Merge pull request #93 from ksenyako/release/ccl_2021.10

Intel(R) oneAPI Collective Communications Library (oneCCL) 2021.10

Author: srirajpaul

CMakeLists.txt

@@ -302,7 +302,7 @@ file(GLOB spv_kernels "${PROJECT_SOURCE_DIR}/src/kernels/kernels.spv")
 endif()
 
 set(CCL_MAJOR_VERSION     "2021")
-set(CCL_MINOR_VERSION     "9")
+set(CCL_MINOR_VERSION     "10")
 set(CCL_UPDATE_VERSION    "0")
 set(CCL_PRODUCT_STATUS    "Gold")
 string(TIMESTAMP CCL_PRODUCT_BUILD_DATE "%Y-%m-%dT %H:%M:%SZ")

INSTALL.md

@@ -5,7 +5,7 @@
 - Ubuntu* 18
 - GNU*: C, C++ 4.8.5 or higher.
 
-Refer to [System Requirements](https://software.intel.com/content/www/us/en/develop/articles/oneapi-collective-communication-library-system-requirements.html) for more details.
+Refer to [System Requirements](https://www.intel.com/content/www/us/en/developer/articles/system-requirements/oneapi-collective-communication-library-system-requirements.html) for more details.
 
 ### SYCL support <!-- omit in toc -->
 Intel(R) oneAPI DPC++/C++ Compiler with L0 v1.0 support

README.md

@@ -1,6 +1,6 @@
 # oneAPI Collective Communications Library (oneCCL) <!-- omit in toc --> <img align="right" width="100" height="100" src="https://spec.oneapi.io/oneapi-logo-white-scaled.jpg">
 
-[Installation](#installation)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Usage](#usage)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Release Notes](https://software.intel.com/content/www/us/en/develop/articles/oneapi-collective-communication-library-ccl-release-notes.html)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Documentation](https://oneapi-src.github.io/oneCCL/)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[How to Contribute](CONTRIBUTING.md)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[License](LICENSE)
+[Installation](#installation)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Usage](#usage)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Release Notes](https://www.intel.com/content/www/us/en/developer/articles/release-notes/oneapi-collective-communication-library-ccl-release-notes.html)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[Documentation](https://oneapi-src.github.io/oneCCL/)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[How to Contribute](CONTRIBUTING.md)&nbsp;&nbsp;&nbsp;|&nbsp;&nbsp;&nbsp;[License](LICENSE)
 
 oneAPI Collective Communications Library (oneCCL) provides an efficient implementation of communication patterns used in deep learning.
 
@@ -30,7 +30,7 @@ oneCCL is part of [oneAPI](https://oneapi.io).
 - Ubuntu* 18
 - GNU*: C, C++ 4.8.5 or higher.
 
-Refer to [System Requirements](https://software.intel.com/content/www/us/en/develop/articles/oneapi-collective-communication-library-system-requirements.html) for more details.
+Refer to [System Requirements](https://www.intel.com/content/www/us/en/developer/articles/system-requirements/oneapi-collective-communication-library-system-requirements.html) for more details.
 
 ### SYCL support <!-- omit in toc -->
 Intel(R) oneAPI DPC++/C++ Compiler with Level Zero v1.0 support.

deps/hwloc/include/hwloc.h

@@ -1,6 +1,6 @@
 /*
  * Copyright © 2009 CNRS
- * Copyright © 2009-2021 Inria.  All rights reserved.
+ * Copyright © 2009-2022 Inria.  All rights reserved.
  * Copyright © 2009-2012 Université Bordeaux
  * Copyright © 2009-2020 Cisco Systems, Inc.  All rights reserved.
  * See COPYING in top-level directory.
@@ -29,7 +29,7 @@
  * THAT IS IN THE PDF/HTML THAT IS ***NOT*** IN hwloc.h!
  *
  * There are entire paragraph-length descriptions, discussions, and
- * pretty prictures to explain subtle corner cases, provide concrete
+ * pretty pictures to explain subtle corner cases, provide concrete
  * examples, etc.
  *
  * Please, go read the documentation.  :-)
@@ -93,7 +93,7 @@ extern "C" {
  * Two stable releases of the same series usually have the same ::HWLOC_API_VERSION
  * even if their HWLOC_VERSION are different.
  */
-#define HWLOC_API_VERSION 0x00020500
+#define HWLOC_API_VERSION 0x00020800
 
 /** \brief Indicate at runtime which hwloc API version was used at build time.
  *
@@ -517,7 +517,7 @@ struct hwloc_obj {
                                           * objects).
                                           *
                                           * If the ::HWLOC_TOPOLOGY_FLAG_INCLUDE_DISALLOWED configuration flag is set,
-                                          * some of these CPUs may not be allowed for binding,
+                                          * some of these CPUs may be online but not allowed for binding,
                                           * see hwloc_topology_get_allowed_cpuset().
                                           *
 					  * \note All objects have non-NULL CPU and node sets except Misc and I/O objects.
@@ -549,7 +549,7 @@ struct hwloc_obj {
                                           * nodes more precisely.
                                           *
                                           * If the ::HWLOC_TOPOLOGY_FLAG_INCLUDE_DISALLOWED configuration flag is set,
-                                          * some of these nodes may not be allowed for allocation,
+                                          * some of these nodes may be online but not allowed for allocation,
                                           * see hwloc_topology_get_allowed_nodeset().
                                           *
                                           * If there are no NUMA nodes in the machine, all the memory is close to this
@@ -642,7 +642,7 @@ union hwloc_obj_attr_u {
     unsigned char revision;
     float linkspeed; /* in GB/s */
   } pcidev;
-  /** \brief Bridge specific Object Attribues */
+  /** \brief Bridge specific Object Attributes */
   struct hwloc_bridge_attr_s {
     union {
       struct hwloc_pcidev_attr_s pci;
@@ -971,7 +971,7 @@ HWLOC_DECLSPEC const char * hwloc_obj_type_string (hwloc_obj_type_t type) __hwlo
  *
  * If \p size is 0, \p string may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 HWLOC_DECLSPEC int hwloc_obj_type_snprintf(char * __hwloc_restrict string, size_t size,
@@ -986,7 +986,7 @@ HWLOC_DECLSPEC int hwloc_obj_type_snprintf(char * __hwloc_restrict string, size_
  *
  * If \p size is 0, \p string may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 HWLOC_DECLSPEC int hwloc_obj_attr_snprintf(char * __hwloc_restrict string, size_t size,
@@ -1089,7 +1089,7 @@ HWLOC_DECLSPEC int hwloc_obj_add_info(hwloc_obj_t obj, const char *name, const c
  *
  * Some operating systems only support binding threads or processes to a single PU.
  * Others allow binding to larger sets such as entire Cores or Packages or
- * even random sets of invididual PUs. In such operating system, the scheduler
+ * even random sets of individual PUs. In such operating system, the scheduler
  * is free to run the task on one of these PU, then migrate it to another PU, etc.
  * It is often useful to call hwloc_bitmap_singlify() on the target CPU set before
  * passing it to the binding function to avoid these expensive migrations.
@@ -1167,7 +1167,7 @@ typedef enum {
    * CPUs are idle, operating systems may execute the thread/process
    * on those other CPUs instead of the designated CPUs, to let them
    * progress anyway.  Strict binding means that the thread/process
-   * will _never_ execute on other cpus than the designated CPUs, even
+   * will _never_ execute on other CPUs than the designated CPUs, even
    * when those are busy with other tasks and other CPUs are idle.
    *
    * \note Depending on the operating system, strict binding may not
@@ -1204,7 +1204,7 @@ typedef enum {
   HWLOC_CPUBIND_NOMEMBIND = (1<<3)
 } hwloc_cpubind_flags_t;
 
-/** \brief Bind current process or thread on cpus given in physical bitmap \p set.
+/** \brief Bind current process or thread on CPUs given in physical bitmap \p set.
  *
  * \return -1 with errno set to ENOSYS if the action is not supported
  * \return -1 with errno set to EXDEV if the binding cannot be enforced
@@ -1219,7 +1219,7 @@ HWLOC_DECLSPEC int hwloc_set_cpubind(hwloc_topology_t topology, hwloc_const_cpus
  */
 HWLOC_DECLSPEC int hwloc_get_cpubind(hwloc_topology_t topology, hwloc_cpuset_t set, int flags);
 
-/** \brief Bind a process \p pid on cpus given in physical bitmap \p set.
+/** \brief Bind a process \p pid on CPUs given in physical bitmap \p set.
  *
  * \note \p hwloc_pid_t is \p pid_t on Unix platforms,
  * and \p HANDLE on native Windows platforms.
@@ -1250,7 +1250,7 @@ HWLOC_DECLSPEC int hwloc_set_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t
 HWLOC_DECLSPEC int hwloc_get_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_cpuset_t set, int flags);
 
 #ifdef hwloc_thread_t
-/** \brief Bind a thread \p thread on cpus given in physical bitmap \p set.
+/** \brief Bind a thread \p thread on CPUs given in physical bitmap \p set.
  *
  * \note \p hwloc_thread_t is \p pthread_t on Unix platforms,
  * and \p HANDLE on native Windows platforms.
@@ -1914,8 +1914,9 @@ HWLOC_DECLSPEC int hwloc_topology_set_components(hwloc_topology_t __hwloc_restri
 enum hwloc_topology_flags_e {
  /** \brief Detect the whole system, ignore reservations, include disallowed objects.
    *
-   * Gather all resources, even if some were disabled by the administrator.
+   * Gather all online resources, even if some were disabled by the administrator.
    * For instance, ignore Linux Cgroup/Cpusets and gather all processors and memory nodes.
+   * However offline PUs and NUMA nodes are still ignored.
    *
    * When this flag is not set, PUs and NUMA nodes that are disallowed are not added to the topology.
    * Parent objects (package, core, cache, etc.) are added only if some of their children are allowed.
@@ -2059,24 +2060,48 @@ enum hwloc_topology_flags_e {
    * not change to due thread binding changes on Windows
    * (see ::HWLOC_TOPOLOGY_FLAG_RESTRICT_TO_CPUBINDING).
    */
-  HWLOC_TOPOLOGY_FLAG_DONT_CHANGE_BINDING = (1UL<<6)
+  HWLOC_TOPOLOGY_FLAG_DONT_CHANGE_BINDING = (1UL<<6),
+
+  /** \brief Ignore distances.
+   *
+   * Ignore distance information from the operating systems (and from XML)
+   * and hence do not use distances for grouping.
+   */
+  HWLOC_TOPOLOGY_FLAG_NO_DISTANCES = (1UL<<7),
+
+  /** \brief Ignore memory attributes.
+   *
+   * Ignore memory attribues from the operating systems (and from XML).
+   */
+  HWLOC_TOPOLOGY_FLAG_NO_MEMATTRS = (1UL<<8),
+
+  /** \brief Ignore CPU Kinds.
+   *
+   * Ignore CPU kind information from the operating systems (and from XML).
+   */
+  HWLOC_TOPOLOGY_FLAG_NO_CPUKINDS = (1UL<<9)
 };
 
 /** \brief Set OR'ed flags to non-yet-loaded topology.
  *
  * Set a OR'ed set of ::hwloc_topology_flags_e onto a topology that was not yet loaded.
  *
- * If this function is called multiple times, the last invokation will erase
+ * If this function is called multiple times, the last invocation will erase
  * and replace the set of flags that was previously set.
  *
- * The flags set in a topology may be retrieved with hwloc_topology_get_flags()
+ * By default, no flags are set (\c 0).
+ *
+ * The flags set in a topology may be retrieved with hwloc_topology_get_flags().
  */
 HWLOC_DECLSPEC int hwloc_topology_set_flags (hwloc_topology_t topology, unsigned long flags);
 
 /** \brief Get OR'ed flags of a topology.
  *
  * Get the OR'ed set of ::hwloc_topology_flags_e of a topology.
  *
+ * If hwloc_topology_set_flags() was not called earlier,
+ * no flags are set (\c 0 is returned).
+ *
  * \return the flags previously set with hwloc_topology_set_flags().
  */
 HWLOC_DECLSPEC unsigned long hwloc_topology_get_flags (hwloc_topology_t topology);

deps/hwloc/include/hwloc/autogen/config.h

@@ -1,7 +1,7 @@
 /* include/hwloc/autogen/config.h.  Generated from config.h.in by configure.  */
 /* -*- c -*-
  * Copyright © 2009 CNRS
- * Copyright © 2009-2020 Inria.  All rights reserved.
+ * Copyright © 2009-2022 Inria.  All rights reserved.
  * Copyright © 2009-2012 Université Bordeaux
  * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
  * See COPYING in top-level directory.
@@ -12,11 +12,20 @@
 #ifndef HWLOC_CONFIG_H
 #define HWLOC_CONFIG_H
 
-#define HWLOC_VERSION "2.7.0rc1-git"
+#define HWLOC_VERSION "2.9.0rc2-git"
 #define HWLOC_VERSION_MAJOR 2
-#define HWLOC_VERSION_MINOR 7
+#define HWLOC_VERSION_MINOR 9
 #define HWLOC_VERSION_RELEASE 0
-#define HWLOC_VERSION_GREEK "rc1"
+#define HWLOC_VERSION_GREEK "rc2"
+
+/* #undef HWLOC_PCI_COMPONENT_BUILTIN */
+/* #undef HWLOC_OPENCL_COMPONENT_BUILTIN */
+/* #undef HWLOC_CUDA_COMPONENT_BUILTIN */
+/* #undef HWLOC_NVML_COMPONENT_BUILTIN */
+/* #undef HWLOC_RSMI_COMPONENT_BUILTIN */
+/* #undef HWLOC_LEVELZERO_COMPONENT_BUILTIN */
+/* #undef HWLOC_GL_COMPONENT_BUILTIN */
+/* #undef HWLOC_XML_LIBXML_COMPONENT_BUILTIN */
 
 #if (__GNUC__ > 2 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 95))
 # define __hwloc_restrict __restrict

deps/hwloc/include/hwloc/bitmap.h

@@ -1,6 +1,6 @@
 /*
  * Copyright © 2009 CNRS
- * Copyright © 2009-2020 Inria.  All rights reserved.
+ * Copyright © 2009-2022 Inria.  All rights reserved.
  * Copyright © 2009-2012 Université Bordeaux
  * Copyright © 2009-2011 Cisco Systems, Inc.  All rights reserved.
  * See COPYING in top-level directory.
@@ -112,7 +112,7 @@ HWLOC_DECLSPEC int hwloc_bitmap_copy(hwloc_bitmap_t dst, hwloc_const_bitmap_t sr
  *
  * If \p buflen is 0, \p buf may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 HWLOC_DECLSPEC int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap);
@@ -137,7 +137,7 @@ HWLOC_DECLSPEC int hwloc_bitmap_sscanf(hwloc_bitmap_t bitmap, const char * __hwl
  *
  * If \p buflen is 0, \p buf may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 HWLOC_DECLSPEC int hwloc_bitmap_list_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap);
@@ -161,7 +161,7 @@ HWLOC_DECLSPEC int hwloc_bitmap_list_sscanf(hwloc_bitmap_t bitmap, const char *
  *
  * If \p buflen is 0, \p buf may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 HWLOC_DECLSPEC int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap);
@@ -357,11 +357,11 @@ HWLOC_DECLSPEC int hwloc_bitmap_last_unset(hwloc_const_bitmap_t bitmap) __hwloc_
  * The loop must start with hwloc_bitmap_foreach_begin() and end
  * with hwloc_bitmap_foreach_end() followed by a terminating ';'.
  *
- * \p index is the loop variable; it should be an unsigned int.  The
- * first iteration will set \p index to the lowest index in the bitmap.
+ * \p id is the loop variable; it should be an unsigned int.  The
+ * first iteration will set \p id to the lowest index in the bitmap.
  * Successive iterations will iterate through, in order, all remaining
  * indexes set in the bitmap.  To be specific: each iteration will return a
- * value for \p index such that hwloc_bitmap_isset(bitmap, index) is true.
+ * value for \p id such that hwloc_bitmap_isset(bitmap, id) is true.
  *
  * The assert prevents the loop from being infinite if the bitmap is infinitely set.
  *

deps/hwloc/include/hwloc/deprecated.h

@@ -1,6 +1,6 @@
 /*
  * Copyright © 2009 CNRS
- * Copyright © 2009-2021 Inria.  All rights reserved.
+ * Copyright © 2009-2022 Inria.  All rights reserved.
  * Copyright © 2009-2012 Université Bordeaux
  * Copyright © 2009-2010 Cisco Systems, Inc.  All rights reserved.
  * See COPYING in top-level directory.
@@ -55,7 +55,7 @@ hwloc_topology_insert_misc_object_by_parent(hwloc_topology_t topology, hwloc_obj
  *
  * If \p size is 0, \p string may safely be \c NULL.
  *
- * \return the number of character that were actually written if not truncating,
+ * \return the number of characters that were actually written if not truncating,
  * or that would have been written (not including the ending \\0).
  */
 static __hwloc_inline int

deps/hwloc/include/hwloc/distances.h

@@ -1,5 +1,5 @@
 /*
- * Copyright © 2010-2021 Inria.  All rights reserved.
+ * Copyright © 2010-2022 Inria.  All rights reserved.
  * See COPYING in top-level directory.
  */
 
@@ -35,8 +35,8 @@ extern "C" {
  * from a core in another node.
  * The corresponding kind is ::HWLOC_DISTANCES_KIND_FROM_OS | ::HWLOC_DISTANCES_KIND_FROM_USER.
  * The name of this distances structure is "NUMALatency".
- * Others distance structures include and "XGMIBandwidth", "XGMIHops"
- * and "NVLinkBandwidth".
+ * 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.
@@ -160,7 +160,8 @@ hwloc_distances_get_by_type(hwloc_topology_t topology, hwloc_obj_type_t type,
  * Usually only one distances structure may match a given name.
  *
  * The name of the most common structure is "NUMALatency".
- * Others include "XGMIBandwidth", "XGMIHops" and "NVLinkBandwidth".
+ * Others include "XGMIBandwidth", "XGMIHops", "XeLinkBandwidth",
+ * and "NVLinkBandwidth".
  */
 HWLOC_DECLSPEC int
 hwloc_distances_get_by_name(hwloc_topology_t topology, const char *name,

deps/hwloc/include/hwloc/helper.h

@@ -1,6 +1,6 @@
 /*
  * Copyright © 2009 CNRS
- * Copyright © 2009-2021 Inria.  All rights reserved.
+ * Copyright © 2009-2022 Inria.  All rights reserved.
  * Copyright © 2009-2012 Université Bordeaux
  * Copyright © 2009-2010 Cisco Systems, Inc.  All rights reserved.
  * See COPYING in top-level directory.
@@ -886,9 +886,6 @@ enum hwloc_distrib_flags_e {
  * \p flags should be 0 or a OR'ed set of ::hwloc_distrib_flags_e.
  *
  * \note This function requires the \p roots objects to have a CPU set.
- *
- * \note This function replaces the now deprecated hwloc_distribute()
- * and hwloc_distributev() functions.
  */
 static __hwloc_inline int
 hwloc_distrib(hwloc_topology_t topology,