Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1465760 > unrolled thread

[PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> - 2016-08-19 03:20 +0200
    Re: [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Tim Chen <tim.c.chen@linux.intel.com> - 2016-08-22 21:10 +0200
    Re: [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Ingo Molnar <mingo@kernel.org> - 2016-08-24 12:40 +0200
      Re: [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Tim Chen <tim.c.chen@linux.intel.com> - 2016-08-24 20:00 +0200
        Re: [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Ingo Molnar <mingo@kernel.org> - 2016-08-24 20:10 +0200
          Re: [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology Peter Zijlstra <peterz@infradead.org> - 2016-08-24 20:30 +0200

#1465760 — [PATCH 04/11] sched,x86: Enable Turbo Boost Max Technology

From: Tim Chen <tim.c.chen@linux.intel.com>

On some Intel cores, they can boosted to a higher turbo frequency than
the other cores on the same die.  So we prefer processes to be run on
them vs other lower frequency ones for extra performance.

We extend the asym packing feature in the scheduler to support packing
task to the higher frequency core at the core sched domain level.

We set up a core priority metric to abstract the core preferences based
on the maximum boost frequency.  The priority is instantiated such that
the core with a higher priority is favored over the core with lower
priority when making scheduling decision using ASYM_PACKING.  The smt
threads that are of higher number are discounted in their priority so
we will not try to pack tasks onto all the threads of a favored core
before using other cpu cores.  The cpu that's of the highese priority
in a sched_group is recorded in sched_group->asym_prefer_cpu during
initialization to save lookup during load balancing.

A sysctl variable /proc/sys/kernel/sched_itmt_enabled is provided so
the scheduling based on favored core can be turned on or off at run time.

Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
---
 arch/x86/Kconfig                |   9 +++
 arch/x86/include/asm/topology.h |  26 +++++++
 arch/x86/kernel/Makefile        |   1 +
 arch/x86/kernel/itmt.c          | 147 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 183 insertions(+)
 create mode 100644 arch/x86/kernel/itmt.c

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index c580d8c..c1d36db 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -928,6 +928,15 @@ config SCHED_MC
 	  making when dealing with multi-core CPU chips at a cost of slightly
 	  increased overhead in some places. If unsure say N here.
 
+config SCHED_ITMT
+	bool "Intel Turbo Boost Max Technology (ITMT) scheduler support"
+	depends on SCHED_MC && CPU_SUP_INTEL && X86_INTEL_PSTATE
+	---help---
+	  ITMT enabled scheduler support improves the CPU scheduler's decision
+	  to move tasks to cpu core that can be boosted to a higher frequency
+	  than others. It will have better performance at a cost of slightly
+	  increased overhead in task migrations. If unsure say N here.
+
 source "kernel/Kconfig.preempt"
 
 config UP_LATE_INIT
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index cf75871..f148843 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -146,4 +146,30 @@ struct pci_bus;
 int x86_pci_root_bus_node(int bus);
 void x86_pci_root_bus_resources(int bus, struct list_head *resources);
 
+#ifdef CONFIG_SCHED_ITMT
+DECLARE_PER_CPU_READ_MOSTLY(int, sched_core_priority);
+
+static inline bool sched_asym_prefer(int a, int b)
+{
+	return per_cpu(sched_core_priority, a) > per_cpu(sched_core_priority, b);
+}
+
+#define sched_asym_prefer sched_asym_prefer
+
+/* Interface to set priority of a cpu */
+void sched_set_itmt_core_prio(int prio, int core_cpu);
+
+/* Interface to notify scheduler that system supports ITMT */
+void set_sched_itmt(bool support_itmt);
+
+#else /* CONFIG_SCHED_ITMT */
+
+static inline void set_sched_itmt(bool support_itmt)
+{
+}
+static inline void sched_set_itmt_core_prio(int prio, int core_cpu)
+{
+}
+#endif /* CONFIG_SCHED_ITMT */
+
 #endif /* _ASM_X86_TOPOLOGY_H */
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 0503f5b..2008335 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -124,6 +124,7 @@ obj-$(CONFIG_EFI)			+= sysfb_efi.o
 
 obj-$(CONFIG_PERF_EVENTS)		+= perf_regs.o
 obj-$(CONFIG_TRACING)			+= tracepoint.o
+obj-$(CONFIG_SCHED_ITMT)		+= itmt.o
 
 ###
 # 64 bit specific files
diff --git a/arch/x86/kernel/itmt.c b/arch/x86/kernel/itmt.c
new file mode 100644
index 0000000..a9470f5
--- /dev/null
+++ b/arch/x86/kernel/itmt.c
@@ -0,0 +1,147 @@
+/*
+ * itmt.c: Functions and data structures for enabling
+ * 	   scheduler to favor scheduling on cores that
+ *	   can be boosted to a higher frequency using
+ *	   Intel Turbo Boost Max Technology 3.0
+ *
+ * (C) Copyright 2016 Intel Corporation
+ * Author: Tim Chen <tim.c.chen@linux.intel.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#include <linux/sched.h>
+#include <linux/cpumask.h>
+#include <linux/cpuset.h>
+#include <asm/mutex.h>
+#include <linux/sched.h>
+#include <linux/sysctl.h>
+
+DEFINE_PER_CPU_READ_MOSTLY(int, sched_core_priority);
+static DEFINE_MUTEX(itmt_update_mutex);
+
+static unsigned int zero = 0;
+static unsigned int one = 1;
+
+/*
+ * Boolean to control whether we want to move processes to cpu capable
+ * of higher turbo frequency for cpus supporting Intel Turbo Boost Max
+ * Technology 3.0.
+ *
+ * It can be set via /proc/sys/kernel/sched_itmt_enabled
+ */
+unsigned int __read_mostly sysctl_sched_itmt_enabled = 0;
+
+/*
+ * The pstate_driver calls set_sched_itmt to indicate if the system
+ * is ITMT capable.
+ */
+static bool __read_mostly sched_itmt_capable;
+
+static void enable_sched_itmt(bool enable_itmt)
+{
+	mutex_lock(&itmt_update_mutex);
+
+	sysctl_sched_itmt_enabled = enable_itmt;
+	regenerate_sched_domains();
+
+	mutex_unlock(&itmt_update_mutex);
+}
+
+static int sched_itmt_update_handler(struct ctl_table *table, int write,
+			      void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	int ret;
+
+	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
+
+	if (ret || !write)
+		return ret;
+
+	enable_sched_itmt(sysctl_sched_itmt_enabled);
+
+	return ret;
+}
+
+static struct ctl_table itmt_kern_table[] = {
+	{
+		.procname	= "sched_itmt_enabled",
+		.data		= &sysctl_sched_itmt_enabled,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= sched_itmt_update_handler,
+		.extra1		= &zero,
+		.extra2		= &one,
+	},
+	{}
+};
+
+static struct ctl_table itmt_root_table[] = {
+	{
+		.procname	= "kernel",
+		.mode		= 0555,
+		.child		= itmt_kern_table,
+	},
+	{}
+};
+
+static struct ctl_table_header *itmt_sysctl_header;
+
+/*
+ * The boot code will find out the max boost frequency
+ * and call this function to set a priority proportional
+ * to the max boost frequency. CPU with higher boost
+ * frequency will receive higher priority.
+ */
+void sched_set_itmt_core_prio(int prio, int core_cpu)
+{
+	int cpu, i = 1;
+
+	for_each_cpu(cpu, topology_sibling_cpumask(core_cpu)) {
+		int smt_prio;
+
+		/*
+		 * Discount the priority of sibling so that we don't
+		 * pack all loads to the same core before using other cores.
+		 */
+		smt_prio = prio * smp_num_siblings / i;
+		i++;
+		per_cpu(sched_core_priority, cpu) = smt_prio;
+	}
+}
+
+/*
+ * During boot up, boot code will detect if the system
+ * is ITMT capable and call set_sched_itmt.
+ *
+ * This should be call after sched_set_itmt_core_prio
+ * has been called to set the cpus' priorities.
+ *
+ * This function should be called without cpu hot plug lock
+ * as we need to acquire the lock to rebuild sched domains
+ * later.
+ */
+void set_sched_itmt(bool itmt_capable)
+{
+	mutex_lock(&itmt_update_mutex);
+
+	if (itmt_capable != sched_itmt_capable) {
+
+		if (itmt_capable) {
+			itmt_sysctl_header =
+				register_sysctl_table(itmt_root_table);
+		} else {
+			if (itmt_sysctl_header)
+				unregister_sysctl_table(itmt_sysctl_header);
+			sysctl_sched_itmt_enabled = false;
+		}
+
+		sched_itmt_capable = itmt_capable;
+		regenerate_sched_domains();
+	}
+
+	mutex_unlock(&itmt_update_mutex);
+}
-- 
2.7.4

[toc] | [next] | [standalone]

#1467966

On Mon, Aug 22, 2016 at 05:01:15PM +0800, kbuild test robot wrote:
> Hi Tim,
> 
> [auto build test ERROR on pm/linux-next]
> [also build test ERROR on v4.8-rc3 next-20160822]
> [if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
> [Suggest to use git(>=2.9.0) format-patch --base=<commit> (or --base=auto for convenience) to record what (public, well-known) commit your patch series was built on]
> [Check https://git-scm.com/docs/git-format-patch for more information]
> 
> url:    https://github.com/0day-ci/linux/commits/Srinivas-Pandruvada/Support-Intel-Turbo-Boost-Max-Technology-3-0/20160819-101955
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next
> config: i386-allyesconfig (attached as .config)
> compiler: gcc-6 (Debian 6.1.1-9) 6.1.1 20160705
> reproduce:
>         # save the attached .config to linux build tree
>         make ARCH=i386 
> 

Should be fixed by the patch below.  Will incorporate the fix in next
rev of the patch series.

Tim

---

diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index f148843..bcf3e85 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -147,6 +147,8 @@ int x86_pci_root_bus_node(int bus);
 void x86_pci_root_bus_resources(int bus, struct list_head *resources);
 
 #ifdef CONFIG_SCHED_ITMT
+#include <asm/percpu.h>
+
 DECLARE_PER_CPU_READ_MOSTLY(int, sched_core_priority);
 
 static inline bool sched_asym_prefer(int a, int b)

[toc] | [prev] | [next] | [standalone]

#1469317

* Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> wrote:

> From: Tim Chen <tim.c.chen@linux.intel.com>
> 
> On some Intel cores, they can boosted to a higher turbo frequency than
> the other cores on the same die.  So we prefer processes to be run on
> them vs other lower frequency ones for extra performance.
> 
> We extend the asym packing feature in the scheduler to support packing
> task to the higher frequency core at the core sched domain level.
> 
> We set up a core priority metric to abstract the core preferences based
> on the maximum boost frequency.  The priority is instantiated such that
> the core with a higher priority is favored over the core with lower
> priority when making scheduling decision using ASYM_PACKING.  The smt
> threads that are of higher number are discounted in their priority so
> we will not try to pack tasks onto all the threads of a favored core
> before using other cpu cores.  The cpu that's of the highese priority
> in a sched_group is recorded in sched_group->asym_prefer_cpu during
> initialization to save lookup during load balancing.
> 
> A sysctl variable /proc/sys/kernel/sched_itmt_enabled is provided so
> the scheduling based on favored core can be turned on or off at run time.

> +/*
> + * Boolean to control whether we want to move processes to cpu capable
> + * of higher turbo frequency for cpus supporting Intel Turbo Boost Max
> + * Technology 3.0.
> + *
> + * It can be set via /proc/sys/kernel/sched_itmt_enabled
> + */
> +unsigned int __read_mostly sysctl_sched_itmt_enabled = 0;

Ugh, no.

We don't add features to the scheduler in the hope that they might or might not 
help. We either enable a new feature by default (and make damn sure it helps!),
or don't add the feature at all.

Thanks,

	Ingo

[toc] | [prev] | [next] | [standalone]

#1469616

On Wed, Aug 24, 2016 at 12:18:53PM +0200, Ingo Molnar wrote:
> 
> * Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> wrote:
> 
> > From: Tim Chen <tim.c.chen@linux.intel.com>
> > 
> > On some Intel cores, they can boosted to a higher turbo frequency than
> > the other cores on the same die.  So we prefer processes to be run on
> > them vs other lower frequency ones for extra performance.
> > 
> > We extend the asym packing feature in the scheduler to support packing
> > task to the higher frequency core at the core sched domain level.
> > 
> > We set up a core priority metric to abstract the core preferences based
> > on the maximum boost frequency.  The priority is instantiated such that
> > the core with a higher priority is favored over the core with lower
> > priority when making scheduling decision using ASYM_PACKING.  The smt
> > threads that are of higher number are discounted in their priority so
> > we will not try to pack tasks onto all the threads of a favored core
> > before using other cpu cores.  The cpu that's of the highese priority
> > in a sched_group is recorded in sched_group->asym_prefer_cpu during
> > initialization to save lookup during load balancing.
> > 
> > A sysctl variable /proc/sys/kernel/sched_itmt_enabled is provided so
> > the scheduling based on favored core can be turned on or off at run time.
> 
> > +/*
> > + * Boolean to control whether we want to move processes to cpu capable
> > + * of higher turbo frequency for cpus supporting Intel Turbo Boost Max
> > + * Technology 3.0.
> > + *
> > + * It can be set via /proc/sys/kernel/sched_itmt_enabled
> > + */
> > +unsigned int __read_mostly sysctl_sched_itmt_enabled = 0;
> 
> Ugh, no.
> 
> We don't add features to the scheduler in the hope that they might or might not 
> help. We either enable a new feature by default (and make damn sure it helps!),
> or don't add the feature at all.
> 
> Thanks,
> 
> 	Ingo

Ingo,

This feature will be a clear benefit for client machines and
less clear on servers.

This feature is most beneficial to single threaded workload running on
a single socket that operates in mostly Turbo mode.  Client platform
like Broadwell High End Desktop is the first one that supports it.
Enablng this feature for such platform by default will be a win as it
runs single threaded workload much of the time (10%-15% peformance
upside).

On the other hand, a heavily loaded server that rarely operates in Turbo
mode will benefit much less from this feature.  There is some overhead
incurred by migrating load to the favored cores.  Some server folks
have asked us to be cautious here and not to turn on ITMT scheduling
by default.   Even so, when the server is lightly loaded, this feature
can still be a win.  That said, this is future looking as we don't have
any server with this feature today.

So if we take the approach to enable this feature by default for only
single node system (using that as a criteria for client), will that seem
reasonable to you?

Thanks.

Tim

[toc] | [prev] | [next] | [standalone]

#1469623

* Tim Chen <tim.c.chen@linux.intel.com> wrote:

> Ingo,
> 
> This feature will be a clear benefit for client machines and
> less clear on servers.
> 
> This feature is most beneficial to single threaded workload running on
> a single socket that operates in mostly Turbo mode.  Client platform
> like Broadwell High End Desktop is the first one that supports it.
> Enablng this feature for such platform by default will be a win as it
> runs single threaded workload much of the time (10%-15% peformance
> upside).
> 
> On the other hand, a heavily loaded server that rarely operates in Turbo
> mode will benefit much less from this feature.  There is some overhead
> incurred by migrating load to the favored cores.  Some server folks
> have asked us to be cautious here and not to turn on ITMT scheduling
> by default.   Even so, when the server is lightly loaded, this feature
> can still be a win.  That said, this is future looking as we don't have
> any server with this feature today.
> 
> So if we take the approach to enable this feature by default for only
> single node system (using that as a criteria for client), will that seem
> reasonable to you?

I suppose that would work. Peter, any objections to such an approach?

Thanks,

	Ingo

[toc] | [prev] | [next] | [standalone]

#1469631

On Wed, Aug 24, 2016 at 08:08:54PM +0200, Ingo Molnar wrote:
> > 
> > So if we take the approach to enable this feature by default for only
> > single node system (using that as a criteria for client), will that seem
> > reasonable to you?
> 
> I suppose that would work. Peter, any objections to such an approach?

Works for me.

Thanks!

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web