-/* Copyright (c) 2006-2020. The SimGrid Team. All rights reserved. */
+/* Copyright (c) 2006-2023. The SimGrid Team. All rights reserved. */
/* This program is free software; you can redistribute it and/or modify it
* under the terms of the license (GNU LGPL) which comes with this package. */
#define SIMGRID_S4U_HOST_HPP
#include <simgrid/forward.h>
+#include <simgrid/kernel/resource/Action.hpp>
#include <xbt/Extendable.hpp>
#include <xbt/signal.hpp>
-#include <xbt/string.hpp>
#include <map>
#include <unordered_map>
* @endrst
*/
class XBT_PUBLIC Host : public xbt::Extendable<Host> {
- friend vm::VMModel; // Use the pimpl_cpu to compute the VM sharing
- friend vm::VirtualMachineImpl; // creates the the pimpl_cpu
+#ifndef DOXYGEN
+ friend kernel::resource::VMModel; // Use the pimpl_cpu to compute the VM sharing
+ friend kernel::resource::VirtualMachineImpl; // creates the the pimpl_cpu
friend kernel::routing::NetZoneImpl;
+ friend kernel::resource::HostImpl; // call destructor from private implementation
+
+ // The private implementation, that never changes
+ kernel::resource::HostImpl* const pimpl_;
+
+ kernel::resource::CpuImpl* pimpl_cpu_ = nullptr;
+ kernel::routing::NetPoint* pimpl_netpoint_ = nullptr;
+#ifndef DOXYGEN
+ friend kernel::resource::CpuAction; // signal exec_state_changed
+#endif
public:
- explicit Host(const std::string& name);
+ explicit Host(kernel::resource::HostImpl* pimpl) : pimpl_(pimpl) {}
- /** Host destruction logic */
protected:
- virtual ~Host();
- void set_netpoint(kernel::routing::NetPoint* netpoint) { pimpl_netpoint_ = netpoint; }
-
-private:
- bool currently_destroying_ = false;
+ virtual ~Host(); // Call destroy() instead of manually deleting it.
+ Host* set_netpoint(kernel::routing::NetPoint* netpoint);
-public:
- /** Called on each newly created host */
static xbt::signal<void(Host&)> on_creation;
- /** Called just before destructing a host */
static xbt::signal<void(Host const&)> on_destruction;
- /** Called when the machine is turned on or off (called AFTER the change) */
+ xbt::signal<void(Host const&)> on_this_destruction;
+ static xbt::signal<void(kernel::resource::CpuAction&, kernel::resource::Action::State)> on_exec_state_change;
+
+public:
+ static xbt::signal<void(Host const&)> on_speed_change;
+ xbt::signal<void(Host const&)> on_this_speed_change;
static xbt::signal<void(Host const&)> on_state_change;
- /** Called when the speed of the machine is changed (called AFTER the change)
+ xbt::signal<void(Host const&)> on_this_state_change;
+
+#endif
+ /** Add a callback fired on each newly created host */
+ static void on_creation_cb(const std::function<void(Host&)>& cb) { on_creation.connect(cb); }
+ /** Add a callback fired when any machine is turned on or off (called AFTER the change) */
+ static void on_state_change_cb(const std::function<void(Host const&)>& cb) { on_state_change.connect(cb); }
+ /** Add a callback fired when this specific machine is turned on or off (called AFTER the change) */
+ void on_this_state_change_cb(const std::function<void(Host const&)>& cb)
+ {
+ on_this_state_change.connect(cb);
+ }
+ /** Add a callback fired when the speed of any machine is changed (called AFTER the change)
* (either because of a pstate switch or because of an external load event coming from the profile) */
- static xbt::signal<void(Host const&)> on_speed_change;
+ static void on_speed_change_cb(const std::function<void(Host const&)>& cb) { on_speed_change.connect(cb); }
+ /** Add a callback fired when the speed of this specific machine is changed (called AFTER the change)
+ * (either because of a pstate switch or because of an external load event coming from the profile) */
+ void on_this_speed_change_cb(const std::function<void(Host const&)>& cb)
+ {
+ on_this_speed_change.connect(cb);
+ }
+ /** Add a callback fired just before destructing any host */
+ static void on_destruction_cb(const std::function<void(Host const&)>& cb) { on_destruction.connect(cb); }
+ /** Add a callback fired just before destructing this specific host */
+ void on_this_destruction_cb(const std::function<void(Host const&)>& cb)
+ {
+ on_this_destruction.connect(cb);
+ }
+ /** Add a callback fired when the state of any exec activity changes */
+ static void on_exec_state_change_cb(
+ const std::function<void(kernel::resource::CpuAction&, kernel::resource::Action::State previous)>& cb)
+ {
+ on_exec_state_change.connect(cb);
+ }
virtual void destroy();
+#ifndef DOXYGEN
// No copy/move
Host(Host const&) = delete;
Host& operator=(Host const&) = delete;
+#endif
/** Retrieve a host from its name, or return nullptr */
static Host* by_name_or_null(const std::string& name);
static Host* current();
/** Retrieves the name of that host as a C++ string */
- xbt::string const& get_name() const { return name_; }
+ std::string const& get_name() const;
/** Retrieves the name of that host as a C string */
- const char* get_cname() const { return name_.c_str(); }
+ const char* get_cname() const;
+ Host* set_cpu(kernel::resource::CpuImpl* cpu);
+ kernel::resource::CpuImpl* get_cpu() const { return pimpl_cpu_; }
kernel::routing::NetPoint* get_netpoint() const { return pimpl_netpoint_; }
+ /**
+ * @brief Callback to set CPU factor
+ *
+ * This callback offers a flexible way to create variability in CPU executions
+ *
+ * @param flops Execution size in flops
+ * @return Multiply factor
+ */
+ using CpuFactorCb = double(double flops);
+ /**
+ * @brief Configure the factor callback to the CPU associated to this host
+ */
+ Host* set_factor_cb(const std::function<CpuFactorCb>& cb);
- int get_actor_count() const;
+ size_t get_actor_count() const;
std::vector<ActorPtr> get_all_actors() const;
/** Turns that host on if it was previously off
bool is_on() const;
const char* get_property(const std::string& key) const;
- void set_property(const std::string& key, const std::string& value);
+ Host* set_property(const std::string& key, const std::string& value);
const std::unordered_map<std::string, std::string>* get_properties() const;
- void set_properties(const std::map<std::string, std::string>& properties);
+ Host* set_properties(const std::unordered_map<std::string, std::string>& properties);
+
+ Host* set_state_profile(kernel::profile::Profile* p);
+ Host* set_speed_profile(kernel::profile::Profile* p);
+
+ /**
+ * @brief Set the max amount of executions that can take place on this host at the same time
+ *
+ * Use -1 to set no limit.
+ */
+ Host* set_concurrency_limit(int limit);
+ int get_concurrency_limit() const;
- void set_state_profile(kernel::profile::Profile* p);
- void set_speed_profile(kernel::profile::Profile* p);
+ /** @brief Convert the CPU's speed from string to double */
+ static std::vector<double> convert_pstate_speed_vector(const std::vector<std::string>& speed_per_state);
+ /**
+ * @brief Set the CPU's speed
+ *
+ * @param speed_per_state list of powers for this processor (default power is at index 0)
+ */
+ Host* set_pstate_speed(const std::vector<double>& speed_per_state);
+ /**
+ * @brief Set the CPU's speed (string version)
+ *
+ * @throw std::invalid_argument if speed format is incorrect.
+ */
+ Host* set_pstate_speed(const std::vector<std::string>& speed_per_state);
+ /** @brief Get the peak computing speed in flops/s at the current pstate, NOT taking the external load into account.
+ *
+ * The amount of flops per second available for computing depends on several things:
+ * - The current pstate determines the maximal peak computing speed (use
+ * @verbatim embed:rst:inline :cpp:func:`get_pstate_speed() <simgrid::s4u::Host::get_pstate_speed>` @endverbatim
+ * to retrieve the computing speed you would get at another pstate)
+ * - If you declared an external load (with
+ * @verbatim embed:rst:inline :cpp:func:`set_speed_profile() <simgrid::s4u::Host::set_speed_profile>` @endverbatim ),
+ * you must multiply the result of
+ * @verbatim embed:rst:inline :cpp:func:`get_speed() <simgrid::s4u::Host::get_speed>` @endverbatim by
+ * @verbatim embed:rst:inline :cpp:func:`get_available_speed() <simgrid::s4u::Host::get_available_speed>` @endverbatim
+ * to retrieve what a new computation would get.
+ *
+ * The remaining speed is then shared between the executions located on this host.
+ * You can retrieve the amount of tasks currently running on this host with
+ * @verbatim embed:rst:inline :cpp:func:`get_load() <simgrid::s4u::Host::get_load>` @endverbatim .
+ *
+ * The host may have multiple cores, and your executions may be able to use more than a single core.
+ *
+ * Finally, executions of priority 2 get twice the amount of flops than executions of priority 1.
+ */
double get_speed() const;
+ /** @brief Get the available speed ratio, between 0 and 1.
+ *
+ * This accounts for external load (see
+ * @verbatim embed:rst:inline :cpp:func:`set_speed_profile() <simgrid::s4u::Host::set_speed_profile>` @endverbatim ).
+ */
double get_available_speed() const;
+
+ /** Returns the number of core of the processor. */
int get_core_count() const;
- double get_load() const;
+ Host* set_core_count(int core_count);
- double get_pstate_speed(int pstate_index) const;
- int get_pstate_count() const;
- void set_pstate(int pstate_index);
- int get_pstate() const;
+ enum class SharingPolicy { NONLINEAR = 1, LINEAR = 0 };
+ /**
+ * @brief Describes how the CPU is shared between concurrent tasks
+ *
+ * Note that the NONLINEAR callback is in the critical path of the solver, so it should be fast.
+ *
+ * @param policy Sharing policy
+ * @param cb Callback for NONLINEAR policies
+ */
+ Host* set_sharing_policy(SharingPolicy policy, const s4u::NonLinearResourceCb& cb = {});
+ SharingPolicy get_sharing_policy() const;
- std::vector<Disk*> get_disks() const;
- void add_disk(Disk* disk);
- void remove_disk(const std::string& disk_name);
+ /** Returns the current computation load (in flops per second)
+ *
+ * The external load (coming from an availability trace) is not taken in account.
+ * You may also be interested in the load plugin.
+ */
+ double get_load() const;
- std::vector<const char*> get_attached_storages() const;
+ unsigned long get_pstate_count() const;
+ unsigned long get_pstate() const;
+ double get_pstate_speed(unsigned long pstate_index) const;
+ Host* set_pstate(unsigned long pstate_index);
+ Host* set_coordinates(const std::string& coords);
- /** Get an associative list [mount point]->[Storage] of all local mount points.
+ std::vector<Disk*> get_disks() const;
+ /**
+ * @brief Create and add disk in the host
*
- * This is defined in the platform file, and cannot be modified programmatically (yet).
+ * @param name Disk name
+ * @param read_bandwidth Reading speed of the disk
+ * @param write_bandwidth Writing speed of the disk
*/
- std::unordered_map<std::string, Storage*> const& get_mounted_storages();
+ Disk* create_disk(const std::string& name, double read_bandwidth, double write_bandwidth);
+ /**
+ * @brief Human-friendly version of create_disk function.
+ *
+ * @throw std::invalid_argument if read/write speeds are incorrect
+ */
+ Disk* create_disk(const std::string& name, const std::string& read_bandwidth, const std::string& write_bandwidth);
+ void add_disk(const Disk* disk);
+ void remove_disk(const std::string& disk_name);
+
+ VirtualMachine* create_vm(const std::string& name, int core_amount);
+ VirtualMachine* create_vm(const std::string& name, int core_amount, size_t ramsize);
+ /** Retrieve a VM running on this host from its name, or return nullptr */
+ VirtualMachine* vm_by_name_or_null(const std::string& name);
void route_to(const Host* dest, std::vector<Link*>& links, double* latency) const;
- void route_to(const Host* dest, std::vector<kernel::resource::LinkImpl*>& links, double* latency) const;
- void send_to(Host* dest, double byte_amount);
+ void route_to(const Host* dest, std::vector<kernel::resource::StandardLinkImpl*>& links, double* latency) const;
+ std::pair<std::vector<Link*>, double> route_to(const Host* dest) const;
- NetZone* get_englobing_zone();
+ /**
+ * @brief Seal this host
+ * No more configuration is allowed after the seal
+ */
+ Host* seal();
+
+ NetZone* get_englobing_zone() const;
/** Block the calling actor on an execution located on the called host
*
* It is not a problem if the actor is not located on the called host.
* The actor will not be migrated in this case. Such remote execution are easy in simulation.
*/
- void execute(double flops);
+ void execute(double flops) const;
/** Start an asynchronous computation on that host (possibly remote) */
- ExecPtr exec_async(double flops_amounts);
+ ExecPtr exec_init(double flops_amounts) const;
+ ExecPtr exec_async(double flops_amounts) const;
/** Block the calling actor on an execution located on the called host (with explicit priority) */
- void execute(double flops, double priority);
-
-private:
- xbt::string name_{"noname"};
- std::unordered_map<std::string, Storage*>* mounts_ = nullptr; // caching
- kernel::routing::NetPoint* pimpl_netpoint_ = nullptr;
-
-public:
-#ifndef DOXYGEN
- /** DO NOT USE DIRECTLY (@todo: these should be protected, once our code is clean) */
- kernel::resource::Cpu* pimpl_cpu = nullptr;
- // TODO, this could be a unique_ptr
- surf::HostImpl* pimpl_ = nullptr;
-#endif
+ void execute(double flops, double priority) const;
+ kernel::resource::HostImpl* get_impl() const { return pimpl_; }
};
} // namespace s4u
} // namespace simgrid