This prevents an infinite loop from D123801, where code trying to reduce
the total number of bitcasts, but also handling constants, could create
the opposite transform. Prevent the transform in these case to let the
bitcast of a constant transform naturally.
Fixes#55345
When processing an entry-stmt in name resolution, attrs_ was
reset before SetBindNameOn was called, causing the symbol to lose
the binding label information.
Differential Revision: https://reviews.llvm.org/D125097
The per-callsite size threshold used today to drive preinline decision is based on hotness/coldness cutoff. The default setup is for callsites with a sample count above the hotness cutoff (99%), a 1500 size threshold is used. Any callsite below 99.99% coldness cutoff uses a zero threshold. This has a couple issues:
1. While both cutoffs and size thoresholds are configurable, different applications may need different setups, making a universal setup impractical.
2. The callsites between hotness cutoff and coldness cutoff are not considered as inline candidates, which could be a missing opportunity.
3. Hot callsites always use the same threshold. In reality we may want a bigger threshold for hotter callsites.
In this change we are introducing a linear threshold regardless of hot/cold cutoffs. Given a sample space, a threshold is computed for a callsite based on the position of that callsite sample in the whole space. With that we no longer need to define what's hot or cold. Callsites with different hotness will get a different threshold. This should overcome the above three issues.
I have seen good results with a universal default setup for two of our internal services.
For one service, 0.2% to 0.5% perf improvement over a baseline with a previous default setup, on-par code size.
For the second service, 0.5% to 0.8% perf improvement over a baseline with a previous default setup, 0.2% code size increase; on-par performance and code size with a baseline that is with a carefully tuned cutoff to cover enough hot functions.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D125023
Adds missing logic in the lowering from NvGPU to NVVM to support fp32
(in an accumulator operand) and tf32 (in multiplicand operand) types.
Fixes logic in one of the helper functions for converting the result
of a mma.sync operation with multiple 8x256bit output tiles, which is
the case for f32 outputs.
Differential Revision: https://reviews.llvm.org/D124533
As Fortran 2018 5.2.2 states, a program shall consist of exactly one
main program. Add this semantic check.
Reviewed By: klausler
Differential Revision: https://reviews.llvm.org/D125186
These are all microcoded/multi-pipe nightmares on Ryzen, but we shouldn't just be using the WriteMicrocoded class which is for REALLY bad microcoded nightmares - instead use the same approximate latencies as znver2 (Agner and uops.info both suggest similar values) - and make sure we use the FPU defs for both
Fixes#53242
\operation ... \endoperation are not valid doxygen commands and cause issues when -Wdocumentation is enabled (Issue #35297)
This patch proposes to replace them with \code{.operation} ... \endcode blocks so that the pseudo-code is correctly retained in any documentation and downstream can use the ".operation" type for its own formatting.
Differential Revision: https://reviews.llvm.org/D125170
Given a commutative reduction leading from a shuffle, the order of the
lanes on the shuffle are not important for the result. This means we can
reorder the shuffle to something simpler, which we try shuffling the
first vector lanes first. This was D123494.
The new shuffle may not be profitable though, and if it is not we can
try the folding of select shuffles from D123911. This, with some
adjustment as the output lane ordering is now unimportant, can allow the
final shuffle to simplify given the inputs to the patterns from D123911.
Where as each transformation on their own are not profitable, the
combination is.
We can only support a single shuffle when called from reductions, but we
are able to sort the ReconstructMask, potentially allowing it to
simplify to an identity or concat mask.
Differential Revision: https://reviews.llvm.org/D125086
Another step toward enabling full -Wsystem-headers testing across all x86 headers
Fix a number of cases where the arg / return value signedness doesn't match the C/C++ intrinsic.
So far I've just added explicit casts as necessary, but we might want to address some of the mismatches directly
Differential Revision: https://reviews.llvm.org/D125164
While I think this is a performance improvement over the original, this actually fixes a correctness issue: For an appendable underlying stream, padToAlignment would fail if the additional padding would have caused the stream to grow since it was doing its own check on bounds. By deferring to the regular writeArray method this takes the same path as everything else, which does the correct bounds check in WritableBinaryStreamRef::checkOffsetForWrite (i.e. skips the extension check if BSF_Append is set). I had started to fix the existing bounds check in BinaryStreamWriter but deferred to this because it layered better and is more efficient/consistent.
It didn't look like this method was tested at all, so I added a unit test.
Differential Revision: https://reviews.llvm.org/D124746
When a PHINode has an incoming block from outside the region, it must be handled specially when assigning a global value number to each incoming value. A PHINode has multiple predecessors, and we must handle this case rather than only the single predecessor case.
Reviewer: paquette
Differential Revision: https://reviews.llvm.org/D124777
As suggested by @foad on D124839
If we're extracting a vector element that originally came from a scalar_to_vector, then avoid the bitcasting of a vector type and perform the shift masking on the (any-extended) scalar source directly, making use of the fact that the upper elements of a scalar_to_vector are all undef.
Differential Revision: https://reviews.llvm.org/D125173
Like other shifts, the type isn't required to match. We shouldn't
assume we can call ZExtPromotedInteger.
I tested the PromoteIntOp_FunnelShift locally by removing the promotion
of the shift amount from PromoteIntRes_FunnelShift. But with the final
version of this patch it is never executed on any tests.
Differential Revision: https://reviews.llvm.org/D125106
This is part of an ongoing effort toward making DAGCombine process the nodes in topological order.
This is able to discover a couple of new optimizations, but also causes a couple of regression. I nevertheless chose to submit this patch for review as to start the discussion with people working on the backend so we can find a good way forward.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D124743
As discussed on D124839, we're almost certainly only ever going to see this from IR directly - which now will create funnel shift intrinsics directly
I've also added a couple of rotl(rotr()) tests to check left/right rotation merging.
The goal of flushing to disk is to keep a reasonable bound on peak memory usage.
With a a default threshold of 512MB (and most BitstreamWriters having no backing
file at all), checking after every byte whether to flush seems excessive.
This change makes clangd's unittests run 5% faster (in opt), so it's not
actually free even in the case with no backing file. Likely there are more
important workloads where it makes some difference.
Differential Revision: https://reviews.llvm.org/D125145
The `LLVMTargetMachineEmitToFile` takes a `char* Filename` right now, but it doesn't modify it.
This is annoying to use in the case where you want to pass a const string, because you either have to remove the const, or copy it somewhere else and pass that. Either way, it's not very nice.
I added a const and clang formatted it. This shouldn't break any ABI in my opinion.
I'm sorry but I didn't know whom to put as reviewer for this, so I chose someone with a lot of commits from the .cpp file.
Reviewed By: deadalnix
Differential Revision: https://reviews.llvm.org/D124453
Given a load without a better order, this patch partially sorts the
elements to form clusters of adjacent elements in memory. These clusters
can potentially be loaded in fewer loads, meaning less overall shuffling
(for example loading v4i8 clusters of a v16i8 as a single f32 loads, as
opposed to multiple independent bytes loads and inserts).
Differential Revision: https://reviews.llvm.org/D122145
As shown in https://github.com/llvm/llvm-project/issues/55150 -
the existing fold may be wrong when converting to a signed value.
This is a quick fix to avoid the miscompile.
I added tests/comments for all of the signed/unsigned combinations
at either side of the boundary width, and tried to confirm with Alive2:
https://alive2.llvm.org/ce/z/3p9DSu
There are already some TODO items in the test file that suggest
possible refinements, so the regression with ui->FP->si is probably ok.
It seems unlikely that we'd see these kind of edge cases with
non-byte-width integer types in real code. The potential miscompile
went undetected for several years.
This and 747c6a0c73fixes#55150.
Differential Revision: https://reviews.llvm.org/D124692
David Green