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:
35 .. code-block:: console
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 - **bmf/max-iterations:** :ref:`cfg=bmf/max-iterations`
88 - **bmf/precision:** :ref:`cfg=bmf/precision`
90 - **contexts/factory:** :ref:`cfg=contexts/factory`
91 - **contexts/guard-size:** :ref:`cfg=contexts/guard-size`
92 - **contexts/nthreads:** :ref:`cfg=contexts/nthreads`
93 - **contexts/stack-size:** :ref:`cfg=contexts/stack-size`
94 - **contexts/synchro:** :ref:`cfg=contexts/synchro`
96 - **cpu/maxmin-selective-update:** :ref:`Cpu Optimization Level <options_model_optim>`
97 - **cpu/model:** :ref:`options_model_select`
98 - **cpu/optim:** :ref:`Cpu Optimization Level <options_model_optim>`
100 - **debug/breakpoint:** :ref:`cfg=debug/breakpoint`
101 - **debug/clean-atexit:** :ref:`cfg=debug/clean-atexit`
102 - **debug/verbose-exit:** :ref:`cfg=debug/verbose-exit`
104 - **exception/cutpath:** :ref:`cfg=exception/cutpath`
106 - **host/model:** :ref:`options_model_select`
108 - **maxmin/precision:** :ref:`cfg=maxmin/precision`
109 - **maxmin/concurrency-limit:** :ref:`cfg=maxmin/concurrency-limit`
111 - **msg/debug-multiple-use:** :ref:`cfg=msg/debug-multiple-use`
113 - **model-check:** :ref:`options_modelchecking`
114 - **model-check/checkpoint:** :ref:`cfg=model-check/checkpoint`
115 - **model-check/communications-determinism:** :ref:`cfg=model-check/communications-determinism`
116 - **model-check/dot-output:** :ref:`cfg=model-check/dot-output`
117 - **model-check/max-depth:** :ref:`cfg=model-check/max-depth`
118 - **model-check/property:** :ref:`cfg=model-check/property`
119 - **model-check/reduction:** :ref:`cfg=model-check/reduction`
120 - **model-check/replay:** :ref:`cfg=model-check/replay`
121 - **model-check/send-determinism:** :ref:`cfg=model-check/send-determinism`
122 - **model-check/setenv:** :ref:`cfg=model-check/setenv`
123 - **model-check/termination:** :ref:`cfg=model-check/termination`
124 - **model-check/timeout:** :ref:`cfg=model-check/timeout`
125 - **model-check/visited:** :ref:`cfg=model-check/visited`
127 - **network/bandwidth-factor:** :ref:`cfg=network/bandwidth-factor`
128 - **network/crosstraffic:** :ref:`cfg=network/crosstraffic`
129 - **network/latency-factor:** :ref:`cfg=network/latency-factor`
130 - **network/loopback-lat:** :ref:`cfg=network/loopback`
131 - **network/loopback-bw:** :ref:`cfg=network/loopback`
132 - **network/maxmin-selective-update:** :ref:`Network Optimization Level <options_model_optim>`
133 - **network/model:** :ref:`options_model_select`
134 - **network/optim:** :ref:`Network Optimization Level <options_model_optim>`
135 - **network/TCP-gamma:** :ref:`cfg=network/TCP-gamma`
136 - **network/weight-S:** :ref:`cfg=network/weight-S`
138 - **ns3/TcpModel:** :ref:`options_pls`
139 - **ns3/seed:** :ref:`options_pls`
140 - **path:** :ref:`cfg=path`
141 - **plugin:** :ref:`cfg=plugin`
143 - **storage/max_file_descriptors:** :ref:`cfg=storage/max_file_descriptors`
145 - **surf/precision:** :ref:`cfg=surf/precision`
147 - **For collective operations of SMPI,** please refer to Section :ref:`cfg=smpi/coll-selector`
148 - **smpi/auto-shared-malloc-thresh:** :ref:`cfg=smpi/auto-shared-malloc-thresh`
149 - **smpi/async-small-thresh:** :ref:`cfg=smpi/async-small-thresh`
150 - **smpi/barrier-finalization:** :ref:`cfg=smpi/barrier-finalization`
151 - **smpi/barrier-collectives:** :ref:`cfg=smpi/barrier-collectives`
152 - **smpi/buffering:** :ref:`cfg=smpi/buffering`
153 - **smpi/coll-selector:** :ref:`cfg=smpi/coll-selector`
154 - **smpi/comp-adjustment-file:** :ref:`cfg=smpi/comp-adjustment-file`
155 - **smpi/cpu-threshold:** :ref:`cfg=smpi/cpu-threshold`
156 - **smpi/display-allocs:** :ref:`cfg=smpi/display-allocs`
157 - **smpi/display-timing:** :ref:`cfg=smpi/display-timing`
158 - **smpi/errors-are-fatal:** :ref:`cfg=smpi/errors-are-fatal`
159 - **smpi/grow-injected-times:** :ref:`cfg=smpi/grow-injected-times`
160 - **smpi/host-speed:** :ref:`cfg=smpi/host-speed`
161 - **smpi/IB-penalty-factors:** :ref:`cfg=smpi/IB-penalty-factors`
162 - **smpi/iprobe:** :ref:`cfg=smpi/iprobe`
163 - **smpi/iprobe-cpu-usage:** :ref:`cfg=smpi/iprobe-cpu-usage`
164 - **smpi/init:** :ref:`cfg=smpi/init`
165 - **smpi/keep-temps:** :ref:`cfg=smpi/keep-temps`
166 - **smpi/ois:** :ref:`cfg=smpi/ois`
167 - **smpi/or:** :ref:`cfg=smpi/or`
168 - **smpi/os:** :ref:`cfg=smpi/os`
169 - **smpi/papi-events:** :ref:`cfg=smpi/papi-events`
170 - **smpi/pedantic:** :ref:`cfg=smpi/pedantic`
171 - **smpi/privatization:** :ref:`cfg=smpi/privatization`
172 - **smpi/privatize-libs:** :ref:`cfg=smpi/privatize-libs`
173 - **smpi/send-is-detached-thresh:** :ref:`cfg=smpi/send-is-detached-thresh`
174 - **smpi/shared-malloc:** :ref:`cfg=smpi/shared-malloc`
175 - **smpi/shared-malloc-hugepage:** :ref:`cfg=smpi/shared-malloc-hugepage`
176 - **smpi/simulate-computation:** :ref:`cfg=smpi/simulate-computation`
177 - **smpi/test:** :ref:`cfg=smpi/test`
178 - **smpi/wtime:** :ref:`cfg=smpi/wtime`
179 - **smpi/list-leaks** :ref:`cfg=smpi/list-leaks`
181 - **Tracing configuration options** can be found in Section :ref:`tracing_tracing_options`
183 - **storage/model:** :ref:`options_model_select`
185 - **vm/model:** :ref:`options_model_select`
189 Configuring the Platform Models
190 -------------------------------
192 .. _options_model_select:
194 Choosing the Platform Models
195 ............................
197 SimGrid comes with several network, CPU and disk models built in,
198 and you can change the used model at runtime by changing the passed
199 configuration. The three main configuration items are given below.
200 For each of these items, passing the special ``help`` value gives you
201 a short description of all possible values (for example,
202 ``--cfg=network/model:help`` will present all provided network
203 models). Also, ``--help-models`` should provide information about all
204 models for all existing resources.
206 - ``network/model``: specify the used network model. Possible values:
208 - **LV08 (default one):** Realistic network analytic model
209 (slow-start modeled by multiplying latency by 13.01, bandwidth by
210 .97; bottleneck sharing uses a payload of S=20537 for evaluating
211 RTT). Described in `Accuracy Study and Improvement of Network
212 Simulation in the SimGrid Framework
213 <http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf>`_.
214 - **Constant:** Simplistic network model where all communication
215 take a constant time (one second). This model provides the lowest
216 realism, but is (marginally) faster.
217 - **SMPI:** Realistic network model specifically tailored for HPC
218 settings (accurate modeling of slow start with correction factors on
219 three intervals: < 1KiB, < 64 KiB, >= 64 KiB). This model can be
220 :ref:`further configured <options_model_network>`.
221 - **IB:** Realistic network model specifically tailored for HPC
222 settings with InfiniBand networks (accurate modeling contention
223 behavior, based on the model explained in `this PhD work
224 <http://mescal.imag.fr/membres/jean-marc.vincent/index.html/PhD/Vienne.pdf>`_.
225 This model can be :ref:`further configured <options_model_network>`.
226 - **CM02:** Legacy network analytic model. Very similar to LV08, but
227 without corrective factors. The timings of small messages are thus
228 poorly modeled. This model is described in `A Network Model for
229 Simulation of Grid Application
230 <https://hal.inria.fr/inria-00071989/document>`_.
231 - **ns-3** (only available if you compiled SimGrid accordingly):
232 Use the packet-level network
233 simulators as network models (see :ref:`model_ns3`).
234 This model can be :ref:`further configured <options_pls>`.
236 - ``cpu/model``: specify the used CPU model. We have only one model
239 - **Cas01:** Simplistic CPU model (time=size/speed)
241 - ``host/model``: The host concept is the aggregation of a CPU with a
242 network card. Three models exists, but actually, only 2 of them are
243 interesting. The "compound" one is simply due to the way our
244 internal code is organized, and can easily be ignored. So at the
245 end, you have two host models: The default one allows aggregation of
246 an existing CPU model with an existing network model, but does not
247 allow parallel tasks because these beasts need some collaboration
248 between the network and CPU model.
250 - **default:** Default host model. Currently, CPU:Cas01 and
251 network:LV08 (with cross traffic enabled)
252 - **compound:** Host model that is automatically chosen if
253 you change the network and CPU models
254 - **ptask_L07:** Host model somehow similar to Cas01+CM02 but
255 allowing "parallel tasks", that are intended to model the moldable
256 tasks of the grid scheduling literature.
258 - ``storage/model``: specify the used storage model. Only one model is
260 - ``vm/model``: specify the model for virtual machines. Only one model
263 .. todo: make 'compound' the default host model.
265 .. _options_model_solver:
270 The different models rely on a linear inequalities solver to share
271 the underlying resources. SimGrid allows you to change the solver, but
272 be cautious, **don't change it unless you are 100% sure**.
274 - items ``cpu/solver``, ``network/solver``, ``disk/solver`` and ``host/solver``
275 allow you to change the solver for each model:
277 - **maxmin:** The default solver for all models except ptask. Provides a
278 max-min fairness allocation.
279 - **fairbottleneck:** The default solver for ptasks. Extends max-min to
280 allow heterogeneous resources.
281 - **bmf:** More realistic solver for heterogeneous resource sharing.
282 Implements BMF (Bottleneck max fairness) fairness. To be used with
283 parallel tasks instead of fair-bottleneck.
285 .. _options_model_optim:
290 The network and CPU models that are based on linear inequalities solver (that
291 is, all our analytical models) accept specific optimization
294 - items ``network/optim`` and ``cpu/optim`` (both default to 'Lazy'):
296 - **Lazy:** Lazy action management (partial invalidation in lmm +
297 heap in action remaining).
298 - **TI:** Trace integration. Highly optimized mode when using
299 availability traces (only available for the Cas01 CPU model for
301 - **Full:** Full update of remaining and variables. Slow but may be
302 useful when debugging.
304 - items ``network/maxmin-selective-update`` and
305 ``cpu/maxmin-selective-update``: configure whether the underlying
306 should be lazily updated or not. It should have no impact on the
307 computed timings, but should speed up the computation. |br| It is
308 still possible to disable this feature because it can reveal
309 counter-productive in very specific scenarios where the
310 interaction level is high. In particular, if all your
311 communication share a given backbone link, you should disable it:
312 without it, a simple regular loop is used to update each
313 communication. With it, each of them is still updated (because of
314 the dependency induced by the backbone), but through a complicated
315 and slow pattern that follows the actual dependencies.
317 .. _cfg=bmf/precision:
318 .. _cfg=maxmin/precision:
319 .. _cfg=surf/precision:
324 **Option** ``maxmin/precision`` **Default:** 1e-5 (in flops or bytes) |br|
325 **Option** ``surf/precision`` **Default:** 1e-9 (in seconds) |br|
326 **Option** ``bmf/precision`` **Default:** 1e-12 (no unit)
328 The analytical models handle a lot of floating point values. It is
329 possible to change the epsilon used to update and compare them through
330 this configuration item. Changing it may speedup the simulation by
331 discarding very small actions, at the price of a reduced numerical
332 precision. You can modify separately the precision used to manipulate
333 timings (in seconds) and the one used to manipulate amounts of work
336 .. _cfg=maxmin/concurrency-limit:
341 **Option** ``maxmin/concurrency-limit`` **Default:** -1 (no limit)
343 The maximum number of variables per resource can be tuned through this
344 option. You can have as many simultaneous actions per resources as you
345 want. If your simulation presents a very high level of concurrency, it
346 may help to use e.g. 100 as a value here. It means that at most 100
347 actions can consume a resource at a given time. The extraneous actions
348 are queued and wait until the amount of concurrency of the considered
349 resource lowers under the given boundary.
351 Such limitations help both to the simulation speed and simulation accuracy
352 on highly constrained scenarios, but the simulation speed suffers of this
353 setting on regular (less constrained) scenarios so it is off by default.
355 .. _cfg=bmf/max-iterations:
360 **Option** ``bmf/max-iterations`` **Default:** 1000
362 It may happen in some settings that the BMF solver fails to converge to
363 a solution, so there is a hard limit on the amount of iteration count to
364 avoid infinite loops.
366 .. _options_model_network:
368 Configuring the Network Model
369 .............................
371 .. _cfg=network/TCP-gamma:
373 Maximal TCP Window Size
374 ^^^^^^^^^^^^^^^^^^^^^^^
376 **Option** ``network/TCP-gamma`` **Default:** 4194304
378 The analytical models need to know the maximal TCP window size to take
379 the TCP congestion mechanism into account. On Linux, this value can
380 be retrieved using the following commands. Both give a set of values,
381 and you should use the last one, which is the maximal size.
383 .. code-block:: console
385 $ cat /proc/sys/net/ipv4/tcp_rmem # gives the sender window
386 $ cat /proc/sys/net/ipv4/tcp_wmem # gives the receiver window
388 .. _cfg=network/bandwidth-factor:
389 .. _cfg=network/latency-factor:
390 .. _cfg=network/weight-S:
392 Manual calibration factors
393 ^^^^^^^^^^^^^^^^^^^^^^^^^^
395 SimGrid can take network irregularities such as a slow startup or changing behavior depending on the message size into account.
396 The values provided by default were computed a long time ago through data fitting one the timings of either packet-level simulators or direct
397 experiments on real platforms. These default values should be OK for most users, but if simulation realism is really important to
398 you, you probably want to recalibrate the models (i.e., devise sensible values for your specific settings). This section only
399 describes how to pass new values to the models while the calibration process involved in the computation of these values is
400 described :ref:`in the relevant chapter <models_calibration>`.
402 We found out that many networking effects can be realistically accounted for with the three following correction factors. They were shown
403 to be enough to capture slow-start effects, the different transmission modes of MPI systems (eager vs. rendez-vous mode), or the
404 non linear effects of wifi sharing.
406 **Option** ``network/latency-factor`` **Default:** 1.0, but overridden by most models
408 This option specifies a multiplier to apply to the *physical* latency (i.e., the one described in the platform) of the set of links involved in a communication. The factor can either be a constant to apply to any
409 communication, or it may depend on the message size. The ``CM02`` model does not use any correction factor, so the
410 latency-factor remains to 1. The ``LV08`` model sets it to 13.01 to model slow-start, while the ``SMPI`` model has several
411 possible values depending on the interval in which the message size falls. The default SMPI setting given below specifies for example that a message smaller than
412 257 bytes will get a latency multiplier of 2.01467 while a message whose size is in [15424, 65472] will get a latency multiplier
413 of 3.48845. The ``wifi`` model goes further and uses a callback in the program to compute the factor that must be non-linear in
416 This multiplier is applied to the latency computed from the platform, that is the sum of all link *physical* latencies over the :ref:`network path <platform_routing>` used by the considered communication, to derive the *effective* end-to-end latency.
418 Constant factors are easy to express, but the interval-based syntax used in SMPI is somewhat complex. It expects a set of
419 factors separated by semicolons, each of the form ``boundary:factor``. For example if your specification is
420 ``0:1;1000:2;5000:3``, it means that on [0, 1000) the factor is 1. On [1000,5000), the factor is 2 while the factor is 3 for
421 5000 and beyond. If your first interval does include size=0, then the default value of 1 is used before. Changing the factor
422 callback is not possible from the command line and must be done from your code, as shown in `this example
423 <https://framagit.org/simgrid/simgrid/tree/master/examples/cpp/network-factors/s4u-network-factors.cpp>`_. Note that the chosen
424 model only provides some default settings. You may pick a ``LV08``
425 model to get some of the settings, and override the latency with interval-based values.
427 SMPI default value: 65472:11.6436; 15424:3.48845; 9376:2.59299; 5776:2.18796; 3484:1.88101; 1426:1.61075; 732:1.9503;
428 257:1.95341;0:2.01467 (interval boundaries are sorted automatically). These values were computed by data fitting on the Stampede
429 Supercomputer at TACC, with optimal deployment of processes on nodes. To accurately model your settings, you should redo the
430 :ref:`calibration <models_calibration>`.
432 **Option** ``network/bandwidth-factor`` **Default:** 1.0, but overridden by most models
434 Setting this option automatically adjusts the *effective* bandwidth (i.e., the one perceived by the application) used by any given communication. As with latency-factor above, the value
435 can be a constant (``CM02`` uses 1 -- no correction -- while ``LV08`` uses 0.97 to discount TCP headers while computing the
436 payload bandwidth), interval-based (as the default provided by the ``SMPI``), or using in-program callbacks (as with ``wifi``).
438 SMPI default value: 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
439 This was also computed on the Stampede Supercomputer.
441 **Option** ``network/weight-S`` **Default:** depends on the model
443 Value used to account for RTT-unfairness when sharing a bottleneck (network connections with a large RTT are generally penalized against those with a small one). Described in `Accuracy Study and Improvement of Network
444 Simulation in the SimGrid Framework <http://mescal.imag.fr/membres/arnaud.legrand/articles/simutools09.pdf>`_
446 Default values for ``CM02`` is 0. ``LV08`` sets it to 20537 while both ``SMPI`` and ``IB`` set it to 8775.
448 .. _cfg=network/loopback:
450 Configuring loopback link
451 ^^^^^^^^^^^^^^^^^^^^^^^^^
453 Several network models provide an implicit loopback link to account for local
454 communication on a host. By default it has a 10GBps bandwidth and a null latency.
455 This can be changed with ``network/loopback-lat`` and ``network/loopback-bw``
456 items. Note that this loopback is conveniently modeled with a :ref:`single FATPIPE link <pf_loopback>`
457 for the whole platform. If modeling contention inside nodes is important then you should
458 rather add such loopback links (one for each host) yourself.
460 .. _cfg=smpi/IB-penalty-factors:
465 InfiniBand network behavior can be modeled through 3 parameters
466 ``smpi/IB-penalty-factors:"βe;βs;γs"``, as explained in `the PhD
467 thesis of Jean-Marc Vincent
468 <http://mescal.imag.fr/membres/jean-marc.vincent/index.html/PhD/Vienne.pdf>`_ (in French)
469 or more concisely in `this paper <https://hal.inria.fr/hal-00953618/document>`_,
470 even if that paper does only describe models for myrinet and ethernet.
471 You can see in Fig 2 some results for Infiniband, for example. This model
472 may be outdated by now for modern infiniband, anyway, so a new
473 validation would be good.
475 The three paramaters are defined as follows:
477 - βs: penalty factor for outgoing messages, computed by running a simple send to
478 two nodes and checking slowdown compared to a single send to one node,
480 - βe: penalty factor for ingoing messages, same computation method but with one
481 node receiving several messages
482 - γr: slowdown factor when communication buffer memory is saturated. It needs a
483 more complicated pattern to run in order to be computed (5.3 in the thesis,
484 page 107), and formula in the end is γr = time(c)/(3×βe×time(ref)), where
485 time(ref) is the time of a single comm with no contention).
487 Once these values are computed, a penalty is assessed for each message (this is
488 the part implemented in the simulator) as shown page 106 of the thesis. Here is
489 a simple translation of this text. First, some notations:
491 - ∆e(e) which corresponds to the incoming degree of node e, that is to say the number of communications having as destination node e.
492 - ∆s (s) which corresponds to the degree outgoing from node s, that is to say the number of communications sent by node s.
493 - Φ (e) which corresponds to the number of communications destined for the node e but coming from a different node.
494 - Ω (s, e) which corresponds to the number of messages coming from node s to node e. If node e only receives communications from different nodes then Φ (e) = ∆e (e). On the other hand if, for example, there are three messages coming from node s and going from node e then Φ (e) 6 = ∆e (e) and Ω (s, e) = 3
496 To determine the penalty for a communication, two values need to be calculated. First, the penalty caused by the conflict in transmission, noted ps.
499 - if ∆s (i) = 1 then ps = 1.
500 - if ∆s (i) ≥ 2 and ∆e (i) ≥ 3 then ps = ∆s (i) × βs × γr
501 - else, ps = ∆s (i) × βs
504 Then, the penalty caused by the conflict in reception (noted pe) should be computed as follows:
506 - if ∆e (i) = 1 then pe = 1
507 - else, pe = Φ (e) × βe × Ω (s, e)
509 Finally, the penalty associated with the communication is:
510 p = max (ps ∈ s, pe)
512 .. _cfg=network/crosstraffic:
514 Simulating Cross-Traffic
515 ^^^^^^^^^^^^^^^^^^^^^^^^
517 Since SimGrid v3.7, cross-traffic effects can be taken into account in
518 analytical simulations. It means that ongoing and incoming
519 communication flows are treated independently. In addition, the LV08
520 model adds 0.05 of usage on the opposite direction for each new
521 created flow. This can be useful to simulate some important TCP
522 phenomena such as ack compression.
524 For that to work, your platform must have two links for each
525 pair of interconnected hosts. An example of usable platform is
526 available in ``examples/platforms/crosstraffic.xml``.
528 This is activated through the ``network/crosstraffic`` item, that
529 can be set to 0 (disable this feature) or 1 (enable it).
531 Note that with the default host model this option is activated by default.
533 .. _cfg=smpi/async-small-thresh:
535 Simulating Asynchronous Send
536 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
538 (this configuration item is experimental and may change or disappear)
540 It is possible to specify that messages below a certain size (in bytes) will be
541 sent as soon as the call to MPI_Send is issued, without waiting for
542 the correspondent receive. This threshold can be configured through
543 the ``smpi/async-small-thresh`` item. The default value is 0. This
544 behavior can also be manually set for mailboxes, by setting the
545 receiving mode of the mailbox with a call to
546 :cpp:func:`MSG_mailbox_set_async`. After this, all messages sent to
547 this mailbox will have this behavior regardless of the message size.
549 This value needs to be smaller than or equals to the threshold set at
550 :ref:`cfg=smpi/send-is-detached-thresh`, because asynchronous messages
551 are meant to be detached as well.
558 **Option** ``ns3/TcpModel`` **Default:** "default" (ns-3 default)
560 When using ns-3, there is an extra item ``ns3/TcpModel``, corresponding
561 to the ``ns3::TcpL4Protocol::SocketType`` configuration item in
562 ns-3. The only valid values (enforced on the SimGrid side) are
563 'default' (no change to the ns-3 configuration), 'NewReno' or 'Reno' or
566 **Option** ``ns3/seed`` **Default:** "" (don't set the seed in ns-3)
568 This option is the random seed to provide to ns-3 with
569 ``ns3::RngSeedManager::SetSeed`` and ``ns3::RngSeedManager::SetRun``.
571 If left blank, no seed is set in ns-3. If the value 'time' is
572 provided, the current amount of seconds since epoch is used as a seed.
573 Otherwise, the provided value must be a number to use as a seed.
575 Configuring the Storage model
576 .............................
578 .. _cfg=storage/max_file_descriptors:
580 File Descriptor Count per Host
581 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
583 **Option** ``storage/max_file_descriptors`` **Default:** 1024
585 Each host maintains a fixed-size array of its file descriptors. You
586 can change its size through this item to either enlarge it if your
587 application requires it or to reduce it to save memory space.
594 SimGrid plugins allow one to extend the framework without changing its
595 source code directly. Read the source code of the existing plugins to
596 learn how to do so (in ``src/plugins``), and ask your questions to the
597 usual channels (Stack Overflow, Mailing list, IRC). The basic idea is
598 that plugins usually register callbacks to some signals of interest.
599 If they need to store some information about a given object (Link, CPU
600 or Actor), they do so through the use of a dedicated object extension.
602 Some of the existing plugins can be activated from the command line,
603 meaning that you can activate them from the command line without any
604 modification to your simulation code. For example, you can activate
605 the host energy plugin by adding ``--cfg=plugin:host_energy`` to your
608 Here is a partial list of plugins that can be activated this way. You can get
609 the full list by passing ``--cfg=plugin:help`` to your simulator.
611 - :ref:`Host Energy <plugin_host_energy>`: models the energy dissipation of the compute units.
612 - :ref:`Link Energy <plugin_link_energy>`: models the energy dissipation of the network.
613 - :ref:`Host Load <plugin_host_load>`: monitors the load of the compute units.
615 .. _options_modelchecking:
617 Configuring the Model-Checking
618 ------------------------------
620 To enable SimGrid's model-checking support, the program should
621 be executed using the simgrid-mc wrapper:
623 .. code-block:: console
625 $ simgrid-mc ./my_program
627 Safety properties are expressed as assertions using the function
628 :cpp:func:`void MC_assert(int prop)`.
630 .. _cfg=smpi/buffering:
632 Specifying the MPI buffering behavior
633 .....................................
635 **Option** ``smpi/buffering`` **Default:** infty
637 Buffering in MPI has a huge impact on the communication semantic. For example,
638 standard blocking sends are synchronous calls when the system buffers are full
639 while these calls can complete immediately without even requiring a matching
640 receive call for small messages sent when the system buffers are empty.
642 In SMPI, this depends on the message size, that is compared against two thresholds:
644 - if (size < :ref:`smpi/async-small-thresh <cfg=smpi/async-small-thresh>`) then
645 MPI_Send returns immediately, even if the corresponding receive has not be issued yet.
646 - 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
647 MPI_Send returns as soon as the corresponding receive has been issued. This is known as the eager mode.
648 - if (:ref:`smpi/send-is-detached-thresh <cfg=smpi/send-is-detached-thresh>` < size) then
649 MPI_Send returns only when the message has actually been sent over the network. This is known as the rendez-vous mode.
651 The ``smpi/buffering`` (only valid with MC) option gives an easier interface to choose between these semantics. It can take two values:
653 - **zero:** means that buffering should be disabled. All communications are actually blocking.
654 - **infty:** means that buffering should be made infinite. All communications are non-blocking.
656 .. _cfg=model-check/property:
658 Specifying a liveness property
659 ..............................
661 **Option** ``model-check/property`` **Default:** unset
663 If you want to specify liveness properties, you have to pass them on
664 the command line, specifying the name of the file containing the
665 property, as formatted by the `ltl2ba <https://github.com/utwente-fmt/ltl2ba>`_ program.
666 Note that ltl2ba is not part of SimGrid and must be installed separately.
668 .. code-block:: console
670 $ simgrid-mc ./my_program --cfg=model-check/property:<filename>
672 .. _cfg=model-check/checkpoint:
674 Going for Stateful Verification
675 ...............................
677 By default, the system is backtracked to its initial state to explore
678 another path, instead of backtracking to the exact step before the fork
679 that we want to explore (this is called stateless verification). This
680 is done this way because saving intermediate states can rapidly
681 exhaust the available memory. If you want, you can change the value of
682 the ``model-check/checkpoint`` item. For example,
683 ``--cfg=model-check/checkpoint:1`` asks to take a checkpoint every
684 step. Beware, this will certainly explode your memory. Larger values
685 are probably better, make sure to experiment a bit to find the right
686 setting for your specific system.
688 .. _cfg=model-check/reduction:
690 Specifying the kind of reduction
691 ................................
693 The main issue when using the model-checking is the state space
694 explosion. You can activate some reduction technique with
695 ``--cfg=model-check/reduction:<technique>``. For now, this
696 configuration variable can take 2 values:
698 - **none:** Do not apply any kind of reduction (mandatory for
699 liveness properties, as our current DPOR algorithm breaks cycles)
700 - **dpor:** Apply Dynamic Partial Ordering Reduction. Only valid if
701 you verify local safety properties (default value for safety
704 Another way to mitigate the state space explosion is to search for
705 cycles in the exploration with the :ref:`cfg=model-check/visited`
706 configuration. Note that DPOR and state-equality reduction may not
707 play well together. You should choose between them.
709 Our current DPOR implementation could be improved in may ways. We are
710 currently improving its efficiency (both in term of reduction ability
711 and computational speed), and future work could make it compatible
712 with liveness properties.
714 .. _cfg=model-check/visited:
716 Size of Cycle Detection Set (state equality reduction)
717 ......................................................
719 Mc SimGrid can be asked to search for cycles during the exploration,
720 i.e. situations where a new explored state is in fact the same state
721 than a previous one.. This can prove useful to mitigate the state
722 space explosion with safety properties, and this is the crux when
723 searching for counter-examples to the liveness properties.
725 Note that this feature may break the current implementation of the
726 DPOR reduction technique.
728 The ``model-check/visited`` item is the maximum number of states, which
729 are stored in memory. If the maximum number of snapshotted state is
730 reached, some states will be removed from the memory and some cycles
731 might be missed. Small values can lead to incorrect verifications, but
732 large values can exhaust your memory and be CPU intensive as each new
733 state must be compared to that amount of older saved states.
735 The default settings depend on the kind of exploration. With safety
736 checking, no state is snapshotted and cycles cannot be detected. With
737 liveness checking, all states are snapshotted because missing a cycle
738 could hinder the exploration soundness.
740 .. _cfg=model-check/termination:
742 Non-Termination Detection
743 .........................
745 The ``model-check/termination`` configuration item can be used to
746 report if a non-termination execution path has been found. This is a
747 path with a cycle, which means that the program might never terminate.
749 This only works in safety mode, not in liveness mode.
751 This options is disabled by default.
753 .. _cfg=model-check/dot-output:
758 If set, the ``model-check/dot-output`` configuration item is the name
759 of a file in which to write a dot file of the path leading to the
760 property violation discovered (safety or liveness violation), as well
761 as the cycle for liveness properties. This dot file can then be fed to the
762 graphviz dot tool to generate a corresponding graphical representation.
764 .. _cfg=model-check/max-depth:
766 Exploration Depth Limit
767 .......................
769 The ``model-check/max-depth`` can set the maximum depth of the
770 exploration graph of the model checker. If this limit is reached, a
771 logging message is sent and the results might not be exact.
773 By default, the exploration is limited to the depth of 1000.
775 .. _cfg=model-check/timeout:
780 By default, the model checker does not handle timeout conditions: the `wait`
781 operations never time out. With the ``model-check/timeout`` configuration item
782 set to **yes**, the model checker will explore timeouts of `wait` operations.
784 .. _cfg=model-check/communications-determinism:
785 .. _cfg=model-check/send-determinism:
787 Communication Determinism
788 .........................
790 The ``model-check/communications-determinism`` and
791 ``model-check/send-determinism`` items can be used to select the
792 communication determinism mode of the model checker, which checks
793 determinism properties of the communications of an application.
795 .. _cfg=model-check/setenv:
797 Passing environment variables
798 .............................
800 You can specify extra environment variables to be set in the verified application
801 with ``model-check/setenv``. For example, you can preload a library as follows:
802 ``-cfg=model-check/setenv:LD_PRELOAD=toto;LD_LIBRARY_PATH=/tmp``.
806 Verification Performance Considerations
807 .......................................
809 The size of the stacks can have a huge impact on the memory
810 consumption when using model-checking. By default, each snapshot will
811 save a copy of the whole stacks and not only of the part that is
812 really meaningful: you should expect the contribution of the memory
813 consumption of the snapshots to be:
814 :math:`\text{number of processes} \times \text{stack size} \times \text{number of states}`.
816 When compiled against the model checker, the stacks are not
817 protected with guards: if the stack size is too small for your
818 application, the stack will silently overflow into other parts of the
819 memory (see :ref:`contexts/guard-size <cfg=contexts/guard-size>`).
821 .. _cfg=model-check/replay:
823 Replaying buggy execution paths from the model checker
824 ......................................................
826 Debugging the problems reported by the model checker is challenging:
827 First, the application under verification cannot be debugged with gdb
828 because the model checker already traces it. Then, the model checker may
829 explore several execution paths before encountering the issue, making it
830 very difficult to understand the output. Fortunately, SimGrid provides
831 the execution path leading to any reported issue so that you can replay
832 this path reported by the model checker, enabling the usage of classical
835 When the model checker finds an interesting path in the application
836 execution graph (where a safety or liveness property is violated), it
837 generates an identifier for this path. Here is an example of the output:
839 .. code-block:: console
841 [ 0.000000] (0:@) Check a safety property
842 [ 0.000000] (0:@) **************************
843 [ 0.000000] (0:@) *** PROPERTY NOT VALID ***
844 [ 0.000000] (0:@) **************************
845 [ 0.000000] (0:@) Counter-example execution trace:
846 [ 0.000000] (0:@) [(1)Tremblay (app)] MC_RANDOM(3)
847 [ 0.000000] (0:@) [(1)Tremblay (app)] MC_RANDOM(4)
848 [ 0.000000] (0:@) Path = 1/3;1/4
849 [ 0.000000] (0:@) Expanded states = 27
850 [ 0.000000] (0:@) Visited states = 68
851 [ 0.000000] (0:@) Executed transitions = 46
853 The interesting line is ``Path = 1/3;1/4``, which means that you should use
854 ``--cfg=model-check/replay:1/3;1/4`` to replay your application on the buggy
855 execution path. All options (but the model checker related ones) must
856 remain the same. In particular, if you ran your application with
857 ``smpirun -wrapper simgrid-mc``, then do it again. Remove all
858 MC-related options, keep non-MC-related ones and add
859 ``--cfg=model-check/replay:???``.
861 Currently, if the path is of the form ``X;Y;Z``, each number denotes
862 the actor's pid that is selected at each indecision point. If it's of
863 the form ``X/a;Y/b``, the X and Y are the selected pids while the a
864 and b are the return values of their simcalls. In the previous
865 example, ``1/3;1/4``, you can see from the full output that the actor
866 1 is doing MC_RANDOM simcalls, so the 3 and 4 simply denote the values
867 that these simcall return on the execution branch leading to the
870 Configuring the User Code Virtualization
871 ----------------------------------------
873 .. _cfg=contexts/factory:
875 Selecting the Virtualization Factory
876 ....................................
878 **Option** contexts/factory **Default:** "raw"
880 In SimGrid, the user code is virtualized in a specific mechanism that
881 allows the simulation kernel to control its execution: when a user
882 process requires a blocking action (such as sending a message), it is
883 interrupted, and only gets released when the simulated clock reaches
884 the point where the blocking operation is done. This is explained
885 graphically in the `relevant tutorial, available online
886 <https://simgrid.org/tutorials/simgrid-simix-101.pdf>`_.
888 In SimGrid, the containers in which user processes are virtualized are
889 called contexts. Several context factory are provided, and you can
890 select the one you want to use with the ``contexts/factory``
891 configuration item. Some of the following may not exist on your
892 machine because of portability issues. In any case, the default one
893 should be the most effcient one (please report bugs if the
894 auto-detection fails for you). They are approximately sorted here from
895 the slowest to the most efficient:
897 - **thread:** very slow factory using full featured threads (either
898 pthreads or windows native threads). They are slow but very
899 standard. Some debuggers or profilers only work with this factory.
900 - **java:** Java applications are virtualized onto java threads (that
901 are regular pthreads registered to the JVM)
902 - **ucontext:** fast factory using System V contexts (Linux and FreeBSD only)
903 - **boost:** This uses the `context
904 implementation <http://www.boost.org/doc/libs/1_59_0/libs/context/doc/html/index.html>`_
905 of the boost library for a performance that is comparable to our
907 |br| Install the relevant library (e.g. with the
908 libboost-contexts-dev package on Debian/Ubuntu) and recompile
910 - **raw:** amazingly fast factory using a context switching mechanism
911 of our own, directly implemented in assembly (only available for x86
912 and amd64 platforms for now) and without any unneeded system call.
914 The main reason to change this setting is when the debugging tools become
915 fooled by the optimized context factories. Threads are the most
916 debugging-friendly contexts, as they allow one to set breakpoints
917 anywhere with gdb and visualize backtraces for all processes, in order
918 to debug concurrency issues. Valgrind is also more comfortable with
919 threads, but it should be usable with all factories (Exception: the
920 callgrind tool really dislikes raw and ucontext factories).
922 .. _cfg=contexts/stack-size:
924 Adapting the Stack Size
925 .......................
927 **Option** ``contexts/stack-size`` **Default:** 8192 KiB
929 Each virtualized used process is executed using a specific system
930 stack. The size of this stack has a huge impact on the simulation
931 scalability, but its default value is rather large. This is because
932 the error messages that you get when the stack size is too small are
933 rather disturbing: this leads to stack overflow (overwriting other
934 stacks), leading to segfaults with corrupted stack traces.
936 If you want to push the scalability limits of your code, you might
937 want to reduce the ``contexts/stack-size`` item. Its default value is
938 8192 (in KiB), while our Chord simulation works with stacks as small
939 as 16 KiB, for example. You can ensure that some actors have a specific
940 size by simply changing the value of this configuration item before
941 creating these actors. The :cpp:func:`simgrid::s4u::Engine::set_config`
942 functions are handy for that.
944 This *setting is ignored* when using the thread factory (because there
945 is no way to modify the stack size with C++ system threads). Instead,
946 you should compile SimGrid and your application with
947 ``-fsplit-stack``. Note that this compilation flag is not compatible
948 with the model checker right now.
950 The operating system should only allocate memory for the pages of the
951 stack which are actually used and you might not need to use this in
952 most cases. However, this setting is very important when using the
953 model checker (see :ref:`options_mc_perf`).
955 .. _cfg=contexts/guard-size:
957 Disabling Stack Guard Pages
958 ...........................
960 **Option** ``contexts/guard-size`` **Default** 1 page in most case (0 pages on Windows or with MC)
962 Unless you use the threads context factory (see
963 :ref:`cfg=contexts/factory`), a stack guard page is usually used
964 which prevents the stack of a given actor from overflowing on another
965 stack. But the performance impact may become prohibitive when the
966 amount of actors increases. The option ``contexts/guard-size`` is the
967 number of stack guard pages used. By setting it to 0, no guard pages
968 will be used: in this case, you should avoid using small stacks (with
969 :ref:`contexts/stack-size <cfg=contexts/stack-size>`) as the stack
970 will silently overflow on other parts of the memory.
972 When no stack guard page is created, stacks may then silently overflow
973 on other parts of the memory if their size is too small for the
976 .. _cfg=contexts/nthreads:
977 .. _cfg=contexts/synchro:
979 Running User Code in Parallel
980 .............................
982 Parallel execution of the user code is only considered stable in
983 SimGrid v3.7 and higher, and mostly for MSG simulations. SMPI
984 simulations may well fail in parallel mode. It is described in
985 `INRIA RR-7653 <http://hal.inria.fr/inria-00602216/>`_.
987 If you are using the **ucontext** or **raw** context factories, you can
988 request to execute the user code in parallel. Several threads are
989 launched, each of them handling the same number of user contexts at each
990 run. To activate this, set the ``contexts/nthreads`` item to the amount
991 of cores that you have in your computer (or lower than 1 to have the
992 amount of cores auto-detected).
994 When parallel execution is activated, you can choose the
995 synchronization schema used with the ``contexts/synchro`` item,
996 which value is either:
998 - **futex:** ultra optimized synchronisation schema, based on futexes
999 (fast user-mode mutexes), and thus only available on Linux systems.
1000 This is the default mode when available.
1001 - **posix:** slow but portable synchronisation using only POSIX
1003 - **busy_wait:** not really a synchronisation: the worker threads
1004 constantly request new contexts to execute. It should be the most
1005 efficient synchronisation schema, but it loads all the cores of
1006 your machine for no good reason. You probably prefer the other less
1009 Configuring the Tracing
1010 -----------------------
1012 The :ref:`tracing subsystem <outcome_vizu>` can be configured in
1013 several different ways depending on the used interface (S4U, SMPI)
1014 and the kind of traces that needs to be obtained. See the
1015 :ref:`Tracing Configuration Options subsection
1016 <tracing_tracing_options>` for a full description of each
1017 configuration option.
1019 We detail here a simple way to get the traces working for you, even if
1020 you never used the tracing API.
1023 - Any SimGrid-based simulator (MSG, SMPI, ...) and raw traces:
1025 .. code-block:: none
1027 --cfg=tracing:yes --cfg=tracing/uncategorized:yes
1029 The first parameter activates the tracing subsystem, and the second
1030 tells it to trace host and link utilization (without any
1033 - MSG-based simulator and categorized traces (you need to
1034 declare categories and classify your tasks according to them)
1036 .. code-block:: none
1038 --cfg=tracing:yes --cfg=tracing/categorized:yes
1040 The first parameter activates the tracing subsystem, and the second
1041 tells it to trace host and link categorized utilization.
1043 - SMPI simulator and traces for a space/time view:
1045 .. code-block:: console
1047 $ smpirun -trace ...
1049 The `-trace` parameter for the smpirun script runs the simulation
1050 with ``--cfg=tracing:yes --cfg=tracing/smpi:yes``. Check the
1051 smpirun's `-help` parameter for additional tracing options.
1053 Sometimes you might want to put additional information on the trace to
1054 correctly identify them later, or to provide data that can be used to
1055 reproduce an experiment. You have two ways to do that:
1057 - Add a string on top of the trace file as comment:
1059 .. code-block:: none
1061 --cfg=tracing/comment:my_simulation_identifier
1063 - Add the contents of a textual file on top of the trace file as comment:
1065 .. code-block:: none
1067 --cfg=tracing/comment-file:my_file_with_additional_information.txt
1069 Please, use these two parameters (for comments) to make reproducible
1070 simulations. For additional details about this and all tracing
1071 options, check See the :ref:`tracing_tracing_options`.
1076 .. _cfg=msg/debug-multiple-use:
1081 **Option** ``msg/debug-multiple-use`` **Default:** off
1083 Sometimes your application may try to send a task that is still being
1084 executed somewhere else, making it impossible to send this task. However,
1085 for debugging purposes, one may want to know what the other host is/was
1086 doing. This option shows a backtrace of the other process.
1091 The SMPI interface provides several specific configuration items.
1092 These are not easy to see with ``--help-cfg``, since SMPI binaries are usually launched through the ``smiprun`` script.
1094 .. _cfg=smpi/host-speed:
1095 .. _cfg=smpi/cpu-threshold:
1096 .. _cfg=smpi/simulate-computation:
1098 Automatic Benchmarking of SMPI Code
1099 ...................................
1101 In SMPI, the sequential code is automatically benchmarked, and these
1102 computations are automatically reported to the simulator. That is to
1103 say that if you have a large computation between a ``MPI_Recv()`` and
1104 a ``MPI_Send()``, SMPI will automatically benchmark the duration of
1105 this code, and create an execution task within the simulator to take
1106 this into account. For that, the actual duration is measured on the
1107 host machine and then scaled to the power of the corresponding
1108 simulated machine. The variable ``smpi/host-speed`` allows one to
1109 specify the computational speed of the host machine (in flop/s by
1110 default) to use when scaling the execution times.
1112 The default value is ``smpi/host-speed=20kf`` (= 20,000 flop/s). This
1113 is probably underestimated for most machines, leading SimGrid to
1114 overestimate the amount of flops in the execution blocks that are
1115 automatically injected in the simulator. As a result, the execution
1116 time of the whole application will probably be overestimated until you
1117 use a realistic value.
1119 When the code consists of numerous consecutive MPI calls, the
1120 previous mechanism feeds the simulation kernel with numerous tiny
1121 computations. The ``smpi/cpu-threshold`` item becomes handy when this
1122 impacts badly on the simulation performance. It specifies a threshold (in
1123 seconds) below which the execution chunks are not reported to the
1124 simulation kernel (default value: 1e-6).
1126 .. note:: The option ``smpi/cpu-threshold`` ignores any computation
1127 time spent below this threshold. SMPI does not consider the
1128 `amount of time` of these computations; there is no offset for
1129 this. Hence, a value that is too small, may lead to unreliable
1132 In some cases, however, one may wish to disable simulation of
1133 the computation of an application. This is the case when SMPI is used not to
1134 simulate an MPI application, but instead an MPI code that performs
1135 "live replay" of another MPI app (e.g., ScalaTrace's replay tool, or
1136 various on-line simulators that run an app at scale). In this case the
1137 computation of the replay/simulation logic should not be simulated by
1138 SMPI. Instead, the replay tool or on-line simulator will issue
1139 "computation events", which correspond to the actual MPI simulation
1140 being replayed/simulated. At the moment, these computation events can
1141 be simulated using SMPI by calling internal smpi_execute*() functions.
1143 To disable the benchmarking/simulation of a computation in the simulated
1144 application, the variable ``smpi/simulate-computation`` should be set
1145 to **no**. This option just ignores the timings in your simulation; it
1146 still executes the computations itself. If you want to stop SMPI from
1147 doing that, you should check the SMPI_SAMPLE macros, documented in
1148 Section :ref:`SMPI_use_faster`.
1150 +------------------------------------+-------------------------+-----------------------------+
1151 | Solution | Computations executed? | Computations simulated? |
1152 +====================================+=========================+=============================+
1153 | --cfg=smpi/simulate-computation:no | Yes | Never |
1154 +------------------------------------+-------------------------+-----------------------------+
1155 | --cfg=smpi/cpu-threshold:42 | Yes, in all cases | If it lasts over 42 seconds |
1156 +------------------------------------+-------------------------+-----------------------------+
1157 | SMPI_SAMPLE() macro | Only once per loop nest | Always |
1158 +------------------------------------+-------------------------+-----------------------------+
1160 .. _cfg=smpi/comp-adjustment-file:
1162 Slow-down or speed-up parts of your code
1163 ........................................
1165 **Option** ``smpi/comp-adjustment-file:`` **Default:** unset
1167 This option allows you to pass a file that contains two columns: The
1168 first column defines the section that will be subject to a speedup;
1169 the second column is the speedup. For instance:
1171 .. code-block:: none
1173 "start:stop","ratio"
1174 "exchange_1.f:30:exchange_1.f:130",1.18244559422142
1176 The first line is the header - you must include it. The following
1177 line means that the code between two consecutive MPI calls on line 30
1178 in exchange_1.f and line 130 in exchange_1.f should receive a speedup
1179 of 1.18244559422142. The value for the second column is therefore a
1180 speedup, if it is larger than 1 and a slowdown if it is smaller
1181 than 1. Nothing will be changed if it is equal to 1.
1183 Of course, you can set any arbitrary filenames you want (so the start
1184 and end don't have to be in the same file), but be aware that this
1185 mechanism only supports `consecutive calls!`
1187 Please note that you must pass the ``-trace-call-location`` flag to
1188 smpicc or smpiff, respectively. This flag activates some internal
1189 macro definitions that help with obtaining the call location.
1191 Bandwidth and latency factors
1192 .............................
1194 Adapting the bandwidth and latency acurately to the network conditions is of a paramount importance to get realistic results.
1195 This is done through the :ref:`network/bandwidth-factor <cfg=network/bandwidth-factor>` and :ref:`network/latency-factor
1196 <cfg=network/latency-factor>` items. You probably also want to read the following section: :ref:`models_calibration`.
1198 .. _cfg=smpi/display-timing:
1200 Reporting Simulation Time
1201 .........................
1203 **Option** ``smpi/display-timing`` **Default:** 0 (false)
1205 Most of the time, you run MPI code with SMPI to compute the time it
1206 would take to run it on a platform. But since the code is run through
1207 the ``smpirun`` script, you don't have any control on the launcher
1208 code, making it difficult to report the simulated time when the
1209 simulation ends. If you enable the ``smpi/display-timing`` item,
1210 ``smpirun`` will display this information when the simulation
1212 SMPI will also display information about the amout of real time spent
1213 in application code and in SMPI internals, to provide hints about the
1214 need to use sampling to reduce simulation time.
1216 .. _cfg=smpi/display-allocs:
1218 Reporting memory allocations
1219 ............................
1221 **Option** ``smpi/display-allocs`` **Default:** 0 (false)
1223 SMPI intercepts malloc and calloc calls performed inside the running
1224 application, if it wasn't compiled with SMPI_NO_OVERRIDE_MALLOC.
1225 With this option, SMPI will show at the end of execution the amount of
1226 memory allocated through these calls, and locate the most expensive one.
1227 This helps finding the targets for manual memory sharing, or the threshold
1228 to use for smpi/auto-shared-malloc-thresh option (see :ref:`cfg=smpi/auto-shared-malloc-thresh`).
1230 .. _cfg=smpi/keep-temps:
1232 Keeping temporary files after simulation
1233 ........................................
1235 **Option** ``smpi/keep-temps`` **default:** 0 (false)
1237 SMPI usually generates a lot of temporary files that are cleaned after
1238 use. This option requests to preserve them, for example to debug or
1239 profile your code. Indeed, the binary files are removed very early
1240 under the dlopen privatization schema, which tends to fool the
1243 .. _cfg=smpi/papi-events:
1245 Trace hardware counters with PAPI
1246 .................................
1248 **Option** ``smpi/papi-events`` **default:** unset
1250 When the PAPI support is compiled into SimGrid, this option takes the
1251 names of PAPI counters and adds their respective values to the trace
1252 files (See Section :ref:`tracing_tracing_options`).
1256 This feature currently requires superuser privileges, as registers
1257 are queried. Only use this feature with code you trust! Call
1258 smpirun for instance via ``smpirun -wrapper "sudo "
1259 <your-parameters>`` or run ``sudo sh -c "echo 0 >
1260 /proc/sys/kernel/perf_event_paranoid"`` In the later case, sudo
1261 will not be required.
1263 It is planned to make this feature available on a per-process (or per-thread?) basis.
1264 The first draft, however, just implements a "global" (i.e., for all processes) set
1265 of counters, the "default" set.
1267 .. code-block:: none
1269 --cfg=smpi/papi-events:"default:PAPI_L3_LDM:PAPI_L2_LDM"
1271 .. _cfg=smpi/privatization:
1273 Automatic Privatization of Global Variables
1274 ...........................................
1276 **Option** ``smpi/privatization`` **default:** "dlopen" (when using smpirun)
1278 MPI executables are usually meant to be executed in separate
1279 processes, but SMPI is executed in only one process. Global variables
1280 from executables will be placed in the same memory region and shared
1281 between processes, causing intricate bugs. Several options are
1282 possible to avoid this, as described in the main `SMPI publication
1283 <https://hal.inria.fr/hal-01415484>`_ and in the :ref:`SMPI
1284 documentation <SMPI_what_globals>`. SimGrid provides two ways of
1285 automatically privatizing the globals, and this option allows one to
1286 choose between them.
1288 - **no** (default when not using smpirun): Do not automatically
1289 privatize variables. Pass ``-no-privatize`` to smpirun to disable
1291 - **dlopen** or **yes** (default when using smpirun): Link multiple
1292 times against the binary.
1293 - **mmap** (slower, but maybe somewhat more stable):
1294 Runtime automatic switching of the data segments.
1297 This configuration option cannot be set in your platform file. You can only
1298 pass it as an argument to smpirun.
1300 .. _cfg=smpi/privatize-libs:
1302 Automatic privatization of global variables inside external libraries
1303 .....................................................................
1305 **Option** ``smpi/privatize-libs`` **default:** unset
1307 **Linux/BSD only:** When using dlopen (default) privatization,
1308 privatize specific shared libraries with internal global variables, if
1309 they can't be linked statically. For example libgfortran is usually
1310 used for Fortran I/O and indexes in files can be mixed up.
1312 Multiple libraries can be given, semicolon separated.
1314 This configuration option can only use either full paths to libraries,
1315 or full names. Check with ldd the name of the library you want to
1318 .. code-block:: console
1322 libgfortran.so.3 => /usr/lib/x86_64-linux-gnu/libgfortran.so.3 (0x00007fbb4d91b000)
1325 Then you can use ``--cfg=smpi/privatize-libs:libgfortran.so.3``
1326 or ``--cfg=smpi/privatize-libs:/usr/lib/x86_64-linux-gnu/libgfortran.so.3``,
1327 but not ``libgfortran`` nor ``libgfortran.so``.
1329 .. _cfg=smpi/send-is-detached-thresh:
1331 Simulating MPI detached send
1332 ............................
1334 **Option** ``smpi/send-is-detached-thresh`` **default:** 65536
1336 This threshold specifies the size in bytes under which the send will
1337 return immediately. This is different from the threshold detailed in
1338 :ref:`cfg=smpi/async-small-thresh` because the message is not
1339 really sent when the send is posted. SMPI still waits for the
1340 corresponding receive to be posted, in order to perform the communication
1343 .. _cfg=smpi/coll-selector:
1345 Simulating MPI collective algorithms
1346 ....................................
1348 **Option** ``smpi/coll-selector`` **Possible values:** naive (default), ompi, mpich
1350 SMPI implements more than 100 different algorithms for MPI collective
1351 communication, to accurately simulate the behavior of most of the
1352 existing MPI libraries. The ``smpi/coll-selector`` item can be used to
1353 select the decision logic either of the OpenMPI or the MPICH libraries. (By
1354 default SMPI uses naive version of collective operations.)
1356 Each collective operation can be manually selected with a
1357 ``smpi/collective_name:algo_name``. Available algorithms are listed in
1358 :ref:`SMPI_use_colls`.
1360 .. TODO:: All available collective algorithms will be made available
1361 via the ``smpirun --help-coll`` command.
1363 .. _cfg=smpi/barrier-collectives:
1365 Add a barrier in all collectives
1366 ................................
1368 **Option** ``smpi/barrier-collectives`` **default:** off
1370 This option adds a simple barrier in some collective operations to catch dangerous
1371 code that may or may not work depending on the MPI implementation: Bcast, Exscan,
1372 Gather, Gatherv, Scan, Scatter, Scatterv and Reduce.
1374 For example, the following code works with OpenMPI while it deadlocks in MPICH and
1375 Intel MPI. Broadcast seem to be "fire and forget" in OpenMPI while other
1376 implementations expect to receive a message.
1381 MPI_Bcast(buf1, buff_size, MPI_CHAR, 0, newcom);
1382 MPI_Send(&buf2, buff_size, MPI_CHAR, 1, tag, newcom);
1383 } else if (rank==1) {
1384 MPI_Recv(&buf2, buff_size, MPI_CHAR, 0, tag, newcom, MPI_STATUS_IGNORE);
1385 MPI_Bcast(buf1, buff_size, MPI_CHAR, 0, newcom);
1388 The barrier is only simulated and does not involve any additional message (it is a S4U barrier).
1389 This option is disabled by default, and activated by the `-analyze` flag of smpirun.
1391 .. _cfg=smpi/barrier-finalization:
1393 Add a barrier in MPI_Finalize
1394 .............................
1396 **Option** ``smpi/finalization-barrier`` **default:** off
1398 By default, SMPI processes are destroyed as soon as soon as their code ends,
1399 so after a successful MPI_Finalize call returns. In some rare cases, some data
1400 might have been attached to MPI objects still active in the remaining processes,
1401 and can be destroyed eagerly by the finished process.
1402 If your code shows issues at finalization, such as segmentation fault, triggering
1403 this option will add an explicit MPI_Barrier(MPI_COMM_WORLD) call inside the
1404 MPI_Finalize, so that all processes will terminate at almost the same point.
1405 It might affect the total timing by the cost of a barrier.
1407 .. _cfg=smpi/errors-are-fatal:
1409 Disable MPI fatal errors
1410 ........................
1412 **Option** ``smpi/errors-are-fatal`` **default:** on
1414 By default, SMPI processes will crash if a MPI error code is returned. MPI allows
1415 to explicitely set MPI_ERRORS_RETURN errhandler to avoid this behaviour. This flag
1416 will turn on this behaviour by default (for all concerned types and errhandlers).
1417 This can ease debugging by going after the first reported error.
1419 .. _cfg=smpi/pedantic:
1421 Disable pedantic MPI errors
1422 ...........................
1424 **Option** ``smpi/pedantic`` **default:** on
1426 By default, SMPI will report all errors it finds in MPI codes. Some of these errors
1427 may not be considered as errors by all developers. This flag can be turned off to
1428 avoid reporting some usually harmless mistakes.
1429 Concerned errors list (will be expanded in the future):
1431 - Calling MPI_Win_fence only once in a program, hence just opening an epoch without
1434 .. _cfg=smpi/iprobe:
1436 Inject constant times for MPI_Iprobe
1437 ....................................
1439 **Option** ``smpi/iprobe`` **default:** 0.0001
1441 The behavior and motivation for this configuration option is identical
1442 with :ref:`smpi/test <cfg=smpi/test>`, but for the function
1445 .. _cfg=smpi/iprobe-cpu-usage:
1447 Reduce speed for iprobe calls
1448 .............................
1450 **Option** ``smpi/iprobe-cpu-usage`` **default:** 1 (no change)
1452 MPI_Iprobe calls can be heavily used in applications. To account
1453 correctly for the energy that cores spend probing, it is necessary to
1454 reduce the load that these calls cause inside SimGrid.
1456 For instance, we measured a maximum power consumption of 220 W for a
1457 particular application but only 180 W while this application was
1458 probing. Hence, the correct factor that should be passed to this
1459 option would be 180/220 = 0.81.
1463 Inject constant times for MPI_Init
1464 ..................................
1466 **Option** ``smpi/init`` **default:** 0
1468 The behavior and motivation for this configuration option is identical
1469 with :ref:`smpi/test <cfg=smpi/test>`, but for the function ``MPI_Init()``.
1473 Inject constant times for MPI_Isend()
1474 .....................................
1476 **Option** ``smpi/ois``
1478 The behavior and motivation for this configuration option is identical
1479 with :ref:`smpi/os <cfg=smpi/os>`, but for the function ``MPI_Isend()``.
1483 Inject constant times for MPI_send()
1484 ....................................
1486 **Option** ``smpi/os``
1488 In several network models such as LogP, send (MPI_Send, MPI_Isend) and
1489 receive (MPI_Recv) operations incur costs (i.e., they consume CPU
1490 time). SMPI can factor these costs in as well, but the user has to
1491 configure SMPI accordingly as these values may vary by machine. This
1492 can be done by using ``smpi/os`` for MPI_Send operations; for MPI_Isend
1493 and MPI_Recv, use ``smpi/ois`` and ``smpi/or``, respectively. These work
1494 exactly as ``smpi/ois``.
1496 This item can consist of multiple sections; each section takes three
1497 values, for example ``1:3:2;10:5:1``. The sections are divided by ";"
1498 so this example contains two sections. Furthermore, each section
1499 consists of three values.
1501 1. The first value denotes the minimum size in bytes for this section to take effect;
1502 read it as "if message size is greater than this value (and other section has a larger
1503 first value that is also smaller than the message size), use this".
1504 In the first section above, this value is "1".
1506 2. The second value is the startup time; this is a constant value that will always
1507 be charged, no matter what the size of the message. In the first section above,
1510 3. The third value is the `per-byte` cost. That is, it is charged for every
1511 byte of the message (incurring cost messageSize*cost_per_byte)
1512 and hence accounts also for larger messages. In the first
1513 section of the example above, this value is "2".
1515 Now, SMPI always checks which section it should use for a given
1516 message; that is, if a message of size 11 is sent with the
1517 configuration of the example above, only the second section will be
1518 used, not the first, as the first value of the second section is
1519 closer to the message size. Hence, when ``smpi/os=1:3:2;10:5:1``, a
1520 message of size 11 incurs the following cost inside MPI_Send:
1521 ``5+11*1`` because 5 is the startup cost and 1 is the cost per byte.
1523 Note that the order of sections can be arbitrary; they will be ordered internally.
1527 Inject constant times for MPI_Recv()
1528 ....................................
1530 **Option** ``smpi/or``
1532 The behavior and motivation for this configuration option is identical
1533 with :ref:`smpi/os <cfg=smpi/os>`, but for the function ``MPI_Recv()``.
1536 .. _cfg=smpi/grow-injected-times:
1538 Inject constant times for MPI_Test
1539 ..................................
1541 **Option** ``smpi/test`` **default:** 0.0001
1543 By setting this option, you can control the amount of time a process
1544 sleeps when MPI_Test() is called; this is important, because SimGrid
1545 normally only advances the time while communication is happening and
1546 thus, MPI_Test will not add to the time, resulting in deadlock if it is
1547 used as a break-condition as in the following example:
1552 MPI_Test(request, flag, status);
1556 To speed up execution, we use a counter to keep track of how often we
1557 checked if the handle is now valid or not. Hence, we actually
1558 use counter*SLEEP_TIME, that is, the time MPI_Test() causes the
1559 process to sleep increases linearly with the number of previously
1560 failed tests. This behavior can be disabled by setting
1561 ``smpi/grow-injected-times`` to **no**. This will also disable this
1562 behavior for MPI_Iprobe.
1564 .. _cfg=smpi/shared-malloc:
1565 .. _cfg=smpi/shared-malloc-hugepage:
1570 **Option** ``smpi/shared-malloc`` **Possible values:** global (default), local
1572 If your simulation consumes too much memory, you may want to modify
1573 your code so that the working areas are shared by all MPI ranks. For
1574 example, in a block-cyclic matrix multiplication, you will only
1575 allocate one set of blocks, and all processes will share them.
1576 Naturally, this will lead to very wrong results, but this will save a
1577 lot of memory. So this is still desirable for some studies. For more on
1578 the motivation for that feature, please refer to the `relevant section
1579 <https://simgrid.github.io/SMPI_CourseWare/topic_understanding_performance/matrixmultiplication>`_
1580 of the SMPI CourseWare (see Activity #2.2 of the pointed
1581 assignment). In practice, change the calls for malloc() and free() into
1582 SMPI_SHARED_MALLOC() and SMPI_SHARED_FREE().
1584 SMPI provides two algorithms for this feature. The first one, called
1585 ``local``, allocates one block per call to SMPI_SHARED_MALLOC()
1586 (each call site gets its own block) ,and this block is shared
1587 among all MPI ranks. This is implemented with the shm_* functions
1588 to create a new POSIX shared memory object (kept in RAM, in /dev/shm)
1589 for each shared block.
1591 With the ``global`` algorithm, each call to SMPI_SHARED_MALLOC()
1592 returns a new address, but it only points to a shadow block: its memory
1593 area is mapped on a 1 MiB file on disk. If the returned block is of size
1594 N MiB, then the same file is mapped N times to cover the whole block.
1595 At the end, no matter how many times you call SMPI_SHARED_MALLOC, this will
1596 only consume 1 MiB in memory.
1598 You can disable this behavior and come back to regular mallocs (for
1599 example for debugging purposes) using ``no`` as a value.
1601 If you want to keep private some parts of the buffer, for instance if these
1602 parts are used by the application logic and should not be corrupted, you
1603 can use SMPI_PARTIAL_SHARED_MALLOC(size, offsets, offsets_count). For example:
1607 mem = SMPI_PARTIAL_SHARED_MALLOC(500, {27,42 , 100,200}, 2);
1609 This will allocate 500 bytes to mem, such that mem[27..41] and
1610 mem[100..199] are shared while other area remain private.
1612 Then, it can be deallocated by calling SMPI_SHARED_FREE(mem).
1614 When smpi/shared-malloc:global is used, the memory consumption problem
1615 is solved, but it may induce too much load on the kernel's pages table.
1616 In this case, you should use huge pages so that the kernel creates only one
1617 entry per MB of malloced data instead of one entry per 4 kB.
1618 To activate this, you must mount a hugetlbfs on your system and allocate
1619 at least one huge page:
1621 .. code-block:: console
1624 $ sudo mount none /home/huge -t hugetlbfs -o rw,mode=0777
1625 $ sudo sh -c 'echo 1 > /proc/sys/vm/nr_hugepages' # echo more if you need more
1627 Then, you can pass the option
1628 ``--cfg=smpi/shared-malloc-hugepage:/home/huge`` to smpirun to
1629 actually activate the huge page support in shared mallocs.
1631 .. _cfg=smpi/auto-shared-malloc-thresh:
1633 Automatically share allocations
1634 ...............................
1636 **Option** ``smpi/auto-shared-malloc-thresh:`` **Default:** 0 (false)
1637 This value in bytes represents the size above which all allocations
1638 will be "shared" by default (as if they were performed through
1639 SMPI_SHARED_MALLOC macros). Default = 0 = disabled feature.
1640 The value must be carefully chosen to only select data buffers which
1641 will not modify execution path or cause crash if their content is false.
1642 Option :ref:`cfg=smpi/display-allocs` can be used to locate the largest
1643 allocation detected in a run, and provide a good starting threshold.
1644 Note : malloc, calloc and free are overridden by smpicc/cxx by default.
1645 This can cause some troubles if codes are already overriding these. If this
1646 is the case, defining SMPI_NO_OVERRIDE_MALLOC in the compilation flags can
1647 help, but will make this feature unusable.
1651 Inject constant times for MPI_Wtime, gettimeofday and clock_gettime
1652 ...................................................................
1654 **Option** ``smpi/wtime`` **default:** 10 ns
1656 This option controls the amount of (simulated) time spent in calls to
1657 MPI_Wtime(), gettimeofday() and clock_gettime(). If you set this value
1658 to 0, the simulated clock is not advanced in these calls, which leads
1659 to issues if your application contains such a loop:
1663 while(MPI_Wtime() < some_time_bound) {
1664 /* some tests, with no communication nor computation */
1667 When the option smpi/wtime is set to 0, the time advances only on
1668 communications and computations. So the previous code results in an
1669 infinite loop: the current [simulated] time will never reach
1670 ``some_time_bound``. This infinite loop is avoided when that option
1671 is set to a small value, as it is by default since SimGrid v3.21.
1673 Note that if your application does not contain any loop depending on
1674 the current time only, then setting this option to a non-zero value
1675 will slow down your simulations by a tiny bit: the simulation loop has
1676 to be broken out of and reset each time your code asks for the current time.
1677 If the simulation speed really matters to you, you can avoid this
1678 extra delay by setting smpi/wtime to 0.
1680 .. _cfg=smpi/list-leaks:
1682 Report leaked MPI objects
1683 .........................
1685 **Option** ``smpi/list-leaks`` **default:** 0
1687 This option controls whether to report leaked MPI objects.
1688 The parameter is the number of leaks to report.
1690 Other Configurations
1691 --------------------
1693 .. _cfg=debug/clean-atexit:
1695 Cleanup at Termination
1696 ......................
1698 **Option** ``debug/clean-atexit`` **default:** on
1700 If your code is segfaulting during its finalization, it may help to
1701 disable this option to request that SimGrid not attempt any cleanups at
1702 the end of the simulation. Since the Unix process is ending anyway,
1703 the operating system will wipe it all.
1710 **Option** ``path`` **default:** . (current dir)
1712 It is possible to specify a list of directories to search in for the
1713 trace files (see :ref:`pf_trace`) by using this configuration
1714 item. To add several directory to the path, set the configuration
1715 item several times, as in ``--cfg=path:toto --cfg=path:tutu``
1717 .. _cfg=debug/breakpoint:
1722 **Option** ``debug/breakpoint`` **default:** unset
1724 This configuration option sets a breakpoint: when the simulated clock
1725 reaches the given time, a SIGTRAP is raised. This can be used to stop
1726 the execution and get a backtrace with a debugger.
1728 It is also possible to set the breakpoint from inside the debugger, by
1729 writing in global variable simgrid::kernel::cfg_breakpoint. For example,
1732 .. code-block:: none
1734 set variable simgrid::kernel::cfg_breakpoint = 3.1416
1736 .. _cfg=debug/verbose-exit:
1741 **Option** ``debug/verbose-exit`` **default:** on
1743 By default, when Ctrl-C is pressed, the status of all existing actors
1744 is displayed before exiting the simulation. This is very useful to
1745 debug your code, but it can become troublesome if you have many
1746 actors. Set this configuration item to **off** to disable this
1749 .. _cfg=exception/cutpath:
1751 Truncate local path from exception backtrace
1752 ............................................
1754 **Option** ``exception/cutpath`` **default:** off
1756 This configuration option is used to remove the path from the
1757 backtrace shown when an exception is thrown. This is mainly useful for
1758 the tests: the full file path would makes the tests non-reproducible because
1759 the paths of source files depend of the build settings. That would
1760 break most of the tests since their output is continually compared.
1764 Logging configuration
1765 ---------------------
1767 As introduced in :ref:`outcome_logs`, the SimGrid logging mechanism allows to configure at runtime the messages that should be displayed and those that should be omitted. Each
1768 message produced in the code is given a category (denoting its topic) and a priority. Then at runtime, each category is given a threshold (only messages of priority higher than
1769 that threshold are displayed), a layout (deciding how the messages in this category are formatted), and an appender (deciding what to do with the message: either print on stderr or
1772 This section explains how to configure this logging features. You can also refer to the documentation of the :ref:`programmer's interface <logging_prog>`, that allows to produce
1773 messages from your code.
1775 Most of the time, the logging mechanism is configured at runtime using the ``--log`` command-line argument, even if you can also use :c:func:`xbt_log_control_set()` to control it from
1776 your program. To pass configure more than one setting, you can either pass several ``--log`` arguments, or separate your settings with spaces, that must be quoted accordingly. In
1777 practice, the following is equivalent to the above settings: ``--log=root.thresh:error --log=s4u_host.thresh:debug``.
1779 If you want to specify more than one setting, you can either pass several ``--log`` argument to your program as above, or separate them with spaces. In this case, you want to quote
1780 your settings, as in ``--log="root.thresh:error s4u_host.thresh:debug"``. The parameters are interpreted in order, from left to right.
1783 Threshold configuration
1784 .......................
1786 The keyword ``threshold`` controls which logging event will get displayed in a given category. For example, ``--log=root.threshold:debug`` displays *every* message produced in the
1787 ``root`` category and its subcategories (i.e., every message produced -- this is *extremely* verbose), while ``--log=root.thres:critical`` turns almost everything off. As you can
1788 see, ``threshold`` can be abbreviated here.
1790 Existing thresholds:
1792 - ``trace`` some functions display a message at this level when entering or returning
1793 - ``debug`` output that is mostly useful when debugging the corresponding module.
1794 - ``verbose`` verbose output that is only mildly interesting and can easily be ignored
1795 - ``info`` usual output (this is the default threshold of all categories)
1796 - ``warning`` minor issue encountered
1797 - ``error`` issue encountered
1798 - ``critical`` major issue encountered, such as assertions failures
1802 Format configuration
1803 ....................
1805 The keyword ``fmt`` controls the layout (the format) of a logging category. For example, ``--log=root.fmt:%m`` reduces the output to the user-message only, removing any decoration such
1806 as the date, or the actor ID, everything. Existing format directives:
1809 - %n: line separator (LOG4J compatible)
1810 - %e: plain old space (SimGrid extension)
1812 - %m: user-provided message
1814 - %c: Category name (LOG4J compatible)
1815 - %p: Priority name (LOG4J compatible)
1817 - %h: Hostname (SimGrid extension)
1818 - %a: Actor name (SimGrid extension -- note that with SMPI this is the integer value of the process rank)
1819 - %i: Actor PID (SimGrid extension -- this is a 'i' as in 'i'dea)
1820 - %t: Thread "name" (LOG4J compatible -- actually the address of the thread in memory)
1822 - %F: file name where the log event was raised (LOG4J compatible)
1823 - %l: location where the log event was raised (LOG4J compatible, like '%%F:%%L' -- this is a l as in 'l'etter)
1824 - %L: line number where the log event was raised (LOG4J compatible)
1825 - %M: function name (LOG4J compatible -- called method name here of course).
1827 - %d: date (UNIX-like epoch)
1828 - %r: application age (time elapsed since the beginning of the application)
1831 ``--log=root.fmt:'[%h:%a:(%i) %r] %l: %m%n'`` gives you the default layout used for info messages while ``--log=root.fmt:'[%h:%a:(%i) %r] %l: [%c/%p] %m%n'`` gives you the default
1832 layout for the other priorities (it adds the source code location). Also, the actor identification is omitted by the default layout for the messages coming directly from the
1833 SimGrid kernel, so info messages are formatted with ``[%r] [%c/%p] %m%n`` in this case. When specifying the layout manually, such distinctions are currently impossible, and the
1834 provided layout is used for every messages.
1836 As with printf, you can specify the precision and width of the fields. For example, ``%.4r`` limits the date precision to four digits while ``%15h`` limits the host name to at most
1840 If you want to have spaces in your log format, you should protect it. Otherwise, SimGrid will consider that this is a space-separated list of several parameters. But you should
1841 also protect it from the shell that also splits command line arguments on spaces. At the end, you should use something such as ``--log="'root.fmt:%l: [%p/%c]: %m%n'"``.
1842 Another option is to use the ``%e`` directive for spaces, as in ``--log=root.fmt:%l:%e[%p/%c]:%e%m%n``.
1847 The keyword ``app`` controls the appended of a logging category. For example ``--log=root.app:file:mylogfile`` redirects every output to the file ``mylogfile``.
1849 With the ``splitfile`` appender, a new file is created when the size of the output reaches the specified size. The format is ``--log=root.app:splitfile:<size>:<file name>``. For
1850 example, ``--log=root.app:splitfile:500:mylog_%`` creates log files of at most 500 bytes, using the names ``mylog_0``, ``mylog_1``, ``mylog_2``, etc.
1852 The ``rollfile`` appender uses one file only, but the file is emptied and recreated when its size reaches the specified maximum. For example, ``--log=root.app:rollfile:500:mylog``
1853 ensures that the log file ``mylog`` will never overpass 500 bytes in size.
1855 Any appender setup this way have its own layout format, that you may change afterward. When specifying a new appender, its additivity is set to false to prevent log event displayed
1856 by this appender to "leak" to any other appender higher in the hierarchy. You can naturally change that if you want your messages to be displayed twice.
1861 The keyword ``add`` controls the additivity of a logging category. By default, the messages are only passed one appender only: the more specific, i.e. the first one found when
1862 climbing the tree from the category in which they were produced. In Log4J parlance, it is said that the default additivity of appenders is false. If you change this setting to
1863 ``on`` (or ``yes`` or ``1``), the produced messages will also be passed to the upper appender.
1865 Let's consider a more complex example: ``--log="root.app:file:all.log s4u.app:file:iface.log xbt.app:file:xbt.log xbt.add:yes``. Here, the logging of s4u will be sent to the
1866 ``iface.log`` file; the logging of the xbt toolbox will be sent to both the ``xbt.log`` file and the ``all.log`` file (because xbt additivity was enabled); and every other loggings
1867 will only be sent to ``all.log``.
1872 ``--help-logs`` displays a complete help message about logging in SimGrid.
1874 ``--help-log-categories`` displays the actual hierarchy of log categories for this binary.
1876 ``--log=no_loc`` hides the source locations (file names and line numbers) from the messages. This is useful to make tests reproducible.