On Tue, 21 Apr 2026 06:36:25 GMT, Xiaohong Gong <[email protected]> wrote:

>> Jatin Bhateja has updated the pull request incrementally with one additional 
>> commit since the last revision:
>> 
>>   Review comments resolutions
>
> src/hotspot/share/opto/vectorIntrinsics.cpp line 321:
> 
>> 319:   return laneType == VectorSupport::LaneType::LT_FLOAT16;
>> 320: }
>> 321: 
> 
> I just noticed that some APIs' intrinsification will check this and back to 
> default implementation if it is the FP16. Is there a plan to support it in 
> future for those APIs? Will the supporting be a kind of backend-level choice? 
> Do you think it's better to merge this type check inside the 
> `arch_supports_vector` or even `match_rule_supported_vector` ?

Hi @XiaohongGong ,
For now only few inline expanders handle float16 type, arch_supports_vector / 
match_rule_supported_vector are dependent on IR opcodes,   idea here is to 
remove unsupported lane type checks one by one once an inline expander, mid-end 
IR and backend fully handles float16 type. Plan is to work on this enablement 
once this base pull request gets integrated.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/28002#discussion_r3116058758

Reply via email to