CL_23-10-25

Limit overclocked Big cluster frequency upto 2.5GHz during heavy multithreaded tasks to avoid thermal throttling.
I think this is safer approach rather than messing around with thermal.

Im using EMS sysbusy handler to detect heavy tasks,

basically there's 4 sysbusy state:

STATE_0 (upto 2.7Ghz):

• Low overall system utilization, active cores are likely running light tasks
• Likely UX/scrolling (boost), etc

STATE_1 (upto 2.7Ghz):

• High utilization on one core, low utilization on others
• Likely happens during first app launch (boost), browsing, camera, etc

STATE_2 (limited to 2.52Ghz):

• Moderate to high balanced utilization across several cores (~80% on each core)
• Likely happens during single/multi core benchmarks and maybe gaming, etc

STATE_3 (limited to 2.4Ghz):

• High utilization across most of all cores (~100% on each core)
• Likely happens during multi core benchmarks and maybe heavy editing/rendering

Undervolt tuning:

Little cluster: -2%

Big cluster:

• 533Mhz ~ 2.0Ghz = -2%
• 2.1Ghz ~ 2.5Ghz = -5%
• 2.6Ghz ~ 2.7Ghz = -2%