8 <object id="TOC" data="graphical-toc.svg" type="image/svg+xml"></object>
10 window.onload=function() { // Wait for the SVG to be loaded before changing it
11 var elem=document.querySelector("#TOC").contentDocument.getElementById("ConfigBox")
12 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";
18 A number of options can be given at runtime to change the default
19 SimGrid behavior. For a complete list of all configuration options
20 accepted by the SimGrid version used in your simulator, simply pass
21 the --help configuration flag to your program. If some of the options
22 are not documented on this page, this is a bug that you should please
23 report so that we can fix it. Note that some of the options presented
24 here may not be available in your simulators, depending on the
25 :ref:`compile-time options <install_src_config>` that you used.
27 Setting Configuration Items
28 ---------------------------
30 There is several way to pass configuration options to the simulators.
31 The most common way is to use the ``--cfg`` command line argument. For
32 example, to set the item ``Item`` to the value ``Value``, simply
33 type the following on the command-line:
37 my_simulator --cfg=Item:Value (other arguments)
39 Several ``--cfg`` command line arguments can naturally be used. If you
40 need to include spaces in the argument, don't forget to quote the
41 argument. You can even escape the included quotes (write ``@'`` for ``'`` if
42 you have your argument between simple quotes).
44 Another solution is to use the ``<config>`` tag in the platform file. The
45 only restriction is that this tag must occur before the first
46 platform element (be it ``<zone>``, ``<cluster>``, ``<peer>`` or whatever).
47 The ``<config>`` tag takes an ``id`` attribute, but it is currently
48 ignored so you don't really need to pass it. The important part is that
49 within that tag, you can pass one or several ``<prop>`` tags to specify
50 the configuration to use. For example, setting ``Item`` to ``Value``
51 can be done by adding the following to the beginning of your platform
57 <prop id="Item" value="Value"/>
60 A last solution is to pass your configuration directly in your program
61 with :cpp:func:`simgrid::s4u::Engine::set_config` or :cpp:func:`MSG_config`.
65 #include <simgrid/s4u.hpp>
67 int main(int argc, char *argv[]) {
68 simgrid::s4u::Engine e(&argc, argv);
70 simgrid::s4u::Engine::set_config("Item:Value");
77 Existing Configuration Items
78 ----------------------------
81 The full list can be retrieved by passing ``--help`` and
82 ``--help-cfg`` to an executable that uses SimGrid. Try passing
83 ``help`` as a value to get the list of values accepted by a given
84 option. For example, ``--cfg=plugin:help`` will give you the list
85 of plugins available in your installation of SimGrid.
87 - **contexts/factory:** :ref:`cfg=contexts/factory`
88 - **contexts/guard-size:** :ref:`cfg=contexts/guard-size`
89 - **contexts/nthreads:** :ref:`cfg=contexts/nthreads`
90 - **contexts/stack-size:** :ref:`cfg=contexts/stack-size`
91 - **contexts/synchro:** :ref:`cfg=contexts/synchro`
93 - **cpu/maxmin-selective-update:** :ref:`Cpu Optimization Level <options_model_optim>`
94 - **cpu/model:** :ref:`options_model_select`
95 - **cpu/optim:** :ref:`Cpu Optimization Level <options_model_optim>`
97 - **debug/breakpoint:** :ref:`cfg=debug/breakpoint`
98 - **debug/clean-atexit:** :ref:`cfg=debug/clean-atexit`
99 - **debug/verbose-exit:** :ref:`cfg=debug/verbose-exit`
101 - **exception/cutpath:** :ref:`cfg=exception/cutpath`
103 - **host/model:** :ref:`options_model_select`
105 - **maxmin/precision:** :ref:`cfg=maxmin/precision`
106 - **maxmin/concurrency-limit:** :ref:`cfg=maxmin/concurrency-limit`
108 - **msg/debug-multiple-use:** :ref:`cfg=msg/debug-multiple-use`
110 - **model-check:** :ref:`options_modelchecking`
111 - **model-check/checkpoint:** :ref:`cfg=model-check/checkpoint`
112 - **model-check/communications-determinism:** :ref:`cfg=model-check/communications-determinism`
113 - **model-check/dot-output:** :ref:`cfg=model-check/dot-output`
114 - **model-check/max-depth:** :ref:`cfg=model-check/max-depth`
115 - **model-check/property:** :ref:`cfg=model-check/property`
116 - **model-check/reduction:** :ref:`cfg=model-check/reduction`
117 - **model-check/replay:** :ref:`cfg=model-check/replay`
118 - **model-check/send-determinism:** :ref:`cfg=model-check/send-determinism`
119 - **model-check/termination:** :ref:`cfg=model-check/termination`
120 - **model-check/timeout:** :ref:`cfg=model-check/timeout`
121 - **model-check/visited:** :ref:`cfg=model-check/visited`
123 - **network/bandwidth-factor:** :ref:`cfg=network/bandwidth-factor`
124 - **network/crosstraffic:** :ref:`cfg=network/crosstraffic`
125 - **network/latency-factor:** :ref:`cfg=network/latency-factor`
126 - **network/loopback-lat:** :ref:`cfg=network/loopback`
127 - **network/loopback-bw:** :ref:`cfg=network/loopback`
128 - **network/maxmin-selective-update:** :ref:`Network Optimization Level <options_model_optim>`
129 - **network/model:** :ref:`options_model_select`
130 - **network/optim:** :ref:`Network Optimization Level <options_model_optim>`
131 - **network/TCP-gamma:** :ref:`cfg=network/TCP-gamma`
132 - **network/weight-S:** :ref:`cfg=network/weight-S`
134 - **ns3/TcpModel:** :ref:`options_pls`
135 - **ns3/seed:** :ref:`options_pls`
136 - **path:** :ref:`cfg=path`
137 - **plugin:** :ref:`cfg=plugin`
139 - **storage/max_file_descriptors:** :ref:`cfg=storage/max_file_descriptors`
141 - **surf/precision:** :ref:`cfg=surf/precision`
143 - **For collective operations of SMPI,** please refer to Section :ref:`cfg=smpi/coll-selector`
144 - **smpi/async-small-thresh:** :ref:`cfg=smpi/async-small-thresh`
145 - **smpi/buffering:** :ref:`cfg=smpi/buffering`
146 - **smpi/bw-factor:** :ref:`cfg=smpi/bw-factor`
147 - **smpi/coll-selector:** :ref:`cfg=smpi/coll-selector`
148 - **smpi/comp-adjustment-file:** :ref:`cfg=smpi/comp-adjustment-file`
149 - **smpi/cpu-threshold:** :ref:`cfg=smpi/cpu-threshold`
150 - **smpi/display-timing:** :ref:`cfg=smpi/display-timing`
151 - **smpi/grow-injected-times:** :ref:`cfg=smpi/grow-injected-times`
152 - **smpi/host-speed:** :ref:`cfg=smpi/host-speed`
153 - **smpi/IB-penalty-factors:** :ref:`cfg=smpi/IB-penalty-factors`
154 - **smpi/iprobe:** :ref:`cfg=smpi/iprobe`
155 - **smpi/iprobe-cpu-usage:** :ref:`cfg=smpi/iprobe-cpu-usage`
156 - **smpi/init:** :ref:`cfg=smpi/init`
157 - **smpi/keep-temps:** :ref:`cfg=smpi/keep-temps`
158 - **smpi/lat-factor:** :ref:`cfg=smpi/lat-factor`
159 - **smpi/ois:** :ref:`cfg=smpi/ois`
160 - **smpi/or:** :ref:`cfg=smpi/or`
161 - **smpi/os:** :ref:`cfg=smpi/os`
162 - **smpi/papi-events:** :ref:`cfg=smpi/papi-events`
163 - **smpi/privatization:** :ref:`cfg=smpi/privatization`
164 - **smpi/privatize-libs:** :ref:`cfg=smpi/privatize-libs`
165 - **smpi/send-is-detached-thresh:** :ref:`cfg=smpi/send-is-detached-thresh`
166 - **smpi/shared-malloc:** :ref:`cfg=smpi/shared-malloc`
167 - **smpi/shared-malloc-hugepage:** :ref:`cfg=smpi/shared-malloc-hugepage`
168 - **smpi/simulate-computation:** :ref:`cfg=smpi/simulate-computation`
169 - **smpi/test:** :ref:`cfg=smpi/test`
170 - **smpi/wtime:** :ref:`cfg=smpi/wtime`
172 - **Tracing configuration options** can be found in Section :ref:`tracing_tracing_options`
174 - **storage/model:** :ref:`options_model_select`
176 - **vm/model:** :ref:`options_model_select`
180 Configuring the Platform Models
181 -------------------------------
183 .. _options_model_select:
185 Choosing the Platform Models
186 ............................
188 SimGrid comes with several network, CPU and disk models built in,
189 and you can change the used model at runtime by changing the passed
190 configuration. The three main configuration items are given below.
191 For each of these items, passing the special ``help`` value gives you
192 a short description of all possible values (for example,
193 ``--cfg=network/model:help`` will present all provided network
194 models). Also, ``--help-models`` should provide information about all
195 models for all existing resources.
197 - ``network/model``: specify the used network model. Possible values:
199 - **LV08 (default one):** Realistic network analytic model
200 (slow-start modeled by multiplying latency by 13.01, bandwidth by
201 .97; bottleneck sharing uses a payload of S=20537 for evaluating
202 RTT). Described in `Accuracy Study and Improvement of Network
203 Simulation in the SimGrid Framework
204 <http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf>`_.
205 - **Constant:** Simplistic network model where all communication
206 take a constant time (one second). This model provides the lowest
207 realism, but is (marginally) faster.
208 - **SMPI:** Realistic network model specifically tailored for HPC
209 settings (accurate modeling of slow start with correction factors on
210 three intervals: < 1KiB, < 64 KiB, >= 64 KiB). This model can be
211 :ref:`further configured <options_model_network>`.
212 - **IB:** Realistic network model specifically tailored for HPC
213 settings with InfiniBand networks (accurate modeling contention
214 behavior, based on the model explained in `this PhD work
215 <http://mescal.imag.fr/membres/jean-marc.vincent/index.html/PhD/Vienne.pdf>`_.
216 This model can be :ref:`further configured <options_model_network>`.
217 - **CM02:** Legacy network analytic model. Very similar to LV08, but
218 without corrective factors. The timings of small messages are thus
219 poorly modeled. This model is described in `A Network Model for
220 Simulation of Grid Application
221 <https://hal.inria.fr/inria-00071989/document>`_.
222 - **ns-3** (only available if you compiled SimGrid accordingly):
223 Use the packet-level network
224 simulators as network models (see :ref:`model_ns3`).
225 This model can be :ref:`further configured <options_pls>`.
227 - ``cpu/model``: specify the used CPU model. We have only one model
230 - **Cas01:** Simplistic CPU model (time=size/speed)
232 - ``host/model``: The host concept is the aggregation of a CPU with a
233 network card. Three models exists, but actually, only 2 of them are
234 interesting. The "compound" one is simply due to the way our
235 internal code is organized, and can easily be ignored. So at the
236 end, you have two host models: The default one allows aggregation of
237 an existing CPU model with an existing network model, but does not
238 allow parallel tasks because these beasts need some collaboration
239 between the network and CPU model. That is why, ptask_07 is used by
240 default when using SimDag.
242 - **default:** Default host model. Currently, CPU:Cas01 and
243 network:LV08 (with cross traffic enabled)
244 - **compound:** Host model that is automatically chosen if
245 you change the network and CPU models
246 - **ptask_L07:** Host model somehow similar to Cas01+CM02 but
247 allowing "parallel tasks", that are intended to model the moldable
248 tasks of the grid scheduling literature.
250 - ``storage/model``: specify the used storage model. Only one model is
252 - ``vm/model``: specify the model for virtual machines. Only one model
255 .. todo: make 'compound' the default host model.
257 .. _options_model_optim:
262 The network and CPU models that are based on lmm_solve (that
263 is, all our analytical models) accept specific optimization
266 - items ``network/optim`` and ``cpu/optim`` (both default to 'Lazy'):
268 - **Lazy:** Lazy action management (partial invalidation in lmm +
269 heap in action remaining).
270 - **TI:** Trace integration. Highly optimized mode when using
271 availability traces (only available for the Cas01 CPU model for
273 - **Full:** Full update of remaining and variables. Slow but may be
274 useful when debugging.
276 - items ``network/maxmin-selective-update`` and
277 ``cpu/maxmin-selective-update``: configure whether the underlying
278 should be lazily updated or not. It should have no impact on the
279 computed timings, but should speed up the computation. |br| It is
280 still possible to disable this feature because it can reveal
281 counter-productive in very specific scenarios where the
282 interaction level is high. In particular, if all your
283 communication share a given backbone link, you should disable it:
284 without it, a simple regular loop is used to update each
285 communication. With it, each of them is still updated (because of
286 the dependency induced by the backbone), but through a complicated
287 and slow pattern that follows the actual dependencies.
289 .. _cfg=maxmin/precision:
290 .. _cfg=surf/precision:
295 **Option** ``maxmin/precision`` **Default:** 0.00001 (in flops or bytes) |br|
296 **Option** ``surf/precision`` **Default:** 0.00001 (in seconds)
298 The analytical models handle a lot of floating point values. It is
299 possible to change the epsilon used to update and compare them through
300 this configuration item. Changing it may speedup the simulation by
301 discarding very small actions, at the price of a reduced numerical
302 precision. You can modify separately the precision used to manipulate
303 timings (in seconds) and the one used to manipulate amounts of work
306 .. _cfg=maxmin/concurrency-limit:
311 **Option** ``maxmin/concurrency-limit`` **Default:** -1 (no limit)
313 The maximum number of variables per resource can be tuned through this
314 option. You can have as many simultaneous actions per resources as you
315 want. If your simulation presents a very high level of concurrency, it
316 may help to use e.g. 100 as a value here. It means that at most 100
317 actions can consume a resource at a given time. The extraneous actions
318 are queued and wait until the amount of concurrency of the considered
319 resource lowers under the given boundary.
321 Such limitations help both to the simulation speed and simulation accuracy
322 on highly constrained scenarios, but the simulation speed suffers of this
323 setting on regular (less constrained) scenarios so it is off by default.
325 .. _options_model_network:
327 Configuring the Network Model
328 .............................
330 .. _cfg=network/TCP-gamma:
332 Maximal TCP Window Size
333 ^^^^^^^^^^^^^^^^^^^^^^^
335 **Option** ``network/TCP-gamma`` **Default:** 4194304
337 The analytical models need to know the maximal TCP window size to take
338 the TCP congestion mechanism into account. On Linux, this value can
339 be retrieved using the following commands. Both give a set of values,
340 and you should use the last one, which is the maximal size.
342 .. code-block:: shell
344 cat /proc/sys/net/ipv4/tcp_rmem # gives the sender window
345 cat /proc/sys/net/ipv4/tcp_wmem # gives the receiver window
347 .. _cfg=smpi/IB-penalty-factors:
348 .. _cfg=network/bandwidth-factor:
349 .. _cfg=network/latency-factor:
350 .. _cfg=network/weight-S:
352 Correcting Important Network Parameters
353 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
355 SimGrid can take network irregularities such as a slow startup or
356 changing behavior depending on the message size into account. You
357 should not change these values unless you really know what you're
358 doing. The corresponding values were computed through data fitting
359 one the timings of packet-level simulators, as described in `Accuracy
360 Study and Improvement of Network Simulation in the SimGrid Framework
361 <http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf>`_.
364 If you are using the SMPI model, these correction coefficients are
365 themselves corrected by constant values depending on the size of the
366 exchange. By default SMPI uses factors computed on the Stampede
367 Supercomputer at TACC, with optimal deployment of processes on
368 nodes. Again, only hardcore experts should bother about this fact.
370 InfiniBand network behavior can be modeled through 3 parameters
371 ``smpi/IB-penalty-factors:"βe;βs;γs"``, as explained in `this PhD
373 <http://mescal.imag.fr/membres/jean-marc.vincent/index.html/PhD/Vienne.pdf>`_.
375 .. todo:: This section should be rewritten, and actually explain the
376 options network/bandwidth-factor, network/latency-factor,
379 .. _cfg=network/crosstraffic:
381 Simulating Cross-Traffic
382 ^^^^^^^^^^^^^^^^^^^^^^^^
384 Since SimGrid v3.7, cross-traffic effects can be taken into account in
385 analytical simulations. It means that ongoing and incoming
386 communication flows are treated independently. In addition, the LV08
387 model adds 0.05 of usage on the opposite direction for each new
388 created flow. This can be useful to simulate some important TCP
389 phenomena such as ack compression.
391 For that to work, your platform must have two links for each
392 pair of interconnected hosts. An example of usable platform is
393 available in ``examples/platforms/crosstraffic.xml``.
395 This is activated through the ``network/crosstraffic`` item, that
396 can be set to 0 (disable this feature) or 1 (enable it).
398 Note that with the default host model this option is activated by default.
400 .. _cfg=network/loopback:
402 Configuring loopback link
403 ^^^^^^^^^^^^^^^^^^^^^^^^^
405 Several network model provide an implicit loopback link to account for local
406 communication on a host. By default it has a 10GBps bandwidth and a null latency.
407 This can be changed with ``network/loopback-lat`` and ``network/loopback-bw``
410 .. _cfg=smpi/async-small-thresh:
412 Simulating Asynchronous Send
413 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
415 (this configuration item is experimental and may change or disappear)
417 It is possible to specify that messages below a certain size (in bytes) will be
418 sent as soon as the call to MPI_Send is issued, without waiting for
419 the correspondent receive. This threshold can be configured through
420 the ``smpi/async-small-thresh`` item. The default value is 0. This
421 behavior can also be manually set for mailboxes, by setting the
422 receiving mode of the mailbox with a call to
423 :cpp:func:`MSG_mailbox_set_async`. After this, all messages sent to
424 this mailbox will have this behavior regardless of the message size.
426 This value needs to be smaller than or equals to the threshold set at
427 :ref:`cfg=smpi/send-is-detached-thresh`, because asynchronous messages
428 are meant to be detached as well.
435 **Option** ``ns3/TcpModel`` **Default:** "default" (ns-3 default)
437 When using ns-3, there is an extra item ``ns3/TcpModel``, corresponding
438 to the ``ns3::TcpL4Protocol::SocketType`` configuration item in
439 ns-3. The only valid values (enforced on the SimGrid side) are
440 'default' (no change to the ns-3 configuration), 'NewReno' or 'Reno' or
443 **Option** ``ns3/seed`` **Default:** "" (don't set the seed in ns-3)
445 This option is the random seed to provide to ns-3 with
446 ``ns3::RngSeedManager::SetSeed`` and ``ns3::RngSeedManager::SetRun``.
448 If left blank, no seed is set in ns-3. If the value 'time' is
449 provided, the current amount of seconds since epoch is used as a seed.
450 Otherwise, the provided value must be a number to use as a seed.
452 Configuring the Storage model
453 .............................
455 .. _cfg=storage/max_file_descriptors:
457 File Descriptor Count per Host
458 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
460 **Option** ``storage/max_file_descriptors`` **Default:** 1024
462 Each host maintains a fixed-size array of its file descriptors. You
463 can change its size through this item to either enlarge it if your
464 application requires it or to reduce it to save memory space.
471 SimGrid plugins allow one to extend the framework without changing its
472 source code directly. Read the source code of the existing plugins to
473 learn how to do so (in ``src/plugins``), and ask your questions to the
474 usual channels (Stack Overflow, Mailing list, IRC). The basic idea is
475 that plugins usually register callbacks to some signals of interest.
476 If they need to store some information about a given object (Link, CPU
477 or Actor), they do so through the use of a dedicated object extension.
479 Some of the existing plugins can be activated from the command line,
480 meaning that you can activate them from the command line without any
481 modification to your simulation code. For example, you can activate
482 the host energy plugin by adding ``--cfg=plugin:host_energy`` to your
485 Here is a partial list of plugins that can be activated this way. You can get
486 the full list by passing ``--cfg=plugin:help`` to your simulator.
488 - :ref:`Host Energy <plugin_host_energy>`: models the energy dissipation of the compute units.
489 - :ref:`Link Energy <plugin_link_energy>`: models the energy dissipation of the network.
490 - :ref:`Host Load <plugin_host_load>`: monitors the load of the compute units.
492 .. _options_modelchecking:
494 Configuring the Model-Checking
495 ------------------------------
497 To enable SimGrid's model-checking support, the program should
498 be executed using the simgrid-mc wrapper:
500 .. code-block:: shell
502 simgrid-mc ./my_program
504 Safety properties are expressed as assertions using the function
505 :cpp:func:`void MC_assert(int prop)`.
507 .. _cfg=smpi/buffering:
509 Specifying the MPI buffering behavior
510 .....................................
512 **Option** ``smpi/buffering`` **Default:** infty
514 Buffering in MPI has a huge impact on the communication semantic. For example,
515 standard blocking sends are synchronous calls when the system buffers are full
516 while these calls can complete immediately without even requiring a matching
517 receive call for small messages sent when the system buffers are empty.
519 In SMPI, this depends on the message size, that is compared against two thresholds:
521 - if (size < :ref:`smpi/async-small-thresh <cfg=smpi/async-small-thresh>`) then
522 MPI_Send returns immediately, even if the corresponding receive has not be issued yet.
523 - if (:ref:`smpi/async-small-thresh <cfg=smpi/async-small-thresh>` < size < :ref:`smpi/send-is-detached-thresh <cfg=smpi/send-is-detached-thresh>`) then
524 MPI_Send returns as soon as the corresponding receive has been issued. This is known as the eager mode.
525 - if (:ref:`smpi/send-is-detached-thresh <cfg=smpi/send-is-detached-thresh>` < size) then
526 MPI_Send returns only when the message has actually been sent over the network. This is known as the rendez-vous mode.
528 The ``smpi/buffering`` (only valid with MC) option gives an easier interface to choose between these semantics. It can take two values:
530 - **zero:** means that buffering should be disabled. All communications are actually blocking.
531 - **infty:** means that buffering should be made infinite. All communications are non-blocking.
533 .. _cfg=model-check/property:
535 Specifying a liveness property
536 ..............................
538 **Option** ``model-check/property`` **Default:** unset
540 If you want to specify liveness properties, you have to pass them on
541 the command line, specifying the name of the file containing the
542 property, as formatted by the `ltl2ba <https://github.com/utwente-fmt/ltl2ba>`_ program.
543 Note that ltl2ba is not part of SimGrid and must be installed separately.
545 .. code-block:: shell
547 simgrid-mc ./my_program --cfg=model-check/property:<filename>
549 .. _cfg=model-check/checkpoint:
551 Going for Stateful Verification
552 ...............................
554 By default, the system is backtracked to its initial state to explore
555 another path, instead of backtracking to the exact step before the fork
556 that we want to explore (this is called stateless verification). This
557 is done this way because saving intermediate states can rapidly
558 exhaust the available memory. If you want, you can change the value of
559 the ``model-check/checkpoint`` item. For example,
560 ``--cfg=model-check/checkpoint:1`` asks to take a checkpoint every
561 step. Beware, this will certainly explode your memory. Larger values
562 are probably better, make sure to experiment a bit to find the right
563 setting for your specific system.
565 .. _cfg=model-check/reduction:
567 Specifying the kind of reduction
568 ................................
570 The main issue when using the model-checking is the state space
571 explosion. You can activate some reduction technique with
572 ``--cfg=model-check/reduction:<technique>``. For now, this
573 configuration variable can take 2 values:
575 - **none:** Do not apply any kind of reduction (mandatory for
576 liveness properties, as our current DPOR algorithm breaks cycles)
577 - **dpor:** Apply Dynamic Partial Ordering Reduction. Only valid if
578 you verify local safety properties (default value for safety
581 Another way to mitigate the state space explosion is to search for
582 cycles in the exploration with the :ref:`cfg=model-check/visited`
583 configuration. Note that DPOR and state-equality reduction may not
584 play well together. You should choose between them.
586 Our current DPOR implementation could be improved in may ways. We are
587 currently improving its efficiency (both in term of reduction ability
588 and computational speed), and future work could make it compatible
589 with liveness properties.
591 .. _cfg=model-check/visited:
593 Size of Cycle Detection Set (state equality reduction)
594 ......................................................
596 Mc SimGrid can be asked to search for cycles during the exploration,
597 i.e. situations where a new explored state is in fact the same state
598 than a previous one.. This can prove useful to mitigate the state
599 space explosion with safety properties, and this is the crux when
600 searching for counter-examples to the liveness properties.
602 Note that this feature may break the current implementation of the
603 DPOR reduction technique.
605 The ``model-check/visited`` item is the maximum number of states, which
606 are stored in memory. If the maximum number of snapshotted state is
607 reached, some states will be removed from the memory and some cycles
608 might be missed. Small values can lead to incorrect verifications, but
609 large values can exhaust your memory and be CPU intensive as each new
610 state must be compared to that amount of older saved states.
612 The default settings depend on the kind of exploration. With safety
613 checking, no state is snapshotted and cycles cannot be detected. With
614 liveness checking, all states are snapshotted because missing a cycle
615 could hinder the exploration soundness.
617 .. _cfg=model-check/termination:
619 Non-Termination Detection
620 .........................
622 The ``model-check/termination`` configuration item can be used to
623 report if a non-termination execution path has been found. This is a
624 path with a cycle, which means that the program might never terminate.
626 This only works in safety mode, not in liveness mode.
628 This options is disabled by default.
630 .. _cfg=model-check/dot-output:
635 If set, the ``model-check/dot-output`` configuration item is the name
636 of a file in which to write a dot file of the path leading to the
637 property violation discovered (safety or liveness violation), as well
638 as the cycle for liveness properties. This dot file can then be fed to the
639 graphviz dot tool to generate a corresponding graphical representation.
641 .. _cfg=model-check/max-depth:
643 Exploration Depth Limit
644 .......................
646 The ``model-checker/max-depth`` can set the maximum depth of the
647 exploration graph of the model checker. If this limit is reached, a
648 logging message is sent and the results might not be exact.
650 By default, there is no depth limit.
652 .. _cfg=model-check/timeout:
657 By default, the model checker does not handle timeout conditions: the `wait`
658 operations never time out. With the ``model-check/timeout`` configuration item
659 set to **yes**, the model checker will explore timeouts of `wait` operations.
661 .. _cfg=model-check/communications-determinism:
662 .. _cfg=model-check/send-determinism:
664 Communication Determinism
665 .........................
667 The ``model-check/communications-determinism`` and
668 ``model-check/send-determinism`` items can be used to select the
669 communication determinism mode of the model checker, which checks
670 determinism properties of the communications of an application.
674 Verification Performance Considerations
675 .......................................
677 The size of the stacks can have a huge impact on the memory
678 consumption when using model-checking. By default, each snapshot will
679 save a copy of the whole stacks and not only of the part that is
680 really meaningful: you should expect the contribution of the memory
681 consumption of the snapshots to be:
682 :math:`\text{number of processes} \times \text{stack size} \times \text{number of states}`.
684 When compiled against the model checker, the stacks are not
685 protected with guards: if the stack size is too small for your
686 application, the stack will silently overflow into other parts of the
687 memory (see :ref:`contexts/guard-size <cfg=contexts/guard-size>`).
689 .. _cfg=model-check/replay:
691 Replaying buggy execution paths from the model checker
692 ......................................................
694 Debugging the problems reported by the model checker is challenging:
695 First, the application under verification cannot be debugged with gdb
696 because the model checker already traces it. Then, the model checker may
697 explore several execution paths before encountering the issue, making it
698 very difficult to understand the output. Fortunately, SimGrid provides
699 the execution path leading to any reported issue so that you can replay
700 this path reported by the model checker, enabling the usage of classical
703 When the model checker finds an interesting path in the application
704 execution graph (where a safety or liveness property is violated), it
705 generates an identifier for this path. Here is an example of the output:
707 .. code-block:: shell
709 [ 0.000000] (0:@) Check a safety property
710 [ 0.000000] (0:@) **************************
711 [ 0.000000] (0:@) *** PROPERTY NOT VALID ***
712 [ 0.000000] (0:@) **************************
713 [ 0.000000] (0:@) Counter-example execution trace:
714 [ 0.000000] (0:@) [(1)Tremblay (app)] MC_RANDOM(3)
715 [ 0.000000] (0:@) [(1)Tremblay (app)] MC_RANDOM(4)
716 [ 0.000000] (0:@) Path = 1/3;1/4
717 [ 0.000000] (0:@) Expanded states = 27
718 [ 0.000000] (0:@) Visited states = 68
719 [ 0.000000] (0:@) Executed transitions = 46
721 The interesting line is ``Path = 1/3;1/4``, which means that you should use
722 ``--cfg=model-check/replay:1/3;1/4`` to replay your application on the buggy
723 execution path. All options (but the model checker related ones) must
724 remain the same. In particular, if you ran your application with
725 ``smpirun -wrapper simgrid-mc``, then do it again. Remove all
726 MC-related options, keep non-MC-related ones and add
727 ``--cfg=model-check/replay:???``.
729 Currently, if the path is of the form ``X;Y;Z``, each number denotes
730 the actor's pid that is selected at each indecision point. If it's of
731 the form ``X/a;Y/b``, the X and Y are the selected pids while the a
732 and b are the return values of their simcalls. In the previous
733 example, ``1/3;1/4``, you can see from the full output that the actor
734 1 is doing MC_RANDOM simcalls, so the 3 and 4 simply denote the values
735 that these simcall return.
737 Configuring the User Code Virtualization
738 ----------------------------------------
740 .. _cfg=contexts/factory:
742 Selecting the Virtualization Factory
743 ....................................
745 **Option** contexts/factory **Default:** "raw"
747 In SimGrid, the user code is virtualized in a specific mechanism that
748 allows the simulation kernel to control its execution: when a user
749 process requires a blocking action (such as sending a message), it is
750 interrupted, and only gets released when the simulated clock reaches
751 the point where the blocking operation is done. This is explained
752 graphically in the `relevant tutorial, available online
753 <https://simgrid.org/tutorials/simgrid-simix-101.pdf>`_.
755 In SimGrid, the containers in which user processes are virtualized are
756 called contexts. Several context factory are provided, and you can
757 select the one you want to use with the ``contexts/factory``
758 configuration item. Some of the following may not exist on your
759 machine because of portability issues. In any case, the default one
760 should be the most effcient one (please report bugs if the
761 auto-detection fails for you). They are approximately sorted here from
762 the slowest to the most efficient:
764 - **thread:** very slow factory using full featured threads (either
765 pthreads or windows native threads). They are slow but very
766 standard. Some debuggers or profilers only work with this factory.
767 - **java:** Java applications are virtualized onto java threads (that
768 are regular pthreads registered to the JVM)
769 - **ucontext:** fast factory using System V contexts (Linux and FreeBSD only)
770 - **boost:** This uses the `context
771 implementation <http://www.boost.org/doc/libs/1_59_0/libs/context/doc/html/index.html>`_
772 of the boost library for a performance that is comparable to our
774 |br| Install the relevant library (e.g. with the
775 libboost-contexts-dev package on Debian/Ubuntu) and recompile
777 - **raw:** amazingly fast factory using a context switching mechanism
778 of our own, directly implemented in assembly (only available for x86
779 and amd64 platforms for now) and without any unneeded system call.
781 The main reason to change this setting is when the debugging tools become
782 fooled by the optimized context factories. Threads are the most
783 debugging-friendly contexts, as they allow one to set breakpoints
784 anywhere with gdb and visualize backtraces for all processes, in order
785 to debug concurrency issues. Valgrind is also more comfortable with
786 threads, but it should be usable with all factories (Exception: the
787 callgrind tool really dislikes raw and ucontext factories).
789 .. _cfg=contexts/stack-size:
791 Adapting the Stack Size
792 .......................
794 **Option** ``contexts/stack-size`` **Default:** 8192 KiB
796 Each virtualized used process is executed using a specific system
797 stack. The size of this stack has a huge impact on the simulation
798 scalability, but its default value is rather large. This is because
799 the error messages that you get when the stack size is too small are
800 rather disturbing: this leads to stack overflow (overwriting other
801 stacks), leading to segfaults with corrupted stack traces.
803 If you want to push the scalability limits of your code, you might
804 want to reduce the ``contexts/stack-size`` item. Its default value is
805 8192 (in KiB), while our Chord simulation works with stacks as small
806 as 16 KiB, for example. You can ensure that some actors have a specific
807 size by simply changing the value of this configuration item before
808 creating these actors. The :cpp:func:`simgrid::s4u::Engine::set_config`
809 functions are handy for that.
811 This *setting is ignored* when using the thread factory (because there
812 is no way to modify the stack size with C++ system threads). Instead,
813 you should compile SimGrid and your application with
814 ``-fsplit-stack``. Note that this compilation flag is not compatible
815 with the model checker right now.
817 The operating system should only allocate memory for the pages of the
818 stack which are actually used and you might not need to use this in
819 most cases. However, this setting is very important when using the
820 model checker (see :ref:`options_mc_perf`).
822 .. _cfg=contexts/guard-size:
824 Disabling Stack Guard Pages
825 ...........................
827 **Option** ``contexts/guard-size`` **Default** 1 page in most case (0 pages on Windows or with MC)
829 Unless you use the threads context factory (see
830 :ref:`cfg=contexts/factory`), a stack guard page is usually used
831 which prevents the stack of a given actor from overflowing on another
832 stack. But the performance impact may become prohibitive when the
833 amount of actors increases. The option ``contexts/guard-size`` is the
834 number of stack guard pages used. By setting it to 0, no guard pages
835 will be used: in this case, you should avoid using small stacks (with
836 :ref:`contexts/stack-size <cfg=contexts/stack-size>`) as the stack
837 will silently overflow on other parts of the memory.
839 When no stack guard page is created, stacks may then silently overflow
840 on other parts of the memory if their size is too small for the
843 .. _cfg=contexts/nthreads:
844 .. _cfg=contexts/synchro:
846 Running User Code in Parallel
847 .............................
849 Parallel execution of the user code is only considered stable in
850 SimGrid v3.7 and higher, and mostly for MSG simulations. SMPI
851 simulations may well fail in parallel mode. It is described in
852 `INRIA RR-7653 <http://hal.inria.fr/inria-00602216/>`_.
854 If you are using the **ucontext** or **raw** context factories, you can
855 request to execute the user code in parallel. Several threads are
856 launched, each of them handling the same number of user contexts at each
857 run. To activate this, set the ``contexts/nthreads`` item to the amount
858 of cores that you have in your computer (or lower than 1 to have the
859 amount of cores auto-detected).
861 When parallel execution is activated, you can choose the
862 synchronization schema used with the ``contexts/synchro`` item,
863 which value is either:
865 - **futex:** ultra optimized synchronisation schema, based on futexes
866 (fast user-mode mutexes), and thus only available on Linux systems.
867 This is the default mode when available.
868 - **posix:** slow but portable synchronisation using only POSIX
870 - **busy_wait:** not really a synchronisation: the worker threads
871 constantly request new contexts to execute. It should be the most
872 efficient synchronisation schema, but it loads all the cores of
873 your machine for no good reason. You probably prefer the other less
876 Configuring the Tracing
877 -----------------------
879 The :ref:`tracing subsystem <outcomes_vizu>` can be configured in
880 several different ways depending on the nature of the simulator (MSG,
881 SimDag, SMPI) and the kind of traces that need to be obtained. See the
882 :ref:`Tracing Configuration Options subsection
883 <tracing_tracing_options>` to get a detailed description of each
884 configuration option.
886 We detail here a simple way to get the traces working for you, even if
887 you never used the tracing API.
890 - Any SimGrid-based simulator (MSG, SimDag, SMPI, ...) and raw traces:
892 .. code-block:: shell
894 --cfg=tracing:yes --cfg=tracing/uncategorized:yes
896 The first parameter activates the tracing subsystem, and the second
897 tells it to trace host and link utilization (without any
900 - MSG or SimDag-based simulator and categorized traces (you need to
901 declare categories and classify your tasks according to them)
903 .. code-block:: shell
905 --cfg=tracing:yes --cfg=tracing/categorized:yes
907 The first parameter activates the tracing subsystem, and the second
908 tells it to trace host and link categorized utilization.
910 - SMPI simulator and traces for a space/time view:
912 .. code-block:: shell
916 The `-trace` parameter for the smpirun script runs the simulation
917 with ``--cfg=tracing:yes --cfg=tracing/smpi:yes``. Check the
918 smpirun's `-help` parameter for additional tracing options.
920 Sometimes you might want to put additional information on the trace to
921 correctly identify them later, or to provide data that can be used to
922 reproduce an experiment. You have two ways to do that:
924 - Add a string on top of the trace file as comment:
926 .. code-block:: shell
928 --cfg=tracing/comment:my_simulation_identifier
930 - Add the contents of a textual file on top of the trace file as comment:
932 .. code-block:: shell
934 --cfg=tracing/comment-file:my_file_with_additional_information.txt
936 Please, use these two parameters (for comments) to make reproducible
937 simulations. For additional details about this and all tracing
938 options, check See the :ref:`tracing_tracing_options`.
943 .. _cfg=msg/debug-multiple-use:
948 **Option** ``msg/debug-multiple-use`` **Default:** off
950 Sometimes your application may try to send a task that is still being
951 executed somewhere else, making it impossible to send this task. However,
952 for debugging purposes, one may want to know what the other host is/was
953 doing. This option shows a backtrace of the other process.
958 The SMPI interface provides several specific configuration items.
959 These are not easy to see, since the code is usually launched through the
960 ``smiprun`` script directly.
962 .. _cfg=smpi/host-speed:
963 .. _cfg=smpi/cpu-threshold:
964 .. _cfg=smpi/simulate-computation:
966 Automatic Benchmarking of SMPI Code
967 ...................................
969 In SMPI, the sequential code is automatically benchmarked, and these
970 computations are automatically reported to the simulator. That is to
971 say that if you have a large computation between a ``MPI_Recv()`` and
972 a ``MPI_Send()``, SMPI will automatically benchmark the duration of
973 this code, and create an execution task within the simulator to take
974 this into account. For that, the actual duration is measured on the
975 host machine and then scaled to the power of the corresponding
976 simulated machine. The variable ``smpi/host-speed`` allows one to
977 specify the computational speed of the host machine (in flop/s by
978 default) to use when scaling the execution times.
980 The default value is ``smpi/host-speed=20kf`` (= 20,000 flop/s). This
981 is probably underestimated for most machines, leading SimGrid to
982 overestimate the amount of flops in the execution blocks that are
983 automatically injected in the simulator. As a result, the execution
984 time of the whole application will probably be overestimated until you
985 use a realistic value.
987 When the code consists of numerous consecutive MPI calls, the
988 previous mechanism feeds the simulation kernel with numerous tiny
989 computations. The ``smpi/cpu-threshold`` item becomes handy when this
990 impacts badly on the simulation performance. It specifies a threshold (in
991 seconds) below which the execution chunks are not reported to the
992 simulation kernel (default value: 1e-6).
994 .. note:: The option ``smpi/cpu-threshold`` ignores any computation
995 time spent below this threshold. SMPI does not consider the
996 `amount of time` of these computations; there is no offset for
997 this. Hence, a value that is too small, may lead to unreliable
1000 In some cases, however, one may wish to disable simulation of
1001 the computation of an application. This is the case when SMPI is used not to
1002 simulate an MPI application, but instead an MPI code that performs
1003 "live replay" of another MPI app (e.g., ScalaTrace's replay tool, or
1004 various on-line simulators that run an app at scale). In this case the
1005 computation of the replay/simulation logic should not be simulated by
1006 SMPI. Instead, the replay tool or on-line simulator will issue
1007 "computation events", which correspond to the actual MPI simulation
1008 being replayed/simulated. At the moment, these computation events can
1009 be simulated using SMPI by calling internal smpi_execute*() functions.
1011 To disable the benchmarking/simulation of a computation in the simulated
1012 application, the variable ``smpi/simulate-computation`` should be set
1013 to **no**. This option just ignores the timings in your simulation; it
1014 still executes the computations itself. If you want to stop SMPI from
1015 doing that, you should check the SMPI_SAMPLE macros, documented in
1016 Section :ref:`SMPI_use_faster`.
1018 +------------------------------------+-------------------------+-----------------------------+
1019 | Solution | Computations executed? | Computations simulated? |
1020 +====================================+=========================+=============================+
1021 | --cfg=smpi/simulate-computation:no | Yes | Never |
1022 +------------------------------------+-------------------------+-----------------------------+
1023 | --cfg=smpi/cpu-threshold:42 | Yes, in all cases | If it lasts over 42 seconds |
1024 +------------------------------------+-------------------------+-----------------------------+
1025 | SMPI_SAMPLE() macro | Only once per loop nest | Always |
1026 +------------------------------------+-------------------------+-----------------------------+
1028 .. _cfg=smpi/comp-adjustment-file:
1030 Slow-down or speed-up parts of your code
1031 ........................................
1033 **Option** ``smpi/comp-adjustment-file:`` **Default:** unset
1035 This option allows you to pass a file that contains two columns: The
1036 first column defines the section that will be subject to a speedup;
1037 the second column is the speedup. For instance:
1039 .. code-block:: shell
1041 "start:stop","ratio"
1042 "exchange_1.f:30:exchange_1.f:130",1.18244559422142
1044 The first line is the header - you must include it. The following
1045 line means that the code between two consecutive MPI calls on line 30
1046 in exchange_1.f and line 130 in exchange_1.f should receive a speedup
1047 of 1.18244559422142. The value for the second column is therefore a
1048 speedup, if it is larger than 1 and a slowdown if it is smaller
1049 than 1. Nothing will be changed if it is equal to 1.
1051 Of course, you can set any arbitrary filenames you want (so the start
1052 and end don't have to be in the same file), but be aware that this
1053 mechanism only supports `consecutive calls!`
1055 Please note that you must pass the ``-trace-call-location`` flag to
1056 smpicc or smpiff, respectively. This flag activates some internal
1057 macro definitions that help with obtaining the call location.
1059 .. _cfg=smpi/bw-factor:
1064 **Option** ``smpi/bw-factor``
1065 |br| **Default:** 65472:0.940694;15424:0.697866;9376:0.58729;5776:1.08739;3484:0.77493;1426:0.608902;732:0.341987;257:0.338112;0:0.812084
1067 The possible throughput of network links is often dependent on the
1068 message sizes, as protocols may adapt to different message sizes. With
1069 this option, a series of message sizes and factors are given, helping
1070 the simulation to be more realistic. For instance, the current default
1071 value means that messages with size 65472 bytes and more will get a total of
1072 MAX_BANDWIDTH*0.940694, messages of size 15424 to 65471 will get
1073 MAX_BANDWIDTH*0.697866, and so on (where MAX_BANDWIDTH denotes the
1074 bandwidth of the link).
1076 An experimental script to compute these factors is available online. See
1077 https://framagit.org/simgrid/platform-calibration/
1078 https://simgrid.org/contrib/smpi-saturation-doc.html
1080 .. _cfg=smpi/display-timing:
1082 Reporting Simulation Time
1083 .........................
1085 **Option** ``smpi/display-timing`` **Default:** 0 (false)
1087 Most of the time, you run MPI code with SMPI to compute the time it
1088 would take to run it on a platform. But since the code is run through
1089 the ``smpirun`` script, you don't have any control on the launcher
1090 code, making it difficult to report the simulated time when the
1091 simulation ends. If you enable the ``smpi/display-timing`` item,
1092 ``smpirun`` will display this information when the simulation
1095 .. _cfg=smpi/keep-temps:
1097 Keeping temporary files after simulation
1098 ........................................
1100 **Option** ``smpi/keep-temps`` **default:** 0 (false)
1102 SMPI usually generates a lot of temporary files that are cleaned after
1103 use. This option requests to preserve them, for example to debug or
1104 profile your code. Indeed, the binary files are removed very early
1105 under the dlopen privatization schema, which tends to fool the
1108 .. _cfg=smpi/lat-factor:
1113 **Option** ``smpi/lat-factor`` |br|
1114 **default:** 65472:11.6436;15424:3.48845;9376:2.59299;5776:2.18796;3484:1.88101;1426:1.61075;732:1.9503;257:1.95341;0:2.01467
1116 The motivation and syntax for this option is identical to the motivation/syntax
1117 of :ref:`cfg=smpi/bw-factor`.
1119 There is an important difference, though: While smpi/bw-factor `reduces` the
1120 actual bandwidth (i.e., values between 0 and 1 are valid), latency factors
1121 increase the latency, i.e., values larger than or equal to 1 are valid here.
1123 .. _cfg=smpi/papi-events:
1125 Trace hardware counters with PAPI
1126 .................................
1128 **Option** ``smpi/papi-events`` **default:** unset
1130 When the PAPI support is compiled into SimGrid, this option takes the
1131 names of PAPI counters and adds their respective values to the trace
1132 files (See Section :ref:`tracing_tracing_options`).
1136 This feature currently requires superuser privileges, as registers
1137 are queried. Only use this feature with code you trust! Call
1138 smpirun for instance via ``smpirun -wrapper "sudo "
1139 <your-parameters>`` or run ``sudo sh -c "echo 0 >
1140 /proc/sys/kernel/perf_event_paranoid"`` In the later case, sudo
1141 will not be required.
1143 It is planned to make this feature available on a per-process (or per-thread?) basis.
1144 The first draft, however, just implements a "global" (i.e., for all processes) set
1145 of counters, the "default" set.
1147 .. code-block:: shell
1149 --cfg=smpi/papi-events:"default:PAPI_L3_LDM:PAPI_L2_LDM"
1151 .. _cfg=smpi/privatization:
1153 Automatic Privatization of Global Variables
1154 ...........................................
1156 **Option** ``smpi/privatization`` **default:** "dlopen" (when using smpirun)
1158 MPI executables are usually meant to be executed in separate
1159 processes, but SMPI is executed in only one process. Global variables
1160 from executables will be placed in the same memory region and shared
1161 between processes, causing intricate bugs. Several options are
1162 possible to avoid this, as described in the main `SMPI publication
1163 <https://hal.inria.fr/hal-01415484>`_ and in the :ref:`SMPI
1164 documentation <SMPI_what_globals>`. SimGrid provides two ways of
1165 automatically privatizing the globals, and this option allows one to
1166 choose between them.
1168 - **no** (default when not using smpirun): Do not automatically
1169 privatize variables. Pass ``-no-privatize`` to smpirun to disable
1171 - **dlopen** or **yes** (default when using smpirun): Link multiple
1172 times against the binary.
1173 - **mmap** (slower, but maybe somewhat more stable):
1174 Runtime automatic switching of the data segments.
1177 This configuration option cannot be set in your platform file. You can only
1178 pass it as an argument to smpirun.
1180 .. _cfg=smpi/privatize-libs:
1182 Automatic privatization of global variables inside external libraries
1183 .....................................................................
1185 **Option** ``smpi/privatize-libs`` **default:** unset
1187 **Linux/BSD only:** When using dlopen (default) privatization,
1188 privatize specific shared libraries with internal global variables, if
1189 they can't be linked statically. For example libgfortran is usually
1190 used for Fortran I/O and indexes in files can be mixed up.
1192 Multiple libraries can be given, semicolon separated.
1194 This configuration option can only use either full paths to libraries,
1195 or full names. Check with ldd the name of the library you want to
1198 .. code-block:: shell
1202 libgfortran.so.3 => /usr/lib/x86_64-linux-gnu/libgfortran.so.3 (0x00007fbb4d91b000)
1205 Then you can use ``--cfg=smpi/privatize-libs:libgfortran.so.3``
1206 or ``--cfg=smpi/privatize-libs:/usr/lib/x86_64-linux-gnu/libgfortran.so.3``,
1207 but not ``libgfortran`` nor ``libgfortran.so``.
1209 .. _cfg=smpi/send-is-detached-thresh:
1211 Simulating MPI detached send
1212 ............................
1214 **Option** ``smpi/send-is-detached-thresh`` **default:** 65536
1216 This threshold specifies the size in bytes under which the send will
1217 return immediately. This is different from the threshold detailed in
1218 :ref:`cfg=smpi/async-small-thresh` because the message is not
1219 really sent when the send is posted. SMPI still waits for the
1220 corresponding receive to be posted, in order to perform the communication
1223 .. _cfg=smpi/coll-selector:
1225 Simulating MPI collective algorithms
1226 ....................................
1228 **Option** ``smpi/coll-selector`` **Possible values:** naive (default), ompi, mpich
1230 SMPI implements more than 100 different algorithms for MPI collective
1231 communication, to accurately simulate the behavior of most of the
1232 existing MPI libraries. The ``smpi/coll-selector`` item can be used to
1233 select the decision logic either of the OpenMPI or the MPICH libraries. (By
1234 default SMPI uses naive version of collective operations.)
1236 Each collective operation can be manually selected with a
1237 ``smpi/collective_name:algo_name``. Available algorithms are listed in
1238 :ref:`SMPI_use_colls`.
1240 .. TODO:: All available collective algorithms will be made available
1241 via the ``smpirun --help-coll`` command.
1243 .. _cfg=smpi/iprobe:
1245 Inject constant times for MPI_Iprobe
1246 ....................................
1248 **Option** ``smpi/iprobe`` **default:** 0.0001
1250 The behavior and motivation for this configuration option is identical
1251 with :ref:`smpi/test <cfg=smpi/test>`, but for the function
1254 .. _cfg=smpi/iprobe-cpu-usage:
1256 Reduce speed for iprobe calls
1257 .............................
1259 **Option** ``smpi/iprobe-cpu-usage`` **default:** 1 (no change)
1261 MPI_Iprobe calls can be heavily used in applications. To account
1262 correctly for the energy that cores spend probing, it is necessary to
1263 reduce the load that these calls cause inside SimGrid.
1265 For instance, we measured a maximum power consumption of 220 W for a
1266 particular application but only 180 W while this application was
1267 probing. Hence, the correct factor that should be passed to this
1268 option would be 180/220 = 0.81.
1272 Inject constant times for MPI_Init
1273 ..................................
1275 **Option** ``smpi/init`` **default:** 0
1277 The behavior and motivation for this configuration option is identical
1278 with :ref:`smpi/test <cfg=smpi/test>`, but for the function ``MPI_Init()``.
1282 Inject constant times for MPI_Isend()
1283 .....................................
1285 **Option** ``smpi/ois``
1287 The behavior and motivation for this configuration option is identical
1288 with :ref:`smpi/os <cfg=smpi/os>`, but for the function ``MPI_Isend()``.
1292 Inject constant times for MPI_send()
1293 ....................................
1295 **Option** ``smpi/os``
1297 In several network models such as LogP, send (MPI_Send, MPI_Isend) and
1298 receive (MPI_Recv) operations incur costs (i.e., they consume CPU
1299 time). SMPI can factor these costs in as well, but the user has to
1300 configure SMPI accordingly as these values may vary by machine. This
1301 can be done by using ``smpi/os`` for MPI_Send operations; for MPI_Isend
1302 and MPI_Recv, use ``smpi/ois`` and ``smpi/or``, respectively. These work
1303 exactly as ``smpi/ois``.
1305 This item can consist of multiple sections; each section takes three
1306 values, for example ``1:3:2;10:5:1``. The sections are divided by ";"
1307 so this example contains two sections. Furthermore, each section
1308 consists of three values.
1310 1. The first value denotes the minimum size in bytes for this section to take effect;
1311 read it as "if message size is greater than this value (and other section has a larger
1312 first value that is also smaller than the message size), use this".
1313 In the first section above, this value is "1".
1315 2. The second value is the startup time; this is a constant value that will always
1316 be charged, no matter what the size of the message. In the first section above,
1319 3. The third value is the `per-byte` cost. That is, it is charged for every
1320 byte of the message (incurring cost messageSize*cost_per_byte)
1321 and hence accounts also for larger messages. In the first
1322 section of the example above, this value is "2".
1324 Now, SMPI always checks which section it should use for a given
1325 message; that is, if a message of size 11 is sent with the
1326 configuration of the example above, only the second section will be
1327 used, not the first, as the first value of the second section is
1328 closer to the message size. Hence, when ``smpi/os=1:3:2;10:5:1``, a
1329 message of size 11 incurs the following cost inside MPI_Send:
1330 ``5+11*1`` because 5 is the startup cost and 1 is the cost per byte.
1332 Note that the order of sections can be arbitrary; they will be ordered internally.
1336 Inject constant times for MPI_Recv()
1337 ....................................
1339 **Option** ``smpi/or``
1341 The behavior and motivation for this configuration option is identical
1342 with :ref:`smpi/os <cfg=smpi/os>`, but for the function ``MPI_Recv()``.
1345 .. _cfg=smpi/grow-injected-times:
1347 Inject constant times for MPI_Test
1348 ..................................
1350 **Option** ``smpi/test`` **default:** 0.0001
1352 By setting this option, you can control the amount of time a process
1353 sleeps when MPI_Test() is called; this is important, because SimGrid
1354 normally only advances the time while communication is happening and
1355 thus, MPI_Test will not add to the time, resulting in deadlock if it is
1356 used as a break-condition as in the following example:
1361 MPI_Test(request, flag, status);
1365 To speed up execution, we use a counter to keep track of how often we
1366 checked if the handle is now valid or not. Hence, we actually
1367 use counter*SLEEP_TIME, that is, the time MPI_Test() causes the
1368 process to sleep increases linearly with the number of previously
1369 failed tests. This behavior can be disabled by setting
1370 ``smpi/grow-injected-times`` to **no**. This will also disable this
1371 behavior for MPI_Iprobe.
1373 .. _cfg=smpi/shared-malloc:
1374 .. _cfg=smpi/shared-malloc-hugepage:
1379 **Option** ``smpi/shared-malloc`` **Possible values:** global (default), local
1381 If your simulation consumes too much memory, you may want to modify
1382 your code so that the working areas are shared by all MPI ranks. For
1383 example, in a block-cyclic matrix multiplication, you will only
1384 allocate one set of blocks, and all processes will share them.
1385 Naturally, this will lead to very wrong results, but this will save a
1386 lot of memory. So this is still desirable for some studies. For more on
1387 the motivation for that feature, please refer to the `relevant section
1388 <https://simgrid.github.io/SMPI_CourseWare/topic_understanding_performance/matrixmultiplication>`_
1389 of the SMPI CourseWare (see Activity #2.2 of the pointed
1390 assignment). In practice, change the calls for malloc() and free() into
1391 SMPI_SHARED_MALLOC() and SMPI_SHARED_FREE().
1393 SMPI provides two algorithms for this feature. The first one, called
1394 ``local``, allocates one block per call to SMPI_SHARED_MALLOC()
1395 (each call site gets its own block) ,and this block is shared
1396 among all MPI ranks. This is implemented with the shm_* functions
1397 to create a new POSIX shared memory object (kept in RAM, in /dev/shm)
1398 for each shared block.
1400 With the ``global`` algorithm, each call to SMPI_SHARED_MALLOC()
1401 returns a new address, but it only points to a shadow block: its memory
1402 area is mapped on a 1 MiB file on disk. If the returned block is of size
1403 N MiB, then the same file is mapped N times to cover the whole block.
1404 At the end, no matter how many times you call SMPI_SHARED_MALLOC, this will
1405 only consume 1 MiB in memory.
1407 You can disable this behavior and come back to regular mallocs (for
1408 example for debugging purposes) using ``no`` as a value.
1410 If you want to keep private some parts of the buffer, for instance if these
1411 parts are used by the application logic and should not be corrupted, you
1412 can use SMPI_PARTIAL_SHARED_MALLOC(size, offsets, offsets_count). For example:
1416 mem = SMPI_PARTIAL_SHARED_MALLOC(500, {27,42 , 100,200}, 2);
1418 This will allocate 500 bytes to mem, such that mem[27..41] and
1419 mem[100..199] are shared while other area remain private.
1421 Then, it can be deallocated by calling SMPI_SHARED_FREE(mem).
1423 When smpi/shared-malloc:global is used, the memory consumption problem
1424 is solved, but it may induce too much load on the kernel's pages table.
1425 In this case, you should use huge pages so that the kernel creates only one
1426 entry per MB of malloced data instead of one entry per 4 kB.
1427 To activate this, you must mount a hugetlbfs on your system and allocate
1428 at least one huge page:
1430 .. code-block:: shell
1433 sudo mount none /home/huge -t hugetlbfs -o rw,mode=0777
1434 sudo sh -c 'echo 1 > /proc/sys/vm/nr_hugepages' # echo more if you need more
1436 Then, you can pass the option
1437 ``--cfg=smpi/shared-malloc-hugepage:/home/huge`` to smpirun to
1438 actually activate the huge page support in shared mallocs.
1442 Inject constant times for MPI_Wtime, gettimeofday and clock_gettime
1443 ...................................................................
1445 **Option** ``smpi/wtime`` **default:** 10 ns
1447 This option controls the amount of (simulated) time spent in calls to
1448 MPI_Wtime(), gettimeofday() and clock_gettime(). If you set this value
1449 to 0, the simulated clock is not advanced in these calls, which leads
1450 to issues if your application contains such a loop:
1454 while(MPI_Wtime() < some_time_bound) {
1455 /* some tests, with no communication nor computation */
1458 When the option smpi/wtime is set to 0, the time advances only on
1459 communications and computations. So the previous code results in an
1460 infinite loop: the current [simulated] time will never reach
1461 ``some_time_bound``. This infinite loop is avoided when that option
1462 is set to a small value, as it is by default since SimGrid v3.21.
1464 Note that if your application does not contain any loop depending on
1465 the current time only, then setting this option to a non-zero value
1466 will slow down your simulations by a tiny bit: the simulation loop has
1467 to be broken out of and reset each time your code asks for the current time.
1468 If the simulation speed really matters to you, you can avoid this
1469 extra delay by setting smpi/wtime to 0.
1471 Other Configurations
1472 --------------------
1474 .. _cfg=debug/clean-atexit:
1476 Cleanup at Termination
1477 ......................
1479 **Option** ``debug/clean-atexit`` **default:** on
1481 If your code is segfaulting during its finalization, it may help to
1482 disable this option to request that SimGrid not attempt any cleanups at
1483 the end of the simulation. Since the Unix process is ending anyway,
1484 the operating system will wipe it all.
1491 **Option** ``path`` **default:** . (current dir)
1493 It is possible to specify a list of directories to search in for the
1494 trace files (see :ref:`pf_trace`) by using this configuration
1495 item. To add several directory to the path, set the configuration
1496 item several times, as in ``--cfg=path:toto --cfg=path:tutu``
1498 .. _cfg=debug/breakpoint:
1503 **Option** ``debug/breakpoint`` **default:** unset
1505 This configuration option sets a breakpoint: when the simulated clock
1506 reaches the given time, a SIGTRAP is raised. This can be used to stop
1507 the execution and get a backtrace with a debugger.
1509 It is also possible to set the breakpoint from inside the debugger, by
1510 writing in global variable simgrid::simix::breakpoint. For example,
1513 .. code-block:: shell
1515 set variable simgrid::simix::breakpoint = 3.1416
1517 .. _cfg=debug/verbose-exit:
1522 **Option** ``debug/verbose-exit`` **default:** on
1524 By default, when Ctrl-C is pressed, the status of all existing actors
1525 is displayed before exiting the simulation. This is very useful to
1526 debug your code, but it can become troublesome if you have many
1527 actors. Set this configuration item to **off** to disable this
1530 .. _cfg=exception/cutpath:
1532 Truncate local path from exception backtrace
1533 ............................................
1535 **Option** ``exception/cutpath`` **default:** off
1537 This configuration option is used to remove the path from the
1538 backtrace shown when an exception is thrown. This is mainly useful for
1539 the tests: the full file path would makes the tests non-reproducible because
1540 the paths of source files depend of the build settings. That would
1541 break most of the tests since their output is continually compared.
1543 Logging Configuration
1544 ---------------------
1546 This can be done by using XBT. Go to :ref:`XBT_log` for more details.