(sometimes after the merge with the ORTE branch), the opal_pointer_array
will became the only pointer_array implementation (the orte_pointer_array
will be removed).
This commit was SVN r17007.
This commit brings over all the work from the /tmp-public/datarep
branch. See commits r16855, r16859, r16860 for the highlights of what
was done.
This commit was SVN r16891.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r16855
r16859
r16860
The following Trac tickets were found above:
Ticket 1029 --> https://svn.open-mpi.org/trac/ompi/ticket/1029
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component.
This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done:
As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in.
In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in.
The incoming changes revamp these procedures in three ways:
1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step.
The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic.
Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure.
2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed.
The size of this data has been reduced in three ways:
(a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes.
To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose.
(b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction.
(c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using.
While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly.
3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup.
It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging.
Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future.
There are a few minor additional changes in the commit that I'll just note in passing:
* propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details.
* requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details.
* cleanup of some stale header files
This commit was SVN r16364.
that it is >= 1, so making it a size_t makes it easier to interact
with all the other size_t variables and removes a compiler warning.
This commit was SVN r15935.
used at nce (up to one unique collective module per collective function).
Matches r15795:15921 of the tmp/bwb-coll-select branch
This commit was SVN r15924.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r15795
r15921
mpi_preconnect_oob_simultaneous > np. Need to scale back
simultaneous to equal np in those cases. Reviewed by Brian.
This commit fixes trac:1064.
This commit was SVN r15916.
The following Trac tickets were found above:
Ticket 1064 --> https://svn.open-mpi.org/trac/ompi/ticket/1064
in the OMPI proc structures. For now, use an extension of the modex that is
keyed on strings. Eventually, this should use the attribute put/get that is
part of the RSL interface.
This commit was SVN r15820.
mpi_show_mpi_alloc_mem_leaks
When activated, MPI_FINALIZE displays a list of memory allocations
from MPI_ALLOC_MEM that were not freed by MPI_FREE_MEM (in each MPI
process).
* If set to a positive integer, display only that many leaks.
* If set to a negative integer, display all leaks.
* If set to 0, do not show any leaks.
This commit was SVN r15736.
* General TCP cleanup for OPAL / ORTE
* Simplifying the OOB by moving much of the logic into the RML
* Allowing the OOB RML component to do routing of messages
* Adding a component framework for handling routing tables
* Moving the xcast functionality from the OOB base to its own framework
Includes merge from tmp/bwb-oob-rml-merge revisions:
r15506, r15507, r15508, r15510, r15511, r15512, r15513
This commit was SVN r15528.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r15506
r15507
r15508
r15510
r15511
r15512
r15513
Cleanup ALL instances of output involving the printing of orte_process_name_t structures using the ORTE_NAME_ARGS macro so that the number of fields and type of data match. Replace those values with a new macro/function pair ORTE_NAME_PRINT that outputs a string (using the new thread safe data capability) so that any future changes to the printing of those structures can be accomplished with a change to a single point.
Note that I could not possibly find outputs that directly print the orte_process_name_t fields, but only dealt with those that used ORTE_NAME_ARGS. Hence, you may still have a few outputs that bark during compilation. Also, I could only verify those that fall within environments I can compile on, so other environments may yield some minor warnings.
This commit was SVN r15517.
* bml.h had a change that introduced a variable named "_order" to
avoid a conflict with a local variable. The namespace starting
with _ belongs to the os/compiler/kernel/not us. So we can't start
symbols with _. So I replaced it with arg_order, and also updated
the threaded equivalent of the macro that was modified.
* in btl_openib_proc.c, one opal_output accidentally had its string
reverted from "ompi_modex_recv..." to
"mca_pml_base_modex_recv....". This was fixed.
* The change to ompi/runtime/ompi_preconnect.c was entirely
reverted; it was an artifact of debugging.
This commit was SVN r15475.
The following SVN revision numbers were found above:
r15474 --> open-mpi/ompi@8ace07efed
1. Galen's fine-grain control of queue pair resources in the openib
BTL.
1. Pasha's new implementation of asychronous HCA event handling.
Pasha's new implementation doesn't take much explanation, but the new
"multifrag" stuff does.
Note that "svn merge" was not used to bring this new code from the
/tmp/ib_multifrag branch -- something Bad happened in the periodic
trunk pulls on that branch making an actual merge back to the trunk
effectively impossible (i.e., lots and lots of arbitrary conflicts and
artifical changes). :-(
== Fine-grain control of queue pair resources ==
Galen's fine-grain control of queue pair resources to the OpenIB BTL
(thanks to Gleb for fixing broken code and providing additional
functionality, Pasha for finding broken code, and Jeff for doing all
the svn work and regression testing).
Prior to this commit, the OpenIB BTL created two queue pairs: one for
eager size fragments and one for max send size fragments. When the
use of the shared receive queue (SRQ) was specified (via "-mca
btl_openib_use_srq 1"), these QPs would use a shared receive queue for
receive buffers instead of the default per-peer (PP) receive queues
and buffers. One consequence of this design is that receive buffer
utilization (the size of the data received as a percentage of the
receive buffer used for the data) was quite poor for a number of
applications.
The new design allows multiple QPs to be specified at runtime. Each
QP can be setup to use PP or SRQ receive buffers as well as giving
fine-grained control over receive buffer size, number of receive
buffers to post, when to replenish the receive queue (low water mark)
and for SRQ QPs, the number of outstanding sends can also be
specified. The following is an example of the syntax to describe QPs
to the OpenIB BTL using the new MCA parameter btl_openib_receive_queues:
{{{
-mca btl_openib_receive_queues \
"P,128,16,4;S,1024,256,128,32;S,4096,256,128,32;S,65536,256,128,32"
}}}
Each QP description is delimited by ";" (semicolon) with individual
fields of the QP description delimited by "," (comma). The above
example therefore describes 4 QPs.
The first QP is:
P,128,16,4
Meaning: per-peer receive buffer QPs are indicated by a starting field
of "P"; the first QP (shown above) is therefore a per-peer based QP.
The second field indicates the size of the receive buffer in bytes
(128 bytes). The third field indicates the number of receive buffers
to allocate to the QP (16). The fourth field indicates the low
watermark for receive buffers at which time the BTL will repost
receive buffers to the QP (4).
The second QP is:
S,1024,256,128,32
Shared receive queue based QPs are indicated by a starting field of
"S"; the second QP (shown above) is therefore a shared receive queue
based QP. The second, third and fourth fields are the same as in the
per-peer based QP. The fifth field is the number of outstanding sends
that are allowed at a given time on the QP (32). This provides a
"good enough" mechanism of flow control for some regular communication
patterns.
QPs MUST be specified in ascending receive buffer size order. This
requirement may be removed prior to 1.3 release.
This commit was SVN r15474.
Short description: major changes include -
1. singletons now fork/exec a local daemon to manage their operations.
2. the orte daemon code now resides in libopen-rte
3. daemons no longer use the orte triggering system during startup. Instead, they directly call back to their parent pls component to report ready to operate. A base function to count the callbacks has been provided.
I have modified all the pls components except xcpu and poe (don't understand either well enough to do it). Full functionality has been verified for rsh, SLURM, and TM systems. Compile has been verified for xgrid and gridengine.
This commit was SVN r15390.
than just the PML/BTLs these days. Also clean up the code so that it
handles the situation where not all nodes register information for a given
node (rather than just spinning until that node sends information, like
we do today).
Includes r15234 and r15265 from the /tmp/bwb-modex branch.
This commit was SVN r15310.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r15234
r15265
Changes paffinity interface to use a cpu mask for available/preferred cpus
rather than the current coarse grained paffinity that lets the OS choose
which processor.
Macros for setting and clearing masks are provided.
Solaris and windows changes have not been made. Solaris subdirectory has some
suggested changes - however the relevant man pages for the Solaris 10 APIs
have some ambiguity regarding order in which one create and sets a processor
set. As we did not have access to a solaris 10 machine we could not test to
see the correct way to do the work under solaris.
This commit was SVN r14887.
This commit moves the initalization/finalization of opal_event and opal_progress
to opal_init/finalize. These were previously init/final in ORTE which is an
abstraction violation. After talking about it we concluded that there are no
ordering issues that require these to be init/final in ORTE instead of OPAL.
I ran the IBM test suite against this commit and it didn't turn up any new
failures so I think it is good to go.
Let us know if this causes problems.
This commit was SVN r14773.
wireup. For small clusters or clusters with decent ARP lookup and
connect times, this will have marginal impact. For systems with either
bad ARP lookup times or long connect times, increasing this number
to something much closer to SOMAXCONN (128 on most modern machines) will
result in a faster OOB wireup. Don't set higher than SOMAXCONN or you
can end up with lots of connect() retries and we'll end up slower.
This commit was SVN r14742.
The primary change that underlies all this is in the OOB. Specifically, the problem in the code until now has been that the OOB attempts to resolve an address when we call the "send" to an unknown recipient. The OOB would then wait forever if that recipient never actually started (and hence, never reported back its OOB contact info). In the case of an orted that failed to start, we would correctly detect that the orted hadn't started, but then we would attempt to order all orteds (including the one that failed to start) to die. This would cause the OOB to "hang" the system.
Unfortunately, revising how the OOB resolves addresses introduced a number of additional problems. Specifically, and most troublesome, was the fact that comm_spawn involved the immediate transmission of the rendezvous point from parent-to-child after the child was spawned. The current code used the OOB address resolution as a "barrier" - basically, the parent would attempt to send the info to the child, and then "hold" there until the child's contact info had arrived (meaning the child had started) and the send could be completed.
Note that this also caused comm_spawn to "hang" the entire system if the child never started... The app-failed-to-start helped improve that behavior - this code provides additional relief.
With this change, the OOB will return an ADDRESSEE_UNKNOWN error if you attempt to send to a recipient whose contact info isn't already in the OOB's hash tables. To resolve comm_spawn issues, we also now force the cross-sharing of connection info between parent and child jobs during spawn.
Finally, to aid in setting triggers to the right values, we introduce the "arith" API for the GPR. This function allows you to atomically change the value in a registry location (either divide, multiply, add, or subtract) by the provided operand. It is equivalent to first fetching the value using a "get", then modifying it, and then putting the result back into the registry via a "put".
This commit was SVN r14711.
* Don't need the 2 process case -- we'll send an extra message, but
at very little cost and less code is better.
* Use COMPLETE sends instead of STANDARD sends so that the connection
is fully established before we move on to the next connection. The
previous code was still causing minor connection flooding for huge
numbers of processes.
* mpi_preconnect_all now connects both OOB and MPI layers. There's
also mpi_preconnect_mpi and mpi_preconnect_oob should you want to
be more specific.
* Since we're only using the MCA parameters once at the beginning
of time, no need for global constants. Just do the quick param
lookup right before the parameter is needed. Save some of that
global variable space for the next guy.
Fixes trac:963
This commit was SVN r14553.
The following Trac tickets were found above:
Ticket 963 --> https://svn.open-mpi.org/trac/ompi/ticket/963
There is a binomial algorithm in the code (i.e., the HNP would send to a subset of the orteds, which then relay it on according to the typical log-2 algo), but that has a bug in it so the code won't let you select it even if you tried (and the mca param doesn't show, so you'd *really* have to try).
This also involved a slight change to the oob.xcast API, so propagated that as required.
Note: this has *only* been tested on rsh, SLURM, and Bproc environments (now that it has been transferred to the OMPI trunk, I'll need to re-test it [only done rsh so far]). It should work fine on any environment that uses the ORTE daemons - anywhere else, you are on your own... :-)
Also, correct a mistake where the orte_debug_flag was declared an int, but the mca param was set as a bool. Move the storage for that flag to the orte/runtime/params.c and orte/runtime/params.h files appropriately.
This commit was SVN r14475.
Fix for memory corruption in the restarted process stack. This stemed from
the brute force method we were previously using. This commit fixes this by
using a lighter weight solution focused in the r2 BML instead of above the PML.
This is a more efficient and flexible solution, and it solves the original
problem.
In the process I pulled out the ft_event function in the tcp BTL and r2 BML
into a set of *_ft.[c|h] files just to keep any updates to these code paths
as isolated as possible to make merging easier on everyone.
This commit was SVN r14371.
The following SVN revision numbers were found above:
r2 --> open-mpi/ompi@58fdc18855
The following Trac tickets were found above:
Ticket 977 --> https://svn.open-mpi.org/trac/ompi/ticket/977
This merge adds Checkpoint/Restart support to Open MPI. The initial
frameworks and components support a LAM/MPI-like implementation.
This commit follows the risk assessment presented to the Open MPI core
development group on Feb. 22, 2007.
This commit closes trac:158
More details to follow.
This commit was SVN r14051.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r13912
The following Trac tickets were found above:
Ticket 158 --> https://svn.open-mpi.org/trac/ompi/ticket/158
support SEND_IN_PLACE causes badness because the BTL tries to use the
not-exactly-complete convertor. Don't need it in this situation anyway.
This commit was SVN r13700.
Add new function opal_get_num_processors() that will return the number
of processors on the local host. Does the Right thing in POSIX
environments (to include a special case for OS X), and will shortly do
the Right Thing for Windows (this commit includes a change to
configure, so I wanted to get that in before the US workday -- the
Windows code can some shortly because it won't involve configury
changes).
This commit was SVN r13506.
The following Trac tickets were found above:
Ticket 853 --> https://svn.open-mpi.org/trac/ompi/ticket/853
1. if the user has specified sched_yield, we simply do what we are told
2. if they didn't specify anything, try to get the number of processors on this node. Note that we already now get the number of local procs in our job that are sharing this node - that now comes in through the proc callback and is stored in the ompi_proc_t structures.
3. if we can get the number of processors, compare that to the number of local procs from my job that are sharing my node. If the number of local procs exceeds the number of processors, then set sched_yield to true. If not, then be a hog and set sched_yield to false
4. if we can't get the number of processors, default to conservative behavior and set sched_yield to true.
Note that I have not yet dealt with the need to dynamically adjust this setting as more processes are added via comm_spawn. So far, we are *only* looking within our own job. Given that we have now moved this logic to mpi_init (and away from the orteds), it isn't yet clear to me how a process will be informed about the number of procs in *other* jobs that are also sharing this node.
Something to continue to ponder.
This commit was SVN r13430.
* The real fix, don't leave the OOB in blocking mode during comm_dyn_init(),
as it means no progressing MPI events while the event library is waiting
for TCP stuff to come in.
* Add many comments explaining the reasons for the current ordering
This commit was SVN r13422.
completed successfully, Bad Things(tm) could happen.
* Now we explicitly check orte_initialized (a new global in ORTE
indicating whether we are between orte_init() and orte_finalize()
or not), and if so, react accordingly.
* If ORTE is initialized, use orte_system_info.nodename; otherwise,
use gethostname().
* Add loop protection to ensure that ompi_mpi_abort() is not invoked
multiple times recursively.
This commit was SVN r13354.
know what my local rank is, and therefore set my paffinity ID as
appropriate. Specifically, we're no longer relying on the
special/secret mpi_paffinity_processor MCA parameter that the orted
would set for us.
This allows processor affinity to be used in environments where the
orted is not used (e.g., bproc, and someday in the hopefully not
too-distant future, SLURM).
This commit was SVN r13352.
The following SVN revision numbers were found above:
r13351 --> open-mpi/ompi@a338b7e533
needlessly registered in multiple different places, and none of them
had a good help string. There was also an inconsistent check for
setting both mpi_leave_pinned and mpi_leave_pinned_pipeline (i.e., it
was only in ob1). This commit moves the registration of these params
to one central place (ompi/runtime/ompi_mpi_params.c, with all other
mpi_* MCA params) and uses globals to propagate the values as
relevant. The error check was also moved to the central location to
ensure that we can consistency everywhere.
This commit was SVN r13226.
- Fix some fpritnf's in ompi_mpi_abort() that incorrectly assumed that
we were always being invoked from MPI_ABORT (ompi_mpi_abort() may be
invoked from a bunch of different places)
- Also try to opal_backtrace_print() if opal_bactrace_buffer() is not
supported.
- Print a message in MPI_ABORT if we're aborting.
This commit was SVN r12998.
* Add line about heterogeneous support to ompi_info output
* Print warning and abort if heterogeneous detected and
no heterogeneous support available.
Refs trac:587
This commit was SVN r12943.
The following Trac tickets were found above:
Ticket 587 --> https://svn.open-mpi.org/trac/ompi/ticket/587
CM PML. The req_mtl structure is cast into a mtl_*_request_structure for
each MTL, which is larger than the req_mtl itself. The cast will cause
the *_request to overwrite parts of the heavy requests if the req_mtl
isn't the *LAST* thing on each structure (hence the comment). This was
moved as an optimization at some point, which caused buffer sends to
fail...
Refs trac:669
This commit was SVN r12871.
The following Trac tickets were found above:
Ticket 669 --> https://svn.open-mpi.org/trac/ompi/ticket/669
- we have to be able to attach a string to an error class, not just to an
error code
- according to MPI-2 the attribute MPI_LASTUSEDCODE has to be updated
everytime you add a new code or a new class. Thus, you have to have single
list for both.
Thus, we got rid of the error_class structure. In the error-code structure, we
can distinguish whether we are dealing with an error code or an error class by
looking at the err->code element of the structure. In case its value is
MPI_UNDEFINED, the according entry is a class, else it is an error code. All
predefined error codes have the code and the class field set to the same
value.
The test MPI_Add_error_class1 passes now.
Fixes trac:418
This commit was SVN r12764.
The following Trac tickets were found above:
Ticket 418 --> https://svn.open-mpi.org/trac/ompi/ticket/418
the same time, remove some of the MPI-related options from OPAL:
- provide mechanism to change at runtime whether sched_yield() should
be called when the progress engine is idle
- provide mechanism for changing the rate at which the event engine
is called when there are "no" users of the event engine (ie, when
using MPI but not TCP)
- fix some function names in the progress engine to better match
their intended use (and remove MPI naming scheme)
- remove progress_mpi_enable / progress_mpi_disable because
we can now use the functions to set the sched_yield and
tick rate interfaces
- rename opal_progress_events() to opal_progress_set_event_flag()
because the first really isn't descriptive of what the function
does and I always got confused by it
This commit was SVN r12645.
Accordingly, there are new APIs to the name service to support the ability to get a job's parent, root, immediate children, and all its descendants. In addition, the terminate_job, terminate_orted, and signal_job APIs for the PLS have been modified to accept attributes that define the extent of their actions. For example, doing a "terminate_job" with an attribute of ORTE_NS_INCLUDE_DESCENDANTS will terminate the given jobid AND all jobs that descended from it.
I have tested this capability on a MacBook under rsh, Odin under SLURM, and LANL's Flash (bproc). It worked successfully on non-MPI jobs (both simple and including a spawn), and MPI jobs (again, both simple and with a spawn).
This commit was SVN r12597.
1. Fix the "hang" condition when an application isn't found. It turned out that the ODLS had some difficulty with the process actually not having been started - hence, it never called the waitpid callback. As a result, the "terminated" trigger didn't fire, and so mpirun didn't wake up. With this change, the HNP's errmgr forces the issue by causing the trigger to fire itself when an abort condition occurs.
2. Shift the recording of the pid and the nodename from mpi_init to the orted launcher. This allows programs such as Eclipse PTP to get the pids even for non-MPI applications. In the case of bproc, the pls handles this chore since we don't use orteds in that system.
This commit was SVN r12558.
1. Added reporting points around the xcasts in MPI_Init. Note that these times will include time spent waiting for a trigger to fire, which is why the times between stage gates did NOT include these times initially. The inter-stage-gate times still do NOT include the xcast time - the xcast time is reported separately.
2. Added the process vpid on the MPI_Init timing reports for clarity.
3. Added a report from the xcast function on the HNP that outputs the number of bytes in the message being sent to the processes.
This commit was SVN r12422.
If you want to look at our launch and MPI process startup times, you can do so with two MCA params:
OMPI_MCA_orte_timing: set it to anything non-zero and you will get the launch time for different steps in the job launch procedure. The degree of detail depends on the launch environment. rsh will provide you with the average, min, and max launch time for the daemons. SLURM block launches the daemon, so you only get the time to launch the daemons and the total time to launch the job. Ditto for bproc. TM looks more like rsh. Only those four environments are currently supported - anyone interested in extending this capability to other environs is welcome to do so. In all cases, you also get the time to setup the job for launch.
OMPI_MCA_ompi_timing: set it to anything non-zero and you will get the time for mpi_init to reach the compound registry command, the time to execute that command, the time to go from our stage1 barrier to the stage2 barrier, and the time to go from the stage2 barrier to the end of mpi_init. This will be output for each process, so you'll have to compile any statistics on your own. Note: if someone develops a nice parser to do so, it would be really appreciated if you could/would share!
This commit was SVN r12302.
I have added a new MCA param (hey, you can't have too many!) called OMPI_MCA_orte_timing. If set to anything other than zero, the system will report out critical timing loops. At the moment, this includes three measurements:
1. Time spent going through the RDS->RAS->RMAPS, setting up triggers, etc. prior to calling the actual PLS launch function. This is reported out as time to setup job.
2. Time spent in MPI_Init from start of that function (well, right after opal_init) to the place where we send all of our info the registry. Reported out as time from start to exec_compound_cmd
3. Time actually spent executing the compound cmd. Reported out as time to exec_compound_cmd.
A few additional timing points will be added shortly.
These may eventually be removed or (better) setup with a conditional compile flag.
This commit was SVN r11892.
should die, according to the MPI standard. It's possible that the
ORTE layer may kill additional processes, but that's beyond our
control and seems to be allowed by the standard (ie, it might also
end up killing all the procs in all the jobs covered by the
communicator).
* update the stack trace printing code to use the framework rather
than calling execinfo directly, so that we should be able to get
stack traces on all the platforms we support stack tracing on
(if the user wants stack traces on abort, of course)
This commit was SVN r11753.
Other changes:
1. Remove the old xcpu components as they are not functional.
2. Fix a "bug" in orterun whereby we called dump_aborted_procs even when we normally terminated. There is still some kind of bug in this procedure, however, as we appear to be calling the orterun job_state_callback function every time a process terminates (instead of only once when they have all terminated). I'll continue digging into that one.
This will require an autogen/configure, I'm afraid.
This commit was SVN r11228.
r10877:
add warm up connection option.. of course this only warms up the first
eager btl but this should be adequate for now..
r10881:
Consulted with Galen and did a few things:
- Fix the algorithm to actually make the connections that we want
- Rename the MCA param to mpi_preconnect_all
- Cleanup the code a bit:
- move the logic to a separate .c file
- check return codes properly
This commit was SVN r11114.
The following SVN revisions from the original message are invalid or
inconsistent and therefore were not cross-referenced:
r10877
r10877
r10881
r10881
did pre-libevent update. The problem is that the behavior of
OPAL_EVLOOP_ONCE was changed by the OMPI team, which them broke things
during the update, so it had to be reverted to the old meaning of
loop until one event occurs. OPAL_EVLOOP_ONELOOP will go through the
event loop once (like EVLOOP_NONBLOCK) but will pause in the event
library for a bit (like EVLOOP_ONCE).
fixes trac:234
This commit was SVN r11081.
The following Trac tickets were found above:
Ticket 234 --> https://svn.open-mpi.org/trac/ompi/ticket/234
users mailing list:
http://www.open-mpi.org/community/lists/users/2006/07/1680.php
Warning: this log message is not for the weak. Read at your own
risk.
The problem was that we had several variables in Fortran common blocks
of various types, but their C counterparts were all of a type
equivalent to a fortran double complex. This didn't seem to matter
for the compilers that we tested, but we never tested static builds
(which is where this problem seems to occur, at least with the Intel
compiler: the linker compilains that the variable in the common block
in the user's .o file was of one size/alignment but the one in the C
library was a different size/alignment).
So this patch fixes the sizes/types of the Fortran common block
variables and their corresponding C instantiations to be of the same
sizes/types.
But wait, there's more.
We recently introduced a fix for the OSX linker where some C versions
of the fortran common block variables (e.g.,
_ompi_fortran_status_ignore) were not being found when linking
ompi_info (!). Further research shows that the code path for
ompi_info to require ompi_fortran_status_ignore is, unfortunately,
necessary (a quirk of how various components pull in different
portions of the code base -- nothing in ompi_info itself requires
fortran or MPI knowledge, of course).
Hence, the real problem was that there was no code path from ompi_info
to the portion of the code base where the C globals corresponding to
the Fortran common block variables were instantiated. This is because
the OSX linker does not automatically pull in .o files that only
contain unintialized global variables; the OSX linker typically only
pulls in a .o file from a library if it either has a function that is
used or have a global variable that is initialized (that's the short
version; lots of details and corner cases omitted). Hence, we changed
the global C variables corresponding to the fortran common blocks to
be initialized, thereby causing the OSX linker to pull them in
automatically -- problem solved. At the same time, we moved the
constants to another .c file with a function, just for good measure.
However, this didn't really solve the problem:
1. The function in the file with the C versions of the fortran common
block variables (ompi/mpi/f77/test_constants_f.c) did not have a
code path that was reachable from ompi_info, so the only reason
that the constants were found (on OSX) was because they were
initialized in the global scope (i.e., causing the OSX compiler to
pull in that .o file).
2. Initializing these variable in the global scope causes problems for
some linkers where -- once all the size/type problems mentioned
above were fixed -- the alignments of fortran common blocks and C
global variables do not match (even though the types of the Fortran
and C variables match -- wow!). Hence, initializing the C
variables would not necessarily match the alignment of what Fortran
expected, and the linker would issue a warning (i.e., the alignment
warnings referenced in the original post).
The solution is two-fold:
1. Move the Fortran variables from test_constants_f.c to
ompi/mpi/runtime/ompi_mpi_init.c where there are other global
constants that *are* initialized (that had nothing to do with
fortran, so the alignment issues described above are not a factor),
and therefore all linkers (including the OSX linker) will pull in
this .o file and find all the symbols that it needs.
2. Do not initialize the C variables corresponding to the Fortran
common blocks in the global scope. Indeed, never initialize them
at all (because we never need their *values* - we only check for
their *locations*). Since nothing is ever written to these
variables (particularly in the global scope), the linker does not
see any alignment differences during initialization, but does make
both the C and Fortran variables have the same addresses (this
method has been working in LAM/MPI for over a decade).
There were some comments here in the OMPI code base and in the LAM
code base that stated/implied that C variables corresponding to
Fortran common blocks had to have the same alignment as the Fortran
common blocks (i.e., 16). There were attempts in both code bases to
ensure that this was true. However, the attempts were wrong (in both
code bases), and I have now read enough Fortran compiler documentation
to convince myself that matching alignments is not required (indeed,
it's beyond our control). As long as C variables corresponding to
Fortran common blocks are not initialized in the global scope, the
linker will "figure it out" and adjust the alignment to whatever is
required (i.e., the greater of the alignments). Specifically (to
counter comments that no longer exist in the OMPI code base but still
exist in the LAM code base):
- there is no need to make attempts to specially align C variables
corresponding to Fortran common blocks
- the types and sizes of C variables corresponding to Fortran common
blocks should match, but do not need to be on any particular
alignment
Finally, as a side effect of this effort, I found a bunch of
inconsistencies with the intent of status/array_of_statuses
parameters. For all the functions that I modified they should be
"out" (not inout).
This commit was SVN r11057.
(which is currently the default, although we may argue over this later
:-) ), a new field in the ompi_proc_t named proc_hostname will have
the string hostname of that peer. If 0, this field will be NULL.
This allows for printing nicer error messages in environments where
peer hostnames are not otherwise easily obtainable, such as the mvapi
BTL (requested by Sandia, who has both a *huge* number of nodes and
6GB of RAM per node, so they don't care about the extra memory usage
;-) ).
This commit was SVN r9902.
- My original patch stands: MPI_FINALIZE directly invokes the
attribute callbacks on MPI_COMM_SELF
- We added some user-level checks to ensure that they don't call
MPI_FINALIZE twice (this isn't really required, but it will prevent
whacky segv's -- they'll at least get a nice error message)
- Removed the attribute callbacks on MPI_COMM_SELF from
ompi_mpi_comm_finalize (i.e., we just moved them from
ompi_mpi_comm_finalize to ompi_mpi_finalize -- we just moved this
process up earlier in the MPI_FINALIZE sequence of events)
- Because there were so many conversations about this, here's the
rationale:
- MPI-2:4.8 says that we have to MPI_COMM_FREE MPI_COMM_SELF so that
the attribute callbacks are invoked.
- After considerable discussion, we came to the conclusion that
FREE'ing COMM_SELF is not the issue -- calling the callbacks is
the issue.
- So it is sufficent for MPI_FINALIZE to directly invoke these
attribute callbacks
- The attribute callbacks are *not* invoked on other communicators
because said communicators are not MPI_COMM_FREE'ed
This commit was SVN r9628.
MPI_ABORT. From the ompi_info output:
MCA mpi: parameter "mpi_abort_delay" (current value: "0")
If nonzero, print out an identifying message when
MPI_ABORT is invoked (hostname, PID of the process
that called MPI_ABORT) and delay for that many seconds
before exiting (a negative delay value means to never
abort). This allows attaching of a debugger before
quitting the job.
MCA mpi: parameter "mpi_abort_print_stack" (current value: "0")
If nonzero, print out a stack trace when MPI_ABORT is
invoked
This commit was SVN r9487.
- move files out of toplevel include/ and etc/, moving it into the
sub-projects
- rather than including config headers with <project>/include,
have them as <project>
- require all headers to be included with a project prefix, with
the exception of the config headers ({opal,orte,ompi}_config.h
mpi.h, and mpif.h)
This commit was SVN r8985.
complete, but stable enough that it will have no impact on general development,
so into the trunk it goes. Changes in this commit include:
- Remove the --with option for disabling MPI-2 onesided support. It
complicated code, and has no real reason for existing
- add a framework osc (OneSided Communication) for encapsulating
all the MPI-2 onesided functionality
- Modify the MPI interface functions for the MPI-2 onesided chapter
to properly call the underlying framework and do the required
error checking
- Created an osc component pt2pt, which is layered over the BML/BTL
for communication (although it also uses the PML for long message
transfers). Currently, all support functions, all communication
functions (Put, Get, Accumulate), and the Fence synchronization
function are implemented. The PWSC active synchronization
functions and Lock/Unlock passive synchronization functions are
still not implemented
This commit was SVN r8836.
