[lworld+fp16] RFR: 8341414: Add support for FP16 conversion routines

[Jatin Bhateja]

On Thu, 31 Oct 2024 13:50:34 GMT, Bhavana Kilambi <bkilambi at openjdk.org> wrote:

>> This patch adds intrinsic support for FP16 conversion routines to int/long/double and also the aarch64 backend support. This patch implements both scalar and vector versions for these conversions.
>> 
>> Performance numbers on aarch64 machine with SVE support :
>> 
>> 
>> Benchmark                         (vectorDim)   Gain
>> Float16OpsBenchmark.fp16ToDouble  1024          18.23
>> Float16OpsBenchmark.fp16ToInt     1024          1.93
>> Float16OpsBenchmark.fp16ToLong    1024          3.95
>> 
>> 
>> The Gain column is the ratio between thrpt of this patch and the thrpt with the intrinsics disabled (which generates FP32 arithmetic).
>
> Hi @jatin-bhateja , I have uploaded a patch addressing your comments. Please review.

Hi @Bhavana-Kilambi , should also consider float16 to integral conversion, i.e. to int/short/long.
ISA supports direct instruction corresponding to these APIs.

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

PR Comment: https://git.openjdk.org/valhalla/pull/1283#issuecomment-2459853719