The five binary operators on Int32x4 (+, -, |, &, ^) were never added
to the recognized-method list as graph intrinsics, so calls to them
were left as runtime calls through external-name bodies. In JIT the
call specializer picked kInt32x4Cid from IC feedback and emitted a
native SimdOpInstr, but AOT has no IC feedback and therefore fell back
to boxed calls, making Int32x4List inner loops 10-70x slower than both
the JIT version and a hand-written scalar equivalent.
This CL wires the same specialization paths that already exist for
Float32x4.+,-,*,/:
- recognize the five operators as graph intrinsics and mark them
with `@pragma("vm:recognized", "graph-intrinsic")` plus an
exact-result-type pragma;
- add Build_Int32x4{Add,Sub,BitAnd,BitOr,BitXor} helpers that
delegate to the existing BuildSimdOp;
- extend SimdOpInstr::KindForOperator and CreateFromCall;
- extend CallSpecializer::InlineSimdOp and TryInlineRecognizedMethod
so the non-speculative null-check path used for Float32x4 operators
in AOT also applies here.
Measured on macOS arm64 (M-series), `dart compile exe`:
Issue 63217 orSimd : 12.58 -> 0.32 us/iter (39x)
Issue 63217 andNotSimd : 23.51 -> 0.34 us/iter (69x)
Issue 53662 mandelbrot : 4038.5 -> 55.5 ms (72x)
A new benchmark benchmarks/SimdInt32x4 exercises all five operators
with a scalar and a SIMD variant so the specialization stays covered
by the benchmark bots; it is registered in Omnibus and OmnibusDeferred.
Existing tests/lib/typed_data/simd_*_test.dart still pass in JIT and
AOT.
TEST=tests/lib/typed_data/int32x4_arithmetic_test; benchmarks/SimdInt32x4
Bug: https://github.com/dart-lang/sdk/issues/53662
Bug: https://github.com/dart-lang/sdk/issues/63217
Change-Id: I9b76ab4fff228ff1a5e3d3c86f4bfc059e66a49a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/497000
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Now that I'm able to open the entire SDK in VSC, I'm fixing some
of the analysis issues in various files (carefully) without changing
their meaning.
In this case, I removed unnecessary imports from benchmarks.
In regexp_benchmark I ignored one warning which likely would
have changed the behavior of the code.
BUG=https://github.com/dart-lang/sdk/issues/52419
Change-Id: I9a195a4e45121313bd9f065f2579a165c3fec05b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/303901
Auto-Submit: Eric Seidel <eric@shorebird.dev>
Reviewed-by: William Hesse <whesse@google.com>
Commit-Queue: William Hesse <whesse@google.com>
A benchmark that is a collection of other benchmarks.
This benchmark has two purposes:
- As a medium sized compilation benchmark
- As a tool for running benchmarks as part of a larger program
This benchmark is not intended to be run for the purpose of generating
benchmark timing results on Golem.
It should be extended over time to include most public benchmarks that
are meaningful on all Dart platforms.
Change-Id: I17f034cfd68d01ff5b4842b89b216fcea0d5bd3c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/170060
Commit-Queue: Stephen Adams <sra@google.com>
Reviewed-by: William Hesse <whesse@google.com>
Modestas Valauskas