a19cf02c2b
Bring in a new version of PLPA (v1.2b2) with some new capabilities for offline processors and mapping of the Nth processor/socket/core to its corresponding Linux processor/socket/core ID. (Sorry for the configure change in the middle of the day, folks -- I need it to be able to continue to integrate paffinity changes for #1435...) This commit was SVN r19282. The following Trac tickets were found above: Ticket 1435 --> https://svn.open-mpi.org/trac/ompi/ticket/1435
116 строки
3.1 KiB
Plaintext
116 строки
3.1 KiB
Plaintext
Copyright (c) 2004-2006 The Trustees of Indiana University and Indiana
|
|
University Research and Technology
|
|
Corporation. All rights reserved.
|
|
Copyright (c) 2004-2005 The Regents of the University of California.
|
|
All rights reserved.
|
|
Copyright (c) 2006-2008 Cisco Systems, Inc. All rights reserved.
|
|
$COPYRIGHT$
|
|
|
|
Additional copyrights may follow
|
|
|
|
$HEADER$
|
|
|
|
===========================================================================
|
|
|
|
This file contains the main features as well as overviews of specific
|
|
bug fixes (and other actions) for each version of PLPA since
|
|
version 1.0.
|
|
|
|
1.2
|
|
---
|
|
|
|
- Allow mapping from simple/contiguous processor, socket, and core
|
|
numbers to the corresponding back-end Linux ID (and clean up
|
|
documentation references between "X number" and "X ID", where X is
|
|
one of processor, socket, core).
|
|
- Support CPU hot plugging.
|
|
- Add explicit control over the PLPA topology information cache.
|
|
|
|
|
|
1.1.1
|
|
-----
|
|
|
|
- Ensure that --enable-included also disables executables by default.
|
|
Thanks to Brian Curtis for identifying the problem and supplying a
|
|
patch.
|
|
- Add ENABLE_EXECUTABLES m4 macro to allow behavior in m4
|
|
configuration analogous to "--enable-included --enable-executables".
|
|
|
|
|
|
1.1
|
|
---
|
|
|
|
- New command: plpa-taskset. It is intended as a drop-in replacement
|
|
for the "taskset" command, except it also understands sockets and
|
|
cores. See "plpa-taskset --help" for more details
|
|
- Renamed "plpa_info" to "plpa-info".
|
|
- Added PLPA_{MAJOR|MINOR|RELEASE}_VERSION integer macros in plpa.h.
|
|
This release, they have the values of 1, 1, and 0, respectively.
|
|
- Add new API functions to map from (socket,core) back and forth from
|
|
the Linux virtual processor ID. Thanks to AMD for the initial code
|
|
contribution that made this possible. See the documentation for
|
|
plpa_map_to_processor_id() as a starting point for more
|
|
information.
|
|
- Various fixes to "included" mode.
|
|
|
|
|
|
1.0.5
|
|
-----
|
|
|
|
- Fix an issue where the PLPA get/set affinity functions would only
|
|
operate on the current process; the PID argument to these functions
|
|
was effectively ignored.
|
|
|
|
|
|
1.0.4
|
|
-----
|
|
|
|
- Fix some 64 bit issues. Thanks to David Creasy for reporting the
|
|
problems.
|
|
- Fix plpa.h to be C++-friendly. Thanks to Emmanuel Paris for
|
|
pointing out this problem.
|
|
|
|
|
|
1.0.3
|
|
-----
|
|
|
|
- Resolve some sizing / units ambiguities that mistakenly did not make
|
|
it into 1.0.1. Thanks to Bert Wesarg for pointing these problems out.
|
|
|
|
|
|
1.0.2
|
|
-----
|
|
|
|
- Ensure that plpa_sched_[set|get]affinity() returns 0 upon success.
|
|
Thanks to David Creasy for bringing this to our attention.
|
|
|
|
|
|
1.0.1
|
|
-----
|
|
|
|
- Specify that cpusetsize should be in units of bytes and add some
|
|
missing "* 8"'s in plpa_dispatch.c.
|
|
|
|
|
|
1.0
|
|
---
|
|
|
|
- Initial release.
|
|
|
|
|
|
0.9a2
|
|
-----
|
|
|
|
- Change the back-end type of the plpa_bitmask_t to unsigned long
|
|
instead of unsigned char to avoid an endian problem.
|
|
- No longer attempt to set the affinity in plpa_api_probe() to avoid a
|
|
possible race condition (setting it after getting it).
|
|
- Fix PLPA_CPU_ZERO macro. Thanks to Bogdan Costescu for pointing
|
|
this out.
|
|
|
|
|
|
0.9a1
|
|
-----
|
|
|
|
Initial public release.
|