Wilco Dijkstra <wilco.dijks...@arm.com> writes: > Hi Kyrill & Richard, > >> I was leaving this to others in case it was obvious to them. On the >> basis that silence suggests it wasn't, :-) could you go into more details? >> Is it expected on first principles that jump alignment doesn't matter >> for Neoverse N1, or is this purely based on experimentation? If it's > > Jump alignment is set to 4 on almost all cores because higher values have > a major codesize cost and yet give no performance gains. > > I suspect any core that set it higher has done so by accident rather than > having benchmarked the cost/benefit. > >> expected, are we sure that the other "32:16" entries are still worthwhile? >> When you say it doesn't make a difference in performance, does that mean >> that no individual test's performance changed significantly, or just that >> the aggregate score didn't? Did you experiment with anything inbetween >> the current 32:16 and 4, such as 32:8 or even 32:4? > > I mean there is no difference above the noise floor for any test you throw at > it.
OK, great. In that case, let's go with the patch as posted. Thanks, Richard > I tried other alignments including 32:16, 32:12, 32:8 but all have a > significant > cost and zero benefit. > > Cheers, > Wilco
