[RFC/RFT][PATCH v3] cpufreq: intel_pstate: Proportional algorithm for Atom
From: Rafael J. Wysocki
Date: Thu Oct 06 2016 - 08:01:26 EST
From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
The PID algorithm used by the intel_pstate driver tends to drive
performance to the minimum for workloads with utilization below the
setpoint, which is undesirable, so replace it with a modified
"proportional" algorithm on Atom.
The new algorithm will set the new P-state to be 1.25 times the
available maximum times the (frequency-invariant) utilization during
the previous sampling period except when the target P-state computed
this way is lower than the average P-state during the previous
sampling period. In the latter case, it will increase the target by
50% of the difference between it and the average P-state to prevent
performance from dropping down too fast in some cases.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
v1 -> v2:
It is better to compare the average P-state to the target (than to
compare the average perf ratio to the utilization), because that takes
turbo into account more accurately.
Plus if the target is below the min, it is better to compare the min
to the average instead of comparing the target to it.
v2 -> v3:
Fix a typo in the ternary operator reported by Doug.
drivers/cpufreq/intel_pstate.c | 22 +++++++++++++++++++++-
1 file changed, 21 insertions(+), 1 deletion(-)
@@ -1232,6 +1232,7 @@ static inline int32_t get_target_pstate_
struct sample *sample = &cpu->sample;
int32_t busy_frac, boost;
+ int target, avg_pstate;
busy_frac = div_fp(sample->mperf, sample->tsc);
@@ -1242,7 +1243,26 @@ static inline int32_t get_target_pstate_
busy_frac = boost;
sample->busy_scaled = busy_frac * 100;
- return get_avg_pstate(cpu) - pid_calc(&cpu->pid, sample->busy_scaled);
+ target = limits->no_turbo || limits->turbo_disabled ?
+ cpu->pstate.max_pstate : cpu->pstate.turbo_pstate;
+ target += target >> 2;
+ target = mul_fp(target, busy_frac);
+ if (target < cpu->pstate.min_pstate)
+ target = cpu->pstate.min_pstate;
+ * If the average P-state during the previous cycle was higher than the
+ * current target, add 50% of the difference to the target to reduce
+ * possible performance oscillations and offset possible performance
+ * loss related to moving the workload from one CPU to another within
+ * a package/module.
+ avg_pstate = get_avg_pstate(cpu);
+ if (avg_pstate > target)
+ target += (avg_pstate - target) >> 1;
+ return target;
static inline int32_t get_target_pstate_use_performance(struct cpudata *cpu)