Re: [PATCH v1 04/20] s390x/tcg: Implement 32/128 bit for VECTOR FP ADD

[David Hildenbrand]

On 01.10.20 18:08, Richard Henderson wrote:
> On 9/30/20 9:55 AM, David Hildenbrand wrote:
>> +        case FPF_LONG:
>> +            fn = se ? gen_helper_gvec_vfa64s : gen_helper_gvec_vfa64;
>> +            break;
> 
> BTW, any reason not to pass SE as data, like you do later for SQ?  Or
> potentially the entire M field as is?

Having a separate implementation for "se" is desirable, because the
compiler can optimize-out the complete loop. If we simply pass the M
field to the helper, I'm not sure how likely it is that the compiler
will specialize (would have to double check).

(if we decide to remove all "s" helpers here, we'd better do it for all
 helpers)

> 
> Just wondering if it would help tidy up here...



-- 
Thanks,

David / dhildenb