mirror of
https://github.com/openembedded/meta-openembedded.git
synced 2026-06-14 17:59:59 +00:00
Liu Yiding
40d15fc4fa
Version 2.12.1
--------------
* Add hwloc-calc's --default-nodes option to hwloc-bind and hwloc-info.
* Improve the --best-memattr "default" fallback, try to use "default"
memory nodes, and add verbose messages and warnings if some
performance info are incomplete or missing.
Thanks to Antoine Morvan for the report.
* Fix CPU and memory binding on different locations,
thanks to Antoine Morvan for the report.
* Add HWLOC_LOCAL_NUMANODE_FLAG_INTERSECT_LOCALITY and enable it by
default in hwloc-calc --local-memory for finding local NUMA nodes
that do not exactly match input locations.
Thanks to Antoine Morvan for the report.
* Fix a possible crash in the x86 backend when Qemu is configured to
expose multicore/thread CPUs that are actually single-core/thread.
Thanks to Georg Pfuetzenreuter.
Version 2.12.0
--------------
* Add hwloc_topology_get_default_nodeset() for the set of default
NUMA nodes.
- hwloc-calc now has --default-nodes option.
* Rework oneAPI LevelZero support to use zesInit() and avoid the need
to set ZES_ENABLE_SYSMAN=1 in the environment.
- zesDriverGetDeviceByUuidExp() is now required in the L0 runtime.
- ZES/Sysman variants were added in hwloc/levelzero.h to specifically
handle ZES/Sysman device handles.
* Fix the locality of AMD GPU partitions, thanks to Edgar Leon for
reporting and debugging the issue.
* Better detect Cray Slingshot NICs, thanks to Edgar Leon.
* Add support for Die objects and Module groups on Windows.
* Only filter-out Dies that are identical to their Packages
when it applies to all Dies.
* Improve hwloc-calc to handle CPU-less NUMA nodes or platforms with
heterogeneous memory without requiring --nodeset-output.
* hwloc-calc now accepts counting/listing cpukinds and memory tiers
with -N and -I cpukind/memorytier.
* The systemd-dbus-api output of hwloc-calc has changed, and
--nodeset-output-format was added, to support NUMA node outputs.
Thanks to Pierre Neyron.
* Update NVLink bandwidth and CUDA capabilities up to NVIDIA Blackwell.
* Fix some NUMA syscalls on Linux for platforms with old libc headers.
* Some minor fixes in distances.
Version 2.11.2
--------------
* Add missing CPU info attrs on aarch64 on Linux.
* Use ACPI CPPC on Linux to get better information about cpukinds,
at least on AMD CPUs.
* Fix crash when manipulating cpukinds after topology
duplication, thanks to Hadrien Grasland for the report.
* Fix missing input target checks in memattr functions,
thanks to Hadrien Grasland for the report.
* Fix a memory leak when ignoring NUMA distances on FreeBSD.
* Fix build failure on old Linux distributions without accessat().
* Fix non-Windows importing of XML topologies and CPUID dumps exported
on Windows.
* hwloc-calc --cpuset-output-format systemd-dbus-api now allows
to generate AllowedCPUs information for systemd slices.
See the hwloc-calc manpage for examples. Thanks to Pierre Neyron.
* Some fixes in manpage EXAMPLES and split them into subsections.
Version 2.11.1
--------------
* Fix bash completions, thanks Tavis Rudd.
Version 2.11.0
--------------
* API
+ Add HWLOC_MEMBIND_WEIGHTED_INTERLEAVE memory binding policy on
Linux 6.9+. Thanks to Honggyu Kim for the patch.
- weighted_interleave_membind is added to membind support bits.
- The "weighted" policy is added to the hwloc-bind tool.
+ Add hwloc_obj_set_subtype(). Thanks to Hadrien Grasland for the report.
* GPU support
+ Don't hide the GPU NUMA node on NVIDIA Grace Hopper.
+ Get Intel GPU OpenCL device locality.
+ Add bandwidths between subdevices in the LevelZero XeLinkBandwidth
matrix.
+ Fix PCI Gen4+ link speed of NVIDIA GPU obtained from NVML,
thanks to Akram Sbaih for the report.
* Windows support
+ Fix Windows support when UNICODE is enabled, several hwloc features
were missing, thanks to Martin for the report.
+ Fix the enabling of CUDA in Windows CMake build,
Thanks to Moritz Kreutzer for the patch.
+ Fix CUDA/OpenCL test source path in Windows CMake.
* Tools
+ Option --best-memattr may now return multiple nodes. Additional
configuration flags may be given to tweak its behavior.
+ hwloc-info has a new --get-attr option to get a single attribute.
+ hwloc-info now supports "levels", "support" and "topology"
special keywords for backward compatibility for hwloc 3.0.
+ The --taskset command-line option is superseded by the new
--cpuset-output-format which also allows to export as list.
+ hwloc-calc may now import bitmasks described as a list of bits
with the new "--cpuset-input-format list".
* Misc
+ The MemoryTiersNr info attribute in the root object now says how many
memory tiers were built. Thanks to Antoine Morvan for the report.
+ Fix the management of infinite cpusets in the bitmap printf/sscanf
API as well as in command-line tools.
+ Add section "Compiling software on top of hwloc's C API" in the
documentation with examples for GNU Make and CMake,
thanks to Florent Pruvost for the help.
Version 2.10.0
--------------
* Heterogeneous Memory core improvements
+ Better heuristics to identify the subtype of memory such as HBM,
DRAM, NVM, CXL-DRAM, etc.
+ Build memory tiers, i.e. sets of NUMA nodes with the same subtype
and similar performance.
- NUMA node tier ranks are exposed in the new MemoryTier info
attribute (starts from 0 for highest bandwidth tier)..
+ See the new Heterogeneous Memory section in the documentation.
* API
+ Add hwloc_topology_free_group_object() to discard a Group created
by hwloc_topology_alloc_group_object().
* Linux backend
+ Fix cpukinds on NVIDIA Grace to report identical cores even if they
actually have very small frequency differences.
Thanks to John C. Linford for the report.
+ Add CXLDevice attributes to CXL DAX objects and NUMA nodes to show
which PCI device implements which window.
+ Ignore buggy memory-side caches and memory attributes when fake NUMA
emulation is enabled on the Linux kernel command-line.
+ Add more info attributes in MemoryModule Misc objects,
thanks to Zubiao Xiong for the patch.
+ Get CPUModel and CPUFamily info attributes on LoongArch platforms.
* x86 backend
+ Add support for new AMD CPUID leaf 0x80000026 for better detection
of Core Complex and Die on Zen4 processors.
+ Improve Zhaoxin CPU topology detection.
* Tools
+ Input locations and many command-line options (e.g. hwloc-calc -I -N -H,
lstopo --only) now accept filters such as "NUMA[HBM]" so that only
objects are that type and subtype are considered.
- NUMA[tier=1] is also accepted for selecting NUMA nodes depending
on their MemoryTier info attribute.
+ Add --object-output to hwloc-calc to report the type as a prefix to
object indexes, e.g. Core:2 instead of 2 in the output of -I.
+ hwloc-info --ancestor and --descendants now accepts kinds of objects
instead of single types.
- The new --first option only shows the first matching object.
+ Add --children-of-pid to hwloc-ps to show a hierarchy of processes.
Thanks to Antoine Morvan for the suggestion.
+ Add --misc-from to lstopo to add Misc objects described in a file.
- To be combined with the new hwloc-ps --lstopo-misc for a customizable
lstopo --top replacement.
* Misc
+ lstopo may now configure the layout of memory object placed above,
for instance with --children-order memory:above:vert.
+ Fix XML import from memory or stdin when using libxml2 2.12.
+ Fix installation failures when configuring with --target,
thanks to Clement Foyer for the patch.
+ Fix support for 128bit pointer architectures.
+ Remove Netloc.
Changelog: https://github.com/open-mpi/hwloc/blob/v2.12/NEWS
Signed-off-by: Liu Yiding <liuyd.fnst@fujitsu.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
meta-oe
This layer depends on:
URI: git://github.com/openembedded/openembedded-core.git branch: master
luajit recipe requires host compiler to be able to generate 32bit code when target is 32bit e.g. arm, so ensure that $CC -m32 is functional on build host, if building this recipe, needed packages to fullfit this might have different names on different host distributions e.g. on archlinux based distributions install prerequisites like below
pacman -S lib32-gcc-libs lib32-glibc
Ubuntu sudo apt-get install gcc-multilib linux-libc-dev:i386
Send pull requests to openembedded-devel@lists.openembedded.org with '[meta-oe]' in the subject'
When sending single patches, please use something like: 'git send-email -M -1 --to openembedded-devel@lists.openembedded.org --subject-prefix="meta-oe][PATCH"'
You are encouraged to fork the mirror on GitHub https://github.com/openembedded/meta-openembedded to share your patches, this is preferred for patch sets consisting of more than one patch.
Note, it is discouraged to send patches via GitHub pull request system. Such patches get less attention from developers and can be mishandled or not reviewed properly. Please use emails instead. For exemple, you can use 'git request-pull' to generate an email referencing your git repository.
Other services like GitLab, repo.or.cz or self-hosted setups are of course accepted as well, 'git fetch ' works the same on all of them. We recommend GitHub because it is free, easy to use, has been proven to be reliable and has a really good web GUI.
layer maintainer: Khem Raj raj.khem@gmail.com