending order. This
never got implemented, and up to now we've been stuck with very poor codegen.
A much simpler approach for achiveing better codegen is to create LDM/STM
instructions with identical sets of virtual registers, let the register
allocator pick arbitrary registers and order register lists when printing an
MCInst. This approach also avoids the need to repeatedly calculate offsets
which ultimately ought to be eliminated pre-RA in order to decrease register
pressure.
This is implemented by lowering the memcpy intrinsic to a series of SD-only
MCOPY pseudo-instructions which performs a memory copy using a given number
of registers. During SD->MI lowering, we lower MCOPY to LDM/STM. This is a
little unusual, but it avoids the need to encode register lists in the SD,
and we can take advantage of SD use lists to decide whether to use the _UPD
variant of the instructions.
Fixes PR9199.
Differential Revision: http://reviews.llvm.org/D9508
llvm-svn: 238473
Summary:
Following on from r209907 which made personality encodings indirect, do the
same for TType encodings. This fixes the case where a try/catch block needs
to generate references to, for example, std::exception in the
.gcc_except_table.
Reviewers: petarj
Reviewed By: petarj
Subscribers: srhines, joerg, tberghammer, llvm-commits
Differential Revision: http://reviews.llvm.org/D9669
llvm-svn: 238427
Extracted from the D6531 patch by Bruno Cardoso Lopes, and re-generated
to reflect the current state of the world. This should let Bruno's D6531
actually show the delta between the approaches by running the x86 test
case update script after re-building.
llvm-svn: 238391
This commit a 3rd attempt at comitting the initial MIR serialization patch.
The first commit (r237708) was reverted in 237730. Then the second commit
(r237954) was reverted in r238007, as the MIR library under CodeGen caused
a circular dependency where the CodeGen library depended on MIR and MIR
library depended on CodeGen.
This commit has fixed the dependencies between CodeGen and MIR by
reorganizing the MIR serialization code - the code that prints out
MIR has been moved to CodeGen, and the MIR library has been renamed
to MIRParser. Now the CodeGen library doesn't depend on the
MIRParser library, thus the circular dependency no longer exists.
--Original Commit Message--
MIR Serialization: print and parse LLVM IR using MIR format.
This commit is the initial commit for the MIR serialization project.
It creates a new library under CodeGen called 'MIR'. This new
library adds a new machine function pass that prints out the LLVM IR
using the MIR format. This pass is then added as a last pass when a
'stop-after' option is used in llc. The new library adds the initial
functionality for parsing of MIR files as well. This commit also
extends the llc tool so that it can recognize and parse MIR input files.
Reviewers: Duncan P. N. Exon Smith, Matthias Braun, Philip Reames
Differential Revision: http://reviews.llvm.org/D9616
llvm-svn: 238341
This broke the llvm-mips-linux builder and several of our out-of-tree builders.
Initial investigations show that the commit probably isn't the problem but
reverting anyway while I investigate.
llvm-svn: 238302
With this patch the x86 backend is now shrink-wrapping capable
and this functionality can be tested by using the
-enable-shrink-wrap switch.
The next step is to make more test and enable shrink-wrapping by
default for x86.
Related to <rdar://problem/20821487>
llvm-svn: 238293
This gets gas and llc -filetype=obj to agree on the order of prefixes.
For llvm-mc we need to fix the asm parser to know that it makes a difference
on which line the "lock" is in.
Part of pr23594.
llvm-svn: 238232
Summary:
Following on from r209907 which made personality encodings indirect, do the
same for TType encodings. This fixes the case where a try/catch block needs
to generate references to, for example, std::exception in the
.gcc_except_table.
This commit uses DW_EH_PE_sdata8 for N64 as far as is possible at the moment.
However, it is possible to end up with DW_EH_PE_sdata4 when a TargetMachine is
not available. There's no risk of issues with inconsistency here since the
tables are self describing but it does mean there is a small chance of the
PC-relative offset being out of range for particularly large programs.
Reviewers: petarj
Reviewed By: petarj
Subscribers: srhines, joerg, tberghammer, llvm-commits
Differential Revision: http://reviews.llvm.org/D9669
llvm-svn: 238190
Part of D9474, this patch extends AVX2 v16i16 types to 2 x 8i32 vectors and uses i32 shift variable shifts before packing back to i16.
Adds AVX2 tests for v8i16 and v16i16
llvm-svn: 238149
in POWER8:
vadduqm
vaddeuqm
vaddcuq
vaddecuq
vsubuqm
vsubeuqm
vsubcuq
vsubecuq
In addition to adding the instructions themselves, it also adds support for the
v1i128 type for intrinsics (Intrinsics.td, Function.cpp, and
IntrinsicEmitter.cpp).
http://reviews.llvm.org/D9081
llvm-svn: 238144
The semantics of the scalar FMA intrinsics are that the high vector elements are copied from the first source.
The existing pattern switches src1 and src2 around, to match the "213" order, which ends up tying the original src2 to the dest. Since the actual scalar fma3 instructions copy the high elements from the dest register, the wrong values are copied.
This modifies the pattern to leave src1 and src2 in their original order.
Differential Revision: http://reviews.llvm.org/D9908
llvm-svn: 238131
On GPU targets, materializing constants is cheap and stores are
expensive, so only doing this for zero vectors was silly.
Most of the new testcases aren't optimally merged, and are for
later improvements.
llvm-svn: 238108
When the compare feeding a branch was in a different BB from the branch, we'd
try to "regenerate" the compare in the block with the branch, possibly trying
to make use of values not available there. Copy a page from AArch64's play book
here to fix the problem (at least in terms of correctness).
Fixes PR23640.
llvm-svn: 238097
This is part of the work to remove TargetMachine::resetTargetOptions.
In this patch, instead of updating global variable NoFramePointerElim in
resetTargetOptions, its use in DisableFramePointerElim is replaced with a call
to TargetFrameLowering::noFramePointerElim. This function determines on a
per-function basis if frame pointer elimination should be disabled.
There is no change in functionality except that cl:opt option "disable-fp-elim"
can now override function attribute "no-frame-pointer-elim".
llvm-svn: 238080
The usual CodeGenPrepare trickery, on a target-specific intrinsic.
Without this, the expansion of atomics will usually have the zext
be hoisted out of the loop, defeating the various patterns we have
to catch this precise case.
Differential Revision: http://reviews.llvm.org/D9930
llvm-svn: 238054
This patch adds a class for processing many recip codegen possibilities.
The TargetRecip class is intended to handle both command-line options to llc as well
as options passed in from a front-end such as clang with the -mrecip option.
The x86 backend is updated to use the new functionality.
Only -mcpu=btver2 with -ffast-math should see a functional change from this patch.
All other CPUs continue to *not* use reciprocal estimates by default with -ffast-math.
Differential Revision: http://reviews.llvm.org/D8982
llvm-svn: 238051
