VSX is an ISA extension supported on the POWER7 and later cores that enhances
floating-point vector and scalar capabilities. Among other things, this adds
<2 x double> support and generally helps to reduce register pressure.
The interesting part of this ISA feature is the register configuration: there
are 64 new 128-bit vector registers, the 32 of which are super-registers of the
existing 32 scalar floating-point registers, and the second 32 of which overlap
with the 32 Altivec vector registers. This makes things like vector insertion
and extraction tricky: this can be free but only if we force a restriction to
the right register subclass when needed. A new "minipass" PPCVSXCopy takes care
of this (although it could do a more-optimal job of it; see the comment about
unnecessary copies below).
Please note that, currently, VSX is not enabled by default when targeting
anything because it is not yet ready for that. The assembler and disassembler
are fully implemented and tested. However:
- CodeGen support causes miscompiles; test-suite runtime failures:
MultiSource/Benchmarks/FreeBench/distray/distray
MultiSource/Benchmarks/McCat/08-main/main
MultiSource/Benchmarks/Olden/voronoi/voronoi
MultiSource/Benchmarks/mafft/pairlocalalign
MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4
SingleSource/Benchmarks/CoyoteBench/almabench
SingleSource/Benchmarks/Misc/matmul_f64_4x4
- The lowering currently falls back to using Altivec instructions far more
than it should. Worse, there are some things that are scalarized through the
stack that shouldn't be.
- A lot of unnecessary copies make it past the optimizers, and this needs to
be fixed.
- Many more regression tests are needed.
Normally, I'd fix these things prior to committing, but there are some
students and other contributors who would like to work this, and so it makes
sense to move this development process upstream where it can be subject to the
regular code-review procedures.
llvm-svn: 203768
This change enables tracking i1 values in the PowerPC backend using the
condition register bits. These bits can be treated on PowerPC as separate
registers; individual bit operations (and, or, xor, etc.) are supported.
Tracking booleans in CR bits has several advantages:
- Reduction in register pressure (because we no longer need GPRs to store
boolean values).
- Logical operations on booleans can be handled more efficiently; we used to
have to move all results from comparisons into GPRs, perform promoted
logical operations in GPRs, and then move the result back into condition
register bits to be used by conditional branches. This can be very
inefficient, because the throughput of these CR <-> GPR moves have high
latency and low throughput (especially when other associated instructions
are accounted for).
- On the POWER7 and similar cores, we can increase total throughput by using
the CR bits. CR bit operations have a dedicated functional unit.
Most of this is more-or-less mechanical: Adjustments were needed in the
calling-convention code, support was added for spilling/restoring individual
condition-register bits, and conditional branch instruction definitions taking
specific CR bits were added (plus patterns and code for generating bit-level
operations).
This is enabled by default when running at -O2 and higher. For -O0 and -O1,
where the ability to debug is more important, this feature is disabled by
default. Individual CR bits do not have assigned DWARF register numbers,
and storing values in CR bits makes them invisible to the debugger.
It is critical, however, that we don't move i1 values that have been promoted
to larger values (such as those passed as function arguments) into bit
registers only to quickly turn around and move the values back into GPRs (such
as happens when values are returned by functions). A pair of target-specific
DAG combines are added to remove the trunc/extends in:
trunc(binary-ops(binary-ops(zext(x), zext(y)), ...)
and:
zext(binary-ops(binary-ops(trunc(x), trunc(y)), ...)
In short, we only want to use CR bits where some of the i1 values come from
comparisons or are used by conditional branches or selects. To put it another
way, if we can do the entire i1 computation in GPRs, then we probably should
(on the POWER7, the GPR-operation throughput is higher, and for all cores, the
CR <-> GPR moves are expensive).
POWER7 test-suite performance results (from 10 runs in each configuration):
SingleSource/Benchmarks/Misc/mandel-2: 35% speedup
MultiSource/Benchmarks/Prolangs-C++/city/city: 21% speedup
MultiSource/Benchmarks/MiBench/automotive-susan: 23% speedup
SingleSource/Benchmarks/CoyoteBench/huffbench: 13% speedup
SingleSource/Benchmarks/Misc-C++/Large/sphereflake: 13% speedup
SingleSource/Benchmarks/Misc-C++/mandel-text: 10% speedup
SingleSource/Benchmarks/Misc-C++-EH/spirit: 10% slowdown
MultiSource/Applications/lemon/lemon: 8% slowdown
llvm-svn: 202451
subsequent changes are easier to review. About to fix some layering
issues, and wanted to separate out the necessary churn.
Also comment and sink the include of "Windows.h" in three .inc files to
match the usage in Memory.inc.
llvm-svn: 198685
This adds a scheduling model for the POWER7 (P7) core, and enables the
machine-instruction scheduler when targeting the P7. Scheduling for the P7,
like earlier ooo PPC cores, requires considering both dispatch group hazards,
and functional unit resources and latencies. These are both modeled in a
combined itinerary. Dispatch group formation is still handled by the post-RA
scheduler (which still needs to be updated for the P7, but nevertheless does a
pretty good job).
One interesting aspect of this change is that I've also enabled to use of AA
duing CodeGen for the P7 (just as it is for the embedded cores). The benchmark
results seem to support this decision (see below), and while this is normally
useful for in-order cores, and not for ooo cores like the P7, I think that the
dispatch slot hazards are enough like in-order resources to make the AA useful.
Test suite significant performance differences (where negative is a speedup,
and positive is a regression) vs. the current situation:
MultiSource/Benchmarks/BitBench/drop3/drop3
with AA: N/A
without AA: -28.7614% +/- 19.8356%
(significantly against AA)
MultiSource/Benchmarks/FreeBench/neural/neural
with AA: -17.7406% +/- 11.2712%
without AA: N/A
(significantly in favor of AA)
MultiSource/Benchmarks/SciMark2-C/scimark2
with AA: -11.2079% +/- 1.80543%
without AA: -11.3263% +/- 2.79651%
MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt
with AA: -41.8649% +/- 17.0053%
without AA: -34.5256% +/- 23.7072%
MultiSource/Benchmarks/mafft/pairlocalalign
with AA: 25.3016% +/- 17.8614%
without AA: 38.6629% +/- 14.9391%
(significantly in favor of AA)
MultiSource/Benchmarks/sim/sim
with AA: N/A
without AA: 13.4844% +/- 7.18195%
(significantly in favor of AA)
SingleSource/Benchmarks/BenchmarkGame/Large/fasta
with AA: 15.0664% +/- 6.70216%
without AA: 12.7747% +/- 8.43043%
SingleSource/Benchmarks/BenchmarkGame/puzzle
with AA: 82.2713% +/- 26.3567%
without AA: 75.7525% +/- 41.1842%
SingleSource/Benchmarks/Misc/flops-2
with AA: -37.1621% +/- 20.7964%
without AA: -35.2342% +/- 20.2999%
(significantly in favor of AA)
These are 99.5% confidence intervals from 5 runs per configuration. Regarding
the choice to turn on AA during CodeGen, of these results, four seem
significantly in favor of using AA, and one seems significantly against. I'm
not making this decision based on these numbers alone, but these results
seem consistent with results I have from other tests, and so I think that, on
balance, using AA is a win.
llvm-svn: 195981
Use the new instruction deprecation feature to mark mftb (now replaced with
mfspr) and dst (along with the other Altivec cache control instructions) as
deprecated when targeting cores supporting at least ISA v2.03.
llvm-svn: 190605
Aggressive anti-dependency breaking is enabled by default for all PPC cores.
This provides a general speedup on the P7 and other platforms (among other
factors, the instruction group formation for the non-embedded PPC cores is done
during post-RA scheduling). In order to do this safely, the incompatibility
between uses of the MFOCRF instruction and anti-dependency breaking are
resolved by marking MFOCRF with hasExtraSrcRegAllocReq. As noted in the removed
FIXME, the problem was that MFOCRF's output is sensitive to the identify of the
source register, and always paired with a shift to undo this effect. Because
anti-dependency breaking is unaware of this hidden dependency of the shift
amount on the source register of the MFOCRF instruction, changing that register
must be inhibited.
Two test cases were adjusted: The SjLj test was made more insensitive to
register choices and scheduling; the saveCR test disabled anti-dependency
breaking because part of what it is testing is proper register reuse.
llvm-svn: 190587
Modern PPC cores support a floating-point copysign instruction, and we can use
this to lower the FCOPYSIGN node (which is created from calls to the libm
copysign function). A couple of extra patterns are necessary because the
operand types of FCOPYSIGN need not agree.
llvm-svn: 188653
This patch provides basic support for powerpc64le as an LLVM target.
However, use of this target will not actually generate little-endian
code. Instead, use of the target will cause the correct little-endian
built-in defines to be generated, so that code that tests for
__LITTLE_ENDIAN__, for example, will be correctly parsed for
syntax-only testing. Code generation will otherwise be the same as
powerpc64 (big-endian), for now.
The patch leaves open the possibility of creating a little-endian
PowerPC64 back end, but there is no immediate intent to create such a
thing.
The LLVM portions of this patch simply add ppc64le coverage everywhere
that ppc64 coverage currently exists. There is nothing of any import
worth testing until such time as little-endian code generation is
implemented. In the corresponding Clang patch, there is a new test
case variant to ensure that correct built-in defines for little-endian
code are generated.
llvm-svn: 187179
This change mirrors the changes that were made to the X86 and ARM targets to
support subtarget feature changing. As indicated in r182899, the mechanism is
still undergoing revision, and so as with the X86 and ARM targets, there is no
test case yet (there is no effective functionality change).
llvm-svn: 186357
When unsafe FP math operations are enabled, we can use the fre[s] and
frsqrte[s] instructions, which generate reciprocal (sqrt) estimates, together
with some Newton iteration, in order to quickly generate floating-point
division and sqrt results. All of these instructions are separately optional,
and so each has its own feature flag (except for the Altivec instructions,
which are covered under the existing Altivec flag). Doing this is not only
faster than using the IEEE-compliant fdiv/fsqrt instructions, but allows these
computations to be pipelined with other computations in order to hide their
overall latency.
I've also added a couple of missing fnmsub patterns which turned out to be
missing (but are necessary for good code generation of the Newton iterations).
Altivec needs a similar fix, but that will probably be more complicated because
fneg is expanded for Altivec's v4f32.
llvm-svn: 178617
The P7 and A2 have additional floating-point conversion instructions which
allow a direct two-instruction sequence (plus load/store) to convert from all
combinations (signed/unsigned i32/i64) <--> (float/double) (on previous cores,
only some combinations were directly available).
llvm-svn: 178480
This instruction is available on modern PPC64 CPUs, and is now used
to improve the SINT_TO_FP lowering (by eliminating the need for the
separate sign extension instruction and decreasing the amount of
needed stack space).
llvm-svn: 178446
These instructions are available on the P5x (and later) and on the A2. They
implement the standard floating-point rounding operations (floor, trunc, etc.).
One caveat: frin (round to nearest) does not implement "ties to even", and so
is only enabled in fast-math mode.
llvm-svn: 178337
These are 64-bit load/store with byte-swap, and available on the P7 and the A2.
Like the similar instructions for 16- and 32-bit words, these are matched in the
target DAG-combine phase against load/store-bswap pairs.
llvm-svn: 178276
PPC ISA 2.06 (P7, A2, etc.) has a popcntd instruction. Add this instruction and
tell TTI about it so that popcount-loop recognition will know about it.
llvm-svn: 178233
into their new header subdirectory: include/llvm/IR. This matches the
directory structure of lib, and begins to correct a long standing point
of file layout clutter in LLVM.
There are still more header files to move here, but I wanted to handle
them in separate commits to make tracking what files make sense at each
layer easier.
The only really questionable files here are the target intrinsic
tablegen files. But that's a battle I'd rather not fight today.
I've updated both CMake and Makefile build systems (I think, and my
tests think, but I may have missed something).
I've also re-sorted the includes throughout the project. I'll be
committing updates to Clang, DragonEgg, and Polly momentarily.
llvm-svn: 171366
Sooooo many of these had incorrect or strange main module includes.
I have manually inspected all of these, and fixed the main module
include to be the nearest plausible thing I could find. If you own or
care about any of these source files, I encourage you to take some time
and check that these edits were sensible. I can't have broken anything
(I strictly added headers, and reordered them, never removed), but they
may not be the headers you'd really like to identify as containing the
API being implemented.
Many forward declarations and missing includes were added to a header
files to allow them to parse cleanly when included first. The main
module rule does in fact have its merits. =]
llvm-svn: 169131
This patch adds initial PPC64 TOC MC object creation using the small mcmodel
(a single 64K TOC) adding the some TOC relocations (R_PPC64_TOC,
R_PPC64_TOC16, and R_PPC64_TOC16DS).
The addition of 'undefinedExplicitRelSym' hook on 'MCELFObjectTargetWriter'
is meant to avoid the creation of an unreferenced ".TOC." symbol (used in
the .odp creation) as well to set the R_PPC64_TOC relocation target as the
temporary ".TOC." symbol. On PPC64 ABI, the R_PPC64_TOC relocation should
not point to any symbol.
llvm-svn: 166677
Original commit message:
Move PPC host-CPU detection logic from PPCSubtarget into sys::getHostCPUName().
Both the new Linux functionality and the old Darwin functions have been moved.
This change also allows this information to be queried directly by clang and
other frontends (clang, for example, will now have real -mcpu=native support).
llvm-svn: 158349
Both the new Linux functionality and the old Darwin functions have been moved.
This change also allows this information to be queried directly by clang and
other frontends (clang, for example, will now have real -mcpu=native support).
llvm-svn: 158337
The PPC target feature gpul (IsGigaProcessor) was only used for one thing:
To enable the generation of the MFOCRF instruction. Furthermore, this
instruction is available on other PPC cores outside of the G5 line. This
feature now corresponds to the HasMFOCRF flag.
No functionality change.
llvm-svn: 158323
Using 'all' instead of 'critical' would be better because it would make it easier to
satisfy the bundling constraints, but, as noted in the FIXME, that is currently not
possible with the crs.
This yields an average 1% speedup over the entire test suite (on Power 7). Largest speedups:
SingleSource/Benchmarks/Shootout-C++/moments - 40%
MultiSource/Benchmarks/McCat/03-testtrie/testtrie - 28%
SingleSource/Benchmarks/BenchmarkGame/nsieve-bits - 26%
SingleSource/Benchmarks/McGill/misr - 23%
MultiSource/Applications/JM/ldecod/ldecod - 22%
Largest slowdowns:
SingleSource/Benchmarks/Shootout-C++/matrix - -29%
SingleSource/Benchmarks/Shootout-C++/ary3 - -22%
MultiSource/Benchmarks/BitBench/uuencode/uuencode - -18%
SingleSource/Benchmarks/Shootout-C++/ary - -17%
MultiSource/Benchmarks/MiBench/automotive-bitcount/automotive-bitcount - -15%
llvm-svn: 158294
Post-RA scheduling gives a significant performance improvement on
the embedded cores, so turn it on. Using full anti-dep. breaking is
important for FP-intensive blocks, so turn it on (just on the
embedded cores for now; this should also be good on the 970s because
post-ra scheduling is all that we have for now, but that should have
more testing first).
llvm-svn: 153843
This adds a full itinerary for IBM's PPC64 A2 embedded core. These
cores form the basis for the CPUs in the new IBM BG/Q supercomputer.
llvm-svn: 153842
and MCSubtargetInfo.
- Added methods to update subtarget features (used when targets automatically
detect subtarget features or switch modes).
- Teach X86Subtarget to update MCSubtargetInfo features bits since the
MCSubtargetInfo layer can be shared with other modules.
- These fixes .code 16 / .code 32 support since mode switch is updated in
MCSubtargetInfo so MC code emitter can do the right thing.
llvm-svn: 134884
CPU, and feature string. Parsing some asm directives can change
subtarget state (e.g. .code 16) and it must be reflected in other
modules (e.g. MCCodeEmitter). That is, the MCSubtargetInfo instance
must be shared.
llvm-svn: 134795
- Each target asm parser now creates its own MCSubtatgetInfo (if needed).
- Changed AssemblerPredicate to take subtarget features which tablegen uses
to generate asm matcher subtarget feature queries. e.g.
"ModeThumb,FeatureThumb2" is translated to
"(Bits & ModeThumb) != 0 && (Bits & FeatureThumb2) != 0".
llvm-svn: 134678
itineraries.
- Refactor TargetSubtarget to be based on MCSubtargetInfo.
- Change tablegen generated subtarget info to initialize MCSubtargetInfo
and hide more details from targets.
llvm-svn: 134257
be the first encoded as the first feature. It then uses the CPU name to look up
features / scheduling itineray even though clients know full well the CPU name
being used to query these properties.
The fix is to just have the clients explictly pass the CPU name!
llvm-svn: 134127
nodes to indicate when ha16/lo16 modifiers should be used. This lets
us pass PowerPC/indirectbr.ll.
The one annoying thing about this patch is that the MCSymbolExpr isn't
expressive enough to represent ha16(label1-label2) which we need on
PowerPC. I have a terrible hack in the meantime, but this will have
to be revisited at some point.
Last major conversion item left is global variable references.
llvm-svn: 119105
See PR5201. There is no way to know if direct calls will be within the allowed
range for BL. Hence emit all calls as indirect when in JIT mode.
Without this long-running applications will fail to JIT on PowerPC with a
relocation failure.
llvm-svn: 110246
Modules and ModuleProviders. Because the "ModuleProvider" simply materializes
GlobalValues now, and doesn't provide modules, it's renamed to
"GVMaterializer". Code that used to need a ModuleProvider to materialize
Functions can now materialize the Functions directly. Functions no longer use a
magic linkage to record that they're materializable; they simply ask the
GVMaterializer.
Because the C ABI must never change, we can't remove LLVMModuleProviderRef or
the functions that refer to it. Instead, because Module now exposes the same
functionality ModuleProvider used to, we store a Module* in any
LLVMModuleProviderRef and translate in the wrapper methods. The bindings to
other languages still use the ModuleProvider concept. It would probably be
worth some time to update them to follow the C++ more closely, but I don't
intend to do it.
Fixes http://llvm.org/PR5737 and http://llvm.org/PR5735.
llvm-svn: 94686
Hal Finkel