On Fri, 4 Mar 2022 18:33:53 +0000, Schmitt, Michael <[email protected]> 
wrote:

>Am I correct in assuming that branch relative (HLASM "jump") instructions are 
>more efficient than branches using register+displacement?

Yes. It´s easier for the Branch Prediction Unit to find the target address 
using relative offset.

>
>C. Kevin Shum's IBM Z System Processor Optimization Primer says:
>
>Use relative branches instead of non-relative (indirect branches) when possible
>
>but "indirect" is what throws me. The only indirect branch I'm aware of is 
>Branch Indirect on Condition (BIC).


"Indirect" because non-relative branch instructions depend on base register 
content at the time of execution.


Regards,

Philippe Leite
LAB Services - IBM

Reply via email to