diff options
| author | Alon Zakai <azakai@google.com> | 2021-07-19 09:07:50 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2021-07-19 09:07:50 -0700 |
| commit | ec6ef26ec4f4fdfc974e7087896ac004ac7c130a (patch) | |
| tree | ef488fa345a18cb40e203960d0ca04d72af8b70a /test/passes/precompute_all-features.wast | |
| parent | 023bbb28f393032a8df6c743319cd835feddf5ba (diff) | |
| download | binaryen-ec6ef26ec4f4fdfc974e7087896ac004ac7c130a.tar.gz binaryen-ec6ef26ec4f4fdfc974e7087896ac004ac7c130a.tar.bz2 binaryen-ec6ef26ec4f4fdfc974e7087896ac004ac7c130a.zip | |
Revert "Partially fix Precompute on SIMD (#3983)" (#4002)
This reverts commit b68691e826a46d1b03b27c552b1f5b7f51f92665.
Instead of applying the workaround to avoid precomputing SIMD in more places,
which prevents things we could optimize before, we should probably focus on
making the workaround not needed - that is, implement full SIMD support in the
intepreter (the reason for that PR), and the other TODO in the comment here,
// Until engines implement v128.const and we have SIMD-aware optimizations
// that can break large v128.const instructions into smaller consts and
// splats, do not try to precompute v128 expressions.
Diffstat (limited to 'test/passes/precompute_all-features.wast')
| -rw-r--r-- | test/passes/precompute_all-features.wast | 7 |
1 files changed, 0 insertions, 7 deletions
diff --git a/test/passes/precompute_all-features.wast b/test/passes/precompute_all-features.wast index e306be495..1275524e0 100644 --- a/test/passes/precompute_all-features.wast +++ b/test/passes/precompute_all-features.wast @@ -329,13 +329,6 @@ (i32.const 0) ) ) - (func $no-simd-precompute_2 (result v128) - (i32x4.extadd_pairwise_i16x8_s - (i32x4.splat - (i32.const 0) - ) - ) - ) (func $no-simd-precompute-if (result v128) (return (i32x4.splat |