/* 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 <ctype.h>
-
#include "surf_private.h"
#include "xbt/module.h"
#include "mc/mc.h"
+#include "simix/smx_host_private.h"
#include "surf/surf_resource.h"
+#include "xbt/xbt_os_thread.h"
+#include "simgrid/sg_config.h"
+
+#include <ctype.h>
+XBT_LOG_NEW_CATEGORY(surf, "All SURF categories");
XBT_LOG_NEW_DEFAULT_SUBCATEGORY(surf_kernel, surf,
"Logging specific to SURF (kernel)");
-/* Additional declarations for Windows potability. */
+/* Additional declarations for Windows portability. */
#ifndef MAX_DRIVE
#define MAX_DRIVE 26
/* Don't forget to update the option description in smx_config when you change this */
s_surf_model_description_t surf_network_model_description[] = {
- {"Constant",
- "Simplistic network model where all communication take a constant time (one second)",
- surf_network_model_init_Constant},
- {"CM02",
- "Legacy network model with lmm_solve (slow-start and bottleneck sharing with small bandwidth poorly modeled).",
- surf_network_model_init_CM02},
{"LV08",
- "Realistic network model (slow-start modeled by multiplying latency by 10.4, bandwidth by .92 and bottleneck sharing uses a payload of S=8775 for evaluating RTT). Uses partial invalidation optimization of lmm system and a heap for action management by default (option --cfg=network/optim:Lazy)",
- surf_network_model_init_LegrandVelho},
- {"LV08_fullupdate",
- "Realistic network model (see LV08) [Deprecated. Use LV08 with option --cfg=network/optim:Full. The Full option updates all actions at every step, which may be useful for debug.]",
+ "Realistic network analytic model (slow-start modeled by multiplying latency by 10.4, bandwidth by .92; bottleneck sharing uses a payload of S=8775 for evaluating RTT). ",
surf_network_model_init_LegrandVelho},
+ {"Constant",
+ "Simplistic network model where all communication take a constant time (one second). This model provides the lowest realism, but is (marginally) faster.",
+ surf_network_model_init_Constant},
{"SMPI",
- "Realistic network model with lmm_solve specifically tailored for HPC setting (accurate modeling of slow start with correction factors on three intervals (< 1KiB, < 64 KiB, >= 64 KiB))",
+ "Realistic network model specifically tailored for HPC settings (accurate modeling of slow start with correction factors on three intervals: < 1KiB, < 64 KiB, >= 64 KiB)",
surf_network_model_init_SMPI},
+ {"CM02",
+ "Legacy network analytic model (Very similar to LV08, but without corrective factors. The timings of small messages are thus poorly modeled).",
+ surf_network_model_init_CM02},
#ifdef HAVE_GTNETS
{"GTNets",
- "Network Pseudo-model using the GTNets simulator instead of an analytic model",
+ "Network pseudo-model using the GTNets simulator instead of an analytic model",
surf_network_model_init_GTNETS},
#endif
#ifdef HAVE_NS3
{"NS3",
- "Use NS3 tcp model",
- surf_network_model_init_NS3},
+ "Network pseudo-model using the NS3 tcp model instead of an analytic model",
+ surf_network_model_init_NS3},
#endif
{"Reno",
- "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info)",
+ "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info).",
surf_network_model_init_Reno},
{"Reno2",
- "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info)",
+ "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info).",
surf_network_model_init_Reno2},
{"Vegas",
- "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info)",
+ "Model from Steven H. Low using lagrange_solve instead of lmm_solve (experts only; check the code for more info).",
surf_network_model_init_Vegas},
{NULL, NULL, NULL} /* this array must be NULL terminated */
};
s_surf_model_description_t surf_cpu_model_description[] = {
- {"Cas01_fullupdate", "CPU classical model time=size/power (Cas01). [Deprecated. Use Cas01 with option --cfg=cpu/optim:Full. The Full option updates all actions at every step, which may be useful for debug.]",
- surf_cpu_model_init_Cas01},
{"Cas01",
- "CPU classical model time=size/power. Default version uses partial invalidation optimization of lmm system and a heap for action management (default option is --cfg=cpu/optim:Lazy)",
+ "Simplistic CPU model (time=size/power).",
surf_cpu_model_init_Cas01},
- {"CpuTI",
- "CPU classical model time=size/power (Cas01). [Deprecated. Use Cas01 with option --cfg=cpu/optim:TI. The TI optimization is highly optimized when using traces]",
- surf_cpu_model_init_ti},
{NULL, NULL, NULL} /* this array must be NULL terminated */
};
s_surf_model_description_t surf_workstation_model_description[] = {
- {"current_default",
- "Curent default workstation model (may change with versions of SimGrid). Currently Cas01 and LV08 (with full duplex support) as CPU and Network",
+ {"default",
+ "Default workstation model. Currently, CPU:Cas01 and network:LV08 (with cross traffic enabled)",
surf_workstation_model_init_current_default},
- {"CLM03",
- "Default workstation model, using Cas01 and CM02 as CPU and Network",
- surf_workstation_model_init_CLM03},
{"compound",
- "Workstation model allowing you to use other network and CPU models",
+ "Workstation model that is automatically chosen if you change the network and CPU models",
surf_workstation_model_init_compound},
- {"ptask_L07", "Workstation model with better parallel task modeling",
+ {"ptask_L07", "Workstation model somehow similar to Cas01+CM02 but allowing parallel tasks",
surf_workstation_model_init_ptask_L07},
{NULL, NULL, NULL} /* this array must be NULL terminated */
};
"Lazy action management (partial invalidation in lmm + heap in action remaining).",
NULL},
{"TI",
- "Trace integration. Highly optimized mode for the Cas01 model when using availability traces.",
+ "Trace integration. Highly optimized mode when using availability traces (only available for the Cas01 CPU model for now).",
NULL},
{"Full",
"Full update of remaining and variables. Slow but may be useful when debugging.",
{NULL, NULL, NULL} /* this array must be NULL terminated */
};
+s_surf_model_description_t surf_storage_model_description[] = {
+ {"default",
+ "Simplistic storage model.",
+ surf_storage_model_init_default},
+ {NULL, NULL, NULL} /* this array must be NULL terminated */
+};
+
+/* ********************************************************************* */
+/* TUTORIAL: New model */
+s_surf_model_description_t surf_new_model_description[] = {
+ {"default",
+ "Tutorial model.",
+ surf_new_model_init_default},
+ {NULL, NULL, NULL} /* this array must be NULL terminated */
+};
+/* ********************************************************************* */
+
+#ifdef CONTEXT_THREADS
+static xbt_parmap_t surf_parmap = NULL; /* parallel map on models */
+#endif
+
+static int surf_nthreads = 1; /* number of threads of the parmap (1 means no parallelism) */
+static double *surf_mins = NULL; /* return value of share_resources for each model */
+static int surf_min_index; /* current index in surf_mins */
+static double min; /* duration determined by surf_solve */
+
+static void surf_share_resources(surf_model_t model);
+static void surf_update_actions_state(surf_model_t model);
+
/** Displays the long description of all registered models, and quit */
void model_help(const char *category, s_surf_model_description_t * table)
{
return min;
}
-XBT_LOG_EXTERNAL_CATEGORY(surf_cpu);
-XBT_LOG_EXTERNAL_CATEGORY(surf_kernel);
-XBT_LOG_EXTERNAL_CATEGORY(surf_lagrange);
-XBT_LOG_EXTERNAL_CATEGORY(surf_lagrange_dichotomy);
-XBT_LOG_EXTERNAL_CATEGORY(surf_maxmin);
-XBT_LOG_EXTERNAL_CATEGORY(surf_network);
-XBT_LOG_EXTERNAL_CATEGORY(surf_trace);
-XBT_LOG_EXTERNAL_CATEGORY(surf_parse);
-XBT_LOG_EXTERNAL_CATEGORY(surf_timer);
-XBT_LOG_EXTERNAL_CATEGORY(surf_workstation);
-XBT_LOG_EXTERNAL_CATEGORY(surf_config);
-XBT_LOG_EXTERNAL_CATEGORY(surf_route);
+double generic_share_resources_lazy(double now, surf_model_t model)
+{
+ surf_action_lmm_t action = NULL;
+ double min = -1;
+ double value;
-#ifdef HAVE_GTNETS
-XBT_LOG_EXTERNAL_CATEGORY(surf_network_gtnets);
-#endif
+ XBT_DEBUG
+ ("Before share resources, the size of modified actions set is %d",
+ xbt_swag_size(model->model_private->modified_set));
-void surf_init(int *argc, char **argv)
+ lmm_solve(model->model_private->maxmin_system);
+
+ XBT_DEBUG
+ ("After share resources, The size of modified actions set is %d",
+ xbt_swag_size(model->model_private->modified_set));
+
+ while((action = xbt_swag_extract(model->model_private->modified_set))) {
+ int max_dur_flag = 0;
+
+ if (action->generic_action.state_set !=
+ model->states.running_action_set)
+ continue;
+
+ /* bogus priority, skip it */
+ if (action->generic_action.priority <= 0)
+ continue;
+
+ generic_update_action_remaining_lazy(action,now);
+
+ min = -1;
+ value = lmm_variable_getvalue(action->variable);
+ if (value > 0) {
+ if (action->generic_action.remains > 0) {
+ value = action->generic_action.remains / value;
+ min = now + value;
+ } else {
+ value = 0.0;
+ min = now;
+ }
+ }
+
+ if ((action->generic_action.max_duration != NO_MAX_DURATION)
+ && (min == -1
+ || action->generic_action.start +
+ action->generic_action.max_duration < min)) {
+ min = action->generic_action.start +
+ action->generic_action.max_duration;
+ max_dur_flag = 1;
+ }
+
+ XBT_DEBUG("Action(%p) Start %lf Finish %lf Max_duration %lf", action,
+ action->generic_action.start, now + value,
+ action->generic_action.max_duration);
+
+ if (min != -1) {
+ surf_action_lmm_heap_remove(model->model_private->action_heap,action);
+ surf_action_lmm_heap_insert(model->model_private->action_heap,action, min, max_dur_flag ? MAX_DURATION : NORMAL);
+ XBT_DEBUG("Insert at heap action(%p) min %lf now %lf", action, min,
+ now);
+ } else DIE_IMPOSSIBLE;
+ }
+
+ //hereafter must have already the min value for this resource model
+ if (xbt_heap_size(model->model_private->action_heap) > 0)
+ min = xbt_heap_maxkey(model->model_private->action_heap) - now;
+ else
+ min = -1;
+
+ XBT_DEBUG("The minimum with the HEAP %lf", min);
+
+ return min;
+}
+static XBT_INLINE void routing_asr_host_free(void *p)
{
- XBT_DEBUG("Create all Libs");
- host_lib = xbt_lib_new();
- link_lib = xbt_lib_new();
- as_router_lib = xbt_lib_new();
-
- XBT_DEBUG("ADD ROUTING LEVEL");
- ROUTING_HOST_LEVEL = xbt_lib_add_level(host_lib,xbt_free);
- ROUTING_ASR_LEVEL = xbt_lib_add_level(as_router_lib,xbt_free);
-
- XBT_DEBUG("ADD SURF LEVELS");
- SURF_CPU_LEVEL = xbt_lib_add_level(host_lib,surf_resource_free);
- SURF_WKS_LEVEL = xbt_lib_add_level(host_lib,surf_resource_free);
- SURF_LINK_LEVEL = xbt_lib_add_level(link_lib,surf_resource_free);
-
- /* Connect our log channels: that must be done manually under windows */
- XBT_LOG_CONNECT(surf_cpu, surf);
- XBT_LOG_CONNECT(surf_kernel, surf);
- XBT_LOG_CONNECT(surf_lagrange, surf);
- XBT_LOG_CONNECT(surf_lagrange_dichotomy, surf_lagrange);
- XBT_LOG_CONNECT(surf_maxmin, surf);
- XBT_LOG_CONNECT(surf_network, surf);
- XBT_LOG_CONNECT(surf_trace, surf);
- XBT_LOG_CONNECT(surf_parse, surf);
- XBT_LOG_CONNECT(surf_timer, surf);
- XBT_LOG_CONNECT(surf_workstation, surf);
- XBT_LOG_CONNECT(surf_config, surf);
- XBT_LOG_CONNECT(surf_route, surf);
+ sg_routing_edge_t elm = p;
+ free(elm->name);
+ xbt_free(elm);
+}
-#ifdef HAVE_GTNETS
- XBT_LOG_CONNECT(surf_network_gtnets, surf);
-#endif
+static XBT_INLINE void routing_asr_prop_free(void *p)
+{
+ xbt_dict_t elm = p;
+ xbt_dict_free(&elm);
+}
+
+void sg_version(int *ver_major,int *ver_minor,int *ver_patch) {
+ *ver_major = SIMGRID_VERSION_MAJOR;
+ *ver_minor = SIMGRID_VERSION_MINOR;
+ *ver_patch = SIMGRID_VERSION_PATCH;
+}
+
+void surf_init(int *argc, char **argv)
+{
+ XBT_DEBUG("Create all Libs");
+ host_lib = xbt_lib_new();
+ link_lib = xbt_lib_new();
+ as_router_lib = xbt_lib_new();
+ storage_lib = xbt_lib_new();
+ storage_type_lib = xbt_lib_new();
+ watched_hosts_lib = xbt_dict_new();
+
+ XBT_DEBUG("Add routing levels");
+ ROUTING_HOST_LEVEL = xbt_lib_add_level(host_lib,routing_asr_host_free);
+ ROUTING_ASR_LEVEL = xbt_lib_add_level(as_router_lib,routing_asr_host_free);
+ ROUTING_PROP_ASR_LEVEL = xbt_lib_add_level(as_router_lib,routing_asr_prop_free);
+
+ XBT_DEBUG("Add SURF levels");
+ SURF_CPU_LEVEL = xbt_lib_add_level(host_lib,surf_resource_free);
+ SURF_WKS_LEVEL = xbt_lib_add_level(host_lib,surf_resource_free);
+ SURF_LINK_LEVEL = xbt_lib_add_level(link_lib,surf_resource_free);
xbt_init(argc, argv);
if (!model_list)
if (!history)
history = tmgr_history_new();
- surf_config_init(argc, argv);
+#ifdef HAVE_TRACING
+ TRACE_add_start_function(TRACE_surf_alloc);
+ TRACE_add_end_function(TRACE_surf_release);
+#endif
+
+ sg_config_init(argc, argv);
+
surf_action_init();
- if (MC_IS_ENABLED)
+ if (MC_is_active())
MC_memory_init();
}
unsigned int iter;
surf_model_t model = NULL;
- surf_config_finalize();
+ sg_config_finalize();
xbt_dynar_foreach(model_list, iter, model)
model->model_private->finalize();
}
surf_action_exit();
+#ifdef CONTEXT_THREADS
+ xbt_parmap_destroy(surf_parmap);
+ xbt_free(surf_mins);
+ surf_mins = NULL;
+#endif
+
xbt_dynar_free(&surf_path);
xbt_lib_free(&host_lib);
xbt_lib_free(&link_lib);
xbt_lib_free(&as_router_lib);
+ xbt_lib_free(&storage_lib);
+ xbt_lib_free(&storage_type_lib);
+ xbt_dict_free(&watched_hosts_lib);
tmgr_finalize();
surf_parse_lex_destroy();
tmgr_history_get_next_event_leq(history, next_event_date,
&value,
(void **) &resource))) {
- resource->model->model_private->update_resource_state(resource,
- event, value,
- NOW);
+ if (value >= 0){
+ resource->model->model_private->update_resource_state(resource,
+ event, value,
+ NOW);
+ }
}
}
xbt_dynar_foreach(model_list, iter, model)
double surf_solve(double max_date)
{
- double min = -1.0; /* duration */
+ min = -1.0; /* duration */
double next_event_date = -1.0;
double model_next_action_end = -1.0;
double value = -1.0;
}
XBT_DEBUG("Looking for next action end for all models except NS3");
- xbt_dynar_foreach(model_list, iter, model) {
- if(strcmp(model->name,"network NS3") ){
- XBT_DEBUG("Running for Resource [%s]", model->name);
- model_next_action_end = model->model_private->share_resources(NOW);
- XBT_DEBUG("Resource [%s] : next action end = %f",
- model->name, model_next_action_end);
- if (((min < 0.0) || (model_next_action_end < min))
- && (model_next_action_end >= 0.0))
- min = model_next_action_end;
+
+ if (surf_mins == NULL) {
+ surf_mins = xbt_new(double, xbt_dynar_length(model_list));
+ }
+ surf_min_index = 0;
+
+
+ if (surf_get_nthreads() > 1) {
+ /* parallel version */
+#ifdef CONTEXT_THREADS
+ xbt_parmap_apply(surf_parmap, (void_f_pvoid_t) surf_share_resources, model_list);
+#else
+ xbt_die("Asked to run in parallel, but no thread at hand...");
+#endif
+ }
+ else {
+ /* sequential version */
+ xbt_dynar_foreach(model_list, iter, model) {
+ surf_share_resources(model);
+ }
+ }
+
+ unsigned i;
+ for (i = 0; i < xbt_dynar_length(model_list); i++) {
+ if ((min < 0.0 || surf_mins[i] < min)
+ && surf_mins[i] >= 0.0) {
+ min = surf_mins[i];
}
}
- XBT_DEBUG("Min for resources (except NS3) : %f", min);
+ XBT_DEBUG("Min for resources (remember that NS3 don't update that value) : %f", min);
XBT_DEBUG("Looking for next trace event");
min = MAX(next_event_date - NOW, min);
}
- XBT_DEBUG("Run for NS3 at most %f", min);
+ XBT_DEBUG("Run for network at most %f", min);
// run until min or next flow
model_next_action_end = surf_network_model->model_private->share_resources(min);
- XBT_DEBUG("Min for NS3 : %f", model_next_action_end);
+ XBT_DEBUG("Min for network : %f", model_next_action_end);
if(model_next_action_end>=0.0)
min = model_next_action_end;
}
- if (next_event_date == -1.0) break;
+ if (next_event_date < 0.0) {
+ XBT_DEBUG("no next TRACE event. Stop searching for it");
+ break;
+ }
- if ((min != -1.0) && (next_event_date > NOW + min)) break;
+ if ((min == -1.0) || (next_event_date > NOW + min)) break;
XBT_DEBUG("Updating models");
while ((event =
}
} while (1);
- /* FIXME: Moved this test to here to avoid stoping simulation if there are actions running on cpus and all cpus are with availability = 0.
+ /* FIXME: Moved this test to here to avoid stopping simulation if there are actions running on cpus and all cpus are with availability = 0.
* This may cause an infinite loop if one cpu has a trace with periodicity = 0 and the other a trace with periodicity > 0.
* The options are: all traces with same periodicity(0 or >0) or we need to change the way how the events are managed */
- if (min < 0.0)
+ if (min == -1.0) {
+ XBT_DEBUG("No next event at all. Bail out now.");
return -1.0;
+ }
XBT_DEBUG("Duration set to %f", min);
NOW = NOW + min;
- xbt_dynar_foreach(model_list, iter, model)
- model->model_private->update_actions_state(NOW, min);
+ if (surf_get_nthreads() > 1) {
+ /* parallel version */
+#ifdef CONTEXT_THREADS
+ xbt_parmap_apply(surf_parmap, (void_f_pvoid_t) surf_update_actions_state, model_list);
+#endif
+ }
+ else {
+ /* sequential version */
+ xbt_dynar_foreach(model_list, iter, model) {
+ surf_update_actions_state(model);
+ }
+ }
#ifdef HAVE_TRACING
TRACE_paje_dump_buffer (0);
{
return NOW;
}
+
+static void surf_share_resources(surf_model_t model)
+{
+ double next_action_end = -1.0;
+ int i = __sync_fetch_and_add(&surf_min_index, 1);
+ if (strcmp(model->name,"network NS3")) {
+ XBT_DEBUG("Running for Resource [%s]", model->name);
+ next_action_end = model->model_private->share_resources(NOW);
+ XBT_DEBUG("Resource [%s] : next action end = %f",
+ model->name, next_action_end);
+ }
+ surf_mins[i] = next_action_end;
+}
+
+static void surf_update_actions_state(surf_model_t model)
+{
+ model->model_private->update_actions_state(NOW, min);
+}
+
+/**
+ * \brief Returns the number of parallel threads used to update the models.
+ * \return the number of threads (1 means no parallelism)
+ */
+int surf_get_nthreads(void) {
+ return surf_nthreads;
+}
+
+/**
+ * \brief Sets the number of parallel threads used to update the models.
+ *
+ * A value of 1 means no parallelism.
+ *
+ * \param nb_threads the number of threads to use
+ */
+void surf_set_nthreads(int nthreads) {
+
+ if (nthreads<=0) {
+ nthreads = xbt_os_get_numcores();
+ XBT_INFO("Auto-setting surf/nthreads to %d",nthreads);
+ }
+
+#ifdef CONTEXT_THREADS
+ xbt_parmap_destroy(surf_parmap);
+ surf_parmap = NULL;
+#endif
+
+ if (nthreads > 1) {
+#ifdef CONTEXT_THREADS
+ surf_parmap = xbt_parmap_new(nthreads, XBT_PARMAP_DEFAULT);
+#else
+ THROWF(arg_error, 0, "Cannot activate parallel threads in Surf: your architecture does not support threads");
+#endif
+ }
+
+ surf_nthreads = nthreads;
+}
+
+/* This function is a pimple that we ought to fix. But it won't be easy.
+ *
+ * The surf_solve() function does properly return the set of actions that changed.
+ * Instead, each model change a global data, and then the caller of surf_solve must
+ * pick into these sets of action_failed and action_done.
+ *
+ * This was not clean but ok as long as we didn't had to restart the processes when the resource comes back up.
+ * We worked by putting sentinel actions on every resources we are interested in,
+ * so that surf informs us if/when the corresponding resource fails.
+ *
+ * But this does not work to get Simix informed of when a resource comes back up, and this is where this pimple comes.
+ * We have a set of resources that are currently down and for which simix needs to know when it comes back up.
+ * And the current function is called *at every simulation step* to sweep over that set, searching for a resource
+ * that was turned back up in the meanwhile. This is UGLY and slow.
+ *
+ * The proper solution would be to not rely on globals for the action_failed and action_done swags.
+ * They must be passed as parameter by the caller (the handling of these actions in simix may let you
+ * think that these two sets can be merged, but their handling in SimDag induce the contrary unless this
+ * simdag code can check by itself whether the action is done of failed -- seems very doable, but yet more
+ * cleanup to do).
+ *
+ * Once surf_solve() is passed the set of actions that changed, you want to add a new set of resources back up
+ * as parameter to this function. You also want to add a boolean field "restart_watched" to each resource, and
+ * make sure that whenever a resource with this field enabled comes back up, it's added to that set so that Simix
+ * sees it and react accordingly. This would kill that need for surf to call simix.
+ *
+ */
+void surf_watched_hosts(void)
+{
+ char *key;
+ void *host;
+ xbt_dict_cursor_t cursor;
+
+ XBT_DEBUG("Check for host SURF_RESOURCE_ON on watched_hosts_lib");
+ xbt_dict_foreach(watched_hosts_lib,cursor,key,host)
+ {
+ if(SIMIX_host_get_state(host) == SURF_RESOURCE_ON){
+ XBT_INFO("Restart processes on host: %s",SIMIX_host_get_name(host));
+ SIMIX_host_autorestart(host);
+ xbt_dict_remove(watched_hosts_lib,key);
+ }
+ else
+ XBT_DEBUG("See SURF_RESOURCE_OFF on host: %s",key);
+ }
+}
