Any possibility of PL improving (usage of) CPU caches?

[Coldblackice]

Is there be any functionality that could be implemented into Process Lasso toward optimizing usage/prioritizing of CPU caches (L1/L2/L3) in favor of a program(s)?

I realize these caches are deeply low-level, tucked away inside the CPU and unbeknownst to software, however these caches have been found recently to bear massive impacts on a number of games' performance, particularly games built on Unity, and performance boosts greater than even GPU + CPU upgrades combined could produce (some cases). It's a large reason why AMD's X3D CPU line has been selling so well.


I realize it's probably not possible beyond overclocking/upgrading, but with all the magic you've unfurled over the years through Process Lasso, if there's anyone who could figure out a way, it'd be you! Any chance you might know of some roundabout manner of juggling queues/threads/interrupts/allocations/IO/etc. to potentially bear positive downstream effects in the form of effectively boosted CPU cache performance, for a particular program or set of programs?

[Jeremy Collake]

Presently, we don't have any magic for that . However, we'll start thinking on it, and keep an eye out for opportunities. Your notion of some roundabout way to affect CPU cache use is not impossible, and could be of utility when properly targeted.

[Coldblackice]

Presently, we don't have any magic for that . However, we'll start thinking on it, and keep an eye out for opportunities. Your notion of some roundabout way to affect CPU cache use is not impossible, and could be of utility when properly targeted.

Awesome! Thanks for considering.