GitLab

Originally the size of IMMED data in RMA packets is 16 bytes
which makes the size of CH3 packet be 56 bytes. Here we reduce
the size of IMMED data in RMA packets to 8 bytes, so that the
size of CH3 packet is reduced to 48 bytes, the same with
mpich-3.1.4 (the old RMA infrastructure).
Signed-off-by: Min Si <msi@il.is.s.u-tokyo.ac.jp>
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

'stream_offset' is used to specify the starting position
(on target window) of the current streaming unit in ACC-like
operations. It is originally put in the RMA packet struct,
which potentially increases the size of CH3 packet size.

In this patch, we move 'stream_offset' out of the RMA
packet as follows: 1. when target data is basic datatype,
we use 'stream_offset' and the starting address for the entire
operation to calculate the starting address for current
streaming unit, and rewrite 'addr' in RMA packet with that
value; 2. when target data is derived datatype, we cannot do
the same thing as basic datatype because the target needs to
know both the starting address for the entire operation and
the starting address for the current streaming unit. Therefore,
we send 'stream_offset' separately to the target side.
Signed-off-by: Min Si <msi@il.is.s.u-tokyo.ac.jp>
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

An assert protecting from a non-null request was happening too late in pkt_COOKIE_handler from
mpid_nem_lmt.c. This patch moves it to an earlier location so that it's checked before it's first
used.

Reported by Dmitry Polyakov.

Update OFI netmod to match portals4 netmod anysource_matched semantics.

This reverts commit 17e31e59.

This fix, along with a pending patch to the Portal4 reference implementation,
should make anysource_matched a more reliable operation for multithreaded apps. We
were seeing a race condition where an ME would unlink successfully, but an
event matching it would still arrive in the queue. CH3 can now reliably search
the netmod queue for matched MPI_ANY_SOURCE requests.

The reason that we no longer assert that an MPI_ANY_SOURCE request was removed
from the CH3 queue is that FDP (find and dequeue posted) operations will remove
the request from the queue, if it is known to be already matched by the netmod,
even if it has not yet completed.

Fixes #2199
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

We now auto-generate mpi_c_interface_types.f90

Fixes #2196
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

We now auto-generate mpi_f08_compile_constants.f90

Fixes #2196
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

The old one defines MPI_MESSAGE_NULL as MPI_REQUEST_NULL, which misleads
F08 buildiface since types of MPI_REQUEST and MPI_MESSAGE are different.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Module files are now only produced under use_mpi_f08 subdirectory.
Also, it now auto-detects cases of names and extensions of module files,
which are compiler-dependant.

Fixes #2163
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Since *.F90 implies the file need preprocess, e.g., by a C preprocessor, which is not
required by the current f08 binding, and will lead to unnecessary compliation
rules with automake.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Ticket #2243 was resolved as a duplicate of ticket #2183.
Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

These tests seem to use a lot of memory per process, causing us to hit
swap space when running with too many processes.  Reducing it to two
processes, allows this test to run on more machines.
Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

We increased the number of cases the bcast test was running in
[e01a20b6].  This is causing it to timeout on some platforms, where
the test now seems to take close to 3 minutes.  This increased timeout
should be sufficient on those platforms.

No reviewer.

 * Rename MPIR_CVAR_DUMP_PROVIDERS to MPIR_CVAR_OFI_DUMP_PROVIDERS
 * Add MPIR_CVAR_OFI_USE_PROVIDER, which takes a string to desired
   provider name

This commit includes multiple fixes:
 - Fixes for MPI_IN_PLACE checking. cudaGetPointerAttributes returns
   true on MPI_IN_PLACE which causes issues. Now we check on MPI_IN_PLACE
   before passing pointer to cuda.
 - Enabling PAMID geometries (in order to get to PAMID collectives) when
   MP_CUDA_AWARE=yes. This allows for intercepting CUDA buffer.
 - Disabling FCA when MP_CUDA_AWARE=yes if user enables FCA.
 - Copying user recv buffer into temp recv host buffer before collective
   starts, especially in MPI_IN_PLACE cases.

(ibm) D203255
Signed-off-by: Tsai-Yang (Alan) Jea <tjea@us.ibm.com>

The linker on Darwin does not allow common symbols, thus libtool adds
the -fno-common option by default for shared libraries. However, the
common symbols defined in different shared libraries and object files
still can not be treated as the same symbol.
For example:
with gfortran, the same common block in the shared libraries and the
object files will have different memory locations separately;
with ifort, the same common block in different shared libraries will get
the same memory location but still get a different location in the
object file.

The -Wl,-commons,use_dylibs option asks linker to check dylibs for
definitions and use them to replace tentative definitions(commons) from
object files, thus it solves the issue of the common symbol mismatch
between the object file and the dylibs (i.e., by setting the address of
a common symbol to the place located in the first dylib that is linked
with the object file and contains this symbol). It needs to be added
only in the linking stage for the final executable file.

The -flat-namespace option allows linker to unify the same common
symbols in different dylibs. It needs to be added in linking stage for
both the shared library and the final executable file.
(see man ld for their definition)

Although gfortran works fine by only adding -flat-namespace, and ifort
works by only adding -Wl,-commons,use_dylibs, we should add both options
here as a generic solution to make sure everything safe.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

When we skip non-comm-world communicators, we need to make sure to
skip them.

No reviewer.

Changes the value of various static limits in the Portals4 netmod, based
on experimentation results and suggestions from collaborators.

1. Bump most ni_limits from 32K to 64K. These limits relate closely to
   queue depth. We can reasonably expect to support a queue depth
   of 64K.

2. Limit issued origin events to 500. This translates to sending ~250
   operations to Portals at a time, which over IB is roughly the
   saturation point. TODO: turn this into a CVAR.

3. Limit per target issued operations to 50. This will give the target a
   better chance to process events without being overwhelmed by a single
   process. TODO: turn this into a CVAR, also.

4. Allocate more buffer space for incoming control messages. Observed
   results, especially with larger messages, showed that more buffer space
   cuts down on flow-control events.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

The 2 commits being reverted introduced a "safe" PtlMEAppend function
that would call MPID_nem_ptl_poll to process some events in case there
was no space to append the match list entry. However the poll function
is not reentrant safe, which could lead to ordering problems.

The increased list entry limit from [c6c0d6f6

] should prevent PTL_NO_SPACE
errors from happening, except in the extreme case. If we still find we are
hitting this error, a proper fix can be done in the Rportals layer.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

Pavan Balaji authored Apr 09, 2015 and Kenneth Raffenetti committed Apr 10, 2015

Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Pavan Balaji authored Apr 08, 2015 and Kenneth Raffenetti committed Apr 10, 2015

The current number of combinations we are checking are too many,
causing the test to take too long on some platforms.  This patch
simplifies the test, so we build two versions of the test.  In the
first version, we run only on COMM_WORLD but go through all datatypes.
In the second version, we run on all communicators, but go through
only a small subset of datatypes.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Pavan Balaji authored Apr 08, 2015 and Kenneth Raffenetti committed Apr 10, 2015

1. Renamed bcast2 to bcast.

2. White-space cleanup for bcast.c
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

Pavan Balaji authored Apr 08, 2015 and Kenneth Raffenetti committed Apr 10, 2015

This test is exactly the same as bcast2.  Originally these two tests
were different, but over time they have become essentially the same.
There's no point testing the same thing twice.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

The datatype size was checked outside the appropriate branches in a couple of places
Signed-off-by: Xin Zhao <xinzhao3@illinois.edu>

This reverts commit b47d95f7.

The previous design for MPICH control messages utilized a small set
of "use once" buffers that could be quickly exhausted. The new approach
processes all control messages via an unexpected queue. Benefits are
a larger incoming message capacity, leading to less flow-control events.
Signed-off-by: Antonio J. Pena <apenya@mcs.anl.gov>

Also change individual author to organization names.
Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>

Signed-off-by: Huiwei Lu <huiweilu@mcs.anl.gov>