Skip to contentSkip to conversation

Scan for an idle sibling in a single pass

PATCH Applied Mel Gorman 3 participants 8 messages Jan 27, 2021 – Feb 17, 2021 Scheduler

lore
Share & tools
lore

Mel Gorman

author
#0 lore raw

v5 cover Scan for an idle sibling in a single pass

Changelog since v4
o Avoid use of intermediate variable during select_idle_cpu

Changelog since v3
o Drop scanning based on cores, SMT4 results showed problems

Changelog since v2
o Remove unnecessary parameters
o Update nr during scan only when scanning for cpus

Changlog since v1
o Move extern declaration to header for coding style
o Remove unnecessary parameter from __select_idle_cpu

This series of 4 patches reposts three patches from Peter entitled "select_idle_sibling() wreckage". It only scans the runqueues in a single pass when searching for an idle sibling.

Three patches from Peter were dropped. The first patch altered how scan depth was calculated. Scan depth deletion is a random number generator with two major limitations. The avg_idle time is based on the time between a CPU going idle and being woken up clamped approximately by 2*sysctl_sched_migration_cost. This is difficult to compare in a sensible fashion to avg_scan_cost. The second issue is that only the avg_scan_cost of scan failures is recorded and it does not decay. This requires deeper surgery that would justify a patch on its own although Peter notes that
https://lkml.kernel.org/r/[email protected] is potentially useful for an alternative avg_idle metric.

The second patch dropped scanned based on cores instead of CPUs as it rationalised the difference between core scanning and CPU scanning. Unfortunately, Vincent reported problems with SMT4 so it's dropped for now until depth searching can be fixed.

The third patch dropped converted the idle core scan throttling mechanism to SIS_PROP. While this would unify the throttling of core and CPU scanning, it was not free of regressions and has_idle_cores is a fairly effective throttling mechanism with the caveat that it can have a lot of false positives for workloads like hackbench.

Peter's series tried to solve three problems at once, this subset addresses one problem.

 kernel/sched/fair.c     | 151 +++++++++++++++++++---------------------
 kernel/sched/features.h |   1 -
 2 files changed, 70 insertions(+), 82 deletions(-)
Signature · 14 lines
2.26.2

Mel Gorman (4):
  sched/fair: Remove SIS_AVG_CPU
  sched/fair: Move avg_scan_cost calculations under SIS_PROP
  sched/fair: Remove select_idle_smt()
  sched/fair: Merge select_idle_core/cpu()

 kernel/sched/fair.c     | 151 +++++++++++++++++++---------------------
 kernel/sched/features.h |   1 -
 2 files changed, 70 insertions(+), 82 deletions(-)

-- 
2.26.2
author
#1 lore raw

PATCH v5 1/4 sched/fair: Remove SIS_AVG_CPU

SIS_AVG_CPU was introduced as a means of avoiding a search when the average search cost indicated that the search would likely fail. It was a blunt instrument and disabled by commit 4c77b18cf8b7 ("sched/fair: Make select_idle_cpu() more aggressive") and later replaced with a proportional search depth by commit 1ad3aaf3fcd2 ("sched/core: Implement new approach to scale select_idle_cpu()").

While there are corner cases where SIS_AVG_CPU is better, it has now been disabled for almost three years. As the intent of SIS_PROP is to reduce the time complexity of select_idle_cpu(), lets drop SIS_AVG_CPU and focus on SIS_PROP as a throttling mechanism.

---
 kernel/sched/fair.c     | 20 +++++++++-----------
 kernel/sched/features.h |  1 -
 2 files changed, 9 insertions(+), 12 deletions(-)
kernel/sched/fair.c +9 -11
index 04a3ce20da67..9f5682aeda2e 100644
@@ -6145,7 +6145,6 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 {
 	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
 	struct sched_domain *this_sd;
-	u64 avg_cost, avg_idle;
 	u64 time;
 	int this = smp_processor_id();
 	int cpu, nr = INT_MAX;
@@ -6154,18 +6153,17 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 	if (!this_sd)
 		return -1;
 
-	/*
-	 * Due to large variance we need a large fuzz factor; hackbench in
-	 * particularly is sensitive here.
-	 */
-	avg_idle = this_rq()->avg_idle / 512;
-	avg_cost = this_sd->avg_scan_cost + 1;
+	if (sched_feat(SIS_PROP)) {
+		u64 avg_cost, avg_idle, span_avg;
 
-	if (sched_feat(SIS_AVG_CPU) && avg_idle < avg_cost)
-		return -1;
+		/*
+		 * Due to large variance we need a large fuzz factor;
+		 * hackbench in particularly is sensitive here.
+		 */
+		avg_idle = this_rq()->avg_idle / 512;
+		avg_cost = this_sd->avg_scan_cost + 1;
 
-	if (sched_feat(SIS_PROP)) {
-		u64 span_avg = sd->span_weight * avg_idle;
+		span_avg = sd->span_weight * avg_idle;
 		if (span_avg > 4*avg_cost)
 			nr = div_u64(span_avg, avg_cost);
 		else
kernel/sched/features.h +0 -1
index 68d369cba9e4..e875eabb6600 100644
@@ -54,7 +54,6 @@ SCHED_FEAT(TTWU_QUEUE, true)
 /*
  * When doing wakeups, attempt to limit superfluous scans of the LLC domain.
  */
-SCHED_FEAT(SIS_AVG_CPU, false)
 SCHED_FEAT(SIS_PROP, true)
 
 /*
Signature · 1 line
2.26.2
author
#2 lore raw

PATCH v5 2/4 sched/fair: Move avg_scan_cost calculations under SIS_PROP

In reply to Mel Gorman

As noted by Vincent Guittot, avg_scan_costs are calculated for SIS_PROP even if SIS_PROP is disabled. Move the time calculations under a SIS_PROP check and while we are at it, exclude the cost of initialising the CPU mask from the average scan cost.

---
 kernel/sched/fair.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)
kernel/sched/fair.c +8 -6
index 9f5682aeda2e..c8d8e185cf3b 100644
@@ -6153,6 +6153,8 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 	if (!this_sd)
 		return -1;
 
+	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
+
 	if (sched_feat(SIS_PROP)) {
 		u64 avg_cost, avg_idle, span_avg;
 
@@ -6168,11 +6170,9 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 			nr = div_u64(span_avg, avg_cost);
 		else
 			nr = 4;
-	}
-
-	time = cpu_clock(this);
 
-	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
+		time = cpu_clock(this);
+	}
 
 	for_each_cpu_wrap(cpu, cpus, target) {
 		if (!--nr)
@@ -6181,8 +6181,10 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 			break;
 	}
 
-	time = cpu_clock(this) - time;
-	update_avg(&this_sd->avg_scan_cost, time);
+	if (sched_feat(SIS_PROP)) {
+		time = cpu_clock(this) - time;
+		update_avg(&this_sd->avg_scan_cost, time);
+	}
 
 	return cpu;
 }
Signature · 1 line
2.26.2
author
#3 lore raw

PATCH v5 3/4 sched/fair: Remove select_idle_smt()

In reply to Mel Gorman

From: Peter Zijlstra (Intel) <[email protected]>

In order to make the next patch more readable, and to quantify the actual effectiveness of this pass, start by removing it.

---
 kernel/sched/fair.c | 30 ------------------------------
 1 file changed, 30 deletions(-)
kernel/sched/fair.c +0 -30
index c8d8e185cf3b..fe587350ea14 100644
@@ -6101,27 +6101,6 @@ static int select_idle_core(struct task_struct *p, struct sched_domain *sd, int
 	return -1;
 }
 
-/*
- * Scan the local SMT mask for idle CPUs.
- */
-static int select_idle_smt(struct task_struct *p, struct sched_domain *sd, int target)
-{
-	int cpu;
-
-	if (!static_branch_likely(&sched_smt_present))
-		return -1;
-
-	for_each_cpu(cpu, cpu_smt_mask(target)) {
-		if (!cpumask_test_cpu(cpu, p->cpus_ptr) ||
-		    !cpumask_test_cpu(cpu, sched_domain_span(sd)))
-			continue;
-		if (available_idle_cpu(cpu) || sched_idle_cpu(cpu))
-			return cpu;
-	}
-
-	return -1;
-}
-
 #else /* CONFIG_SCHED_SMT */
 
 static inline int select_idle_core(struct task_struct *p, struct sched_domain *sd, int target)
@@ -6129,11 +6108,6 @@ static inline int select_idle_core(struct task_struct *p, struct sched_domain *s
 	return -1;
 }
 
-static inline int select_idle_smt(struct task_struct *p, struct sched_domain *sd, int target)
-{
-	return -1;
-}
-
 #endif /* CONFIG_SCHED_SMT */
 
 /*
@@ -6323,10 +6297,6 @@ static int select_idle_sibling(struct task_struct *p, int prev, int target)
 	if ((unsigned)i < nr_cpumask_bits)
 		return i;
 
-	i = select_idle_smt(p, sd, target);
-	if ((unsigned)i < nr_cpumask_bits)
-		return i;
-
 	return target;
 }
 
Signature · 1 line
2.26.2
author
#4 lore raw

PATCH v5 4/4 sched/fair: Merge select_idle_core/cpu()

In reply to Mel Gorman

From: Peter Zijlstra (Intel) <[email protected]>

Both select_idle_core() and select_idle_cpu() do a loop over the same cpumask. Observe that by clearing the already visited CPUs, we can fold the iteration and iterate a core at a time.

All we need to do is remember any non-idle CPU we encountered while scanning for an idle core. This way we'll only iterate every CPU once.

---
 kernel/sched/fair.c | 99 +++++++++++++++++++++++++++------------------
 1 file changed, 59 insertions(+), 40 deletions(-)
kernel/sched/fair.c +59 -40
index fe587350ea14..01591e24a95f 100644
@@ -6006,6 +6006,14 @@ static inline int find_idlest_cpu(struct sched_domain *sd, struct task_struct *p
 	return new_cpu;
 }
 
+static inline int __select_idle_cpu(int cpu)
+{
+	if (available_idle_cpu(cpu) || sched_idle_cpu(cpu))
+		return cpu;
+
+	return -1;
+}
+
 #ifdef CONFIG_SCHED_SMT
 DEFINE_STATIC_KEY_FALSE(sched_smt_present);
 EXPORT_SYMBOL_GPL(sched_smt_present);
@@ -6064,48 +6072,51 @@ void __update_idle_core(struct rq *rq)
  * there are no idle cores left in the system; tracked through
  * sd_llc->shared->has_idle_cores and enabled through update_idle_core() above.
  */
-static int select_idle_core(struct task_struct *p, struct sched_domain *sd, int target)
+static int select_idle_core(struct task_struct *p, int core, struct cpumask *cpus, int *idle_cpu)
 {
-	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
-	int core, cpu;
+	bool idle = true;
+	int cpu;
 
 	if (!static_branch_likely(&sched_smt_present))
-		return -1;
-
-	if (!test_idle_cores(target, false))
-		return -1;
-
-	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
+		return __select_idle_cpu(core);
 
-	for_each_cpu_wrap(core, cpus, target) {
-		bool idle = true;
-
-		for_each_cpu(cpu, cpu_smt_mask(core)) {
-			if (!available_idle_cpu(cpu)) {
-				idle = false;
-				break;
+	for_each_cpu(cpu, cpu_smt_mask(core)) {
+		if (!available_idle_cpu(cpu)) {
+			idle = false;
+			if (*idle_cpu == -1) {
+				if (sched_idle_cpu(cpu) && cpumask_test_cpu(cpu, p->cpus_ptr)) {
+					*idle_cpu = cpu;
+					break;
+				}
+				continue;
 			}
+			break;
 		}
-
-		if (idle)
-			return core;
-
-		cpumask_andnot(cpus, cpus, cpu_smt_mask(core));
+		if (*idle_cpu == -1 && cpumask_test_cpu(cpu, p->cpus_ptr))
+			*idle_cpu = cpu;
 	}
 
-	/*
-	 * Failed to find an idle core; stop looking for one.
-	 */
-	set_idle_cores(target, 0);
+	if (idle)
+		return core;
 
+	cpumask_andnot(cpus, cpus, cpu_smt_mask(core));
 	return -1;
 }
 
 #else /* CONFIG_SCHED_SMT */
 
-static inline int select_idle_core(struct task_struct *p, struct sched_domain *sd, int target)
+static inline void set_idle_cores(int cpu, int val)
 {
-	return -1;
+}
+
+static inline bool test_idle_cores(int cpu, bool def)
+{
+	return def;
+}
+
+static inline int select_idle_core(struct task_struct *p, int core, struct cpumask *cpus, int *idle_cpu)
+{
+	return __select_idle_cpu(core);
 }
 
 #endif /* CONFIG_SCHED_SMT */
@@ -6118,10 +6129,11 @@ static inline int select_idle_core(struct task_struct *p, struct sched_domain *s
 static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int target)
 {
 	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
+	int i, cpu, idle_cpu = -1, nr = INT_MAX;
+	bool smt = test_idle_cores(target, false);
+	int this = smp_processor_id();
 	struct sched_domain *this_sd;
 	u64 time;
-	int this = smp_processor_id();
-	int cpu, nr = INT_MAX;
 
 	this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
 	if (!this_sd)
@@ -6129,7 +6141,7 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 
 	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
 
-	if (sched_feat(SIS_PROP)) {
+	if (sched_feat(SIS_PROP) && !smt) {
 		u64 avg_cost, avg_idle, span_avg;
 
 		/*
@@ -6149,18 +6161,29 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 	}
 
 	for_each_cpu_wrap(cpu, cpus, target) {
-		if (!--nr)
-			return -1;
-		if (available_idle_cpu(cpu) || sched_idle_cpu(cpu))
-			break;
+		if (smt) {
+			i = select_idle_core(p, cpu, cpus, &idle_cpu);
+			if ((unsigned int)i < nr_cpumask_bits)
+				return i;
+
+		} else {
+			if (!--nr)
+				return -1;
+			idle_cpu = __select_idle_cpu(cpu);
+			if ((unsigned int)idle_cpu < nr_cpumask_bits)
+				break;
+		}
 	}
 
-	if (sched_feat(SIS_PROP)) {
+	if (smt)
+		set_idle_cores(this, false);
+
+	if (sched_feat(SIS_PROP) && !smt) {
 		time = cpu_clock(this) - time;
 		update_avg(&this_sd->avg_scan_cost, time);
 	}
 
-	return cpu;
+	return idle_cpu;
 }
 
 /*
@@ -6289,10 +6312,6 @@ static int select_idle_sibling(struct task_struct *p, int prev, int target)
 	if (!sd)
 		return target;
 
-	i = select_idle_core(p, sd, target);
-	if ((unsigned)i < nr_cpumask_bits)
-		return i;
-
 	i = select_idle_cpu(p, sd, target);
 	if ((unsigned)i < nr_cpumask_bits)
 		return i;
Signature · 1 line
2.26.2
merge bot Applied
#5 lore raw

[tip: sched/core] sched/fair: Merge select_idle_core/cpu()

The following commit has been merged into the sched/core branch of tip:

The following commit has been merged into the sched/core branch of tip:

Commit-ID:     9fe1f127b913318c631d0041ecf71486e38c2c2d
Gitweb:        https://git.kernel.org/tip/9fe1f127b913318c631d0041ecf71486e38c2c2d
Author:        Mel Gorman <[email protected]>
AuthorDate:    Wed, 27 Jan 2021 13:52:03 
Committer:     Ingo Molnar <[email protected]>
CommitterDate: Wed, 17 Feb 2021 14:07:25 +01:00

sched/fair: Merge select_idle_core/cpu()

Both select_idle_core() and select_idle_cpu() do a loop over the same cpumask. Observe that by clearing the already visited CPUs, we can fold the iteration and iterate a core at a time.

All we need to do is remember any non-idle CPU we encountered while scanning for an idle core. This way we'll only iterate every CPU once.

---
 kernel/sched/fair.c |  99 +++++++++++++++++++++++++------------------
 1 file changed, 59 insertions(+), 40 deletions(-)
kernel/sched/fair.c +59 -40
index 6a0fc8a..c73d588 100644
@@ -6019,6 +6019,14 @@ static inline int find_idlest_cpu(struct sched_domain *sd, struct task_struct *p
 	return new_cpu;
 }
 
+static inline int __select_idle_cpu(int cpu)
+{
+	if (available_idle_cpu(cpu) || sched_idle_cpu(cpu))
+		return cpu;
+
+	return -1;
+}
+
 #ifdef CONFIG_SCHED_SMT
 DEFINE_STATIC_KEY_FALSE(sched_smt_present);
 EXPORT_SYMBOL_GPL(sched_smt_present);
@@ -6077,48 +6085,51 @@ unlock:
  * there are no idle cores left in the system; tracked through
  * sd_llc->shared->has_idle_cores and enabled through update_idle_core() above.
  */
-static int select_idle_core(struct task_struct *p, struct sched_domain *sd, int target)
+static int select_idle_core(struct task_struct *p, int core, struct cpumask *cpus, int *idle_cpu)
 {
-	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
-	int core, cpu;
+	bool idle = true;
+	int cpu;
 
 	if (!static_branch_likely(&sched_smt_present))
-		return -1;
-
-	if (!test_idle_cores(target, false))
-		return -1;
-
-	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
+		return __select_idle_cpu(core);
 
-	for_each_cpu_wrap(core, cpus, target) {
-		bool idle = true;
-
-		for_each_cpu(cpu, cpu_smt_mask(core)) {
-			if (!available_idle_cpu(cpu)) {
-				idle = false;
-				break;
+	for_each_cpu(cpu, cpu_smt_mask(core)) {
+		if (!available_idle_cpu(cpu)) {
+			idle = false;
+			if (*idle_cpu == -1) {
+				if (sched_idle_cpu(cpu) && cpumask_test_cpu(cpu, p->cpus_ptr)) {
+					*idle_cpu = cpu;
+					break;
+				}
+				continue;
 			}
+			break;
 		}
-
-		if (idle)
-			return core;
-
-		cpumask_andnot(cpus, cpus, cpu_smt_mask(core));
+		if (*idle_cpu == -1 && cpumask_test_cpu(cpu, p->cpus_ptr))
+			*idle_cpu = cpu;
 	}
 
-	/*
-	 * Failed to find an idle core; stop looking for one.
-	 */
-	set_idle_cores(target, 0);
+	if (idle)
+		return core;
 
+	cpumask_andnot(cpus, cpus, cpu_smt_mask(core));
 	return -1;
 }
 
 #else /* CONFIG_SCHED_SMT */
 
-static inline int select_idle_core(struct task_struct *p, struct sched_domain *sd, int target)
+static inline void set_idle_cores(int cpu, int val)
 {
-	return -1;
+}
+
+static inline bool test_idle_cores(int cpu, bool def)
+{
+	return def;
+}
+
+static inline int select_idle_core(struct task_struct *p, int core, struct cpumask *cpus, int *idle_cpu)
+{
+	return __select_idle_cpu(core);
 }
 
 #endif /* CONFIG_SCHED_SMT */
@@ -6131,10 +6142,11 @@ static inline int select_idle_core(struct task_struct *p, struct sched_domain *s
 static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int target)
 {
 	struct cpumask *cpus = this_cpu_cpumask_var_ptr(select_idle_mask);
+	int i, cpu, idle_cpu = -1, nr = INT_MAX;
+	bool smt = test_idle_cores(target, false);
+	int this = smp_processor_id();
 	struct sched_domain *this_sd;
 	u64 time;
-	int this = smp_processor_id();
-	int cpu, nr = INT_MAX;
 
 	this_sd = rcu_dereference(*this_cpu_ptr(&sd_llc));
 	if (!this_sd)
@@ -6142,7 +6154,7 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 
 	cpumask_and(cpus, sched_domain_span(sd), p->cpus_ptr);
 
-	if (sched_feat(SIS_PROP)) {
+	if (sched_feat(SIS_PROP) && !smt) {
 		u64 avg_cost, avg_idle, span_avg;
 
 		/*
@@ -6162,18 +6174,29 @@ static int select_idle_cpu(struct task_struct *p, struct sched_domain *sd, int t
 	}
 
 	for_each_cpu_wrap(cpu, cpus, target) {
-		if (!--nr)
-			return -1;
-		if (available_idle_cpu(cpu) || sched_idle_cpu(cpu))
-			break;
+		if (smt) {
+			i = select_idle_core(p, cpu, cpus, &idle_cpu);
+			if ((unsigned int)i < nr_cpumask_bits)
+				return i;
+
+		} else {
+			if (!--nr)
+				return -1;
+			idle_cpu = __select_idle_cpu(cpu);
+			if ((unsigned int)idle_cpu < nr_cpumask_bits)
+				break;
+		}
 	}
 
-	if (sched_feat(SIS_PROP)) {
+	if (smt)
+		set_idle_cores(this, false);
+
+	if (sched_feat(SIS_PROP) && !smt) {
 		time = cpu_clock(this) - time;
 		update_avg(&this_sd->avg_scan_cost, time);
 	}
 
-	return cpu;
+	return idle_cpu;
 }
 
 /*
@@ -6302,10 +6325,6 @@ static int select_idle_sibling(struct task_struct *p, int prev, int target)
 	if (!sd)
 		return target;
 
-	i = select_idle_core(p, sd, target);
-	if ((unsigned)i < nr_cpumask_bits)
-		return i;
-
 	i = select_idle_cpu(p, sd, target);
 	if ((unsigned)i < nr_cpumask_bits)
 		return i;
#6 lore raw
In reply to Mel Gorman
Quoted 40 lines · Mel Gorman

Changelog since v4
o Avoid use of intermediate variable during select_idle_cpu

Changelog since v3
o Drop scanning based on cores, SMT4 results showed problems

Changelog since v2
o Remove unnecessary parameters
o Update nr during scan only when scanning for cpus

Changlog since v1
o Move extern declaration to header for coding style
o Remove unnecessary parameter from __select_idle_cpu

This series of 4 patches reposts three patches from Peter entitled "select_idle_sibling() wreckage". It only scans the runqueues in a single pass when searching for an idle sibling.

Three patches from Peter were dropped. The first patch altered how scan depth was calculated. Scan depth deletion is a random number generator with two major limitations. The avg_idle time is based on the time between a CPU going idle and being woken up clamped approximately by 2*sysctl_sched_migration_cost. This is difficult to compare in a sensible fashion to avg_scan_cost. The second issue is that only the avg_scan_cost of scan failures is recorded and it does not decay. This requires deeper surgery that would justify a patch on its own although Peter notes that
https://lkml.kernel.org/r/[email protected] is potentially useful for an alternative avg_idle metric.

The second patch dropped scanned based on cores instead of CPUs as it rationalised the difference between core scanning and CPU scanning. Unfortunately, Vincent reported problems with SMT4 so it's dropped for now until depth searching can be fixed.

The third patch dropped converted the idle core scan throttling mechanism to SIS_PROP. While this would unify the throttling of core and CPU scanning, it was not free of regressions and has_idle_cores is a fairly effective throttling mechanism with the caveat that it can have a lot of false positives for workloads like hackbench.

Quoted 7 lines

Peter's series tried to solve three problems at once, this subset addresses one problem.

 kernel/sched/fair.c     | 151 +++++++++++++++++++---------------------
 kernel/sched/features.h |   1 -
 2 files changed, 70 insertions(+), 82 deletions(-)

4 benchmarks measured on a x86 4s system with 24 cores per socket and 2 HTs per core, total 192 CPUs.

The load level is [25%, 50%, 75%, 100%].

- hackbench almost has a universal win.
- netperf high load has notable changes, as well as tbench 50% load.

Details below:

hackbench: 10 iterations, 10000 loops, 40 fds per group
======================================================

- pipe process

	group	base	%std	v5	%std
	3	1	19.18	1.0266	9.06
	6	1	9.17	0.987	13.03
	9	1	7.11	1.0195	4.61
	12	1	1.07	0.9927	1.43

- pipe thread

	group	base	%std	v5	%std
	3	1	11.14	0.9742	7.27
	6	1	9.15	0.9572	7.48
	9	1	2.95	0.986	4.05
	12	1	1.75	0.9992	1.68

- socket process

	group	base	%std	v5	%std
	3	1	2.9	0.9586	2.39
	6	1	0.68	0.9641	1.3
	9	1	0.64	0.9388	0.76
	12	1	0.56	0.9375	0.55

- socket thread

	group	base	%std	v5	%std
	3	1	3.82	0.9686	2.97
	6	1	2.06	0.9667	1.91
	9	1	0.44	0.9354	1.25
	12	1	0.54	0.9362	0.6

netperf: 10 iterations x 100 seconds, transactions rate / sec
=============================================================

- tcp request/response performance

	thread	base	%std	v4	%std
	25%	1	5.34	1.0039	5.13
	50%	1	4.97	1.0115	6.3
	75%	1	5.09	0.9257	6.75
	100%	1	4.53	0.908	4.83

- udp request/response performance

	thread	base	%std	v4	%std
	25%	1	6.18	0.9896	6.09
	50%	1	5.88	1.0198	8.92
	75%	1	24.38	0.9236	29.14
	100%	1	26.16	0.9063	22.16

tbench: 10 iterations x 100 seconds, throughput / sec
=====================================================

	thread	base	%std	v4	%std
	25%	1	0.45	1.003	1.48
	50%	1	1.71	0.9286	0.82
	75%	1	0.84	0.9928	0.94
	100%	1	0.76	0.9762	0.59

schbench: 10 iterations x 100 seconds, 99th percentile latency
==============================================================

	mthread	base	%std	v4	%std
	25%	1	2.89	0.9884	7.34
	50%	1	40.38	1.0055	38.37
	75%	1	4.76	1.0095	4.62
	100%	1	10.09	1.0083	8.03

Thanks,
-Aubrey

author
#7 lore raw
Quoted 8 lines · Aubrey
Quoted 7 lines

Peter's series tried to solve three problems at once, this subset addresses one problem.

 kernel/sched/fair.c     | 151 +++++++++++++++++++---------------------
 kernel/sched/features.h |   1 -
 2 files changed, 70 insertions(+), 82 deletions(-)
Quoted 8 lines

4 benchmarks measured on a x86 4s system with 24 cores per socket and 2 HTs per core, total 192 CPUs.

The load level is [25%, 50%, 75%, 100%].

- hackbench almost has a universal win.
- netperf high load has notable changes, as well as tbench 50% load.

Ok, both netperf and tbench are somewhat expected as at those loads are rapidly idling. Previously I observed that rapidly idling loads can allow the has_idle_cores test pass for short durations and the double scanning means there is a greater chance of finding an idle CPU over the two passes. I think overall it's better to avoid double scanning even if there are counter examples as it's possible we'll get that back through better depth selection in the future.

Thanks.

Signature · 2 lines
Mel Gorman
SUSE Labs