[PATCH 1/1] Implement vstore_half{,n}

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
---

vstore_half(float) version passes newly posted piglit on Kaveri.
vstore_half(double) fails, but the generated llvm IR appears good

For vstore_half(double) do all vector sizes fail?

-Tom

>
> Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
> ---
>
> vstore_half(float) version passes newly posted piglit on Kaveri.
> vstore_half(double) fails, but the generated llvm IR appears good
>

For vstore_half(double) do all vector sizes fail?

yes, conversion from double->half fails. I have a patch for that[0],
it's basically trunc-store-f64-to-f16.ll lit test.

Jan

https://github.com/jvesely/llvm/commit/bcf2cff1ef584a0543a9bd486dfd1411
365157c0

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>

ping. Even if the llvm patch needs more work, this one should be safe.

Jan

> Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>

ping. Even if the llvm patch needs more work, this one should be safe.

LGTM.