[vectorIntrinsics] RFR: 8262492: Add cast nodes between float types implementation for Arm SVE

[Wang Huang]

On Mon, 1 Mar 2021 06:43:44 GMT, Ningsheng Jian <njian at openjdk.org> wrote:

>> Add cast nodes like castf2d castd2f
>> 
>> * add `sve_fcvt`
>> * reuse macro `VECTOR_CAST_I2F_NARROW1` `VECTOR_CAST_I2F_EXTEND1` (Should we change these macro names?)
>> 
>> other issues :
>> * precision loss
>
> src/hotspot/cpu/aarch64/assembler_aarch64.hpp line 3438:
> 
>> 3436:   }
>> 3437: 
>> 3438: // SVE floating-point convert precision(predicated)
> 
> space between precision and '(' ?

OK. I will change that.

> src/hotspot/cpu/aarch64/assembler_aarch64.hpp line 3450:
> 
>> 3448:   }
>> 3449: 
>> 3450:   INSN(sve_fcvt);
> 
> Since there's only one use of the INSN macro, I think it's no need to define a macro here.

OK. I will clear that.

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

PR: https://git.openjdk.java.net/panama-vector/pull/41