On Wed, 11 Nov 2026 12:49:58 GMT, Emanuel Peter <[email protected]> wrote:

>> Jatin Bhateja has updated the pull request incrementally with one additional 
>> commit since the last revision:
>> 
>>   Review comments resolution
>
> src/hotspot/share/opto/vectorIntrinsics.cpp line 2015:
> 
>> 2013:   }
>> 2014: 
>> 2015:   if (!is_supported_lane_type(vltype)) {
> 
> What exactly is the semantics of `is_supported_lane_type`? It seems the only 
> lane type that it does not support is `LT_FLOAT16`. But now we do actually 
> support that lane type? Or do we not actually "support" it, but only its 
> carrier `LT_SHORT`?
> 
> A comment at `is_supported_lane_type` would help me here.

Hi @eme64 , 
As the name suggest, is_supported_lane_type return true for lane type supported 
by inline expander, For `rearrange` the distinction between "supporting the 
lane type" and "supporting only its carrier" collapses, because `rearrange` is 
pure data movement. It selects lanes by index and never interprets the lane 
bits, so the carrier width is the whole story and the existing 16-bit permute 
path is already correct for Float16. That is why this intrinsic checks only 
`is_primitive_lane_type` and `is_supported_lane_type` is removed.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/33198#discussion_r4559984785

Reply via email to