3 <object id="TOC" data="graphical-toc.svg" type="image/svg+xml"></object>
5 window.onload=function() { // Wait for the SVG to be loaded before changing it
6 var elem=document.querySelector("#TOC").contentDocument.getElementById("ReferenceBox")
7 elem.style="opacity:0.93999999;fill:#ff0000;fill-opacity:0.1;stroke:#000000;stroke-width:0.35277778;stroke-linecap:round;stroke-linejoin:round;stroke-miterlimit:4;stroke-dasharray:none;stroke-dashoffset:0;stroke-opacity:1";
13 .. _platform_reference:
15 Complete XML Reference
16 **********************
18 Your platform description should follow the specification presented in the
19 `simgrid.dtd <https://simgrid.org/simgrid.dtd>`_ DTD file. The same DTD is used for both platform and deployment files.
21 -------------------------------------------------------------------------------
28 This complex tag builds a full zone, with some hosts, links, a router, and the relevant routing algorithm. There is several
29 kind of specifically tailored cluster types: crossbar clusters (contention-free internal network), backbone clusters
30 (constrained internal network), fat-trees, DragonFly and Torus (star clusters are similar, but can only be created from the
31 C++ code). The ``topology`` attribute is used to choose the type of cluster while the ``topo_parameters`` attribute is used
32 to configure the topology. Please refer to the :ref:`examples in the documentation <platform_examples>` for all details.
34 Regardless of its topology, each cluster comes with a router that can be used for inter-zone routing. Its name is defined
35 as ``${prefix}${clusterId}_router${suffix}``.
37 **Parent tags:** :ref:`pf_tag_platform`, :ref:`pf_tag_zone` |br|
38 **Children tags:** none |br|
41 :``id``: The identifier of the cluster. Facilitates referring to this cluster.
42 :``prefix``: Each node of the cluster has to have a name. This name will be prefixed with this prefix.
43 :``suffix``: Each node of the cluster will be suffixed with this suffix
44 :``radical``: Regexp used to generate cluster nodes name.
46 Syntax: "10-20" will give you 11 machines numbered from 10 to 20, "10-20;2" will give you 12 machines, one with the number 2, others numbered as before.
48 The produced number is concatenated between prefix and suffix to form machine names.
49 :``speed``: Same as the ``speed`` attribute of the :ref:`pf_tag_host` tag.
50 :``core``: Same as the ``core`` attribute of the :ref:`pf_tag_host` tag.
51 :``bw``: Bandwidth for the links between nodes and backbone (if any). See :ref:`pf_tag_link` for syntax/details.
52 :``lat``: Latency for the links between nodes and backbone (if any). See :ref:`pf_tag_link` for syntax/details.
53 :``sharing_policy``: Sharing policy for the links between nodes and backbone (if any). See :ref:`pf_tag_link` for syntax/details.
54 :``bb_bw``: Bandwidth for backbone (if any). See :ref:`pf_tag_link` for syntax/details.
56 If bb_bw and bb_lat attributes are omitted, no backbone is created (alternative cluster architecture described earlier).
57 :``bb_lat``: Latency for backbone (if any). See :ref:`pf_tag_link` section for syntax/details.
59 If bb_lat and bb_bw attributes are omitted, no backbone is created (alternative cluster architecture described earlier).
60 :``bb_sharing_policy``: Sharing policy for the backbone (if any). See :ref:`pf_tag_link` section for syntax/details.
61 :``limiter_link``: Bandwidth for limiter link (if any).
63 This adds a specific link for each node, to set the maximum bandwidth reached when communicating in both directions at the same time.
65 In theory this value should be 2*bw for splitduplex links, but in reality this might be less. This value will depend heavily on the communication model, and on the cluster's hardware, so no default value can be set, this has to be measured.
67 More details can be obtained in `Toward Better Simulation of MPI Applications on Ethernet/TCP Networks <https://hal.inria.fr/hal-00919507/>`_
68 :``loopback_bw``: Bandwidth for loopback (if any). See :ref:`pf_tag_link` section for syntax/details.
70 If loopback_bw and loopback_lat attributes are omitted, no loopback link is created and all intra-node communication will use the main network link of the node.
72 The sharing policy of a loopback link is **FATPIPE** :ref:`pf_tag_link`.
73 :``loopback_lat``: Latency for loopback (if any). See loopback_bw for more info.
74 :``topology``: Network topology to use.
76 SimGrid currently supports FLAT (with or without backbone, as described before), TORUS, FAT_TREE and DRAGONFLY attributes for this tag.
78 See :ref:`platform_examples` for more details.
80 :``topo_parameters``: Specific parameters to pass for the topology defined in the topology tag.
82 For torus networks, comma-separated list of the number of nodes in each dimension of the torus.
84 Please refer to :ref:`platform_examples`.
86 -------------------------------------------------------------------------------
92 Adding configuration flags directly into the platform file becomes particularly
93 useful when the realism of the described platform depends on some specific
94 flags. For example, this could help you to finely tune SMPI. Almost all
95 :ref:`command-line configuration items <options_list>` can be configured this
98 Each configuration flag is described as a :ref:`pf_tag_prop` whose ``id`` is the
99 name of the flag and ``value`` is what it has to be set to.
101 **Parent tags:** :ref:`pf_tag_platform` (must appear before any other tags) |br|
102 **Children tags:** :ref:`pf_tag_prop` |br|
107 <?xml version = '1.0'?>
108 <!DOCTYPE platform SYSTEM "https://simgrid.org/simgrid.dtd">
109 <platform version = "4.1">
111 <prop id = "maxmin/precision" value = "0.000010" />
112 <prop id = "cpu/optim" value = "TI" />
113 <prop id = "network/model" value = "SMPI" />
114 <prop id = "smpi/bw-factor" value = "65472:0.940694;15424:0.697866;9376:0.58729" />
117 <!-- The rest of your platform -->
120 -------------------------------------------------------------------------------
127 SimGrid can simulate the time it takes to read or write data on disk, even if the stored data is not made persistent in
128 any way by SimGrid. This means that your application will correctly be slowed down when doing simulated I/O, but there
129 is no way to get the data stored this way.
131 We decided to not model anything beyond raw access in SimGrid because we believe that there is not single way of doing so.
132 We provide an example model of file system as a plugin, (sparsely) documented in :ref:`plugin_filesystem`.
134 **Parent tags:** :ref:`pf_tag_host` |br|
135 **Children tags:** :ref:`pf_tag_prop` |br|
138 :``id``: A name of your choice (must be unique on this host).
139 :``read_bw``: Read bandwidth for this disk. You must specify a unit as follows.
141 **Units in bytes and powers of 2** (1 KiBps = 1,024 Bps):
142 Bps, KiBps, MiBps, GiBps, TiBps, PiBps, or EiBps. |br|
143 **Units in bits and powers of 2** (1 Bps = 8 bps):
144 bps, Kibps, Mibps, Gibps, Tibps, Pibps, or Eibps. |br|
145 **Units in bytes and powers of 10** (1 KBps = 1,000 Bps):
146 Bps, KBps, MBps, GBps, TBps, PBps, or EBps. |br|
147 **Units in bits and powers of 10:**
148 bps, Kbps, Mbps, Gbps, Tbps, Pbps, or Ebps.
150 :``write_bw``: Write bandwidth for this disk. You must specify a unit as for the read bandwidth.
154 <host id="alice" speed="1Gf">
155 <disk id="Disk1" read_bw="200MBps" write_bw="80MBps">
156 <!-- you can add properties for anything you want: they are not used by SimGrid -->
157 <prop id="content" value="storage/content/small_content.txt"/>
159 <prop id="ram" value="100B" />
162 -------------------------------------------------------------------------------
169 A host is the computing resource on which an actor can run. See :cpp:class:`simgrid::s4u::Host`.
171 **Parent tags:** :ref:`pf_tag_zone` (only leaf zones, i.e., zones containing neither inner zones nor clusters) |br|
172 **Children tags:** :ref:`pf_tag_mount`, :ref:`pf_tag_prop`, :ref:`pf_tag_disk` |br|
176 Must be unique over the whole platform.
177 :``speed``: Computational power (per core, in flop/s).
178 If you use DVFS, provide a comma-separated list of values for each pstate (see :ref:`howto_dvfs`).
179 :``core``: Amount of cores (default: 1).
180 See :ref:`howto_multicore`.
181 :``availability_file``:
182 File containing the availability profile.
183 Almost every lines of such files describe timed events as ``date ratio``.
186 .. code-block:: python
193 - At time t = 1, half of the host computational power (0.5 means 50%) is used to process some background load, hence
194 only 50% of this initial power remains available to your own simulation.
195 - At time t = 2, the available power drops at 20% of the initial value.
196 - At time t = 5, the host can compute at full speed again.
197 - At time t = 10, the profile is reset (as we are 5 seconds after the last event). Then the available speed will drop
198 again to 50% at time t = 11.
200 If your profile does not contain any LOOPAFTER line, then it will
201 be executed only once and not repeated.
203 .. warning:: Don't get fooled: Bandwidth and Latency profiles of a :ref:`pf_tag_link` contain absolute values, while
204 Availability profiles of a :ref:`pf_tag_host` contain ratios.
205 :``state_file``: File containing the state profile.
206 Almost every lines of such files describe timed events as ``date boolean``.
209 .. code-block:: python
215 - At time t = 1, the host is turned off (a zero value means OFF)
216 - At time t = 2, the host is turned back on (any other value than zero means ON)
217 - At time t = 10, the profile is reset (as we are 8 seconds after the last event). Then the host will be turned off
218 again at time t = 11.
220 If your profile does not contain any LOOPAFTER line, then it will
221 be executed only once and not repeated.
223 :``coordinates``: Vivaldi coordinates (meaningful for Vivaldi zones only).
224 See :ref:`pf_tag_peer`.
225 :``pstate``: Initial pstate (default: 0, the first one).
226 See :ref:`howto_dvfs`.
228 -------------------------------------------------------------------------------
235 SimGrid links usually represent one-hop network connections (see :cpp:class:`simgrid::s4u::Link`), i.e., a single wire.
236 They can also be used to abstract a larger network interconnect, e.g., the entire transcontinental network, into a
237 single element. Links are characterized by their bandwidth and latency, and their sharing is realistic wrt TCP connexions.
238 Another unusual point is that SimGrid links can be used to connect more than two elements, just like
239 hyperlinks in an `hypergraph <https://en.wikipedia.org/wiki/Hypergraph>`_.
241 **Parent tags:** :ref:`pf_tag_zone` (both leaf zones and inner zones) |br|
242 **Children tags:** :ref:`pf_tag_prop` |br|
245 :``id``: Link name. Must be unique over the whole platform.
246 :``bandwidth``: Maximum bandwidth for this link. You must specify a unit as follows.
248 **Units in bytes and powers of 2** (1 KiBps = 1,024 Bps):
249 Bps, KiBps, MiBps, GiBps, TiBps, PiBps, or EiBps. |br|
250 **Units in bits and powers of 2** (1 Bps = 8 bps):
251 bps, Kibps, Mibps, Gibps, Tibps, Pibps, or Eibps. |br|
252 **Units in bytes and powers of 10** (1 KBps = 1,000 Bps):
253 Bps, KBps, MBps, GBps, TBps, PBps, or EBps. |br|
254 **Units in bits and powers of 10:**
255 bps, Kbps, Mbps, Gbps, Tbps, Pbps, or Ebps.
257 :``latency``: Latency for this link (default: 0.0). You must specify a unit as follows.
259 ==== =========== ======================
260 Unit Meaning Duration in seconds
261 ==== =========== ======================
262 ps picosecond 10⁻¹² = 0.000000000001
263 ns nanosecond 10⁻⁹ = 0.000000001
264 us microsecond 10⁻⁶ = 0.000001
265 ms millisecond 10⁻³ = 0.001
270 w week 60 * 60 * 24 * 7
271 ==== =========== ======================
273 :``sharing_policy``: Sharing policy for the link. Possible values are ``SHARED``, ``FATPIPE`` or ``SPLITDUPLEX``
274 (default: ``SPLITDUPLEX``).
276 If set to ``SPLITDUPLEX``, the link models the full-duplex
277 behavior, as meant in TCP or UDP. To that extend, the link is
278 actually split in two links whose names are suffixed with "_UP" and
279 "_DOWN". You should then specify the direction to use when
280 referring to that link in a :ref:`pf_tag_link_ctn`.
282 If set to ``FATPIPE``, flows have no impact on each other, hence
283 each flow can exploit the full bandwidth. This models Internet
284 backbones that cannot get saturated by your application. From your
285 application point of view, there is no congestion on these
288 If set to ``SHARED``, the available bandwidth is fairly shared
289 among ALL flows traversing this link. The resulting link is not
290 full-duplex (as UDP or TCP would be): communications in both
291 directions share the same link. Prefer ``SPLITDUPLEX`` for TCP flows.
293 :``bandwidth_file``: File containing the bandwidth profile.
294 Almost every lines of such files describe timed events as ``date
295 bandwidth`` (in bytes per second).
298 .. code-block:: python
304 - At time t = 4, the bandwidth is of 40 MBps.
305 - At time t = 8, it raises to 60MBps.
306 - At time t = 24, it drops at 40 MBps again.
308 If your profile does not contain any LOOPAFTER line, then it will
309 be executed only once and not repeated.
311 .. warning:: Don't get fooled: Bandwidth and Latency profiles of a :ref:`pf_tag_link` contain absolute values, while
312 Availability profiles of a :ref:`pf_tag_host` contain ratios.
314 :``latency_file``: File containing the latency profile.
315 Almost every lines of such files describe timed events as ``date
316 latency`` (in seconds).
319 .. code-block:: python
325 - At time t = 1, the latency is of 1ms (0.001 second)
326 - At time t = 3, the latency is of 100ms (0.1 second)
327 - At time t = 8 (5 seconds after the last event), the profile loops.
328 - At time t = 9 (1 second after the loop reset), the latency is back at 1ms.
330 If your profile does not contain any LOOPAFTER line, then it will
331 be executed only once and not repeated.
333 .. warning:: Don't get fooled: Bandwidth and Latency profiles of a :ref:`pf_tag_link` contain absolute values, while
334 Availability profiles of a :ref:`pf_tag_host` contain ratios.
336 :``state_file``: File containing the state profile. See :ref:`pf_tag_host`.
338 -------------------------------------------------------------------------------
345 An element in a route, representing a previously defined link.
347 **Parent tags:** :ref:`pf_tag_route` |br|
348 **Children tags:** none |br|
351 :``id``: Link that is to be included in this route.
352 :``direction``: either ``UP`` (by default) or ``DOWN``, specifying whether to
353 use the uplink or downlink component of the link (that must
354 follow the ``SPLITDUPLEX`` sharing policy). |br|
355 Please refer to the ``sharing_policy`` attribute in
358 -------------------------------------------------------------------------------
365 This tag represents a peer, as in Peer-to-Peer (P2P) networks. It is
366 handy to model situations where hosts have an asymmetric
367 connectivity. Computers connected through set-top-boxes usually have a
368 much better download rate than their upload rate. To model this,
369 <peer> creates and connects several elements: a host, an upload link
372 **Parent tags:** :ref:`pf_tag_zone` (only with Vivaldi routing) |br|
373 **Children tags:** none |br|
376 :``id``: Name of the host. Must be unique on the whole platform.
377 :``speed``: Computational power (in flop/s).
379 If you use DVFS, provide a comma-separated list of values for each pstate (see :ref:`howto_dvfs`).
380 :``bw_in``: Bandwidth of the private downstream link, along with its
381 unit. See :ref:`pf_tag_link`.
382 :``bw_out``: Bandwidth of the private upstream link, along with its
383 unit. See :ref:`pf_tag_link`.
384 :``lat``: Latency of both private links. See :ref:`pf_tag_link`.
385 :``coordinates``: Coordinates of the gateway for this peer.
387 The communication latency between a host A = (xA,yA,zA) and a host B = (xB,yB,zB) is computed as follows:
389 latency = sqrt( (xA-xB)² + (yA-yB)² ) + zA + zB
391 See the documentation of
392 :cpp:class:`simgrid::kernel::routing::VivaldiZone` for details on
393 how the latency is computed from the coordinates, and on how the up
394 and down bandwidth are used.
395 :``availability_file``: File containing the availability profile.
396 See the full description in :ref:`pf_tag_host`
397 :``state_file``: File containing the state profile.
398 See the full description in :ref:`pf_tag_host`
400 -------------------------------------------------------------------------------
407 **Parent tags:** none (this is the root tag of every file) |br|
408 **Children tags:** :ref:`pf_tag_config` (must come first),
409 :ref:`pf_tag_cluster`, :ref:`pf_tag_cabinet`, :ref:`pf_tag_peer`,
410 :ref:`pf_tag_zone`, :ref:`pf_tag_trace`, :ref:`pf_tag_trace_connect`, or
411 :ref:`pf_tag_actor` in :ref:`deployment <deploy>` files. |br|
414 :``version``: Version of the DTD, describing the whole XML format.
415 This versioning allow future evolutions, even if we
416 avoid backward-incompatible changes. The current version
417 is **4.1**. The ``simgrid_update_xml`` program can
418 upgrade most of the past platform files to the most recent
421 -------------------------------------------------------------------------------
428 This tag can be used to attach user-defined properties to some
429 platform elements. Both the name and the value can be any string of
430 your wish. You can use this to pass extra parameters to your code and
433 From your code, you can interact with these properties using the
436 - Actor: :cpp:func:`simgrid::s4u::Actor::get_property` or :cpp:func:`MSG_process_get_property_value`
437 - Cluster: this is a zone, see below.
438 - Host: :cpp:func:`simgrid::s4u::Host::get_property` or :cpp:func:`MSG_host_get_property_value`
439 - Link: :cpp:func:`simgrid::s4u::Link::get_property`
440 - Disk: :cpp:func:`simgrid::s4u::Disk::get_property`
441 - Zone: :cpp:func:`simgrid::s4u::NetZone::get_property` of :cpp:func:`MSG_zone_get_property_value`
443 **Parent tags:** :ref:`pf_tag_actor`, :ref:`pf_tag_config`, :ref:`pf_tag_cluster`, :ref:`pf_tag_host`,
444 :ref:`pf_tag_link`, :ref:`pf_tag_disk`,:ref:`pf_tag_zone` |br|
445 **Children tags:** none |br|
448 :``id``: Name of the defined property.
449 :``value``: Value of the defined property.
451 -------------------------------------------------------------------------------
458 A path between two network locations, composed of several occurrences of :ref:`pf_tag_link`. It is only relevant for
459 intra-:ref:`pf_tag_zone` routes. Inter-zone routes must be declared with :ref:`pf_tag_zoneRoute`. More detail can be
460 found in the following sections :ref:`pf_routes` and :ref:`pf_route_usage`.
462 **Parent tags:** :ref:`pf_tag_zone` |br|
463 **Children tags:** :ref:`pf_tag_link_ctn` |br|
466 :``src``: Host from which this route starts. Must be the name of an existing host.
467 :``dst``: Host to which this route leads. Must be the name of an existing host.
468 :``symmetrical``: Whether this route is symmetrical, ie, whether we
469 are defining the route ``dst -> src`` at the same
470 time. Valid values: ``yes``, ``no``, ``YES``, ``NO``
473 -------------------------------------------------------------------------------
480 A router is similar to a :ref:`pf_tag_host`, but it cannot contain any actor. It is only useful to some routing
481 algorithms. In particular, they are useful when you want to use the NS3 bindings to break the routes that are longer
484 **Parent tags:** :ref:`pf_tag_zone` (only leaf zones, i.e., zones containing neither inner zones nor clusters) |br|
487 :``id``: Router name.
488 No other host or router may have the same name over the whole platform.
489 :``coordinates``: Vivaldi coordinates. See :ref:`pf_tag_peer`.
491 -------------------------------------------------------------------------------
498 A networking zone is an area in which elements are located. See :cpp:class:`simgrid::s4u::NetZone`.
500 **Parent tags:** :ref:`pf_tag_platform`, :ref:`pf_tag_zone` (only internal nodes, i.e., zones
501 containing only inner zones or clusters but no basic
502 elements such as host or peer) |br|
503 **Children tags (if internal zone):** :ref:`pf_tag_cluster`, :ref:`pf_tag_link`, :ref:`pf_tag_zone` |br|
504 **Children tags (if leaf zone):** :ref:`pf_tag_host`, :ref:`pf_tag_link`, :ref:`pf_tag_peer` |br|
508 No other zone may have the same name over the whole platform.
509 :``routing``: Routing algorithm to use.
511 -------------------------------------------------------------------------------
513 .. _pf_tag_zoneRoute:
518 All information to factorize the routes between two zones; intra-zone paths must be declared with :ref:`pf_tag_route`.
519 Since zones form a tree, :ref:`pf_tag_zoneRoute` is probably best used to connect sibling zones.
521 When declaring a zoneRoute, you must give the source and destination zones, along with the gateway in each zone (i.e.,
522 the point to reach within that zone to reach the declared route), and the list of links to go from one zone to another.
523 More detail can be found in the following sections :ref:`pf_routes` and :ref:`pf_route_usage`.
525 **Parent tags:** :ref:`pf_tag_zone` |br|
526 **Children tags:** :ref:`pf_tag_link_ctn` |br|
529 :``src``: Zone from which this route starts. Must be an existing zone.
530 :``dst``: Zone to which this route leads. Must be an existing zone.
531 :``gw_src``: Netpoint (within src zone) from which this route starts. Must be an existing host/router.
532 :``gw_dst``: Netpoint (within dst zone) to which this route leads. Must be an existing host/router.
533 :``symmetrical``: Whether this route is symmetrical, ie, whether we are defining the route ``dst -> src`` at the same
534 time. Valid values: ``yes``, ``no``, ``YES``, ``NO``.
536 -------------------------------------------------------------------------------
545 This tag is only available when the routing mode of the network zone
546 is set to ``Cluster``.
548 The cabinet tag is, like the :ref:`pf_tag_cluster` tag,
549 a meta-tag. This means that it is simply a shortcut for creating a set of (homogenous) hosts and links quickly;
550 unsurprisingly, this tag was introduced to setup cabinets in data centers. Unlike
551 :ref:`pf_tag_cluster`, however, the :ref:`pf_tag_cabinet` assumes that you create the backbone
552 and routers yourself; see our examples below.
554 **Parent tags:** :ref:`pf_tag_zone` |br|
555 **Children tags:** none |br|
558 :``id``: The identifier of the cabinet. Facilitates referring to it.
559 :``prefix``: Each node of the cabinet has to have a name. This name will be prefixed with this prefix.
560 :``suffix``: Each node of the cabinet will be suffixed with this suffix.
561 :``radical``: Regexp used to generate cabinet nodes name.
562 Syntax: "10-20" will give you 11 machines numbered from 10 to 20, "10-20;2" will give you 12 machines, one with the number 2, others numbered as before.
564 The produced number is concatenated between prefix and suffix to form machine names.
565 :``speed``: Same as the ``speed`` attribute of the :ref:`pf_tag_host` tag.
566 :``bw``: Bandwidth for the links between nodes and backbone (if any). See the :ref:`pf_tag_link` for syntax/details.
567 :``lat``: Latency for the links between nodes and backbone (if any). See the :ref:`pf_tag_link` for syntax/details.
570 Please note that as of now, it is impossible to change attributes such as,
571 amount of cores (always set to 1), the initial state of hosts/links
572 (always set to ON), the sharing policy of the links (always set to **SPLITDUPLEX** :ref:`pf_tag_link` ).
577 The following example shows how to use the cabinet tag.
579 .. literalinclude:: ../../examples/platforms/meta_cluster.xml
582 Please note that you must specify the :ref:`pf_backbone` tag by yourself;
583 this is not done automatically and there are no checks
584 that ensure this backbone was defined.
586 The hosts generated in the above example are named host-1.cluster, host-2.cluster1
591 Missing tags: <host_link> <backbone> <bypassRoute> <bypassZoneRoute> <actor><argument>