On Wed, Mar 14, 2018 at 06:12:46PM +0000, Ionela Voinescu wrote:
[...]
Although broader scope items would make some narrow scope items unnecessary, they have value on their own if the boarder scope items cannot be proven in a reasonable amount of time, as they will provide fixes and support earlier than it takes to make a full re-factor of code.
- 1 will provide valuable fixes now.
- 2 will simplify find_best_target decision logic to facilitate and simplify 3.
- 3 will add support for tri-gear platforms for which the current order of CPUs will be incorrect in some cases.
Here means tri-clusters case? If so, fbt() is hard to handle energy model overlap issue within them, this issue exists in two clusters case, and will get worse for tri-clusters cases.
It means having three groups of CPUs (preferably in their own DVFS domains) each with different capacity ranges. They might or might not all share a last level cache, that's why I was avoiding the term tri-cluster.
Thanks for clarification.
When you say energy model overlap, I suppose you mean power-performance curves overlap, right?
Yes.
[...]
Thanks, Leo Yan