mirror of
https://github.com/torvalds/linux.git
synced 2026-06-07 14:04:54 +02:00
The new sysctl sched_pelt_multiplier allows a user to set a clock multiplier x2 or x4 (x1 being the default). This clock multiplier artificially speed-up PELT ramp up/down similarly to a faster half-life. Indeed, if we write PELT as a first order filter: y(t) = G * (1 - exp(t/tau)) Then we can see that multiplying the time by a constant X, is the same as dividing the time constant tau by X. y(t) = G * (1 - exp((t*X)/tau)) y(t) = G * (1 - exp(t/(tau/X))) Tau being half-life*ln(2), multiplying the PELT time is the same as dividing the half-life: - x1: 32ms half-life - x2: 16ms half-life - x4: 8ms half-life Internally, a new clock is created: rq->clock_task_mult. It sits in the clock hierarchy between rq->clock_task and rq->clock_pelt. Bug: 177593580 Bug: 237219700 Change-Id: I67e6ca7994bebea22bf75732ee11d2b10e0d6b7e Suggested-by: Morten Rasmussen <morten.rasmussen@arm.com> Signed-off-by: Vincent Donnefort <vincent.donnefort@arm.com> Signed-off-by: JianMin Liu <jian-min.liu@mediatek.com> |
||
|---|---|---|
| .. | ||
| autogroup.c | ||
| autogroup.h | ||
| clock.c | ||
| completion.c | ||
| core.c | ||
| cpuacct.c | ||
| cpudeadline.c | ||
| cpudeadline.h | ||
| cpufreq_schedutil.c | ||
| cpufreq.c | ||
| cpupri.c | ||
| cpupri.h | ||
| cputime.c | ||
| deadline.c | ||
| debug.c | ||
| fair.c | ||
| features.h | ||
| idle.c | ||
| isolation.c | ||
| loadavg.c | ||
| Makefile | ||
| membarrier.c | ||
| pelt.c | ||
| pelt.h | ||
| psi.c | ||
| rt.c | ||
| sched-pelt.h | ||
| sched.h | ||
| smp.h | ||
| stats.c | ||
| stats.h | ||
| stop_task.c | ||
| swait.c | ||
| topology.c | ||
| wait_bit.c | ||
| wait.c | ||