+++ /dev/null
-/* Copyright (c) 2007-2021. 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. */
-
-#include "ptask_L07.hpp"
-#include "simgrid/kernel/routing/NetZoneImpl.hpp"
-#include "simgrid/s4u/Engine.hpp"
-#include "src/kernel/EngineImpl.hpp"
-#include "src/kernel/resource/profile/Event.hpp"
-#include "surf/surf.hpp"
-#include "xbt/config.hpp"
-
-#include <unordered_set>
-
-XBT_LOG_EXTERNAL_DEFAULT_CATEGORY(res_host);
-XBT_LOG_EXTERNAL_CATEGORY(xbt_cfg);
-
-/**************************************/
-/*** Resource Creation & Destruction **/
-/**************************************/
-void surf_host_model_init_ptask_L07()
-{
- XBT_CINFO(xbt_cfg, "Switching to the L07 model to handle parallel tasks.");
-
- auto host_model = std::make_shared<simgrid::surf::HostL07Model>("Host_Ptask");
- simgrid::kernel::EngineImpl::get_instance()->add_model(host_model);
- simgrid::s4u::Engine::get_instance()->get_netzone_root()->get_impl()->set_host_model(host_model);
-}
-
-namespace simgrid {
-namespace surf {
-
-HostL07Model::HostL07Model(const std::string& name) : HostModel(name)
-{
- auto* maxmin_system = new simgrid::kernel::lmm::FairBottleneck(true /* selective update */);
- set_maxmin_system(maxmin_system);
-
- auto net_model = std::make_shared<NetworkL07Model>("Network_Ptask", this, maxmin_system);
- auto engine = simgrid::kernel::EngineImpl::get_instance();
- engine->add_model(net_model);
- simgrid::s4u::Engine::get_instance()->get_netzone_root()->get_impl()->set_network_model(net_model);
-
- auto cpu_model = std::make_shared<CpuL07Model>("Cpu_Ptask", this, maxmin_system);
- engine->add_model(cpu_model);
- simgrid::s4u::Engine::get_instance()->get_netzone_root()->get_impl()->set_cpu_pm_model(cpu_model);
-}
-
-CpuL07Model::CpuL07Model(const std::string& name, HostL07Model* hmodel, kernel::lmm::System* sys)
- : CpuModel(name), hostModel_(hmodel)
-{
- set_maxmin_system(sys);
-}
-
-CpuL07Model::~CpuL07Model()
-{
- set_maxmin_system(nullptr);
-}
-
-NetworkL07Model::NetworkL07Model(const std::string& name, HostL07Model* hmodel, kernel::lmm::System* sys)
- : NetworkModel(name), hostModel_(hmodel)
-{
- set_maxmin_system(sys);
- loopback_ = NetworkL07Model::create_link(
- "__loopback__", std::vector<double>{simgrid::config::get_value<double>("network/loopback-bw")},
- s4u::Link::SharingPolicy::FATPIPE)
- ->set_latency(simgrid::config::get_value<double>("network/loopback-lat"));
- loopback_->seal();
-}
-
-NetworkL07Model::~NetworkL07Model()
-{
- set_maxmin_system(nullptr);
-}
-
-double HostL07Model::next_occurring_event(double now)
-{
- double min = HostModel::next_occurring_event_full(now);
- for (kernel::resource::Action const& action : *get_started_action_set()) {
- const auto& net_action = static_cast<const L07Action&>(action);
- if (net_action.get_latency() > 0 && (min < 0 || net_action.get_latency() < min)) {
- min = net_action.get_latency();
- XBT_DEBUG("Updating min with %p (start %f): %f", &net_action, net_action.get_start_time(), min);
- }
- }
- XBT_DEBUG("min value: %f", min);
-
- return min;
-}
-
-void HostL07Model::update_actions_state(double /*now*/, double delta)
-{
- for (auto it = std::begin(*get_started_action_set()); it != std::end(*get_started_action_set());) {
- auto& action = static_cast<L07Action&>(*it);
- ++it; // increment iterator here since the following calls to action.finish() may invalidate it
- if (action.get_latency() > 0) {
- if (action.get_latency() > delta) {
- action.update_latency(delta, sg_surf_precision);
- } else {
- action.set_latency(0.0);
- }
- if ((action.get_latency() <= 0.0) && (action.is_suspended() == 0)) {
- action.updateBound();
- get_maxmin_system()->update_variable_penalty(action.get_variable(), 1.0);
- action.set_last_update();
- }
- }
- XBT_DEBUG("Action (%p) : remains (%g) updated by %g.", &action, action.get_remains(),
- action.get_variable()->get_value() * delta);
- action.update_remains(action.get_variable()->get_value() * delta);
- action.update_max_duration(delta);
-
- XBT_DEBUG("Action (%p) : remains (%g).", &action, action.get_remains());
-
- /* In the next if cascade, the action can be finished either because:
- * - The amount of remaining work reached 0
- * - The max duration was reached
- * If it's not done, it may have failed.
- */
-
- if (((action.get_remains() <= 0) && (action.get_variable()->get_penalty() > 0)) ||
- ((action.get_max_duration() != NO_MAX_DURATION) && (action.get_max_duration() <= 0))) {
- action.finish(kernel::resource::Action::State::FINISHED);
- continue;
- }
-
- /* Need to check that none of the model has failed */
- int i = 0;
- const kernel::lmm::Constraint* cnst = action.get_variable()->get_constraint(i);
- while (cnst != nullptr) {
- i++;
- const kernel::resource::Resource* constraint_id = cnst->get_id();
- if (not constraint_id->is_on()) {
- XBT_DEBUG("Action (%p) Failed!!", &action);
- action.finish(kernel::resource::Action::State::FAILED);
- break;
- }
- cnst = action.get_variable()->get_constraint(i);
- }
- }
-}
-
-kernel::resource::CpuAction* HostL07Model::execute_parallel(const std::vector<s4u::Host*>& host_list,
- const double* flops_amount, const double* bytes_amount,
- double rate)
-{
- return new L07Action(this, host_list, flops_amount, bytes_amount, rate);
-}
-
-L07Action::L07Action(kernel::resource::Model* model, const std::vector<s4u::Host*>& host_list,
- const double* flops_amount, const double* bytes_amount, double rate)
- : CpuAction(model, 1.0, false), computationAmount_(flops_amount), communicationAmount_(bytes_amount), rate_(rate)
-{
- size_t link_nb = 0;
- size_t used_host_nb = 0; /* Only the hosts with something to compute (>0 flops) are counted) */
- double latency = 0.0;
- this->set_last_update();
-
- hostList_.insert(hostList_.end(), host_list.begin(), host_list.end());
-
- if (flops_amount != nullptr)
- used_host_nb += std::count_if(flops_amount, flops_amount + host_list.size(), [](double x) { return x > 0.0; });
-
- /* Compute the number of affected resources... */
- if (bytes_amount != nullptr) {
- std::unordered_set<const char*> affected_links;
-
- for (size_t k = 0; k < host_list.size() * host_list.size(); k++) {
- if (bytes_amount[k] <= 0)
- continue;
-
- double lat = 0.0;
- std::vector<kernel::resource::LinkImpl*> route;
- hostList_[k / host_list.size()]->route_to(hostList_[k % host_list.size()], route, &lat);
- latency = std::max(latency, lat);
-
- for (auto const& link : route)
- affected_links.insert(link->get_cname());
- }
-
- link_nb = affected_links.size();
- }
-
- XBT_DEBUG("Creating a parallel task (%p) with %zu hosts and %zu unique links.", this, host_list.size(), link_nb);
- latency_ = latency;
-
- set_variable(
- model->get_maxmin_system()->variable_new(this, 1.0, (rate > 0 ? rate : -1.0), host_list.size() + link_nb));
-
- if (latency_ > 0)
- model->get_maxmin_system()->update_variable_penalty(get_variable(), 0.0);
-
- /* Expand it for the CPUs even if there is nothing to compute, to make sure that it gets expended even if there is no
- * communication either */
- double bound = std::numeric_limits<double>::max();
- for (size_t i = 0; i < host_list.size(); i++) {
- model->get_maxmin_system()->expand(host_list[i]->pimpl_cpu->get_constraint(), get_variable(),
- (flops_amount == nullptr ? 0.0 : flops_amount[i]));
- if (flops_amount && flops_amount[i] > 0)
- bound = std::min(bound, host_list[i]->pimpl_cpu->get_speed(1.0) * host_list[i]->pimpl_cpu->get_speed_ratio() /
- flops_amount[i]);
- }
- if (bound < std::numeric_limits<double>::max())
- model->get_maxmin_system()->update_variable_bound(get_variable(), bound);
-
- if (bytes_amount != nullptr) {
- for (size_t k = 0; k < host_list.size() * host_list.size(); k++) {
- if (bytes_amount[k] <= 0.0)
- continue;
- std::vector<kernel::resource::LinkImpl*> route;
- hostList_[k / host_list.size()]->route_to(hostList_[k % host_list.size()], route, nullptr);
-
- for (auto const& link : route)
- model->get_maxmin_system()->expand_add(link->get_constraint(), this->get_variable(), bytes_amount[k]);
- }
- }
-
- if (link_nb + used_host_nb == 0) {
- this->set_cost(1.0);
- this->set_remains(0.0);
- }
-}
-
-kernel::resource::Action* NetworkL07Model::communicate(s4u::Host* src, s4u::Host* dst, double size, double rate)
-{
- std::vector<s4u::Host*> host_list = {src, dst};
- const auto* flops_amount = new double[2]();
- auto* bytes_amount = new double[4]();
-
- bytes_amount[1] = size;
-
- kernel::resource::Action* res = hostModel_->execute_parallel(host_list, flops_amount, bytes_amount, rate);
- static_cast<L07Action*>(res)->free_arrays_ = true;
- return res;
-}
-
-kernel::resource::Cpu* CpuL07Model::create_cpu(s4u::Host* host, const std::vector<double>& speed_per_pstate)
-{
- return (new CpuL07(host, speed_per_pstate))->set_model(this);
-}
-
-kernel::resource::LinkImpl* NetworkL07Model::create_link(const std::string& name, const std::vector<double>& bandwidths,
- s4u::Link::SharingPolicy policy)
-{
- xbt_assert(bandwidths.size() == 1, "Non WIFI link must have only 1 bandwidth.");
- auto link = new LinkL07(name, bandwidths[0], policy, get_maxmin_system());
- link->set_model(this);
- return link;
-}
-
-/************
- * Resource *
- ************/
-
-kernel::resource::CpuAction* CpuL07::execution_start(double size)
-{
- std::vector<s4u::Host*> host_list = {get_iface()};
-
- auto* flops_amount = new double[host_list.size()]();
- flops_amount[0] = size;
-
- kernel::resource::CpuAction* res =
- static_cast<CpuL07Model*>(get_model())->hostModel_->execute_parallel(host_list, flops_amount, nullptr, -1);
- static_cast<L07Action*>(res)->free_arrays_ = true;
- return res;
-}
-
-kernel::resource::CpuAction* CpuL07::sleep(double duration)
-{
- auto* action = static_cast<L07Action*>(execution_start(1.0));
- action->set_max_duration(duration);
- action->set_suspend_state(kernel::resource::Action::SuspendStates::SLEEPING);
- get_model()->get_maxmin_system()->update_variable_penalty(action->get_variable(), 0.0);
-
- return action;
-}
-
-bool CpuL07::is_used() const
-{
- return get_model()->get_maxmin_system()->constraint_used(get_constraint());
-}
-
-/** @brief take into account changes of speed (either load or max) */
-void CpuL07::on_speed_change()
-{
- const kernel::lmm::Element* elem = nullptr;
-
- get_model()->get_maxmin_system()->update_constraint_bound(get_constraint(), speed_.peak * speed_.scale);
- while (const auto* var = get_constraint()->get_variable(&elem)) {
- const kernel::resource::Action* action = var->get_id();
-
- get_model()->get_maxmin_system()->update_variable_bound(action->get_variable(), speed_.scale * speed_.peak);
- }
-
- Cpu::on_speed_change();
-}
-
-LinkL07::LinkL07(const std::string& name, double bandwidth, s4u::Link::SharingPolicy policy,
- kernel::lmm::System* system)
- : LinkImpl(name)
-{
- this->set_constraint(system->constraint_new(this, bandwidth));
- bandwidth_.peak = bandwidth;
-
- if (policy == s4u::Link::SharingPolicy::FATPIPE)
- get_constraint()->unshare();
-}
-
-bool LinkL07::is_used() const
-{
- return get_model()->get_maxmin_system()->constraint_used(get_constraint());
-}
-
-void CpuL07::apply_event(kernel::profile::Event* triggered, double value)
-{
- XBT_DEBUG("Updating cpu %s (%p) with value %g", get_cname(), this, value);
- if (triggered == speed_.event) {
- speed_.scale = value;
- on_speed_change();
- tmgr_trace_event_unref(&speed_.event);
-
- } else if (triggered == state_event_) {
- if (value > 0) {
- if (not is_on()) {
- XBT_VERB("Restart actors on host %s", get_iface()->get_cname());
- get_iface()->turn_on();
- }
- } else
- get_iface()->turn_off();
- tmgr_trace_event_unref(&state_event_);
-
- } else {
- xbt_die("Unknown event!\n");
- }
-}
-
-void LinkL07::apply_event(kernel::profile::Event* triggered, double value)
-{
- XBT_DEBUG("Updating link %s (%p) with value=%f", get_cname(), this, value);
- if (triggered == bandwidth_.event) {
- set_bandwidth(value);
- tmgr_trace_event_unref(&bandwidth_.event);
-
- } else if (triggered == latency_.event) {
- set_latency(value);
- tmgr_trace_event_unref(&latency_.event);
-
- } else if (triggered == state_event_) {
- if (value > 0)
- turn_on();
- else
- turn_off();
- tmgr_trace_event_unref(&state_event_);
-
- } else {
- xbt_die("Unknown event ! \n");
- }
-}
-
-void LinkL07::set_bandwidth(double value)
-{
- bandwidth_.peak = value;
- LinkImpl::on_bandwidth_change();
-
- get_model()->get_maxmin_system()->update_constraint_bound(get_constraint(), bandwidth_.peak * bandwidth_.scale);
-}
-
-kernel::resource::LinkImpl* LinkL07::set_latency(double value)
-{
- latency_check(value);
- const kernel::lmm::Element* elem = nullptr;
-
- latency_.peak = value;
- while (const auto* var = get_constraint()->get_variable(&elem)) {
- auto* action = static_cast<L07Action*>(var->get_id());
- action->updateBound();
- }
- return this;
-}
-LinkL07::~LinkL07() = default;
-
-/**********
- * Action *
- **********/
-
-L07Action::~L07Action()
-{
- if (free_arrays_) {
- delete[] computationAmount_;
- delete[] communicationAmount_;
- }
-}
-
-void L07Action::updateBound()
-{
- double lat_current = 0.0;
-
- size_t host_count = hostList_.size();
-
- if (communicationAmount_ != nullptr) {
- for (size_t i = 0; i < host_count; i++) {
- for (size_t j = 0; j < host_count; j++) {
- if (communicationAmount_[i * host_count + j] > 0) {
- double lat = 0.0;
- std::vector<kernel::resource::LinkImpl*> route;
- hostList_.at(i)->route_to(hostList_.at(j), route, &lat);
-
- lat_current = std::max(lat_current, lat * communicationAmount_[i * host_count + j]);
- }
- }
- }
- }
- double lat_bound = kernel::resource::NetworkModel::cfg_tcp_gamma / (2.0 * lat_current);
- XBT_DEBUG("action (%p) : lat_bound = %g", this, lat_bound);
- if ((latency_ <= 0.0) && is_running()) {
- if (rate_ < 0)
- get_model()->get_maxmin_system()->update_variable_bound(get_variable(), lat_bound);
- else
- get_model()->get_maxmin_system()->update_variable_bound(get_variable(), std::min(rate_, lat_bound));
- }
-}
-
-} // namespace surf
-} // namespace simgrid
