[jdk21u-dev] RFR: 8345146: [PPC64] Make intrinsic conversions between bit representations of half precision values and floats

[Martin Doerr]

On Fri, 27 Dec 2024 18:35:12 GMT, Martin Doerr <mdoerr at openjdk.org> wrote:

> Almost clean backport of [JDK-8345146](https://bugs.openjdk.org/browse/JDK-8345146). Two hunks had to be integrated manually due to unrelated differences in the context. The test updates had to be integrated manually because the RISCV part is not backported. TestFloat16VectorConvChain.java is not in 21u (introduced by https://github.com/openjdk/jdk/commit/55fd1ed228ea3c42aaf92579e5dcb818fe14351d).
> 
> Original PR with performance numbers: https://github.com/openjdk/jdk/pull/22433

We'll ship it later.

-------------

PR Comment: https://git.openjdk.org/jdk21u-dev/pull/1295#issuecomment-2654694879