tree a9086872e507ebffc924ae1bdea6f64eedefd2bc
parent ff82e4227c938236aa8302269e02a838ed6aa3fe
author Scott Todd <scotttodd@google.com> 1647294719 -0700
committer GitHub <noreply@github.com> 1647294719 -0700
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsBcBAABCAAQBQJiL7j/CRBK7hj4Ov3rIwAAFrUIABE10QB0OiYegOL7LJHc6xZY
 +xtquSej403A/Tg2pK7nrwlLjryvxoc9RYe0qNZwTmOlEysyduxmtXwF4jrvgcy+
 xDAJuZW0oMX9oJqFvqG2AFgWV4MwvlVA8aEnWIqtHD4g6+8gLFYHWzviVG2JpXqP
 iGWBRCRB1E2NdFaN0SCkU2RJ0bp3CGRpEhIFs6WCYDi9CNlAL5Sc5YmAIlO0f5+9
 BboovFXZxXjdp3sinanyTGJXzS4qK3WAO290cSUfjCiu+2AmcGUGCElDTl9wKRh8
 zprK039nx+INK+RIgy3CYO0NfPyQiOM7AgnpVlv+z4lZYOd/wBYir0cnJ2VjFRc=
 =Mhiy
 -----END PGP SIGNATURE-----
 

Update test/e2e/xla_ops/round.mlir test status. (#8524)

Progress on https://github.com/google/iree/issues/7618

https://github.com/google/iree/blob/c96372348654eea8548509d38df5c8e8b179aef3/iree/test/e2e/xla_ops/round.mlir#L1-L6

`mhlo.round_nearest_afz` lowers to `math.copysign`, which has this support in IREE today:

backend | result
-- | --
LLVM CPU | ✔️ success
LLVM WebAssembly | ❌  same error as on https://github.com/google/iree/issues/8514
CUDA | ✔️ success, `ConvertToNVVM` lowers to `llvm.intr.copysign`
SPIR-V | ❌ missing lowering in `ConvertToSPIRV`
VMVX | ❌ missing lowering in `IREE::VM::ConversionPass`

This test was disabled on all backends/drivers, so this PR enables it where it passes and updates the comments where it is marked as failing.