Add more manual aarch64 load/store intrinsic tests#2086
Add more manual aarch64 load/store intrinsic tests#2086folkertdev merged 2 commits intorust-lang:mainfrom
Conversation
|
r? @folkertdev rustbot has assigned @folkertdev. Use Why was this reviewer chosen?The reviewer was selected based on:
|
adamgemmell
changed the title
adamgemmell
force-pushed
the
dev/adagem01/new-neon-tests
branch
3 times, most recently
from
aa38fd0 to
4aba691
Compare
| macro_rules! lane_wide_store_load_roundtrip_fp16 { | ||
| ($( $name:ident $args:tt);* $(;)?) => { | ||
| $( | ||
| #[simd_test(enable = "neon,fp16")] |
There was a problem hiding this comment.
Any idea why these f16 tests pass on miri but not the regular ones above? They're all implemented with llvm builtins and aren't shimmed in miri as far as I can see
There was a problem hiding this comment.
what works and what doesn't work?
There was a problem hiding this comment.
Ah, got there - the simd_test macro was silently filtering out these tests under miri because the fp16 feature wasn't enabled. I've added brought some target features from a recent cpu in the github actions job, which causes these to fail, and then ignored them again 😄
adamgemmell
force-pushed
the
dev/adagem01/new-neon-tests
branch
from
4aba691 to
076805d
Compare
adamgemmell
force-pushed
the
dev/adagem01/new-neon-tests
branch
from
076805d to
2d49c7b
Compare
3 participants
This was originally larger, but a fair number of the intrinsics I wrote tests for were already tested under different test names!