Folding of transfer_write into transfer_read is already supported but
this requires the read and write to have the same permuation map.
After linalg vectorization it is common to have different ppermuation
map for write followed by read even though the cases could be
propagated.
This canonicalization handle cases where the permuation maps are
different but the data read and written match and replace the transfer
ops with broadcast and permuation
Differential Revision: https://reviews.llvm.org/D130135
This patch aims to create a webpage to document
Flang's command line options on https://flang.llvm.org/docs/
in a similar way to Clang's
https://clang.llvm.org/docs/ClangCommandLineReference.html
This is done by using clang_tablegen to generate an .rst
file from Options.td (which is current shared with Clang)
For this to work, ClangOptionDocEmitter.cpp was updated
to allow specific Flang flags to be included,
rather than bulk excluding clang flags.
Note:
Some headings in the generated documentation will incorrectly
contain references to Clang, e.g.
"Flags controlling the behaviour of Clang during compilation"
This is because Options.td (Which is shared between both Clang and Flang)
contains hard-coded DocBrief sections. I couldn't find a non-intrusive way
to make this target-dependant, as such I've left this as is, and it will need revisiting later.
Reviewed By: awarzynski
Differential Revision: https://reviews.llvm.org/D129864
The standard says two equal pointers must compare equal
so there is no need to call the user comparator function
in this case.
Differential Revision: https://reviews.llvm.org/D130310
This patch modifies only test cases.
Converted the MMA test cases to use opaque pointers.
Reviewed By: lei, amyk
Differential Revision: https://reviews.llvm.org/D130090
As noticed on D129765 and reported on Issue #56531 - aarch64 targets can use the neon ctpop + add-reduce instructions to speed up scalar ctpop instructions, but we fail to do this for parity calculations.
I'm not sure where the cutoff should be for specific CPUs, but i64 (+ i128 special case) shows a definite reduction in instruction count. i32 is about the same (but scalar <-> neon transfers are probably more costly?), and sub-i32 promotion looks to be a definite regression compared to parity expansion optimized for those widths.
Differential Revision: https://reviews.llvm.org/D130246
The name `getEntrySamples` was misleading for 2 reasons. One, it's
close in name to `Function::getEntryCount`, but the equivalent here is
`getHeadSamples`; second, as opposed to the other get* APIs in
`FunctionSamples`, it performs an estimate/heuristic rather than just
retrieving raw data (or a non-heuristic derivate off that data, like
`getMaxCountInside`)
The new name should more clearly communicate its intent; and, being
close (in name) to `getHeadSamples`, it should allow the reader discover
the relation between them.
Also updated the doc comments for both `getHeadSamples[Estimate]` so a
reader may better understand the relation between them.
Differential Revision: https://reviews.llvm.org/D130281
This commit changes how math intrinsics are lowered: we, first,
try to lower them into MLIR operations or libm calls via
mathOperations table and only then fallback to pgmath runtime calls.
The pgmath fallback is needed, because mathOperations does not
support all intrinsics that pgmath supports. The main purpose
of this change is to get rid of llvmIntrinsics table so that
we do not have to update both llvmIntrinsics and mathOperations
when adding new intrinsic support.
mathOperations lowering should phase out pgmath lowering, when
more operations are available (e.g. power operations being
added in D129809 and D129811; complex type operations from
Complex dialect).
Differential Revision: https://reviews.llvm.org/D130129
Multiple calls to `omp_get_wtime` could be optimized out due to the function
is mistakenly marked as `readnone`. This patch fixes the issue, and also add the
support to run optimization on `libomptarget` tests.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D130179
We were looking for loads or any_extend+load. reduceLoadWidth
hasn't known how to look through such an any_extend to find the
load since D40667 almost 5 years ago.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D130333
Reorganize the code to make it clear what is and isn't handle, and why.
Restructure bailout to remove (false and confusing) dependence on
CM_Scalarize; just return invalid cost and propagate, that's what it
is for.
This warning was added because using attribute or type assembly formats
with `skipDefaultBuilders` set could cause compilation errors, since the
required builder prototype may not necessarily be generated and would
need to be checked by hand. This patch removes the warning because a
warning that the generated C++ "might" not compile is not particularly
useful. Attempting to address the TODO (i.e. detect whether a builder of
the correct prototype is provided) would be fragile since it would not
be possible to account for implicit conversions, etc.
In general, ODS should not be emitting warnings in cases like these.
Reviewed By: rriddle, wrengr
Differential Revision: https://reviews.llvm.org/D130102
Depends On D130304
This patch pulls the `ApplyBuiltinTransfer` from the `TypeErasedDataflowAnalysis` class into a new `DataflowAnalysisOptions` struct, to allow us to add additional options later without breaking existing code.
Reviewed By: gribozavr2, sgatev
Differential Revision: https://reviews.llvm.org/D130305
The latest perf tool can return non-empty buffer when executing
buildid-list command, even when perf.data was recorded with -B flag.
Some binaries will be listed without the ID, while others may have a
recorded ID. Allow invalid entires on the input, while checking the
valid ones for the match.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D130223
This patch adds a command line flag to be able to test
the type based cost-model analysis for Intrinsics.
Differential Revision: https://reviews.llvm.org/D129109
With correctly rounded implementations, these functions will be tested for all
rounding modes. Since fegetround and fesetround are not implemented for arm32,
these tests will fail in all non-default rounding modes. We will re-enable
these entrypoints and tests once fegetround and fesetround are implemented for
arm32.
GEP's across basic blocks were not getting splitted due to EnableGEPOpt
which was turned off by default. Hence, EarlyCSE missed the opportunity
to eliminate common part of GEP's. This can be achieved by simply
turning GEP pass on.
- This patch moves SeparateConstOffsetFromGEPPass() just before LSR.
- It enables EnableGEPOpt by default.
Resolves - https://github.com/llvm/llvm-project/issues/50528
Added an unit test.
Differential Revision: https://reviews.llvm.org/D128582
Scope ops file to ops. Used canonicalization as grouping for canonicalization
patterns and folders (also considered OpTransforms but that felt too generic
and the former two are used together).
Reviewed By: silvas, rsuderman
Differential Revision: https://reviews.llvm.org/D130297
This patch moves `Analysis/FlowSensitive/NoopAnalysis.h` from `clang/unittests/` to `clang/include/clang/`, so that we can use it for doing context-sensitive analysis.
Reviewed By: ymandel, gribozavr2, sgatev
Differential Revision: https://reviews.llvm.org/D130304
Move this out of the switch, so that different branches can
indicate an error by breaking out of the switch. This becomes
important if there are more than the two current error cases.
Summary:
1. Added a new option object mode -X for llvm-ar. In AIX OS , there is a object mode option -X for ar command.
please see the "-X mode" part of https://www.ibm.com/docs/ko/aix/7.1?topic=ar-command
Specifies the type of object file ar should examine. The mode must be one of the following:
32
Processes only 32-bit object files
64
Processes only 64-bit object files
32_64
Processes both 32-bit and 64-bit object files
any
Processes all of the supported object files.
The default is to process 32-bit object files (ignore 64-bit objects). The mode can also be set with the OBJECT_MODE environment variable. For example, OBJECT_MODE=64 causes ar to process any 64-bit objects and ignore 32-bit objects. The -X flag overrides the OBJECT_MODE variable.
2. Before adding the new option -X, the default behaviors of llvm-ar like -Xany, but after the adding the new option -X, the default behaviors of llvm-ar change to -X32 ,in order to let some test cases which has 32bit and 64bit object file in the same llvm-ar command, we need to add the "export OBJECT_MODE=any" into test case to change the default behaviors of llvm-ar's object mode.
Reviewers: James Henderson, Owen Reynolds, Fangrui Song
Differential Revision: https://reviews.llvm.org/D127864
Previously we made `libomptarget` link as an LLVM library so we have
access to the LLVM core libraries. After the initial patch stuck we can
now apply the same changes to the plugins. This will allow us to use
LLVM in all of `libomptarget` when we have uses for them. In the future
this should allow us to remove the dependencies on `libelf`, `libffi`,
and `dl`.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D130262
Add float type and flag for nearest integer to automatically test with
and without SSE4.2 flag.
Reviewed By: sivachandra
Differential Revision: https://reviews.llvm.org/D129916
Add a constraint to ensure that the operand and result of the
threadprivate operation are the same.
Reviewed By: peixin
Differential Revision: https://reviews.llvm.org/D128609
Opaque pointers support is complete and default. Specify ptr as
the normal pointer type and i8* as something supported under
non-default options.
A larger update of examples in LangRef is still needed.
This could crash when our heuristic picks the wrong function. Make sure
there is enough parameters in the candidate to prevent those crashes.
Also special case copy/move constructors to make the heuristic work in
presence of those.
Fixes https://github.com/llvm/llvm-project/issues/56620
Differential Revision: https://reviews.llvm.org/D130260
When -fexperimental-library is passed, libc++ will now pick up the
appropriate __has_feature flag defined by Clang to enable the
experimental library features.
As a fly-by, also update the documentation for the various TSes.
Differential Revision: https://reviews.llvm.org/D130176
In D125283, we ensured that integer distributions would not compile when
used with arbitrary unsupported types. This effectively enforced what
the Standard mentions here: http://eel.is/c++draft/rand#req.genl-1.5.
However, this also had the effect of breaking some users that were
using integer distributions with unsupported types like int8_t. Since we
already support using __int128_t in those distributions, it is reasonable
to also support smaller types like int8_t and its unsigned variant. This
commit implements that, adds tests and documents the extension. Note that
we voluntarily don't add support for instantiating these distributions
with bool and char, since those are not integer types. However, it is
trivial to replace uses of these random distributions on char using int8_t.
It is also interesting to note that in the process of adding tests
for smaller types, I discovered that our distributions sometimes don't
provide as faithful a distribution when instantiated with smaller types,
so I had to relax a couple of tests. In particular, we do a really bad
job at implementing the negative binomial, geometric and poisson distributions
for small types. I think this all boils down to the algorithm we use in
std::poisson_distribution, however I am running out of time to investigate
that and changing the algorithm would be an ABI break (which might be
reasonable).
As part of this patch, I also added a mitigation for a very likely
integer overflow bug we were hitting in our tests in negative_binomial_distribution.
I also filed http://llvm.org/PR56656 to track fixing the problematic
distributions with int8_t and uint8_t.
Supersedes D125283.
Differential Revision: https://reviews.llvm.org/D126823
This patch adds the use of the `-internalize-public-api-file` option in
the internalization pass to internalize any definition that isn't
explicitly needed for the interface. This will allow us to perform more
optimizations on the file that normally would not have been possible
with functions internal to the library not being internal.
Depends on D130293
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D130298
The internalize pass supports an option to provide a list of symbols
that should not be internalized. THis is useful retaining certain
defintions that should be kept alive. However, this interface is
somewhat difficult to use as it requires knowing every single symbol's
name and specifying it. Many APIs provide common prefixes for the
symbols exported by the library, so it would make sense to be able to
match these using a simple glob pattern. This patch changes the handling
from a simple string comparison to a glob pattern match.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D130319
Currently the bitcode library is build using the clang front-end
manually. This was originally done because we did not support device
only compilation. Now we support device only compilation, at least for a
single offloading toolchain, so we can instead use clang directly rather
than using the front-end. This saves us needing to define things like
`aux_triple`.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D130293
This is useful for building small test cases and will be utilized in a subsequent commit that adds a fusion example.
Differential Revision: https://reviews.llvm.org/D130344
This op fuses a given payload op into a given container op. Inside the container, all uses of the producer are replaced (fused) with the newly inserted op. If the producer is tileable and accessed via a tensor.extract_slice, the new op computes only the requested slice ("tile and fuse"). Otherwise, the entire tensor value is computed inside the container ("clone and fuse").
Differential Revision: https://reviews.llvm.org/D130244
Thomas Raoux