| Age | Commit message (Collapse) | Author |
|---|
|
Currently if architectures want to support HOTPLUG_SMT they need to
provide a topology_is_primary_thread() telling the framework which
thread in the SMT cannot offline. However arm64 doesn't have a
restriction on which thread in the SMT cannot offline, a simplest
choice is that just make 1st thread as the "primary" thread. So
just make this as the default implementation in the framework and
let architectures like x86 that have special primary thread to
override this function (which they've already done).
There's no need to provide a stub function if !CONFIG_SMP or
!CONFIG_HOTPLUG_SMT. In such case the testing CPU is already
the 1st CPU in the SMT so it's always the primary thread.
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Pierre Gondois <pierre.gondois@arm.com>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>
Link: https://lore.kernel.org/r/20250311075143.61078-2-yangyicong@huawei.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull scheduler updates from Ingo Molnar:
"Fair scheduler (SCHED_FAIR) enhancements:
- Behavioral improvements:
- Untangle NEXT_BUDDY and pick_next_task() (Peter Zijlstra)
- Delayed-dequeue enhancements & fixes: (Vincent Guittot)
- Rename h_nr_running into h_nr_queued
- Add new cfs_rq.h_nr_runnable
- Use the new cfs_rq.h_nr_runnable
- Removed unsued cfs_rq.h_nr_delayed
- Rename cfs_rq.idle_h_nr_running into h_nr_idle
- Remove unused cfs_rq.idle_nr_running
- Rename cfs_rq.nr_running into nr_queued
- Do not try to migrate delayed dequeue task
- Fix variable declaration position
- Encapsulate set custom slice in a __setparam_fair() function
- Fixes:
- Fix race between yield_to() and try_to_wake_up() (Tianchen Ding)
- Fix CPU bandwidth limit bypass during CPU hotplug (Vishal
Chourasia)
- Cleanups:
- Clean up in migrate_degrades_locality() to improve readability
(Peter Zijlstra)
- Mark m*_vruntime() with __maybe_unused (Andy Shevchenko)
- Update comments after sched_tick() rename (Sebastian Andrzej
Siewior)
- Remove CONFIG_CFS_BANDWIDTH=n definition of cfs_bandwidth_used()
(Valentin Schneider)
Deadline scheduler (SCHED_DL) enhancements:
- Restore dl_server bandwidth on non-destructive root domain changes
(Juri Lelli)
- Correctly account for allocated bandwidth during hotplug (Juri
Lelli)
- Check bandwidth overflow earlier for hotplug (Juri Lelli)
- Clean up goto label in pick_earliest_pushable_dl_task() (John
Stultz)
- Consolidate timer cancellation (Wander Lairson Costa)
Load-balancer enhancements:
- Improve performance by prioritizing migrating eligible tasks in
sched_balance_rq() (Hao Jia)
- Do not compute NUMA Balancing stats unnecessarily during
load-balancing (K Prateek Nayak)
- Do not compute overloaded status unnecessarily during
load-balancing (K Prateek Nayak)
Generic scheduling code enhancements:
- Use READ_ONCE() in task_on_rq_queued(), to consistently use the
WRITE_ONCE() updated ->on_rq field (Harshit Agarwal)
Isolated CPUs support enhancements: (Waiman Long)
- Make "isolcpus=nohz" equivalent to "nohz_full"
- Consolidate housekeeping cpumasks that are always identical
- Remove HK_TYPE_SCHED
- Unify HK_TYPE_{TIMER|TICK|MISC} to HK_TYPE_KERNEL_NOISE
RSEQ enhancements:
- Validate read-only fields under DEBUG_RSEQ config (Mathieu
Desnoyers)
PSI enhancements:
- Fix race when task wakes up before psi_sched_switch() adjusts flags
(Chengming Zhou)
IRQ time accounting performance enhancements: (Yafang Shao)
- Define sched_clock_irqtime as static key
- Don't account irq time if sched_clock_irqtime is disabled
Virtual machine scheduling enhancements:
- Don't try to catch up excess steal time (Suleiman Souhlal)
Heterogenous x86 CPU scheduling enhancements: (K Prateek Nayak)
- Convert "sysctl_sched_itmt_enabled" to boolean
- Use guard() for itmt_update_mutex
- Move the "sched_itmt_enabled" sysctl to debugfs
- Remove x86_smt_flags and use cpu_smt_flags directly
- Use x86_sched_itmt_flags for PKG domain unconditionally
Debugging code & instrumentation enhancements:
- Change need_resched warnings to pr_err() (David Rientjes)
- Print domain name in /proc/schedstat (K Prateek Nayak)
- Fix value reported by hot tasks pulled in /proc/schedstat (Peter
Zijlstra)
- Report the different kinds of imbalances in /proc/schedstat
(Swapnil Sapkal)
- Move sched domain name out of CONFIG_SCHED_DEBUG (Swapnil Sapkal)
- Update Schedstat version to 17 (Swapnil Sapkal)"
* tag 'sched-core-2025-01-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (48 commits)
rseq: Fix rseq unregistration regression
psi: Fix race when task wakes up before psi_sched_switch() adjusts flags
sched, psi: Don't account irq time if sched_clock_irqtime is disabled
sched: Don't account irq time if sched_clock_irqtime is disabled
sched: Define sched_clock_irqtime as static key
sched/fair: Do not compute overloaded status unnecessarily during lb
sched/fair: Do not compute NUMA Balancing stats unnecessarily during lb
x86/topology: Use x86_sched_itmt_flags for PKG domain unconditionally
x86/topology: Remove x86_smt_flags and use cpu_smt_flags directly
x86/itmt: Move the "sched_itmt_enabled" sysctl to debugfs
x86/itmt: Use guard() for itmt_update_mutex
x86/itmt: Convert "sysctl_sched_itmt_enabled" to boolean
sched/core: Prioritize migrating eligible tasks in sched_balance_rq()
sched/debug: Change need_resched warnings to pr_err
sched/fair: Encapsulate set custom slice in a __setparam_fair() function
sched: Fix race between yield_to() and try_to_wake_up()
docs: Update Schedstat version to 17
sched/stats: Print domain name in /proc/schedstat
sched: Move sched domain name out of CONFIG_SCHED_DEBUG
sched: Report the different kinds of imbalances in /proc/schedstat
...
|
|
In preparation to move "sysctl_sched_itmt_enabled" to debugfs, convert
the unsigned int to bool since debugfs readily exposes boolean fops
primitives (debugfs_read_file_bool, debugfs_write_file_bool) which can
streamline the conversion.
Since the current ctl_table initializes extra1 and extra2 to SYSCTL_ZERO
and SYSCTL_ONE respectively, the value of "sysctl_sched_itmt_enabled"
can only be 0 or 1 and this datatype conversion should not cause any
functional changes.
Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
Link: https://lore.kernel.org/r/20241223043407.1611-2-kprateek.nayak@amd.com
|
|
On x86, topology_core_id() returns a unique core ID within the PKG
domain. Looking at match_smt() suggests that a core ID just needs to be
unique within a LLC domain. For use cases such as the core RAPL PMU,
there exists a need for a unique core ID across the entire system with
multiple PKG domains. Introduce topology_logical_core_id() to derive a
unique core ID across the system.
Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
Signed-off-by: Dhananjay Ugwekar <Dhananjay.Ugwekar@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Zhang Rui <rui.zhang@intel.com>
Reviewed-by: "Gautham R. Shenoy" <gautham.shenoy@amd.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Link: https://lore.kernel.org/r/20241115060805.447565-3-Dhananjay.Ugwekar@amd.com
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 cpuid updates from Borislav Petkov:
- Add a feature flag which denotes AMD CPUs supporting workload
classification with the purpose of using such hints when making
scheduling decisions
- Determine the boost enumerator for each AMD core based on its type:
efficiency or performance, in the cppc driver
- Add the type of a CPU to the topology CPU descriptor with the goal of
supporting and making decisions based on the type of the respective
core
- Add a feature flag to denote AMD cores which have heterogeneous
topology and enable SD_ASYM_PACKING for those
- Check microcode revisions before disabling PCID on Intel
- Cleanups and fixlets
* tag 'x86_cpu_for_v6.13' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Remove redundant CONFIG_NUMA guard around numa_add_cpu()
x86/cpu: Fix FAM5_QUARK_X1000 to use X86_MATCH_VFM()
x86/cpu: Fix formatting of cpuid_bits[] in scattered.c
x86/cpufeatures: Add X86_FEATURE_AMD_WORKLOAD_CLASS feature bit
x86/amd: Use heterogeneous core topology for identifying boost numerator
x86/cpu: Add CPU type to struct cpuinfo_topology
x86/cpu: Enable SD_ASYM_PACKING for PKG domain on AMD
x86/cpufeatures: Add X86_FEATURE_AMD_HETEROGENEOUS_CORES
x86/cpufeatures: Rename X86_FEATURE_FAST_CPPC to have AMD prefix
x86/mm: Don't disable PCID when INVLPG has been fixed by microcode
|
|
arch_init_invariance_cppc() is called at the end of
acpi_cppc_processor_probe() in order to configure frequency invariance
based upon the values from _CPC.
This however doesn't work on AMD CPPC shared memory designs that have
AMD preferred cores enabled because _CPC needs to be analyzed from all
cores to judge if preferred cores are enabled.
This issue manifests to users as a warning since commit 21fb59ab4b97
("ACPI: CPPC: Adjust debug messages in amd_set_max_freq_ratio() to warn"):
```
Could not retrieve highest performance (-19)
```
However the warning isn't the cause of this, it was actually
commit 279f838a61f9 ("x86/amd: Detect preferred cores in
amd_get_boost_ratio_numerator()") which exposed the issue.
To fix this problem, change arch_init_invariance_cppc() into a new weak
symbol that is called at the end of acpi_processor_driver_init().
Each architecture that supports it can declare the symbol to override
the weak one.
Define it for x86, in arch/x86/kernel/acpi/cppc.c, and for all of the
architectures using the generic arch_topology.c code.
Fixes: 279f838a61f9 ("x86/amd: Detect preferred cores in amd_get_boost_ratio_numerator()")
Reported-by: Ivan Shapovalov <intelfx@intelfx.name>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=219431
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://patch.msgid.link/20241104222855.3959267-1-superm1@kernel.org
[ rjw: Changelog edit ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
Sometimes it is required to take actions based on if a CPU is a performance or
efficiency core. As an example, intel_pstate driver uses the Intel core-type
to determine CPU scaling. Also, some CPU vulnerabilities only affect
a specific CPU type, like RFDS only affects Intel Atom. Hybrid systems that
have variants P+E, P-only(Core) and E-only(Atom), it is not straightforward to
identify which variant is affected by a type specific vulnerability.
Such processors do have CPUID field that can uniquely identify them. Like,
P+E, P-only and E-only enumerates CPUID.1A.CORE_TYPE identification, while P+E
additionally enumerates CPUID.7.HYBRID. Based on this information, it is
possible for boot CPU to identify if a system has mixed CPU types.
Add a new field hw_cpu_type to struct cpuinfo_topology that stores the
hardware specific CPU type. This saves the overhead of IPIs to get the CPU
type of a different CPU. CPU type is populated early in the boot process,
before vulnerabilities are enumerated.
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Co-developed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20241025171459.1093-5-mario.limonciello@amd.com
|
|
In order be able to compute the sizes of tasks consistently across all
CPUs in a hybrid system, it is necessary to provide CPU capacity scaling
information to the scheduler via arch_scale_cpu_capacity(). Moreover,
the value returned by arch_scale_freq_capacity() for the given CPU must
correspond to the arch_scale_cpu_capacity() return value for it, or
utilization computations will be inaccurate.
Add support for it through per-CPU variables holding the capacity and
maximum-to-base frequency ratio (times SCHED_CAPACITY_SCALE) that will
be returned by arch_scale_cpu_capacity() and used by scale_freq_tick()
to compute arch_freq_scale for the current CPU, respectively.
In order to avoid adding measurable overhead for non-hybrid x86 systems,
which are the vast majority in the field, whether or not the new hybrid
CPU capacity scaling will be in effect is controlled by a static key.
This static key is set by calling arch_enable_hybrid_capacity_scale()
which also allocates memory for the per-CPU data and initializes it.
Next, arch_set_cpu_capacity() is used to set the per-CPU variables
mentioned above for each CPU and arch_rebuild_sched_domains() needs
to be called for the scheduler to realize that capacity-aware
scheduling can be used going forward.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Tested-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com> # scale invariance
Link: https://patch.msgid.link/10523497.nUPlyArG6x@rjwysocki.net
[ rjw: Added parens to function kerneldoc comments ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC updates from Borislav Petkov:
- Add a FRU (Field Replaceable Unit) memory poison manager which
collects and manages previously encountered hw errors in order to
save them to persistent storage across reboots. Previously recorded
errors are "replayed" upon reboot in order to poison memory which has
caused said errors in the past.
The main use case is stacked, on-chip memory which cannot simply be
replaced so poisoning faulty areas of it and thus making them
inaccessible is the only strategy to prolong its lifetime.
- Add an AMD address translation library glue which converts the
reported addresses of hw errors into system physical addresses in
order to be used by other subsystems like memory failure, for
example. Add support for MI300 accelerators to that library.
- igen6: Add support for Alder Lake-N SoC
- i10nm: Add Grand Ridge support
- The usual fixlets and cleanups
* tag 'edac_updates_for_v6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/versal: Convert to platform remove callback returning void
RAS/AMD/FMPM: Fix off by one when unwinding on error
RAS/AMD/FMPM: Add debugfs interface to print record entries
RAS/AMD/FMPM: Save SPA values
RAS: Export helper to get ras_debugfs_dir
RAS/AMD/ATL: Fix bit overflow in denorm_addr_df4_np2()
RAS: Introduce a FRU memory poison manager
RAS/AMD/ATL: Add MI300 row retirement support
Documentation: Move RAS section to admin-guide
EDAC/versal: Make the bit position of injected errors configurable
EDAC/i10nm: Add Intel Grand Ridge micro-server support
EDAC/igen6: Add one more Intel Alder Lake-N SoC support
RAS/AMD/ATL: Add MI300 DRAM to normalized address translation support
RAS/AMD/ATL: Fix array overflow in get_logical_coh_st_fabric_id_mi300()
RAS/AMD/ATL: Add MI300 support
Documentation: RAS: Add index and address translation section
EDAC/amd64: Use new AMD Address Translation Library
RAS: Introduce AMD Address Translation Library
EDAC/synopsys: Convert to devm_platform_ioremap_resource()
|
|
Expose properly accounted information and accessors so the fiddling with
other topology variables can be replaced.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210253.120958987@linutronix.de
|
|
The plural of die is dies.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210253.065874205@linutronix.de
|
|
It's really a non-intuitive name. Rename it to __max_threads_per_core which
is obvious.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210253.011307973@linutronix.de
|
|
Replace the logical package and die management functionality and retrieve
the logical IDs from the topology bitmaps.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210252.901865302@linutronix.de
|
|
With the topology bitmaps in place the logical package and die IDs can
trivially be retrieved by determining the bitmap weight of the relevant
topology domain level up to and including the physical ID in question.
Provide a function to that effect.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210252.846136196@linutronix.de
|
|
Now that all possible APIC IDs are tracked in the topology bitmaps, its
trivial to retrieve the real information from there.
This gets rid of the guesstimates for the maximal packages and dies per
package as the actual numbers can be determined before a single AP has been
brought up.
The number of SMT threads can now be determined correctly from the bitmaps
in all situations. Up to now a system which has SMT disabled in the BIOS
will still claim that it is SMT capable, because the lowest APIC ID bit is
reserved for that and CPUID leaf 0xb/0x1f still enumerates the SMT domain
accordingly. By calculating the bitmap weights of the SMT and the CORE
domain and setting them into relation the SMT disabled in BIOS situation
reports correctly that the system is not SMT capable.
It also handles the situation correctly when a hybrid systems boot CPU does
not have SMT as it takes the SMT capability of the APs fully into account.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210252.681709880@linutronix.de
|
|
Managing possible CPUs is an unreadable and uncomprehensible maze. Aside of
that it's backwards because it applies command line limits after
registering all APICs.
Rewrite it so that it:
- Applies the command line limits upfront so that only the allowed amount
of APIC IDs can be registered.
- Applies eventual late restrictions in an understandable way
- Uses simple min_t() calculations which are trivial to follow.
- Provides a separate function for resetting to UP mode late in the
bringup process.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Link: https://lore.kernel.org/r/20240213210252.290098853@linutronix.de
|
|
The package shift has been already evaluated by the early CPU init.
Put the mindless copy right next to the original leaf 0xb parser.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20240212153625.637385562@linutronix.de
|
|
Now that everything is converted switch it over and remove the intermediate
operation.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20240212153625.334185785@linutronix.de
|
|
cpuinfo::topo::x86_coreid_bits is about to be phased out. Use the core
domain size from the topology information.
Add a comment why the early MPTABLE parsing is required and decrapify the
loop which sets the APIC ID to node map.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20240212153625.270320718@linutronix.de
|
|
AMD/HYGON uses various methods for topology evaluation:
- Leaf 0x80000008 and 0x8000001e based with an optional leaf 0xb,
which is the preferred variant for modern CPUs.
Leaf 0xb will be superseded by leaf 0x80000026 soon, which is just
another variant of the Intel 0x1f leaf for whatever reasons.
- Subleaf 0x80000008 and NODEID_MSR base
- Legacy fallback
That code is following the principle of random bits and pieces all over the
place which results in multiple evaluations and impenetrable code flows in
the same way as the Intel parsing did.
Provide a sane implementation by clearly separating the three variants and
bringing them in the proper preference order in one place.
This provides the parsing for both AMD and HYGON because there is no point
in having a separate HYGON parser which only differs by 3 lines of
code. Any further divergence between AMD and HYGON can be handled in
different functions, while still sharing the existing parsers.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20240212153625.020038641@linutronix.de
|
|
AMD (ab)uses topology_die_id() to store the Node ID information and
topology_max_dies_per_pkg to store the number of nodes per package.
This collides with the proper processor die level enumeration which is
coming on AMD with CPUID 8000_0026, unless there is a correlation between
the two. There is zero documentation about that.
So provide new storage and new accessors which for now still access die_id
and topology_max_die_per_pkg(). Will be mopped up after AMD and HYGON are
converted over.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20240212153624.956116738@linutronix.de
|
|
Topology evaluation is a complete disaster and impenetrable mess. It's
scattered all over the place with some vendor implementations doing early
evaluation and some not. The most horrific part is the permanent
overwriting of smt_max_siblings and __max_die_per_package, instead of
establishing them once on the boot CPU and validating the result on the
APs.
The goals are:
- One topology evaluation entry point
- Proper sharing of pointlessly duplicated code
- Proper structuring of the evaluation logic and preferences.
- Evaluating important system wide information only once on the boot CPU
- Making the 0xb/0x1f leaf parsing less convoluted and actually fixing
the short comings of leaf 0x1f evaluation.
Start to consolidate the topology evaluation code by providing the entry
points for the early boot CPU evaluation and for the final parsing on the
boot CPU and the APs.
Move the trivial pieces into that new code:
- The initialization of cpuinfo_x86::topo
- The evaluation of CPUID leaf 1, which presets topo::initial_apicid
- topo_apicid is set to topo::initial_apicid when invoked from early
boot. When invoked for the final evaluation on the boot CPU it reads
the actual APIC ID, which makes apic_get_initial_apicid() obsolete
once everything is converted over.
Provide a temporary helper function topo_converted() which shields off the
not yet converted CPU vendors from invoking code which would break them.
This shielding covers all vendor CPUs which support SMP, but not the
historical pure UP ones as they only need the topology info init and
eventually the initial APIC initialization.
Provide two new members in cpuinfo_x86::topo to store the maximum number of
SMT siblings and the number of dies per package and add them to the debugfs
readout. These two members will be used to populate this information on the
boot CPU and to validate the APs against it.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Tested-by: Wang Wendy <wendy.wang@intel.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20240212153624.581436579@linutronix.de
|
|
The topology IDs which identify the LLC and L2 domains clearly belong to
the per CPU topology information.
Move them into cpuinfo_x86::cpuinfo_topo and get rid of the extra per CPU
data and the related exports.
This also paves the way to do proper topology evaluation during early boot
because it removes the only per CPU dependency for that.
No functional change.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Reviewed-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.803864641@linutronix.de
|
|
Yet another topology related data pair. Rename logical_proc_id to
logical_pkg_id so it fits the common naming conventions.
No functional change.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.745139505@linutronix.de
|
|
Rename it to core_id and stick it to the other ID fields.
No functional change.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.566519388@linutronix.de
|
|
Move the next member.
No functional change.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.388185134@linutronix.de
|
|
Rename it to pkg_id which is the terminology used in the kernel.
No functional change.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.329006989@linutronix.de
|
|
Since the maximum number of threads is now passed to cpu_smt_set_num_threads(),
checking that value is enough to know whether SMT is supported.
Suggested-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Laurent Dufour <ldufour@linux.ibm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Link: https://lore.kernel.org/r/20230705145143.40545-6-ldufour@linux.ibm.com
|
|
In order to export the cpuhp_smt_control enum as part of the interface
between generic and architecture code, the architecture code needs to
include asm/topology.h.
But that leads to circular header dependencies. So split the enum and
related declarations into a separate header.
[ ldufour: Reworded the commit's description ]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Laurent Dufour <ldufour@linux.ibm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Link: https://lore.kernel.org/r/20230705145143.40545-3-ldufour@linux.ibm.com
|
|
Make the primary thread tracking CPU mask based in preparation for simpler
handling of parallel bootup.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Tested-by: Helge Deller <deller@gmx.de> # parisc
Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com> # Steam Deck
Link: https://lore.kernel.org/r/20230512205257.186599880@linutronix.de
|
|
Make topology_phys_to_logical_pkg_die() static as it's only used in
smpboot.c and fixup the kernel-doc warnings for both functions.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Tested-by: Helge Deller <deller@gmx.de> # parisc
Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com> # Steam Deck
Link: https://lore.kernel.org/r/20230512205255.493750666@linutronix.de
|
|
The utilization of arch_scale_freq_tick() for CPU frequency readouts is
incomplete as it failed to move the function prototype and the define
out of the CONFIG_SMP && CONFIG_X86_64 #ifdef.
Make them unconditionally available.
Fixes: bb6e89df9028 ("x86/aperfmperf: Make parts of the frequency invariance code unconditional")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/lkml/202205010106.06xRBR2C-lkp@intel.com
|
|
The frequency invariance support is currently limited to x86/64 and SMP,
which is the vast majority of machines.
arch_scale_freq_tick() is called every tick on all CPUs and reads the APERF
and MPERF MSRs. The CPU frequency getters function do the same via dedicated
IPIs.
While it could be argued that on systems where frequency invariance support
is disabled (32bit, !SMP) the per tick read of the APERF and MPERF MSRs can
be avoided, it does not make sense to keep the extra code and the resulting
runtime issues of mass IPIs around.
As a first step split out the non frequency invariance specific
initialization code and the read MSR portion of arch_scale_freq_tick(). The
rest of the code is still conditional and guarded with a static key.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Link: https://lore.kernel.org/r/20220415161206.761988704@linutronix.de
|
|
AMD boot CPU initialization happens late via ACPI/CPPC which prevents the
Intel parts from being marked __init.
Split out the common code and provide a dedicated interface for the AMD
initialization and mark the Intel specific code and data __init.
The remaining text size is almost cut in half:
text: 2614 -> 1350
init.text: 0 -> 786
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Link: https://lore.kernel.org/r/20220415161206.592465719@linutronix.de
|
|
This code is convoluted and because it can be invoked post init via the
ACPI/CPPC code, all of the initialization functionality is built in instead
of being part of init text and init data.
As a first step create separate calls for the boot and the application
processors.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Link: https://lore.kernel.org/r/20220415161206.536733494@linutronix.de
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull ACPI updates from Rafael Wysocki:
"From the new functionality perspective, the most significant items
here are the new driver for the 'ARM Generic Diagnostic Dump and
Reset' device, the extension of fine grain fan control in the ACPI fan
driver, and the change making it possible to use CPPC information to
obtain CPU capacity.
There are also a few new quirks, a bunch of fixes, including the
platform-level _OSC handling change to make it actually take the
platform firmware response into account, some code and documentation
cleanups, and a notable update of the ACPI device enumeration
documentation.
Specifics:
- Use uintptr_t and offsetof() in the ACPICA code to avoid compiler
warnings regarding NULL pointer arithmetic (Rafael Wysocki).
- Fix possible NULL pointer dereference in acpi_ns_walk_namespace()
when passed "acpi=off" in the command line (Rafael Wysocki).
- Fix and clean up acpi_os_read/write_port() (Rafael Wysocki).
- Introduce acpi_bus_for_each_dev() and use it for walking all ACPI
device objects in the Type C code (Rafael Wysocki).
- Fix the _OSC platform capabilities negotioation and prevent CPPC
from being used if the platform firmware indicates that it not
supported via _OSC (Rafael Wysocki).
- Use ida_alloc() instead of ida_simple_get() for ACPI enumeration of
devices (Rafael Wysocki).
- Add AGDI and CEDT to the list of known ACPI table signatures (Ilkka
Koskinen, Robert Kiraly).
- Add power management debug messages related to suspend-to-idle in
two places (Rafael Wysocki).
- Fix __acpi_node_get_property_reference() return value and clean up
that function (Andy Shevchenko, Sakari Ailus).
- Fix return value of the __setup handler in the ACPI PM timer clock
source driver (Randy Dunlap).
- Clean up double words in two comments (Tom Rix).
- Add "skip i2c clients" quirks for Lenovo Yoga Tablet 1050F/L and
Nextbook Ares 8 (Hans de Goede).
- Clean up frequency invariance handling on x86 in the ACPI CPPC
library (Huang Rui).
- Work around broken XSDT on the Advantech DAC-BJ01 board (Mark
Cilissen).
- Make wakeup events checks in the ACPI EC driver more
straightforward and clean up acpi_ec_submit_event() (Rafael
Wysocki).
- Make it possible to obtain the CPU capacity with the help of CPPC
information (Ionela Voinescu).
- Improve fine grained fan control in the ACPI fan driver and
document it (Srinivas Pandruvada).
- Add device HID and quirk for Microsoft Surface Go 3 to the ACPI
battery driver (Maximilian Luz).
- Make the ACPI driver for Intel SoCs (LPSS) let the SPI driver know
the exact type of the controller (Andy Shevchenko).
- Force native backlight mode on Clevo NL5xRU and NL5xNU (Werner
Sembach).
- Fix return value of __setup handlers in the APEI code (Randy
Dunlap).
- Add Arm Generic Diagnostic Dump and Reset device driver (Ilkka
Koskinen).
- Limit printable size of BERT table data (Darren Hart).
- Fix up HEST and GHES initialization (Shuai Xue).
- Update the ACPI device enumeration documentation and unify the ASL
style in GPIO-related examples (Andy Shevchenko)"
* tag 'acpi-5.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (52 commits)
clocksource: acpi_pm: fix return value of __setup handler
ACPI: bus: Avoid using CPPC if not supported by firmware
Revert "ACPI: Pass the same capabilities to the _OSC regardless of the query flag"
ACPI: video: Force backlight native for Clevo NL5xRU and NL5xNU
arm64, topology: enable use of init_cpu_capacity_cppc()
arch_topology: obtain cpu capacity using information from CPPC
x86, ACPI: rename init_freq_invariance_cppc() to arch_init_invariance_cppc()
ACPI: AGDI: Add driver for Arm Generic Diagnostic Dump and Reset device
ACPI: tables: Add AGDI to the list of known table signatures
ACPI/APEI: Limit printable size of BERT table data
ACPI: docs: gpio-properties: Unify ASL style for GPIO examples
ACPI / x86: Work around broken XSDT on Advantech DAC-BJ01 board
ACPI: APEI: fix return value of __setup handlers
x86/ACPI: CPPC: Move init_freq_invariance_cppc() into x86 CPPC
x86: Expose init_freq_invariance() to topology header
x86/ACPI: CPPC: Move AMD maximum frequency ratio setting function into x86 CPPC
x86/ACPI: CPPC: Rename cppc_msr.c to cppc.c
ACPI / x86: Add skip i2c clients quirk for Lenovo Yoga Tablet 1050F/L
ACPI / x86: Add skip i2c clients quirk for Nextbook Ares 8
ACPICA: Avoid walking the ACPI Namespace if it is not there
...
|
|
init_freq_invariance_cppc() was called in acpi_cppc_processor_probe(),
after CPU performance information and controls were populated from the
per-cpu _CPC objects.
But these _CPC objects provide information that helps with both CPU
(u-arch) and frequency invariance. Therefore, change the function name
to a more generic one, while adding the arch_ prefix, as this function
is expected to be defined differently by different architectures.
Signed-off-by: Ionela Voinescu <ionela.voinescu@arm.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
The init_freq_invariance_cppc code actually doesn't need the SMP
functionality. So setting the CONFIG_SMP as the check condition for
init_freq_invariance_cppc may cause the confusion to misunderstand the
CPPC. And the x86 CPPC file is better space to store the CPPC related
functions, while the init_freq_invariance_cppc is out of smpboot, that
means, the CONFIG_SMP won't be mandatory condition any more. And It's more
clear than before.
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ rjw: Subject adjustment ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
The function init_freq_invariance will be used on x86 CPPC, so expose it in
the topology header.
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ rjw: Subject adjustment ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
The AMD maximum frequency ratio setting function depends on CPPC, so the
x86 CPPC implementation file is better space for this function.
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ rjw: Subject adjustment ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
PPIN is the Protected Processor Identification Number.
This is used to identify the socket as a Field Replaceable Unit (FRU).
Existing code only displays this when reporting errors. But this makes
it inconvenient for large clusters to use it for its intended purpose
of inventory control.
Add ppin to /sys/devices/system/cpu/cpu*/topology to make what
is already available using RDMSR more easily accessible. Make
the file read only for root in case there are still people
concerned about making a unique system "serial number" available.
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://lore.kernel.org/r/20220131230111.2004669-6-tony.luck@intel.com
|
|
The init_freq_invariance_cppc function is implemented in smpboot and depends on
CONFIG_SMP.
MODPOST vmlinux.symvers
MODINFO modules.builtin.modinfo
GEN modules.builtin
LD .tmp_vmlinux.kallsyms1
ld: drivers/acpi/cppc_acpi.o: in function `acpi_cppc_processor_probe':
/home/ray/brahma3/linux/drivers/acpi/cppc_acpi.c:819: undefined reference to `init_freq_invariance_cppc'
make: *** [Makefile:1161: vmlinux] Error 1
See https://lore.kernel.org/lkml/484af487-7511-647e-5c5b-33d4429acdec@infradead.org/.
Fixes: 41ea667227ba ("x86, sched: Calculate frequency invariance for AMD systems")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ rjw: Subject edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
There are x86 CPU architectures (e.g. Jacobsville) where L2 cahce is
shared among a cluster of cores instead of being exclusive to one
single core.
To prevent oversubscription of L2 cache, load should be balanced
between such L2 clusters, especially for tasks with no shared data.
On benchmark such as SPECrate mcf test, this change provides a boost
to performance especially on medium load system on Jacobsville. on a
Jacobsville that has 24 Atom cores, arranged into 6 clusters of 4
cores each, the benchmark number is as follow:
Improvement over baseline kernel for mcf_r
copies run time base rate
1 -0.1% -0.2%
6 25.1% 25.1%
12 18.8% 19.0%
24 0.3% 0.3%
So this looks pretty good. In terms of the system's task distribution,
some pretty bad clumping can be seen for the vanilla kernel without
the L2 cluster domain for the 6 and 12 copies case. With the extra
domain for cluster, the load does get evened out between the clusters.
Note this patch isn't an universal win as spreading isn't necessarily
a win, particually for those workload who can benefit from packing.
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20210924085104.44806-4-21cnbao@gmail.com
|
|
Move it outside of CONFIG_SMP in order to avoid ifdeffery at the usage
sites.
Fixes: 76e2fc63ca40 ("x86/cpu/amd: Set __max_die_per_package on AMD")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lkml.kernel.org/r/20210114111814.5346-1-bp@alien8.de
|
|
This is the first pass in creating the ability to calculate the
frequency invariance on AMD systems. This approach uses the CPPC
highest performance and nominal performance values that range from
0 - 255 instead of a high and base frquency. This is because we do
not have the ability on AMD to get a highest frequency value.
On AMD systems the highest performance and nominal performance
vaues do correspond to the highest and base frequencies for the system
so using them should produce an appropriate ratio but some tweaking
is likely necessary.
Due to CPPC being initialized later in boot than when the frequency
invariant calculation is currently made, I had to create a callback
from the CPPC init code to do the calculation after we have CPPC
data.
Special thanks to "kernel test robot <lkp@intel.com>" for reporting that
compilation of drivers/acpi/cppc_acpi.c is conditional to
CONFIG_ACPI_CPPC_LIB, not just CONFIG_ACPI.
[ ggherdovich@suse.cz: made safe under CPU hotplug, edited changelog. ]
Signed-off-by: Nathan Fontenot <nathan.fontenot@amd.com>
Signed-off-by: Giovanni Gherdovich <ggherdovich@suse.cz>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lkml.kernel.org/r/20201112182614.10700-2-ggherdovich@suse.cz
|
|
The product mcnt * arch_max_freq_ratio can overflows u64.
For context, a large value for arch_max_freq_ratio would be 5000,
corresponding to a turbo_freq/base_freq ratio of 5 (normally it's more like
1500-2000). A large increment frequency for the MPERF counter would be 5GHz
(the base clock of all CPUs on the market today is less than that). With
these figures, a CPU would need to go without a scheduler tick for around 8
days for the u64 overflow to happen. It is unlikely, but the check is
warranted.
Under similar conditions, the difference acnt of two consecutive APERF
readings can overflow as well.
In these circumstances is appropriate to disable frequency invariant
accounting: the feature relies on measures of the clock frequency done at
every scheduler tick, which need to be "fresh" to be at all meaningful.
A note on i386: prior to version 5.1, the GCC compiler didn't have the
builtin function __builtin_mul_overflow. In these GCC versions the macro
check_mul_overflow needs __udivdi3() to do (u64)a/b, which the kernel
doesn't provide. For this reason this change fails to build on i386 if
GCC<5.1, and we protect the entire frequency invariant code behind
CONFIG_X86_64 (special thanks to "kbuild test robot" <lkp@intel.com>).
Fixes: 1567c3e3467c ("x86, sched: Add support for frequency invariance")
Signed-off-by: Giovanni Gherdovich <ggherdovich@suse.cz>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lkml.kernel.org/r/20200531182453.15254-2-ggherdovich@suse.cz
|
|
invariance
On some platforms such as the Dell XPS 13 laptop the firmware disables turbo
when the machine is disconnected from AC, and viceversa it enables it again
when it's reconnected. In these cases a _PPC ACPI notification is issued.
The scheduler needs to know freq_max for frequency-invariant calculations.
To account for turbo availability to come and go, record freq_max at boot as
if turbo was available and store it in a helper variable. Use a setter
function to swap between freq_base and freq_max every time turbo goes off or on.
Signed-off-by: Giovanni Gherdovich <ggherdovich@suse.cz>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lkml.kernel.org/r/20200122151617.531-7-ggherdovich@suse.cz
|
|
Implement arch_scale_freq_capacity() for 'modern' x86. This function
is used by the scheduler to correctly account usage in the face of
DVFS.
The present patch addresses Intel processors specifically and has positive
performance and performance-per-watt implications for the schedutil cpufreq
governor, bringing it closer to, if not on-par with, the powersave governor
from the intel_pstate driver/framework.
Large performance gains are obtained when the machine is lightly loaded and
no regression are observed at saturation. The benchmarks with the largest
gains are kernel compilation, tbench (the networking version of dbench) and
shell-intensive workloads.
1. FREQUENCY INVARIANCE: MOTIVATION
* Without it, a task looks larger if the CPU runs slower
2. PECULIARITIES OF X86
* freq invariance accounting requires knowing the ratio freq_curr/freq_max
2.1 CURRENT FREQUENCY
* Use delta_APERF / delta_MPERF * freq_base (a.k.a "BusyMHz")
2.2 MAX FREQUENCY
* It varies with time (turbo). As an approximation, we set it to a
constant, i.e. 4-cores turbo frequency.
3. EFFECTS ON THE SCHEDUTIL FREQUENCY GOVERNOR
* The invariant schedutil's formula has no feedback loop and reacts faster
to utilization changes
4. KNOWN LIMITATIONS
* In some cases tasks can't reach max util despite how hard they try
5. PERFORMANCE TESTING
5.1 MACHINES
* Skylake, Broadwell, Haswell
5.2 SETUP
* baseline Linux v5.2 w/ non-invariant schedutil. Tested freq_max = 1-2-3-4-8-12
active cores turbo w/ invariant schedutil, and intel_pstate/powersave
5.3 BENCHMARK RESULTS
5.3.1 NEUTRAL BENCHMARKS
* NAS Parallel Benchmark (HPC), hackbench
5.3.2 NON-NEUTRAL BENCHMARKS
* tbench (10-30% better), kernbench (10-15% better),
shell-intensive-scripts (30-50% better)
* no regressions
5.3.3 SELECTION OF DETAILED RESULTS
5.3.4 POWER CONSUMPTION, PERFORMANCE-PER-WATT
* dbench (5% worse on one machine), kernbench (3% worse),
tbench (5-10% better), shell-intensive-scripts (10-40% better)
6. MICROARCH'ES ADDRESSED HERE
* Xeon Core before Scalable Performance processors line (Xeon Gold/Platinum
etc have different MSRs semantic for querying turbo levels)
7. REFERENCES
* MMTests performance testing framework, github.com/gormanm/mmtests
+-------------------------------------------------------------------------+
| 1. FREQUENCY INVARIANCE: MOTIVATION
+-------------------------------------------------------------------------+
For example; suppose a CPU has two frequencies: 500 and 1000 Mhz. When
running a task that would consume 1/3rd of a CPU at 1000 MHz, it would
appear to consume 2/3rd (or 66.6%) when running at 500 MHz, giving the
false impression this CPU is almost at capacity, even though it can go
faster [*]. In a nutshell, without frequency scale-invariance tasks look
larger just because the CPU is running slower.
[*] (footnote: this assumes a linear frequency/performance relation; which
everybody knows to be false, but given realities its the best approximation
we can make.)
+-------------------------------------------------------------------------+
| 2. PECULIARITIES OF X86
+-------------------------------------------------------------------------+
Accounting for frequency changes in PELT signals requires the computation of
the ratio freq_curr / freq_max. On x86 neither of those terms is readily
available.
2.1 CURRENT FREQUENCY
====================
Since modern x86 has hardware control over the actual frequency we run
at (because amongst other things, Turbo-Mode), we cannot simply use
the frequency as requested through cpufreq.
Instead we use the APERF/MPERF MSRs to compute the effective frequency
over the recent past. Also, because reading MSRs is expensive, don't
do so every time we need the value, but amortize the cost by doing it
every tick.
2.2 MAX FREQUENCY
=================
Obtaining freq_max is also non-trivial because at any time the hardware can
provide a frequency boost to a selected subset of cores if the package has
enough power to spare (eg: Turbo Boost). This means that the maximum frequency
available to a given core changes with time.
The approach taken in this change is to arbitrarily set freq_max to a constant
value at boot. The value chosen is the "4-cores (4C) turbo frequency" on most
microarchitectures, after evaluating the following candidates:
* 1-core (1C) turbo frequency (the fastest turbo state available)
* around base frequency (a.k.a. max P-state)
* something in between, such as 4C turbo
To interpret these options, consider that this is the denominator in
freq_curr/freq_max, and that ratio will be used to scale PELT signals such as
util_avg and load_avg. A large denominator will undershoot (util_avg looks a
bit smaller than it really is), viceversa with a smaller denominator PELT
signals will tend to overshoot. Given that PELT drives frequency selection
in the schedutil governor, we will have:
freq_max set to | effect on DVFS
--------------------+------------------
1C turbo | power efficiency (lower freq choices)
base freq | performance (higher util_avg, higher freq requests)
4C turbo | a bit of both
4C turbo proves to be a good compromise in a number of benchmarks (see below).
+-------------------------------------------------------------------------+
| 3. EFFECTS ON THE SCHEDUTIL FREQUENCY GOVERNOR
+-------------------------------------------------------------------------+
Once an architecture implements a frequency scale-invariant utilization (the
PELT signal util_avg), schedutil switches its frequency selection formula from
freq_next = 1.25 * freq_curr * util [non-invariant util signal]
to
freq_next = 1.25 * freq_max * util [invariant util signal]
where, in the second formula, freq_max is set to the 1C turbo frequency (max
turbo). The advantage of the second formula, whose usage we unlock with this
patch, is that freq_next doesn't depend on the current frequency in an
iterative fashion, but can jump to any frequency in a single update. This
absence of feedback in the formula makes it quicker to react to utilization
changes and more robust against pathological instabilities.
Compare it to the update formula of intel_pstate/powersave:
freq_next = 1.25 * freq_max * Busy%
where again freq_max is 1C turbo and Busy% is the percentage of time not spent
idling (calculated with delta_MPERF / delta_TSC); essentially the same as
invariant schedutil, and largely responsible for intel_pstate/powersave good
reputation. The non-invariant schedutil formula is derived from the invariant
one by approximating util_inv with util_raw * freq_curr / freq_max, but this
has limitations.
Testing shows improved performances due to better frequency selections when
the machine is lightly loaded, and essentially no change in behaviour at
saturation / overutilization.
+-------------------------------------------------------------------------+
| 4. KNOWN LIMITATIONS
+-------------------------------------------------------------------------+
It's been shown that it is possible to create pathological scenarios where a
CPU-bound task cannot reach max utilization, if the normalizing factor
freq_max is fixed to a constant value (see [Lelli-2018]).
If freq_max is set to 4C turbo as we do here, one needs to peg at least 5
cores in a package doing some busywork, and observe that none of those task
will ever reach max util (1024) because they're all running at less than the
4C turbo frequency.
While this concern still applies, we believe the performance benefit of
frequency scale-invariant PELT signals outweights the cost of this limitation.
[Lelli-2018]
https://lore.kernel.org/lkml/20180517150418.GF22493@localhost.localdomain/
+-------------------------------------------------------------------------+
| 5. PERFORMANCE TESTING
+-------------------------------------------------------------------------+
5.1 MACHINES
============
We tested the patch on three machines, with Skylake, Broadwell and Haswell
CPUs. The details are below, together with the available turbo ratios as
reported by the appropriate MSRs.
* 8x-SKYLAKE-UMA:
Single socket E3-1240 v5, Skylake 4 cores/8 threads
Max EFFiciency, BASE frequency and available turbo levels (MHz):
EFFIC 800 |********
BASE 3500 |***********************************
4C 3700 |*************************************
3C 3800 |**************************************
2C 3900 |***************************************
1C 3900 |***************************************
* 80x-BROADWELL-NUMA:
Two sockets E5-2698 v4, 2x Broadwell 20 cores/40 threads
Max EFFiciency, BASE frequency and available turbo levels (MHz):
EFFIC 1200 |************
BASE 2200 |**********************
8C 2900 |*****************************
7C 3000 |******************************
6C 3100 |*******************************
5C 3200 |********************************
4C 3300 |*********************************
3C 3400 |**********************************
2C 3600 |************************************
1C 3600 |************************************
* 48x-HASWELL-NUMA
Two sockets E5-2670 v3, 2x Haswell 12 cores/24 threads
Max EFFiciency, BASE frequency and available turbo levels (MHz):
EFFIC 1200 |************
BASE 2300 |***********************
12C 2600 |**************************
11C 2600 |**************************
10C 2600 |**************************
9C 2600 |**************************
8C 2600 |**************************
7C 2600 |**************************
6C 2600 |**************************
5C 2700 |***************************
4C 2800 |****************************
3C 2900 |*****************************
2C 3100 |*******************************
1C 3100 |*******************************
5.2 SETUP
=========
* The baseline is Linux v5.2 with schedutil (non-invariant) and the intel_pstate
driver in passive mode.
* The rationale for choosing the various freq_max values to test have been to
try all the 1-2-3-4C turbo levels (note that 1C and 2C turbo are identical
on all machines), plus one more value closer to base_freq but still in the
turbo range (8C turbo for both 80x-BROADWELL-NUMA and 48x-HASWELL-NUMA).
* In addition we've run all tests with intel_pstate/powersave for comparison.
* The filesystem is always XFS, the userspace is openSUSE Leap 15.1.
* 8x-SKYLAKE-UMA is capable of HWP (Hardware-Managed P-States), so the runs
with active intel_pstate on this machine use that.
This gives, in terms of combinations tested on each machine:
* 8x-SKYLAKE-UMA
* Baseline: Linux v5.2, non-invariant schedutil, intel_pstate passive
* intel_pstate active + powersave + HWP
* invariant schedutil, freq_max = 1C turbo
* invariant schedutil, freq_max = 3C turbo
* invariant schedutil, freq_max = 4C turbo
* both 80x-BROADWELL-NUMA and 48x-HASWELL-NUMA
* [same as 8x-SKYLAKE-UMA, but no HWP capable]
* invariant schedutil, freq_max = 8C turbo
(which on 48x-HASWELL-NUMA is the same as 12C turbo, or "all cores turbo")
5.3 BENCHMARK RESULTS
=====================
5.3.1 NEUTRAL BENCHMARKS
------------------------
Tests that didn't show any measurable difference in performance on any of the
test machines between non-invariant schedutil and our patch are:
* NAS Parallel Benchmarks (NPB) using either MPI or openMP for IPC, any
computational kernel
* flexible I/O (FIO)
* hackbench (using threads or processes, and using pipes or sockets)
5.3.2 NON-NEUTRAL BENCHMARKS
----------------------------
What follow are summary tables where each benchmark result is given a score.
* A tilde (~) means a neutral result, i.e. no difference from baseline.
* Scores are computed with the ratio result_new / result_baseline, so a tilde
means a score of 1.00.
* The results in the score ratio are the geometric means of results running
the benchmark with different parameters (eg: for kernbench: using 1, 2, 4,
... number of processes; for pgbench: varying the number of clients, and so
on).
* The first three tables show higher-is-better kind of tests (i.e. measured in
operations/second), the subsequent three show lower-is-better kind of tests
(i.e. the workload is fixed and we measure elapsed time, think kernbench).
* "gitsource" is a name we made up for the test consisting in running the
entire unit tests suite of the Git SCM and measuring how long it takes. We
take it as a typical example of shell-intensive serialized workload.
* In the "I_PSTATE" column we have the results for intel_pstate/powersave. Other
columns show invariant schedutil for different values of freq_max. 4C turbo
is circled as it's the value we've chosen for the final implementation.
80x-BROADWELL-NUMA (comparison ratio; higher is better)
+------+
I_PSTATE 1C 3C | 4C | 8C
pgbench-ro 1.14 ~ ~ | 1.11 | 1.14
pgbench-rw ~ ~ ~ | ~ | ~
netperf-udp 1.06 ~ 1.06 | 1.05 | 1.07
netperf-tcp ~ 1.03 ~ | 1.01 | 1.02
tbench4 1.57 1.18 1.22 | 1.30 | 1.56
+------+
8x-SKYLAKE-UMA (comparison ratio; higher is better)
+------+
I_PSTATE/HWP 1C 3C | 4C |
pgbench-ro ~ ~ ~ | ~ |
pgbench-rw ~ ~ ~ | ~ |
netperf-udp ~ ~ ~ | ~ |
netperf-tcp ~ ~ ~ | ~ |
tbench4 1.30 1.14 1.14 | 1.16 |
+------+
48x-HASWELL-NUMA (comparison ratio; higher is better)
+------+
I_PSTATE 1C 3C | 4C | 12C
pgbench-ro 1.15 ~ ~ | 1.06 | 1.16
pgbench-rw ~ ~ ~ | ~ | ~
netperf-udp 1.05 0.97 1.04 | 1.04 | 1.02
netperf-tcp 0.96 1.01 1.01 | 1.01 | 1.01
tbench4 1.50 1.05 1.13 | 1.13 | 1.25
+------+
In the table above we see that active intel_pstate is slightly better than our
4C-turbo patch (both in reference to the baseline non-invariant schedutil) on
read-only pgbench and much better on tbench. Both cases are notable in which
it shows that lowering our freq_max (to 8C-turbo and 12C-turbo on
80x-BROADWELL-NUMA and 48x-HASWELL-NUMA respectively) helps invariant
schedutil to get closer.
If we ignore active intel_pstate and focus on the comparison with baseline
alone, there are several instances of double-digit performance improvement.
80x-BROADWELL-NUMA (comparison ratio; lower is better)
+------+
I_PSTATE 1C 3C | 4C | 8C
dbench4 1.23 0.95 0.95 | 0.95 | 0.95
kernbench 0.93 0.83 0.83 | 0.83 | 0.82
gitsource 0.98 0.49 0.49 | 0.49 | 0.48
+------+
8x-SKYLAKE-UMA (comparison ratio; lower is better)
+------+
I_PSTATE/HWP 1C 3C | 4C |
dbench4 ~ ~ ~ | ~ |
kernbench ~ ~ ~ | ~ |
gitsource 0.92 0.55 0.55 | 0.55 |
+------+
48x-HASWELL-NUMA (comparison ratio; lower is better)
+------+
I_PSTATE 1C 3C | 4C | 8C
dbench4 ~ ~ ~ | ~ | ~
kernbench 0.94 0.90 0.89 | 0.90 | 0.90
gitsource 0.97 0.69 0.69 | 0.69 | 0.69
+------+
dbench is not very remarkable here, unless we notice how poorly active
intel_pstate is performing on 80x-BROADWELL-NUMA: 23% regression versus
non-invariant schedutil. We repeated that run getting consistent results. Out
of scope for the patch at hand, but deserving future investigation. Other than
that, we previously ran this campaign with Linux v5.0 and saw the patch doing
better on dbench a the time. We haven't checked closely and can only speculate
at this point.
On the NUMA boxes kernbench gets 10-15% improvements on average; we'll see in
the detailed tables that the gains concentrate on low process counts (lightly
loaded machines).
The test we call "gitsource" (running the git unit test suite, a long-running
single-threaded shell script) appears rather spectacular in this table (gains
of 30-50% depending on the machine). It is to be noted, however, that
gitsource has no adjustable parameters (such as the number of jobs in
kernbench, which we average over in order to get a single-number summary
score) and is exactly the kind of low-parallelism workload that benefits the
most from this patch. When looking at the detailed tables of kernbench or
tbench4, at low process or client counts one can see similar numbers.
5.3.3 SELECTION OF DETAILED RESULTS
-----------------------------------
Machine : 48x-HASWELL-NUMA
Benchmark : tbench4 (i.e. dbench4 over the network, actually loopback)
Varying parameter : number of clients
Unit : MB/sec (higher is better)
5.2.0 vanilla (BASELINE) 5.2.0 intel_pstate 5.2.0 1C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Hmean 1 126.73 +- 0.31% ( ) 315.91 +- 0.66% ( 149.28%) 125.03 +- 0.76% ( -1.34%)
Hmean 2 258.04 +- 0.62% ( ) 614.16 +- 0.51% ( 138.01%) 269.58 +- 1.45% ( 4.47%)
Hmean 4 514.30 +- 0.67% ( ) 1146.58 +- 0.54% ( 122.94%) 533.84 +- 1.99% ( 3.80%)
Hmean 8 1111.38 +- 2.52% ( ) 2159.78 +- 0.38% ( 94.33%) 1359.92 +- 1.56% ( 22.36%)
Hmean 16 2286.47 +- 1.36% ( ) 3338.29 +- 0.21% ( 46.00%) 2720.20 +- 0.52% ( 18.97%)
Hmean 32 4704.84 +- 0.35% ( ) 4759.03 +- 0.43% ( 1.15%) 4774.48 +- 0.30% ( 1.48%)
Hmean 64 7578.04 +- 0.27% ( ) 7533.70 +- 0.43% ( -0.59%) 7462.17 +- 0.65% ( -1.53%)
Hmean 128 6998.52 +- 0.16% ( ) 6987.59 +- 0.12% ( -0.16%) 6909.17 +- 0.14% ( -1.28%)
Hmean 192 6901.35 +- 0.25% ( ) 6913.16 +- 0.10% ( 0.17%) 6855.47 +- 0.21% ( -0.66%)
5.2.0 3C-turbo 5.2.0 4C-turbo 5.2.0 12C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Hmean 1 128.43 +- 0.28% ( 1.34%) 130.64 +- 3.81% ( 3.09%) 153.71 +- 5.89% ( 21.30%)
Hmean 2 311.70 +- 6.15% ( 20.79%) 281.66 +- 3.40% ( 9.15%) 305.08 +- 5.70% ( 18.23%)
Hmean 4 641.98 +- 2.32% ( 24.83%) 623.88 +- 5.28% ( 21.31%) 906.84 +- 4.65% ( 76.32%)
Hmean 8 1633.31 +- 1.56% ( 46.96%) 1714.16 +- 0.93% ( 54.24%) 2095.74 +- 0.47% ( 88.57%)
Hmean 16 3047.24 +- 0.42% ( 33.27%) 3155.02 +- 0.30% ( 37.99%) 3634.58 +- 0.15% ( 58.96%)
Hmean 32 4734.31 +- 0.60% ( 0.63%) 4804.38 +- 0.23% ( 2.12%) 4674.62 +- 0.27% ( -0.64%)
Hmean 64 7699.74 +- 0.35% ( 1.61%) 7499.72 +- 0.34% ( -1.03%) 7659.03 +- 0.25% ( 1.07%)
Hmean 128 6935.18 +- 0.15% ( -0.91%) 6942.54 +- 0.10% ( -0.80%) 7004.85 +- 0.12% ( 0.09%)
Hmean 192 6901.62 +- 0.12% ( 0.00%) 6856.93 +- 0.10% ( -0.64%) 6978.74 +- 0.10% ( 1.12%)
This is one of the cases where the patch still can't surpass active
intel_pstate, not even when freq_max is as low as 12C-turbo. Otherwise, gains are
visible up to 16 clients and the saturated scenario is the same as baseline.
The scores in the summary table from the previous sections are ratios of
geometric means of the results over different clients, as seen in this table.
Machine : 80x-BROADWELL-NUMA
Benchmark : kernbench (kernel compilation)
Varying parameter : number of jobs
Unit : seconds (lower is better)
5.2.0 vanilla (BASELINE) 5.2.0 intel_pstate 5.2.0 1C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Amean 2 379.68 +- 0.06% ( ) 330.20 +- 0.43% ( 13.03%) 285.93 +- 0.07% ( 24.69%)
Amean 4 200.15 +- 0.24% ( ) 175.89 +- 0.22% ( 12.12%) 153.78 +- 0.25% ( 23.17%)
Amean 8 106.20 +- 0.31% ( ) 95.54 +- 0.23% ( 10.03%) 86.74 +- 0.10% ( 18.32%)
Amean 16 56.96 +- 1.31% ( ) 53.25 +- 1.22% ( 6.50%) 48.34 +- 1.73% ( 15.13%)
Amean 32 34.80 +- 2.46% ( ) 33.81 +- 0.77% ( 2.83%) 30.28 +- 1.59% ( 12.99%)
Amean 64 26.11 +- 1.63% ( ) 25.04 +- 1.07% ( 4.10%) 22.41 +- 2.37% ( 14.16%)
Amean 128 24.80 +- 1.36% ( ) 23.57 +- 1.23% ( 4.93%) 21.44 +- 1.37% ( 13.55%)
Amean 160 24.85 +- 0.56% ( ) 23.85 +- 1.17% ( 4.06%) 21.25 +- 1.12% ( 14.49%)
5.2.0 3C-turbo 5.2.0 4C-turbo 5.2.0 8C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Amean 2 284.08 +- 0.13% ( 25.18%) 283.96 +- 0.51% ( 25.21%) 285.05 +- 0.21% ( 24.92%)
Amean 4 153.18 +- 0.22% ( 23.47%) 154.70 +- 1.64% ( 22.71%) 153.64 +- 0.30% ( 23.24%)
Amean 8 87.06 +- 0.28% ( 18.02%) 86.77 +- 0.46% ( 18.29%) 86.78 +- 0.22% ( 18.28%)
Amean 16 48.03 +- 0.93% ( 15.68%) 47.75 +- 1.99% ( 16.17%) 47.52 +- 1.61% ( 16.57%)
Amean 32 30.23 +- 1.20% ( 13.14%) 30.08 +- 1.67% ( 13.57%) 30.07 +- 1.67% ( 13.60%)
Amean 64 22.59 +- 2.02% ( 13.50%) 22.63 +- 0.81% ( 13.32%) 22.42 +- 0.76% ( 14.12%)
Amean 128 21.37 +- 0.67% ( 13.82%) 21.31 +- 1.15% ( 14.07%) 21.17 +- 1.93% ( 14.63%)
Amean 160 21.68 +- 0.57% ( 12.76%) 21.18 +- 1.74% ( 14.77%) 21.22 +- 1.00% ( 14.61%)
The patch outperform active intel_pstate (and baseline) by a considerable
margin; the summary table from the previous section says 4C turbo and active
intel_pstate are 0.83 and 0.93 against baseline respectively, so 4C turbo is
0.83/0.93=0.89 against intel_pstate (~10% better on average). There is no
noticeable difference with regard to the value of freq_max.
Machine : 8x-SKYLAKE-UMA
Benchmark : gitsource (time to run the git unit test suite)
Varying parameter : none
Unit : seconds (lower is better)
5.2.0 vanilla 5.2.0 intel_pstate/hwp 5.2.0 1C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Amean 858.85 +- 1.16% ( ) 791.94 +- 0.21% ( 7.79%) 474.95 ( 44.70%)
5.2.0 3C-turbo 5.2.0 4C-turbo
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Amean 475.26 +- 0.20% ( 44.66%) 474.34 +- 0.13% ( 44.77%)
In this test, which is of interest as representing shell-intensive
(i.e. fork-intensive) serialized workloads, invariant schedutil outperforms
intel_pstate/powersave by a whopping 40% margin.
5.3.4 POWER CONSUMPTION, PERFORMANCE-PER-WATT
---------------------------------------------
The following table shows average power consumption in watt for each
benchmark. Data comes from turbostat (package average), which in turn is read
from the RAPL interface on CPUs. We know the patch affects CPU frequencies so
it's reasonable to ignore other power consumers (such as memory or I/O). Also,
we don't have a power meter available in the lab so RAPL is the best we have.
turbostat sampled average power every 10 seconds for the entire duration of
each benchmark. We took all those values and averaged them (i.e. with don't
have detail on a per-parameter granularity, only on whole benchmarks).
80x-BROADWELL-NUMA (power consumption, watts)
+--------+
BASELINE I_PSTATE 1C 3C | 4C | 8C
pgbench-ro 130.01 142.77 131.11 132.45 | 134.65 | 136.84
pgbench-rw 68.30 60.83 71.45 71.70 | 71.65 | 72.54
dbench4 90.25 59.06 101.43 99.89 | 101.10 | 102.94
netperf-udp 65.70 69.81 66.02 68.03 | 68.27 | 68.95
netperf-tcp 88.08 87.96 88.97 88.89 | 88.85 | 88.20
tbench4 142.32 176.73 153.02 163.91 | 165.58 | 176.07
kernbench 92.94 101.95 114.91 115.47 | 115.52 | 115.10
gitsource 40.92 41.87 75.14 75.20 | 75.40 | 75.70
+--------+
8x-SKYLAKE-UMA (power consumption, watts)
+--------+
BASELINE I_PSTATE/HWP 1C 3C | 4C |
pgbench-ro 46.49 46.68 46.56 46.59 | 46.52 |
pgbench-rw 29.34 31.38 30.98 31.00 | 31.00 |
dbench4 27.28 27.37 27.49 27.41 | 27.38 |
netperf-udp 22.33 22.41 22.36 22.35 | 22.36 |
netperf-tcp 27.29 27.29 27.30 27.31 | 27.33 |
tbench4 41.13 45.61 43.10 43.33 | 43.56 |
kernbench 42.56 42.63 43.01 43.01 | 43.01 |
gitsource 13.32 13.69 17.33 17.30 | 17.35 |
+--------+
48x-HASWELL-NUMA (power consumption, watts)
+--------+
BASELINE I_PSTATE 1C 3C | 4C | 12C
pgbench-ro 128.84 136.04 129.87 132.43 | 132.30 | 134.86
pgbench-rw 37.68 37.92 37.17 37.74 | 37.73 | 37.31
dbench4 28.56 28.73 28.60 28.73 | 28.70 | 28.79
netperf-udp 56.70 60.44 56.79 57.42 | 57.54 | 57.52
netperf-tcp 75.49 75.27 75.87 76.02 | 76.01 | 75.95
tbench4 115.44 139.51 119.53 123.07 | 123.97 | 130.22
kernbench 83.23 91.55 95.58 95.69 | 95.72 | 96.04
gitsource 36.79 36.99 39.99 40.34 | 40.35 | 40.23
+--------+
A lower power consumption isn't necessarily better, it depends on what is done
with that energy. Here are tables with the ratio of performance-per-watt on
each machine and benchmark. Higher is always better; a tilde (~) means a
neutral ratio (i.e. 1.00).
80x-BROADWELL-NUMA (performance-per-watt ratios; higher is better)
+------+
I_PSTATE 1C 3C | 4C | 8C
pgbench-ro 1.04 1.06 0.94 | 1.07 | 1.08
pgbench-rw 1.10 0.97 0.96 | 0.96 | 0.97
dbench4 1.24 0.94 0.95 | 0.94 | 0.92
netperf-udp ~ 1.02 1.02 | ~ | 1.02
netperf-tcp ~ 1.02 ~ | ~ | 1.02
tbench4 1.26 1.10 1.06 | 1.12 | 1.26
kernbench 0.98 0.97 0.97 | 0.97 | 0.98
gitsource ~ 1.11 1.11 | 1.11 | 1.13
+------+
8x-SKYLAKE-UMA (performance-per-watt ratios; higher is better)
+------+
I_PSTATE/HWP 1C 3C | 4C |
pgbench-ro ~ ~ ~ | ~ |
pgbench-rw 0.95 0.97 0.96 | 0.96 |
dbench4 ~ ~ ~ | ~ |
netperf-udp ~ ~ ~ | ~ |
netperf-tcp ~ ~ ~ | ~ |
tbench4 1.17 1.09 1.08 | 1.10 |
kernbench ~ ~ ~ | ~ |
gitsource 1.06 1.40 1.40 | 1.40 |
+------+
48x-HASWELL-NUMA (performance-per-watt ratios; higher is better)
+------+
I_PSTATE 1C 3C | 4C | 12C
pgbench-ro 1.09 ~ 1.09 | 1.03 | 1.11
pgbench-rw ~ 0.86 ~ | ~ | 0.86
dbench4 ~ 1.02 1.02 | 1.02 | ~
netperf-udp ~ 0.97 1.03 | 1.02 | ~
netperf-tcp 0.96 ~ ~ | ~ | ~
tbench4 1.24 ~ 1.06 | 1.05 | 1.11
kernbench 0.97 0.97 0.98 | 0.97 | 0.96
gitsource 1.03 1.33 1.32 | 1.32 | 1.33
+------+
These results are overall pleasing: in plenty of cases we observe
performance-per-watt improvements. The few regressions (read/write pgbench and
dbench on the Broadwell machine) are of small magnitude. kernbench loses a few
percentage points (it has a 10-15% performance improvement, but apparently the
increase in power consumption is larger than that). tbench4 and gitsource, which
benefit the most from the patch, keep a positive score in this table which is
a welcome surprise; that suggests that in those particular workloads the
non-invariant schedutil (and active intel_pstate, too) makes some rather
suboptimal frequency selections.
+-------------------------------------------------------------------------+
| 6. MICROARCH'ES ADDRESSED HERE
+-------------------------------------------------------------------------+
The patch addresses Xeon Core processors that use MSR_PLATFORM_INFO and
MSR_TURBO_RATIO_LIMIT to advertise their base frequency and turbo frequencies
respectively. This excludes the recent Xeon Scalable Performance processors
line (Xeon Gold, Platinum etc) whose MSRs have to be parsed differently.
Subsequent patches will address:
* Xeon Scalable Performance processors and Atom Goldmont/Goldmont Plus
* Xeon Phi (Knights Landing, Knights Mill)
* Atom Silvermont
+-------------------------------------------------------------------------+
| 7. REFERENCES
+-------------------------------------------------------------------------+
Tests have been run with the help of the MMTests performance testing
framework, see github.com/gormanm/mmtests. The configuration file names for
the benchmark used are:
db-pgbench-timed-ro-small-xfs
db-pgbench-timed-rw-small-xfs
io-dbench4-async-xfs
network-netperf-unbound
network-tbench
scheduler-unbound
workload-kerndevel-xfs
workload-shellscripts-xfs
hpc-nas-c-class-mpi-full-xfs
hpc-nas-c-class-omp-full
All those benchmarks are generally available on the web:
pgbench: https://www.postgresql.org/docs/10/pgbench.html
netperf: https://hewlettpackard.github.io/netperf/
dbench/tbench: https://dbench.samba.org/
gitsource: git unit test suite, github.com/git/git
NAS Parallel Benchmarks: https://www.nas.nasa.gov/publications/npb.html
hackbench: https://people.redhat.com/mingo/cfs-scheduler/tools/hackbench.c
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Giovanni Gherdovich <ggherdovich@suse.cz>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Doug Smythies <dsmythies@telus.net>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lkml.kernel.org/r/20200122151617.531-2-ggherdovich@suse.cz
|
|
Create CPU topology sysfs attributes: "core_cpus" and "core_cpus_list"
These attributes represent all of the logical CPUs that share the
same core.
These attriutes is synonymous with the existing "thread_siblings" and
"thread_siblings_list" attribute, which will be deprecated.
Create CPU topology sysfs attributes: "die_cpus" and "die_cpus_list".
These attributes represent all of the logical CPUs that share the
same die.
Suggested-by: Brice Goglin <Brice.Goglin@inria.fr>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/071c23a298cd27ede6ed0b6460cae190d193364f.1557769318.git.len.brown@intel.com
|
|
Define topology_logical_die_id() ala existing topology_logical_package_id()
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/2f3526e25ae14fbeff26fb26e877d159df8946d9.1557769318.git.len.brown@intel.com
|
