XBT_LOG_NEW_DEFAULT_CATEGORY(dag_scheduling, "Logging specific to this example");
namespace sg4 = simgrid::s4u;
-struct HostAttribute {
- /* Earliest time at which a host is ready to execute a task */
- double available_at = 0.0;
- sg4::Exec* last_scheduled_task = nullptr;
-};
-
-static double sg_host_get_available_at(const sg4::Host* host)
-{
- return host->get_data<HostAttribute>()->available_at;
-}
-
-static void sg_host_set_available_at(const sg4::Host* host, double time)
-{
- host->get_data<HostAttribute>()->available_at = time;
-}
-
-static sg4::Exec* sg_host_get_last_scheduled_task(const sg4::Host* host)
-{
- return host->get_data<HostAttribute>()->last_scheduled_task;
-}
-
-static void sg_host_set_last_scheduled_task(const sg4::Host* host, sg4::ExecPtr task)
-{
- host->get_data<HostAttribute>()->last_scheduled_task = task.get();
-}
-
-static bool dependency_exists(const sg4::Exec* src, sg4::Exec* dst)
-{
- const auto& dependencies = src->get_dependencies();
- const auto& successors = src->get_successors();
- return (std::find(successors.begin(), successors.end(), dst) != successors.end() ||
- dependencies.find(dst) != dependencies.end());
-}
-
static std::vector<sg4::Exec*> get_ready_tasks(const std::vector<sg4::ActivityPtr>& dax)
{
std::vector<sg4::Exec*> ready_tasks;
std::map<sg4::Exec*, unsigned int> candidate_execs;
- for (auto& a : dax) {
+ for (const auto& a : dax) {
// Only look at activity that have their dependencies solved but are not assigned
if (a->dependencies_solved() && not a->is_assigned()) {
// if it is an exec, it's ready
return ready_tasks;
}
-static double finish_on_at(const sg4::ExecPtr task, const sg4::Host* host)
+static sg4::Host* get_best_host(const sg4::ExecPtr exec, double* min_finish_time)
{
- double data_available = 0.;
- double last_data_available = -1.0;
- /* compute last_data_available */
- for (const auto& parent : task->get_dependencies()) {
- /* normal case */
- if (const auto* comm = dynamic_cast<sg4::Comm*>(parent.get())) {
- auto source = comm->get_source();
- XBT_DEBUG("transfer from %s to %s", source->get_cname(), host->get_cname());
- /* Estimate the redistribution time from this parent */
- double redist_time;
- if (comm->get_remaining() <= 1e-6) {
- redist_time = 0;
- } else {
- redist_time =
- sg_host_get_route_latency(source, host) + comm->get_remaining() / sg_host_get_route_bandwidth(source, host);
+ sg4::Host* best_host = nullptr;
+ *min_finish_time = std::numeric_limits<double>::max();
+
+ for (const auto& host : sg4::Engine::get_instance()->get_all_hosts()) {
+ double data_available = 0.;
+ double last_data_available = -1.0;
+ /* compute last_data_available */
+ for (const auto& parent : exec->get_dependencies()) {
+ /* normal case */
+ if (const auto* comm = dynamic_cast<sg4::Comm*>(parent.get())) {
+ const auto* source = comm->get_source();
+ XBT_DEBUG("transfer from %s to %s", source->get_cname(), host->get_cname());
+ /* Estimate the redistribution time from this parent */
+ double redist_time;
+ if (comm->get_remaining() <= 1e-6) {
+ redist_time = 0;
+ } else {
+ double bandwidth = std::numeric_limits<double>::max();
+ auto [links, latency] = source->route_to(host);
+ for (auto const& link : links)
+ bandwidth = std::min(bandwidth, link->get_bandwidth());
+
+ redist_time = latency + comm->get_remaining() / bandwidth;
+ }
+ // We use the user data field to store the finish time of the predecessor of the comm, i.e., its potential
+ // start time
+ data_available = *comm->get_data<double>() + redist_time;
}
- // We use the user data field to store the finish time of the predecessor of the comm, i.e., its potential start
- // time
- data_available = *comm->get_data<double>() + redist_time;
- }
- /* no transfer, control dependency */
- if (const auto* exec = dynamic_cast<sg4::Exec*>(parent.get()))
- data_available = exec->get_finish_time();
-
- if (last_data_available < data_available)
- last_data_available = data_available;
- }
+ /* no transfer, control dependency */
+ if (const auto* parent_exec = dynamic_cast<sg4::Exec*>(parent.get()))
+ data_available = parent_exec->get_finish_time();
- return std::max(sg_host_get_available_at(host), last_data_available) + task->get_remaining() / host->get_speed();
-}
+ if (last_data_available < data_available)
+ last_data_available = data_available;
+ }
-static sg4::Host* get_best_host(const sg4::ExecPtr exec)
-{
- std::vector<sg4::Host*> hosts = sg4::Engine::get_instance()->get_all_hosts();
- auto best_host = hosts.front();
- double min_EFT = finish_on_at(exec, best_host);
+ double finish_time = std::max(*host->get_data<double>(), last_data_available) +
+ exec->get_remaining() / host->get_speed();
- for (const auto& h : hosts) {
- double EFT = finish_on_at(exec, h);
- XBT_DEBUG("%s finishes on %s at %f", exec->get_cname(), h->get_cname(), EFT);
+ XBT_DEBUG("%s finishes on %s at %f", exec->get_cname(), host->get_cname(), finish_time);
- if (EFT < min_EFT) {
- min_EFT = EFT;
- best_host = h;
+ if (finish_time < *min_finish_time) {
+ *min_finish_time = finish_time;
+ best_host = host;
}
}
+
return best_host;
}
-static void schedule_on(sg4::ExecPtr exec, sg4::Host* host)
+static void schedule_on(sg4::ExecPtr exec, sg4::Host* host, double busy_until = 0.0)
{
exec->set_host(host);
+ // We use the user data field to store up to when the host is busy
+ delete host->get_data<double>(); // In case we're erasing a previous value
+ host->set_data(new double(busy_until));
// we can also set the destination of all the input comms of this exec
for (const auto& pred : exec->get_dependencies()) {
auto* comm = dynamic_cast<sg4::Comm*>(pred.get());
std::set<sg4::Activity*> vetoed;
e.track_vetoed_activities(&vetoed);
- sg4::Activity::on_completion_cb([](sg4::Activity const& activity) {
+ sg4::Exec::on_completion_cb([](sg4::Exec const& exec) {
// when an Exec completes, we need to set the potential start time of all its ouput comms
- const auto* exec = dynamic_cast<sg4::Exec const*>(&activity);
- if (exec == nullptr) // Only Execs are concerned here
- return;
- for (const auto& succ : exec->get_successors()) {
+ for (const auto& succ : exec.get_successors()) {
auto* comm = dynamic_cast<sg4::Comm*>(succ.get());
if (comm != nullptr) {
- auto* finish_time = new double(exec->get_finish_time());
+ auto* finish_time = new double(exec.get_finish_time());
// We use the user data field to store the finish time of the predecessor of the comm, i.e., its potential start
// time
comm->set_data(finish_time);
});
e.load_platform(argv[1]);
- const auto hosts = e.get_all_hosts();
/* Mark all hosts as sequential, as it ought to be in such a scheduling example.
*
* It means that the hosts can only compute one thing at a given time. If an execution already takes place on a given
* host, any subsequently started execution will be queued until after the first execution terminates */
- for (auto const& host : hosts)
+ for (auto const& host : e.get_all_hosts()) {
host->set_concurrency_limit(1);
-
- /* Allocating the host attribute */
- unsigned long total_nhosts = e.get_host_count();
- std::vector<HostAttribute> host_attributes(total_nhosts);
- for (unsigned long i = 0; i < total_nhosts; i++)
- hosts[i]->set_data(&host_attributes[i]);
-
+ host->set_data(new double(0.0));
+ }
/* load the DAX file */
auto dax = sg4::create_DAG_from_DAX(argv[2]);
/* Schedule the root first */
+ double root_finish_time;
auto* root = static_cast<sg4::Exec*>(dax.front().get());
- auto host = get_best_host(root);
+ auto* host = get_best_host(root, &root_finish_time);
schedule_on(root, host);
e.run();
vetoed.clear();
if (ready_tasks.empty()) {
- /* there is no ready task, let advance the simulation */
+ /* there is no ready exec, let advance the simulation */
e.run();
continue;
}
- /* For each ready task:
+ /* For each ready exec:
* get the host that minimizes the completion time.
- * select the task that has the minimum completion time on its best host.
+ * select the exec that has the minimum completion time on its best host.
*/
- double min_finish_time = -1.0;
- sg4::Exec* selected_task = nullptr;
- sg4::Host* selected_host = nullptr;
-
- for (auto task : ready_tasks) {
- XBT_DEBUG("%s is ready", task->get_cname());
- host = get_best_host(task);
- double finish_time = finish_on_at(task, host);
- if (min_finish_time < 0 || finish_time < min_finish_time) {
+ double min_finish_time = std::numeric_limits<double>::max();
+ sg4::Exec* selected_task = nullptr;
+ sg4::Host* selected_host = nullptr;
+
+ for (auto* exec : ready_tasks) {
+ XBT_DEBUG("%s is ready", exec->get_cname());
+ double finish_time;
+ host = get_best_host(exec, &finish_time);
+ if (finish_time < min_finish_time) {
min_finish_time = finish_time;
- selected_task = task;
+ selected_task = exec;
selected_host = host;
}
}
XBT_INFO("Schedule %s on %s", selected_task->get_cname(), selected_host->get_cname());
- schedule_on(selected_task, selected_host);
-
- /*
- * tasks can be executed concurrently when they can by default.
- * Yet schedulers take decisions assuming that tasks wait for resource availability to start.
- * The solution (well crude hack is to keep track of the last task scheduled on a host and add a special type of
- * dependency if needed to force the sequential execution meant by the scheduler.
- * If the last scheduled task is already done, has failed or is a predecessor of the current task, no need for a
- * new dependency
- */
-
- if (auto last_scheduled_task = sg_host_get_last_scheduled_task(selected_host);
- last_scheduled_task && (last_scheduled_task->get_state() != sg4::Activity::State::FINISHED) &&
- (last_scheduled_task->get_state() != sg4::Activity::State::FAILED) &&
- not dependency_exists(sg_host_get_last_scheduled_task(selected_host), selected_task))
- last_scheduled_task->add_successor(selected_task);
-
- sg_host_set_last_scheduled_task(selected_host, selected_task);
- sg_host_set_available_at(selected_host, min_finish_time);
+ schedule_on(selected_task, selected_host, min_finish_time);
ready_tasks.clear();
e.run();
}
+ /* Cleanup memory */
+ for (auto const* h : e.get_all_hosts())
+ delete h->get_data<double>();
+
XBT_INFO("Simulation Time: %f", simgrid_get_clock());
return 0;