Merge branch 'dev'

2024-10-23 12:52:16 +07:00 · 2024-10-23 12:52:16 +07:00 · 1f75d198d8
commit 1f75d198d8
parent ef14d55aa5 5cf2422766
39 changed files with 917 additions and 352 deletions
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@ -1,3 +1,10 @@
+# v6.22.1
+- [#3531](https://github.com/xmrig/xmrig/pull/3531) Always reset nonce on RandomX dataset change.
+- [#3534](https://github.com/xmrig/xmrig/pull/3534) Fixed threads auto-config on Zen5.
+- [#3535](https://github.com/xmrig/xmrig/pull/3535) RandomX: tweaks for Zen5.
+- [#3539](https://github.com/xmrig/xmrig/pull/3539) Added Zen5 to `randomx_boost.sh`.
+- [#3540](https://github.com/xmrig/xmrig/pull/3540) Detect AMD engineering samples in `randomx_boost.sh`.
+
 # v6.22.0
 - [#2411](https://github.com/xmrig/xmrig/pull/2411) Added support for [Yada](https://yadacoin.io/) (`rx/yada` algorithm).
 - [#3492](https://github.com/xmrig/xmrig/pull/3492) Fixed `--background` option on Unix systems.
--- a/scripts/build.hwloc.sh
+++ b/scripts/build.hwloc.sh
@ -1,8 +1,8 @@
 #!/bin/sh -e

 HWLOC_VERSION_MAJOR="2"
-HWLOC_VERSION_MINOR="10"
-HWLOC_VERSION_PATCH="0"
+HWLOC_VERSION_MINOR="11"
+HWLOC_VERSION_PATCH="2"

 HWLOC_VERSION="${HWLOC_VERSION_MAJOR}.${HWLOC_VERSION_MINOR}.${HWLOC_VERSION_PATCH}"

--- a/scripts/build.openssl3.sh
+++ b/scripts/build.openssl3.sh
@ -1,6 +1,6 @@
 #!/bin/sh -e

-OPENSSL_VERSION="3.0.14"
+OPENSSL_VERSION="3.0.15"

 mkdir -p deps
 mkdir -p deps/include
--- a/scripts/build.uv.sh
+++ b/scripts/build.uv.sh
@ -1,6 +1,6 @@
 #!/bin/sh -e

-UV_VERSION="1.48.0"
+UV_VERSION="1.49.2"

 mkdir -p deps
 mkdir -p deps/include
--- a/scripts/randomx_boost.sh
+++ b/scripts/randomx_boost.sh
@ -8,7 +8,7 @@ else
 	modprobe msr allow_writes=on
 fi

-if grep -E 'AMD Ryzen|AMD EPYC' /proc/cpuinfo > /dev/null;
+if grep -E 'AMD Ryzen|AMD EPYC|AuthenticAMD' /proc/cpuinfo > /dev/null;
 	then
 	if grep "cpu family[[:space:]]\{1,\}:[[:space:]]25" /proc/cpuinfo > /dev/null;
 		then
@ -28,6 +28,14 @@ if grep -E 'AMD Ryzen|AMD EPYC' /proc/cpuinfo > /dev/null;
 					wrmsr -a 0xc001102b 0x2000cc10
 					echo "MSR register values for Zen3 applied"
 				fi
+		elif grep "cpu family[[:space:]]\{1,\}:[[:space:]]26" /proc/cpuinfo > /dev/null;
+			then
+				echo "Detected Zen5 CPU"
+				wrmsr -a 0xc0011020 0x4400000000000
+				wrmsr -a 0xc0011021 0x4000000000040
+				wrmsr -a 0xc0011022 0x8680000401570000
+				wrmsr -a 0xc001102b 0x2040cc10
+				echo "MSR register values for Zen5 applied"
 		else
 			echo "Detected Zen1/Zen2 CPU"
 			wrmsr -a 0xc0011020 0
--- a/src/3rdparty/hwloc/NEWS
+++ b/src/3rdparty/hwloc/NEWS
@ -1,5 +1,5 @@
 Copyright © 2009 CNRS
-Copyright © 2009-2023 Inria.  All rights reserved.
+Copyright © 2009-2024 Inria.  All rights reserved.
 Copyright © 2009-2013 Université Bordeaux
 Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
 Copyright © 2020 Hewlett Packard Enterprise.  All rights reserved.
@ -17,6 +17,71 @@ bug fixes (and other actions) for each version of hwloc since version
 0.9.


+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
--- a/src/3rdparty/hwloc/README
+++ b/src/3rdparty/hwloc/README
@ -418,14 +418,8 @@ return 0;
 }

 hwloc provides a pkg-config executable to obtain relevant compiler and linker
-flags. For example, it can be used thusly to compile applications that utilize
-the hwloc library (assuming GNU Make):
-
-CFLAGS += $(shell pkg-config --cflags hwloc)
-LDLIBS += $(shell pkg-config --libs hwloc)
-
-hwloc-hello: hwloc-hello.c
-     $(CC) hwloc-hello.c $(CFLAGS) -o hwloc-hello $(LDLIBS)
+flags. See Compiling software on top of hwloc's C API for details on building
+program on top of hwloc's API using GNU Make or CMake.

 On a machine 2 processor packages -- each package of which has two processing
 cores -- the output from running hwloc-hello could be something like the
--- a/src/3rdparty/hwloc/VERSION
+++ b/src/3rdparty/hwloc/VERSION
@ -8,8 +8,8 @@
 # Please update HWLOC_VERSION* in contrib/windows/hwloc_config.h too.

 major=2
-minor=10
-release=0
+minor=11
+release=2

 # greek is used for alpha or beta release tags.  If it is non-empty,
 # it will be appended to the version number.  It does not have to be
@ -22,7 +22,7 @@ greek=

 # The date when this release was created

-date="Dec 04, 2023"
+date="Sep 26, 2024"

 # If snapshot=1, then use the value from snapshot_version as the
 # entire hwloc version (i.e., ignore major, minor, release, and
@ -41,6 +41,6 @@ snapshot_version=${major}.${minor}.${release}${greek}-git
 # 2. Version numbers are described in the Libtool current:revision:age
 # format.

-libhwloc_so_version=22:0:7
+libhwloc_so_version=23:1:8

 # Please also update the <TargetName> lines in contrib/windows/libhwloc.vcxproj
--- a/src/3rdparty/hwloc/include/hwloc.h
+++ b/src/3rdparty/hwloc/include/hwloc.h
--- a/src/3rdparty/hwloc/include/hwloc/autogen/config.h
+++ b/src/3rdparty/hwloc/include/hwloc/autogen/config.h
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2023 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2012 Université Bordeaux
 * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -11,10 +11,10 @@
 #ifndef HWLOC_CONFIG_H
 #define HWLOC_CONFIG_H

-#define HWLOC_VERSION "2.10.0"
+#define HWLOC_VERSION "2.11.2"
 #define HWLOC_VERSION_MAJOR 2
-#define HWLOC_VERSION_MINOR 10
-#define HWLOC_VERSION_RELEASE 0
+#define HWLOC_VERSION_MINOR 11
+#define HWLOC_VERSION_RELEASE 2
 #define HWLOC_VERSION_GREEK ""

 #define __hwloc_restrict
--- a/src/3rdparty/hwloc/include/hwloc/distances.h
+++ b/src/3rdparty/hwloc/include/hwloc/distances.h
@ -1,5 +1,5 @@
 /*
- * Copyright © 2010-2023 Inria.  All rights reserved.
+ * Copyright © 2010-2024 Inria.  All rights reserved.
 * See COPYING in top-level directory.
 */

@ -28,18 +28,18 @@ extern "C" {

 /** \brief Matrix of distances between a set of objects.
 *
- * This matrix often contains latencies between NUMA nodes
+ * The most common matrix contains latencies between NUMA nodes
 * (as reported in the System Locality Distance Information Table (SLIT)
 * in the ACPI specification), which may or may not be physically accurate.
 * It corresponds to the latency for accessing the memory of one node
 * from a core in another node.
- * The corresponding kind is ::HWLOC_DISTANCES_KIND_FROM_OS | ::HWLOC_DISTANCES_KIND_FROM_USER.
+ * The corresponding kind is ::HWLOC_DISTANCES_KIND_MEANS_LATENCY | ::HWLOC_DISTANCES_KIND_FROM_USER.
 * The name of this distances structure is "NUMALatency".
- * Others distance structures include and "XGMIBandwidth", "XGMIHops",
- * "XeLinkBandwidth" and "NVLinkBandwidth".
 *
 * The matrix may also contain bandwidths between random sets of objects,
 * possibly provided by the user, as specified in the \p kind attribute.
+ * Others common distance structures include and "XGMIBandwidth", "XGMIHops",
+ * "XeLinkBandwidth" and "NVLinkBandwidth".
 *
 * Pointers \p objs and \p values should not be replaced, reallocated, freed, etc.
 * However callers are allowed to modify \p kind as well as the contents
@ -70,11 +70,10 @@ struct hwloc_distances_s {
 * The \p kind attribute of struct hwloc_distances_s is a OR'ed set
 * of kinds.
 *
- * A kind of format HWLOC_DISTANCES_KIND_FROM_* specifies where the
- * distance information comes from, if known.
- *
- * A kind of format HWLOC_DISTANCES_KIND_MEANS_* specifies whether
- * values are latencies or bandwidths, if applicable.
+ * Each distance matrix may have only one kind among HWLOC_DISTANCES_KIND_FROM_*
+ * specifying where distance information comes from,
+ * and one kind among HWLOC_DISTANCES_KIND_MEANS_* specifying
+ * whether values are latencies or bandwidths.
 */
 enum hwloc_distances_kind_e {
  /** \brief These distances were obtained from the operating system or hardware.
@ -357,6 +356,8 @@ typedef void * hwloc_distances_add_handle_t;
 * Otherwise, it will be copied internally and may later be freed by the caller.
 *
 * \p kind specifies the kind of distance as a OR'ed set of ::hwloc_distances_kind_e.
+ * Only one kind of meaning and one kind of provenance may be given if appropriate
+ * (e.g. ::HWLOC_DISTANCES_KIND_MEANS_BANDWIDTH and ::HWLOC_DISTANCES_KIND_FROM_USER).
 * Kind ::HWLOC_DISTANCES_KIND_HETEROGENEOUS_TYPES will be automatically set
 * according to objects having different types in hwloc_distances_add_values().
 *
@ -403,7 +404,8 @@ HWLOC_DECLSPEC int hwloc_distances_add_values(hwloc_topology_t topology,
 /** \brief Flags for adding a new distances to a topology. */
 enum hwloc_distances_add_flag_e {
  /** \brief Try to group objects based on the newly provided distance information.
-   * This is ignored for distances between objects of different types.
+   * Grouping is only performed when the distances structure contains latencies,
+   * and when all objects are of the same type.
   * \hideinitializer
   */
  HWLOC_DISTANCES_ADD_FLAG_GROUP = (1UL<<0),
--- a/src/3rdparty/hwloc/include/hwloc/helper.h
+++ b/src/3rdparty/hwloc/include/hwloc/helper.h
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2023 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2012 Université Bordeaux
 * Copyright © 2009-2010 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -946,6 +946,14 @@ enum hwloc_distrib_flags_e {
 *
 * \return 0 on success, -1 on error.
 *
+ * \note On hybrid CPUs (or asymmetric platforms), distribution may be suboptimal
+ * since the number of cores or PUs inside packages or below caches may vary
+ * (the top-down recursive partitioning ignores these numbers until reaching their levels).
+ * Hence it is recommended to distribute only inside a single homogeneous domain.
+ * For instance on a CPU with energy-efficient E-cores and high-performance P-cores,
+ * one should distribute separately N tasks on E-cores and M tasks on P-cores
+ * instead of trying to distribute directly M+N tasks on the entire CPUs.
+ *
 * \note This function requires the \p roots objects to have a CPU set.
 */
 static __hwloc_inline int
@ -960,7 +968,7 @@ hwloc_distrib(hwloc_topology_t topology,
  unsigned given, givenweight;
  hwloc_cpuset_t *cpusetp = set;

-  if (flags & ~HWLOC_DISTRIB_FLAG_REVERSE) {
+  if (!n || (flags & ~HWLOC_DISTRIB_FLAG_REVERSE)) {
    errno = EINVAL;
    return -1;
  }
--- a/src/3rdparty/hwloc/include/hwloc/memattrs.h
+++ b/src/3rdparty/hwloc/include/hwloc/memattrs.h
--- a/src/3rdparty/hwloc/include/hwloc/opencl.h
+++ b/src/3rdparty/hwloc/include/hwloc/opencl.h
@ -41,6 +41,15 @@ extern "C" {
 */
 /* Copyright (c) 2008-2018 The Khronos Group Inc. */

+/* needs "cl_khr_pci_bus_info" device extension, but not strictly required for clGetDeviceInfo() */
+typedef struct {
+    cl_uint pci_domain;
+    cl_uint pci_bus;
+    cl_uint pci_device;
+    cl_uint pci_function;
+} hwloc_cl_device_pci_bus_info_khr;
+#define HWLOC_CL_DEVICE_PCI_BUS_INFO_KHR 0x410F
+
 /* needs "cl_amd_device_attribute_query" device extension, but not strictly required for clGetDeviceInfo() */
 #define HWLOC_CL_DEVICE_TOPOLOGY_AMD 0x4037
 typedef union {
@ -78,9 +87,19 @@ hwloc_opencl_get_device_pci_busid(cl_device_id device,
                               unsigned *domain, unsigned *bus, unsigned *dev, unsigned *func)
 {
 	hwloc_cl_device_topology_amd amdtopo;
+	hwloc_cl_device_pci_bus_info_khr khrbusinfo;
 	cl_uint nvbus, nvslot, nvdomain;
 	cl_int clret;

+	clret = clGetDeviceInfo(device, HWLOC_CL_DEVICE_PCI_BUS_INFO_KHR, sizeof(khrbusinfo), &khrbusinfo, NULL);
+	if (CL_SUCCESS == clret) {
+		*domain = (unsigned) khrbusinfo.pci_domain;
+		*bus = (unsigned) khrbusinfo.pci_bus;
+		*dev = (unsigned) khrbusinfo.pci_device;
+		*func = (unsigned) khrbusinfo.pci_function;
+		return 0;
+	}
+
 	clret = clGetDeviceInfo(device, HWLOC_CL_DEVICE_TOPOLOGY_AMD, sizeof(amdtopo), &amdtopo, NULL);
 	if (CL_SUCCESS == clret
 	    && HWLOC_CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD == amdtopo.raw.type) {
--- a/src/3rdparty/hwloc/include/hwloc/plugins.h
+++ b/src/3rdparty/hwloc/include/hwloc/plugins.h
@ -1,5 +1,5 @@
 /*
- * Copyright © 2013-2022 Inria.  All rights reserved.
+ * Copyright © 2013-2024 Inria.  All rights reserved.
 * Copyright © 2016 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
 */
@ -645,6 +645,19 @@ HWLOC_DECLSPEC struct hwloc_obj * hwloc_pci_find_parent_by_busid(struct hwloc_to
 */
 HWLOC_DECLSPEC struct hwloc_obj * hwloc_pci_find_by_busid(struct hwloc_topology *topology, unsigned domain, unsigned bus, unsigned dev, unsigned func);

+
+/** @} */
+
+
+
+
+/** \defgroup hwlocality_components_distances Components and Plugins: distances
+ *
+ * \note These structures and functions may change when ::HWLOC_COMPONENT_ABI is modified.
+ *
+ * @{
+ */
+
 /** \brief Handle to a new distances structure during its addition to the topology. */
 typedef void * hwloc_backend_distances_add_handle_t;

--- a/src/3rdparty/hwloc/include/hwloc/rename.h
+++ b/src/3rdparty/hwloc/include/hwloc/rename.h
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
- * Copyright © 2010-2022 Inria.  All rights reserved.
+ * Copyright © 2010-2024 Inria.  All rights reserved.
 * See COPYING in top-level directory.
 */

@ -210,6 +210,7 @@ extern "C" {

 #define hwloc_obj_get_info_by_name HWLOC_NAME(obj_get_info_by_name)
 #define hwloc_obj_add_info HWLOC_NAME(obj_add_info)
+#define hwloc_obj_set_subtype HWLOC_NAME(obj_set_subtype)

 #define HWLOC_CPUBIND_PROCESS HWLOC_NAME_CAPS(CPUBIND_PROCESS)
 #define HWLOC_CPUBIND_THREAD HWLOC_NAME_CAPS(CPUBIND_THREAD)
@ -232,6 +233,7 @@ extern "C" {
 #define HWLOC_MEMBIND_FIRSTTOUCH HWLOC_NAME_CAPS(MEMBIND_FIRSTTOUCH)
 #define HWLOC_MEMBIND_BIND HWLOC_NAME_CAPS(MEMBIND_BIND)
 #define HWLOC_MEMBIND_INTERLEAVE HWLOC_NAME_CAPS(MEMBIND_INTERLEAVE)
+#define HWLOC_MEMBIND_WEIGHTED_INTERLEAVE HWLOC_NAME_CAPS(MEMBIND_WEIGHTED_INTERLEAVE)
 #define HWLOC_MEMBIND_NEXTTOUCH HWLOC_NAME_CAPS(MEMBIND_NEXTTOUCH)
 #define HWLOC_MEMBIND_MIXED HWLOC_NAME_CAPS(MEMBIND_MIXED)

@ -560,6 +562,7 @@ extern "C" {

 /* opencl.h */

+#define hwloc_cl_device_pci_bus_info_khr HWLOC_NAME(cl_device_pci_bus_info_khr)
 #define hwloc_cl_device_topology_amd HWLOC_NAME(cl_device_topology_amd)
 #define hwloc_opencl_get_device_pci_busid HWLOC_NAME(opencl_get_device_pci_ids)
 #define hwloc_opencl_get_device_cpuset HWLOC_NAME(opencl_get_device_cpuset)
@ -715,6 +718,8 @@ extern "C" {
 #define hwloc__obj_type_is_dcache HWLOC_NAME(_obj_type_is_dcache)
 #define hwloc__obj_type_is_icache HWLOC_NAME(_obj_type_is_icache)

+#define hwloc__pci_link_speed HWLOC_NAME(_pci_link_speed)
+
 /* private/cpuid-x86.h */

 #define hwloc_have_x86_cpuid HWLOC_NAME(have_x86_cpuid)
--- a/src/3rdparty/hwloc/include/private/autogen/config.h
+++ b/src/3rdparty/hwloc/include/private/autogen/config.h
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009, 2011, 2012 CNRS.  All rights reserved.
- * Copyright © 2009-2021 Inria.  All rights reserved.
+ * Copyright © 2009-2020 Inria.  All rights reserved.
 * Copyright © 2009, 2011, 2012, 2015 Université Bordeaux.  All rights reserved.
 * Copyright © 2009-2020 Cisco Systems, Inc.  All rights reserved.
 * $COPYRIGHT$
@ -17,6 +17,10 @@

 #define HWLOC_HAVE_MSVC_CPUIDEX 1

+/* #undef HAVE_MKSTEMP */
+
+#define HWLOC_HAVE_X86_CPUID 1
+
 /* Define to 1 if the system has the type `CACHE_DESCRIPTOR'. */
 #define HAVE_CACHE_DESCRIPTOR 0

@ -128,8 +132,7 @@
 #define HAVE_DECL__SC_PAGE_SIZE 0

 /* Define to 1 if you have the <dirent.h> header file. */
-/* #define HAVE_DIRENT_H 1 */
-#undef HAVE_DIRENT_H
+/* #undef HAVE_DIRENT_H */

 /* Define to 1 if you have the <dlfcn.h> header file. */
 /* #undef HAVE_DLFCN_H */
@ -282,7 +285,7 @@
 #define HAVE_STRING_H 1

 /* Define to 1 if you have the `strncasecmp' function. */
-#define HAVE_STRNCASECMP 1
+/* #undef HAVE_STRNCASECMP */

 /* Define to '1' if sysctl is present and usable */
 /* #undef HAVE_SYSCTL */
@ -323,8 +326,7 @@
 /* #undef HAVE_UNAME */

 /* Define to 1 if you have the <unistd.h> header file. */
-/* #define HAVE_UNISTD_H 1 */
-#undef HAVE_UNISTD_H
+/* #undef HAVE_UNISTD_H */

 /* Define to 1 if you have the `uselocale' function. */
 /* #undef HAVE_USELOCALE */
@ -659,7 +661,7 @@
 #define hwloc_pid_t HANDLE

 /* Define this to either strncasecmp or strncmp */
-#define hwloc_strncasecmp strncasecmp
+/* #undef hwloc_strncasecmp */

 /* Define this to the thread ID type */
 #define hwloc_thread_t HANDLE
--- a/src/3rdparty/hwloc/include/private/cpuid-x86.h
+++ b/src/3rdparty/hwloc/include/private/cpuid-x86.h
@ -11,6 +11,22 @@
 #ifndef HWLOC_PRIVATE_CPUID_X86_H
 #define HWLOC_PRIVATE_CPUID_X86_H

+/* A macro for annotating memory as uninitialized when building with MSAN
+ * (and otherwise having no effect). See below for why this is used with
+ * our custom assembly.
+ */
+#ifdef __has_feature
+#define HWLOC_HAS_FEATURE(name) __has_feature(name)
+#else
+#define HWLOC_HAS_FEATURE(name) 0
+#endif
+#if HWLOC_HAS_FEATURE(memory_sanitizer) || defined(MEMORY_SANITIZER)
+#include <sanitizer/msan_interface.h>
+#define HWLOC_ANNOTATE_MEMORY_IS_INITIALIZED(ptr, len) __msan_unpoison(ptr, len)
+#else
+#define HWLOC_ANNOTATE_MEMORY_IS_INITIALIZED(ptr, len)
+#endif
+
 #if (defined HWLOC_X86_32_ARCH) && (!defined HWLOC_HAVE_MSVC_CPUIDEX)
 static __hwloc_inline int hwloc_have_x86_cpuid(void)
 {
@ -71,12 +87,18 @@ static __hwloc_inline void hwloc_x86_cpuid(unsigned *eax, unsigned *ebx, unsigne
  "movl %k2,%1\n\t"
  : "+a" (*eax), "=m" (*ebx), "=&r"(sav_rbx),
    "+c" (*ecx), "=&d" (*edx));
+  /* MSAN does not recognize the effect of the above assembly on the memory operand
+   * (`"=m"(*ebx)`). This may get improved in MSAN at some point in the future, e.g.
+   * see https://github.com/llvm/llvm-project/pull/77393. */
+  HWLOC_ANNOTATE_MEMORY_IS_INITIALIZED(ebx, sizeof *ebx);
 #elif defined(HWLOC_X86_32_ARCH)
  __asm__(
  "mov %%ebx,%1\n\t"
  "cpuid\n\t"
  "xchg %%ebx,%1\n\t"
  : "+a" (*eax), "=&SD" (*ebx), "+c" (*ecx), "=&d" (*edx));
+  /* See above. */
+  HWLOC_ANNOTATE_MEMORY_IS_INITIALIZED(ebx, sizeof *ebx);
 #else
 #error unknown architecture
 #endif
--- a/src/3rdparty/hwloc/include/private/misc.h
+++ b/src/3rdparty/hwloc/include/private/misc.h
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2019 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2012 Université Bordeaux
 * Copyright © 2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -573,4 +573,35 @@ typedef SSIZE_T ssize_t;
 #  endif
 #endif

+static __inline float
+hwloc__pci_link_speed(unsigned generation, unsigned lanes)
+{
+  float lanespeed;
+  /*
+   * These are single-direction bandwidths only.
+   *
+   * Gen1 used NRZ with 8/10 encoding.
+   * PCIe Gen1 = 2.5GT/s signal-rate per lane x 8/10    =  0.25GB/s data-rate per lane
+   * PCIe Gen2 = 5  GT/s signal-rate per lane x 8/10    =  0.5 GB/s data-rate per lane
+   * Gen3 switched to NRZ with 128/130 encoding.
+   * PCIe Gen3 = 8  GT/s signal-rate per lane x 128/130 =  1   GB/s data-rate per lane
+   * PCIe Gen4 = 16 GT/s signal-rate per lane x 128/130 =  2   GB/s data-rate per lane
+   * PCIe Gen5 = 32 GT/s signal-rate per lane x 128/130 =  4   GB/s data-rate per lane
+   * Gen6 switched to PAM with with 242/256 FLIT (242B payload protected by 8B CRC + 6B FEC).
+   * PCIe Gen6 = 64 GT/s signal-rate per lane x 242/256 =  8   GB/s data-rate per lane
+   * PCIe Gen7 = 128GT/s signal-rate per lane x 242/256 = 16   GB/s data-rate per lane
+   */
+
+  /* lanespeed in Gbit/s */
+  if (generation <= 2)
+    lanespeed = 2.5f * generation * 0.8f;
+  else if (generation <= 5)
+    lanespeed = 8.0f * (1<<(generation-3)) * 128/130;
+  else
+    lanespeed = 8.0f * (1<<(generation-3)) * 242/256; /* assume Gen8 will be 256 GT/s and so on */
+
+  /* linkspeed in GB/s */
+  return lanespeed * lanes / 8;
+}
+
 #endif /* HWLOC_PRIVATE_MISC_H */
--- a/src/3rdparty/hwloc/src/bind.c
+++ b/src/3rdparty/hwloc/src/bind.c
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2020 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2010, 2012 Université Bordeaux
 * Copyright © 2011-2015 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -287,6 +287,7 @@ static __hwloc_inline int hwloc__check_membind_policy(hwloc_membind_policy_t pol
      || policy == HWLOC_MEMBIND_FIRSTTOUCH
      || policy == HWLOC_MEMBIND_BIND
      || policy == HWLOC_MEMBIND_INTERLEAVE
+      || policy == HWLOC_MEMBIND_WEIGHTED_INTERLEAVE
      || policy == HWLOC_MEMBIND_NEXTTOUCH)
    return 0;
  return -1;
--- a/src/3rdparty/hwloc/src/bitmap.c
+++ b/src/3rdparty/hwloc/src/bitmap.c
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2020 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2011 Université Bordeaux
 * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -245,6 +245,7 @@ int hwloc_bitmap_copy(struct hwloc_bitmap_s * dst, const struct hwloc_bitmap_s *
 /* Strings always use 32bit groups */
 #define HWLOC_PRIxSUBBITMAP		"%08lx"
 #define HWLOC_BITMAP_SUBSTRING_SIZE	32
+#define HWLOC_BITMAP_SUBSTRING_FULL_VALUE 0xFFFFFFFFUL
 #define HWLOC_BITMAP_SUBSTRING_LENGTH	(HWLOC_BITMAP_SUBSTRING_SIZE/4)
 #define HWLOC_BITMAP_STRING_PER_LONG	(HWLOC_BITS_PER_LONG/HWLOC_BITMAP_SUBSTRING_SIZE)

@ -261,6 +262,7 @@ int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, const stru
  const unsigned long accum_mask = ~0UL;
 #else /* HWLOC_BITS_PER_LONG != HWLOC_BITMAP_SUBSTRING_SIZE */
  const unsigned long accum_mask = ((1UL << HWLOC_BITMAP_SUBSTRING_SIZE) - 1) << (HWLOC_BITS_PER_LONG - HWLOC_BITMAP_SUBSTRING_SIZE);
+  int merge_with_infinite_prefix = 0;
 #endif /* HWLOC_BITS_PER_LONG != HWLOC_BITMAP_SUBSTRING_SIZE */

  HWLOC__BITMAP_CHECK(set);
@ -279,6 +281,9 @@ int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, const stru
      res = size>0 ? (int)size - 1 : 0;
    tmp += res;
    size -= res;
+#if HWLOC_BITS_PER_LONG > HWLOC_BITMAP_SUBSTRING_SIZE
+    merge_with_infinite_prefix = 1;
+#endif
  }

  i=(int) set->ulongs_count-1;
@ -294,16 +299,24 @@ int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, const stru
  }

  while (i>=0 || accumed) {
+    unsigned long value;
+
    /* Refill accumulator */
    if (!accumed) {
      accum = set->ulongs[i--];
      accumed = HWLOC_BITS_PER_LONG;
    }
+    value = (accum & accum_mask) >> (HWLOC_BITS_PER_LONG - HWLOC_BITMAP_SUBSTRING_SIZE);

-    if (accum & accum_mask) {
+#if HWLOC_BITS_PER_LONG > HWLOC_BITMAP_SUBSTRING_SIZE
+    if (merge_with_infinite_prefix && value == HWLOC_BITMAP_SUBSTRING_FULL_VALUE) {
+      /* first full subbitmap merged with infinite prefix */
+      res = 0;
+    } else
+#endif
+    if (value) {
      /* print the whole subset if not empty */
-        res = hwloc_snprintf(tmp, size, needcomma ? ",0x" HWLOC_PRIxSUBBITMAP : "0x" HWLOC_PRIxSUBBITMAP,
-		     (accum & accum_mask) >> (HWLOC_BITS_PER_LONG - HWLOC_BITMAP_SUBSTRING_SIZE));
+      res = hwloc_snprintf(tmp, size, needcomma ? ",0x" HWLOC_PRIxSUBBITMAP : "0x" HWLOC_PRIxSUBBITMAP, value);
      needcomma = 1;
    } else if (i == -1 && accumed == HWLOC_BITMAP_SUBSTRING_SIZE) {
      /* print a single 0 to mark the last subset */
@ -323,6 +336,7 @@ int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, const stru
 #else
    accum <<= HWLOC_BITMAP_SUBSTRING_SIZE;
    accumed -= HWLOC_BITMAP_SUBSTRING_SIZE;
+    merge_with_infinite_prefix = 0;
 #endif

    if (res >= size)
@ -362,7 +376,8 @@ int hwloc_bitmap_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc_restric
 {
  const char * current = string;
  unsigned long accum = 0;
-  int count=0;
+  int count = 0;
+  int ulongcount;
  int infinite = 0;

  /* count how many substrings there are */
@ -383,9 +398,20 @@ int hwloc_bitmap_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc_restric
    count--;
  }

-  if (hwloc_bitmap_reset_by_ulongs(set, (count + HWLOC_BITMAP_STRING_PER_LONG - 1) / HWLOC_BITMAP_STRING_PER_LONG) < 0)
+  ulongcount = (count + HWLOC_BITMAP_STRING_PER_LONG - 1) / HWLOC_BITMAP_STRING_PER_LONG;
+  if (hwloc_bitmap_reset_by_ulongs(set, ulongcount) < 0)
    return -1;
-  set->infinite = 0;
+
+  set->infinite = 0; /* will be updated later */
+
+#if HWLOC_BITS_PER_LONG != HWLOC_BITMAP_SUBSTRING_SIZE
+  if (infinite && (count % HWLOC_BITMAP_STRING_PER_LONG) != 0) {
+    /* accumulate substrings of the first ulong that are hidden in the infinite prefix */
+    int i;
+    for(i = (count % HWLOC_BITMAP_STRING_PER_LONG); i < HWLOC_BITMAP_STRING_PER_LONG; i++)
+      accum |= (HWLOC_BITMAP_SUBSTRING_FULL_VALUE << (i*HWLOC_BITMAP_SUBSTRING_SIZE));
+  }
+#endif

  while (*current != '\0') {
    unsigned long val;
@ -544,6 +570,9 @@ int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, co
  ssize_t size = buflen;
  char *tmp = buf;
  int res, ret = 0;
+#if HWLOC_BITS_PER_LONG == 64
+  int merge_with_infinite_prefix = 0;
+#endif
  int started = 0;
  int i;

@ -563,6 +592,9 @@ int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, co
      res = size>0 ? (int)size - 1 : 0;
    tmp += res;
    size -= res;
+#if HWLOC_BITS_PER_LONG == 64
+    merge_with_infinite_prefix = 1;
+#endif
  }

  i=set->ulongs_count-1;
@ -582,7 +614,11 @@ int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, co
    if (started) {
      /* print the whole subset */
 #if HWLOC_BITS_PER_LONG == 64
-      res = hwloc_snprintf(tmp, size, "%016lx", val);
+      if (merge_with_infinite_prefix && (val & 0xffffffff00000000UL) == 0xffffffff00000000UL) {
+        res = hwloc_snprintf(tmp, size, "%08lx", val & 0xffffffffUL);
+      } else  {
+        res = hwloc_snprintf(tmp, size, "%016lx", val);
+      }
 #else
      res = hwloc_snprintf(tmp, size, "%08lx", val);
 #endif
@ -599,6 +635,9 @@ int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, co
      res = size>0 ? (int)size - 1 : 0;
    tmp += res;
    size -= res;
+#if HWLOC_BITS_PER_LONG == 64
+    merge_with_infinite_prefix = 0;
+#endif
  }

  /* if didn't display anything, display 0x0 */
@ -679,6 +718,10 @@ int hwloc_bitmap_taskset_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc
      goto failed;

    set->ulongs[count-1] = val;
+    if (infinite && tmpchars != HWLOC_BITS_PER_LONG/4) {
+      /* infinite prefix with partial substring, fill remaining bits */
+      set->ulongs[count-1] |= (~0ULL)<<(4*tmpchars);
+    }

    current += tmpchars;
    chars -= tmpchars;
--- a/src/3rdparty/hwloc/src/cpukinds.c
+++ b/src/3rdparty/hwloc/src/cpukinds.c
@ -1,5 +1,5 @@
 /*
- * Copyright © 2020-2022 Inria.  All rights reserved.
+ * Copyright © 2020-2024 Inria.  All rights reserved.
 * See COPYING in top-level directory.
 */

@ -50,6 +50,7 @@ hwloc_internal_cpukinds_dup(hwloc_topology_t new, hwloc_topology_t old)
    return -1;
  new->cpukinds = kinds;
  new->nr_cpukinds = old->nr_cpukinds;
+  new->nr_cpukinds_allocated = old->nr_cpukinds;
  memcpy(kinds, old->cpukinds, old->nr_cpukinds * sizeof(*kinds));

  for(i=0;i<old->nr_cpukinds; i++) {
--- a/src/3rdparty/hwloc/src/distances.c
+++ b/src/3rdparty/hwloc/src/distances.c
@ -1,5 +1,5 @@
 /*
- * Copyright © 2010-2022 Inria.  All rights reserved.
+ * Copyright © 2010-2024 Inria.  All rights reserved.
 * Copyright © 2011-2012 Université Bordeaux
 * Copyright © 2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -624,8 +624,8 @@ void * hwloc_distances_add_create(hwloc_topology_t topology,
    return NULL;
  }
  if ((kind & ~HWLOC_DISTANCES_KIND_ALL)
-      || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_FROM_ALL) != 1
-      || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_MEANS_ALL) != 1) {
+      || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_FROM_ALL) > 1
+      || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_MEANS_ALL) > 1) {
    errno = EINVAL;
    return NULL;
  }
--- a/src/3rdparty/hwloc/src/memattrs.c
+++ b/src/3rdparty/hwloc/src/memattrs.c
@ -1,5 +1,5 @@
 /*
- * Copyright © 2020-2023 Inria.  All rights reserved.
+ * Copyright © 2020-2024 Inria.  All rights reserved.
 * See COPYING in top-level directory.
 */

@ -14,13 +14,26 @@
 */

 static __hwloc_inline
-hwloc_uint64_t hwloc__memattr_get_convenience_value(hwloc_memattr_id_t id,
-                                                    hwloc_obj_t node)
+int hwloc__memattr_get_convenience_value(hwloc_memattr_id_t id,
+                                         hwloc_obj_t node,
+                                         hwloc_uint64_t *valuep)
 {
-  if (id == HWLOC_MEMATTR_ID_CAPACITY)
-    return node->attr->numanode.local_memory;
-  else if (id == HWLOC_MEMATTR_ID_LOCALITY)
-    return hwloc_bitmap_weight(node->cpuset);
+  if (id == HWLOC_MEMATTR_ID_CAPACITY) {
+    if (node->type != HWLOC_OBJ_NUMANODE) {
+      errno = EINVAL;
+      return -1;
+    }
+    *valuep = node->attr->numanode.local_memory;
+    return 0;
+  }
+  else if (id == HWLOC_MEMATTR_ID_LOCALITY) {
+    if (!node->cpuset) {
+      errno = EINVAL;
+      return -1;
+    }
+    *valuep = hwloc_bitmap_weight(node->cpuset);
+    return 0;
+  }
  else
    assert(0);
  return 0; /* shut up the compiler */
@ -622,7 +635,7 @@ hwloc_memattr_get_targets(hwloc_topology_t topology,
      if (found<max) {
        targets[found] = node;
        if (values)
-          values[found] = hwloc__memattr_get_convenience_value(id, node);
+          hwloc__memattr_get_convenience_value(id, node, &values[found]);
      }
      found++;
    }
@ -748,7 +761,7 @@ hwloc_memattr_get_initiators(hwloc_topology_t topology,
  struct hwloc_internal_memattr_target_s *imtg;
  unsigned i, max;

-  if (flags) {
+  if (flags || !target_node) {
    errno = EINVAL;
    return -1;
  }
@ -810,7 +823,7 @@ hwloc_memattr_get_value(hwloc_topology_t topology,
  struct hwloc_internal_memattr_s *imattr;
  struct hwloc_internal_memattr_target_s *imtg;

-  if (flags) {
+  if (flags || !target_node) {
    errno = EINVAL;
    return -1;
  }
@ -823,8 +836,7 @@ hwloc_memattr_get_value(hwloc_topology_t topology,

  if (imattr->iflags & HWLOC_IMATTR_FLAG_CONVENIENCE) {
    /* convenience attributes */
-    *valuep = hwloc__memattr_get_convenience_value(id, target_node);
-    return 0;
+    return hwloc__memattr_get_convenience_value(id, target_node, valuep);
  }

  /* normal attributes */
@ -936,7 +948,7 @@ hwloc_memattr_set_value(hwloc_topology_t topology,
 {
  struct hwloc_internal_location_s iloc, *ilocp;

-  if (flags) {
+  if (flags || !target_node) {
    errno = EINVAL;
    return -1;
  }
@ -1007,10 +1019,10 @@ hwloc_memattr_get_best_target(hwloc_topology_t topology,
    /* convenience attributes */
    for(j=0; ; j++) {
      hwloc_obj_t node = hwloc_get_obj_by_type(topology, HWLOC_OBJ_NUMANODE, j);
-      hwloc_uint64_t value;
+      hwloc_uint64_t value = 0;
      if (!node)
        break;
-      value = hwloc__memattr_get_convenience_value(id, node);
+      hwloc__memattr_get_convenience_value(id, node, &value);
      hwloc__update_best_target(&best, &best_value, &found,
                                node, value,
                                imattr->flags & HWLOC_MEMATTR_FLAG_HIGHER_FIRST);
@ -1093,7 +1105,7 @@ hwloc_memattr_get_best_initiator(hwloc_topology_t topology,
  int found;
  unsigned i;

-  if (flags) {
+  if (flags || !target_node) {
    errno = EINVAL;
    return -1;
  }
@ -1806,6 +1818,12 @@ hwloc__apply_memory_tiers_subtypes(hwloc_topology_t topology,
      }
    }
  }
+  if (nr_tiers > 1) {
+    hwloc_obj_t root = hwloc_get_root_obj(topology);
+    char tmp[20];
+    snprintf(tmp, sizeof(tmp), "%u", nr_tiers);
+    hwloc__add_info_nodup(&root->infos, &root->infos_count, "MemoryTiersNr", tmp, 1);
+  }
 }

 int
--- a/src/3rdparty/hwloc/src/pci-common.c
+++ b/src/3rdparty/hwloc/src/pci-common.c
@ -1,5 +1,5 @@
 /*
- * Copyright © 2009-2022 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * See COPYING in top-level directory.
 */

@ -886,36 +886,12 @@ hwloc_pcidisc_find_linkspeed(const unsigned char *config,
 			     unsigned offset, float *linkspeed)
 {
  unsigned linksta, speed, width;
-  float lanespeed;

  memcpy(&linksta, &config[offset + HWLOC_PCI_EXP_LNKSTA], 4);
  speed = linksta & HWLOC_PCI_EXP_LNKSTA_SPEED; /* PCIe generation */
  width = (linksta & HWLOC_PCI_EXP_LNKSTA_WIDTH) >> 4; /* how many lanes */
-  /*
-   * These are single-direction bandwidths only.
-   *
-   * Gen1 used NRZ with 8/10 encoding.
-   * PCIe Gen1 = 2.5GT/s signal-rate per lane x 8/10    =  0.25GB/s data-rate per lane
-   * PCIe Gen2 = 5  GT/s signal-rate per lane x 8/10    =  0.5 GB/s data-rate per lane
-   * Gen3 switched to NRZ with 128/130 encoding.
-   * PCIe Gen3 = 8  GT/s signal-rate per lane x 128/130 =  1   GB/s data-rate per lane
-   * PCIe Gen4 = 16 GT/s signal-rate per lane x 128/130 =  2   GB/s data-rate per lane
-   * PCIe Gen5 = 32 GT/s signal-rate per lane x 128/130 =  4   GB/s data-rate per lane
-   * Gen6 switched to PAM with with 242/256 FLIT (242B payload protected by 8B CRC + 6B FEC).
-   * PCIe Gen6 = 64 GT/s signal-rate per lane x 242/256 =  8   GB/s data-rate per lane
-   * PCIe Gen7 = 128GT/s signal-rate per lane x 242/256 = 16   GB/s data-rate per lane
-   */

-  /* lanespeed in Gbit/s */
-  if (speed <= 2)
-    lanespeed = 2.5f * speed * 0.8f;
-  else if (speed <= 5)
-    lanespeed = 8.0f * (1<<(speed-3)) * 128/130;
-  else
-    lanespeed = 8.0f * (1<<(speed-3)) * 242/256; /* assume Gen8 will be 256 GT/s and so on */
-
-  /* linkspeed in GB/s */
-  *linkspeed = lanespeed * width / 8;
+  *linkspeed = hwloc__pci_link_speed(speed, width);
  return 0;
 }

--- a/src/3rdparty/hwloc/src/topology-windows.c
+++ b/src/3rdparty/hwloc/src/topology-windows.c
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2023 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2012, 2020 Université Bordeaux
 * Copyright © 2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -220,7 +220,7 @@ static void hwloc_win_get_function_ptrs(void)
 #pragma GCC diagnostic ignored "-Wcast-function-type"
 #endif

-    kernel32 = LoadLibrary("kernel32.dll");
+    kernel32 = LoadLibrary(TEXT("kernel32.dll"));
    if (kernel32) {
      GetActiveProcessorGroupCountProc =
 	(PFN_GETACTIVEPROCESSORGROUPCOUNT) GetProcAddress(kernel32, "GetActiveProcessorGroupCount");
@ -249,12 +249,12 @@ static void hwloc_win_get_function_ptrs(void)
    }

    if (!QueryWorkingSetExProc) {
-      HMODULE psapi = LoadLibrary("psapi.dll");
+      HMODULE psapi = LoadLibrary(TEXT("psapi.dll"));
      if (psapi)
        QueryWorkingSetExProc = (PFN_QUERYWORKINGSETEX) GetProcAddress(psapi, "QueryWorkingSetEx");
    }

-    ntdll = GetModuleHandle("ntdll");
+    ntdll = GetModuleHandle(TEXT("ntdll"));
    RtlGetVersionProc = (PFN_RTLGETVERSION) GetProcAddress(ntdll, "RtlGetVersion");

 #if HWLOC_HAVE_GCC_W_CAST_FUNCTION_TYPE
--- a/src/3rdparty/hwloc/src/topology-x86.c
+++ b/src/3rdparty/hwloc/src/topology-x86.c
@ -1,11 +1,11 @@
 /*
- * Copyright © 2010-2023 Inria.  All rights reserved.
+ * Copyright © 2010-2024 Inria.  All rights reserved.
 * Copyright © 2010-2013 Université Bordeaux
 * Copyright © 2010-2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
 *
 *
- * This backend is only used when the operating system does not export
+ * This backend is mostly used when the operating system does not export
 * the necessary hardware topology information to user-space applications.
 * Currently, FreeBSD and NetBSD only add PUs and then fallback to this
 * backend for CPU/Cache discovery.
@ -15,6 +15,7 @@
 * on various architectures, without having to use this x86-specific code.
 * But this backend is still used after them to annotate some objects with
 * additional details (CPU info in Package, Inclusiveness in Caches).
+ * It may also be enabled manually to work-around bugs in native OS discovery.
 */

 #include "private/autogen/config.h"
@ -487,7 +488,7 @@ static void read_amd_cores_legacy(struct procinfo *infos, struct cpuiddump *src_
 }

 /* AMD unit/node from CPUID 0x8000001e leaf (topoext) */
-static void read_amd_cores_topoext(struct hwloc_x86_backend_data_s *data, struct procinfo *infos, unsigned long flags, struct cpuiddump *src_cpuiddump)
+static void read_amd_cores_topoext(struct hwloc_x86_backend_data_s *data, struct procinfo *infos, unsigned long flags __hwloc_attribute_unused, struct cpuiddump *src_cpuiddump)
 {
  unsigned apic_id, nodes_per_proc = 0;
  unsigned eax, ebx, ecx, edx;
@ -496,7 +497,6 @@ static void read_amd_cores_topoext(struct hwloc_x86_backend_data_s *data, struct
  cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump);
  infos->apicid = apic_id = eax;

-  if (flags & HWLOC_X86_DISC_FLAG_TOPOEXT_NUMANODES) {
    if (infos->cpufamilynumber == 0x16) {
      /* ecx is reserved */
      infos->ids[NODE] = 0;
@ -511,7 +511,6 @@ static void read_amd_cores_topoext(struct hwloc_x86_backend_data_s *data, struct
        || (infos->cpufamilynumber == 0x19 && nodes_per_proc > 1)) {
      hwloc_debug("warning: undefined nodes_per_proc value %u, assuming it means %u\n", nodes_per_proc, nodes_per_proc);
    }
-  }

  if (infos->cpufamilynumber <= 0x16) { /* topoext appeared in 0x15 and compute-units were only used in 0x15 and 0x16 */
    unsigned cores_per_unit;
@ -533,9 +532,9 @@ static void read_amd_cores_topoext(struct hwloc_x86_backend_data_s *data, struct
 }

 /* Intel core/thread or even die/module/tile from CPUID 0x0b or 0x1f leaves (v1 and v2 extended topology enumeration)
- * or AMD complex/ccd from CPUID 0x80000026 (extended CPU topology)
+ * or AMD core/thread or even complex/ccd from CPUID 0x0b or 0x80000026 (extended CPU topology)
 */
-static void read_extended_topo(struct hwloc_x86_backend_data_s *data, struct procinfo *infos, unsigned leaf, enum cpuid_type cpuid_type, struct cpuiddump *src_cpuiddump)
+static void read_extended_topo(struct hwloc_x86_backend_data_s *data, struct procinfo *infos, unsigned leaf, enum cpuid_type cpuid_type __hwloc_attribute_unused, struct cpuiddump *src_cpuiddump)
 {
  unsigned level, apic_nextshift, apic_type, apic_id = 0, apic_shift = 0, id;
  unsigned threadid __hwloc_attribute_unused = 0; /* shut-up compiler */
@ -547,20 +546,15 @@ static void read_extended_topo(struct hwloc_x86_backend_data_s *data, struct pro
    eax = leaf;
    cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump);
    /* Intel specifies that the 0x0b/0x1f loop should stop when we get "invalid domain" (0 in ecx[8:15])
-     * (if so, we also get 0 in eax/ebx for invalid subleaves).
+     * (if so, we also get 0 in eax/ebx for invalid subleaves). Zhaoxin implements this too.
     * However AMD rather says that the 0x80000026/0x0b loop should stop when we get "no thread at this level" (0 in ebx[0:15]).
-     * Zhaoxin follows the Intel specs but also returns "no thread at this level" for the last *valid* level (at least on KH-4000).
-     * From the Linux kernel code, it's very likely that AMD also returns "invalid domain"
-     * (because detect_extended_topology() uses that for all x86 CPUs)
-     * but keep with the official doc until AMD can clarify that (see #593).
+     *
+     * Linux kernel <= 6.8 used "invalid domain" for both Intel and AMD (in detect_extended_topology())
+     * but x86 discovery revamp in 6.9 now properly checks both Intel and AMD conditions (in topo_subleaf()).
+     * So let's assume we are allowed to break-out once one of the Intel+AMD conditions is met.
     */
-    if (cpuid_type == amd) {
-      if (!(ebx & 0xffff))
-        break;
-    } else {
-      if (!(ecx & 0xff00))
-        break;
-    }
+    if (!(ebx & 0xffff) || !(ecx & 0xff00))
+      break;
    apic_packageshift = eax & 0x1f;
  }

@ -572,13 +566,8 @@ static void read_extended_topo(struct hwloc_x86_backend_data_s *data, struct pro
 	ecx = level;
 	eax = leaf;
 	cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump);
-        if (cpuid_type == amd) {
-          if (!(ebx & 0xffff))
-            break;
-        } else {
-          if (!(ecx & 0xff00))
-            break;
-        }
+        if (!(ebx & 0xffff) || !(ecx & 0xff00))
+          break;
 	apic_nextshift = eax & 0x1f;
 	apic_type = (ecx & 0xff00) >> 8;
 	apic_id = edx;
@ -1825,7 +1814,7 @@ hwloc_x86_check_cpuiddump_input(const char *src_cpuiddump_path, hwloc_bitmap_t s
    goto out_with_path;
  }
  fclose(file);
-  if (strcmp(line, "Architecture: x86\n")) {
+  if (strncmp(line, "Architecture: x86", 17)) {
    fprintf(stderr, "hwloc/x86: Found non-x86 dumped cpuid summary in %s: %s\n", path, line);
    goto out_with_path;
  }
--- a/src/3rdparty/hwloc/src/topology-xml-nolibxml.c
+++ b/src/3rdparty/hwloc/src/topology-xml-nolibxml.c
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2020 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2011 Université Bordeaux
 * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -41,7 +41,7 @@ typedef struct hwloc__nolibxml_import_state_data_s {
 static char *
 hwloc__nolibxml_import_ignore_spaces(char *buffer)
 {
-  return buffer + strspn(buffer, " \t\n");
+  return buffer + strspn(buffer, " \t\n\r");
 }

 static int
--- a/src/3rdparty/hwloc/src/topology-xml.c
+++ b/src/3rdparty/hwloc/src/topology-xml.c
@ -1,6 +1,6 @@
 /*
 * Copyright © 2009 CNRS
- * Copyright © 2009-2023 Inria.  All rights reserved.
+ * Copyright © 2009-2024 Inria.  All rights reserved.
 * Copyright © 2009-2011, 2020 Université Bordeaux
 * Copyright © 2009-2018 Cisco Systems, Inc.  All rights reserved.
 * See COPYING in top-level directory.
@ -872,6 +872,10 @@ hwloc__xml_import_object(hwloc_topology_t topology,
 	  /* deal with possible future type */
 	  obj->type = HWLOC_OBJ_GROUP;
 	  obj->attr->group.kind = HWLOC_GROUP_KIND_INTEL_MODULE;
+	} else if (!strcasecmp(attrvalue, "Cluster")) {
+	  /* deal with possible future type */
+	  obj->type = HWLOC_OBJ_GROUP;
+	  obj->attr->group.kind = HWLOC_GROUP_KIND_LINUX_CLUSTER;
 	} else if (!strcasecmp(attrvalue, "MemCache")) {
 	  /* ignore possible future type */
 	  obj->type = _HWLOC_OBJ_FUTURE;
@ -1344,7 +1348,7 @@ hwloc__xml_v2import_support(hwloc_topology_t topology,
    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_support) == 4*sizeof(void*));
    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_discovery_support) == 6);
    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_cpubind_support) == 11);
-    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_membind_support) == 15);
+    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_membind_support) == 16);
    HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_misc_support) == 1);
 #endif

@ -1378,6 +1382,7 @@ hwloc__xml_v2import_support(hwloc_topology_t topology,
    else DO(membind,firsttouch_membind);
    else DO(membind,bind_membind);
    else DO(membind,interleave_membind);
+    else DO(membind,weighted_interleave_membind);
    else DO(membind,nexttouch_membind);
    else DO(membind,migrate_membind);
    else DO(membind,get_area_memlocation);
@ -1436,6 +1441,10 @@ hwloc__xml_v2import_distances(hwloc_topology_t topology,
    }
    else if (!strcmp(attrname, "kind")) {
      kind = strtoul(attrvalue, NULL, 10);
+      /* forward compat with "HOPS" kind in v3 */
+      if (kind & (1UL<<5))
+        /* hops becomes latency */
+        kind = (kind & ~(1UL<<5)) | HWLOC_DISTANCES_KIND_MEANS_LATENCY;
    }
    else if (!strcmp(attrname, "name")) {
      name = attrvalue;
@ -3087,7 +3096,7 @@ hwloc__xml_v2export_support(hwloc__xml_export_state_t parentstate, hwloc_topolog
  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_support) == 4*sizeof(void*));
  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_discovery_support) == 6);
  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_cpubind_support) == 11);
-  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_membind_support) == 15);
+  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_membind_support) == 16);
  HWLOC_BUILD_ASSERT(sizeof(struct hwloc_topology_misc_support) == 1);
 #endif

@ -3132,6 +3141,7 @@ hwloc__xml_v2export_support(hwloc__xml_export_state_t parentstate, hwloc_topolog
  DO(membind,firsttouch_membind);
  DO(membind,bind_membind);
  DO(membind,interleave_membind);
+  DO(membind,weighted_interleave_membind);
  DO(membind,nexttouch_membind);
  DO(membind,migrate_membind);
  DO(membind,get_area_memlocation);
--- a/src/3rdparty/hwloc/src/topology.c
+++ b/src/3rdparty/hwloc/src/topology.c
@ -465,6 +465,20 @@ hwloc_debug_print_objects(int indent __hwloc_attribute_unused, hwloc_obj_t obj)
 #define hwloc_debug_print_objects(indent, obj) do { /* nothing */ } while (0)
 #endif /* !HWLOC_DEBUG */

+int hwloc_obj_set_subtype(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj, const char *subtype)
+{
+  char *new = NULL;
+  if (subtype) {
+    new = strdup(subtype);
+    if (!new)
+      return -1;
+  }
+  if (obj->subtype)
+    free(obj->subtype);
+  obj->subtype = new;
+  return 0;
+}
+
 void hwloc__free_infos(struct hwloc_info_s *infos, unsigned count)
 {
  unsigned i;
--- a/Show More
+++ b/Show More