merge from master

Author: rscohn2

.github/workflows/main.yml

@@ -1,6 +1,6 @@
 name: CI
 
-on: [push, pull_request]
+on: push
 
 jobs:
   build:
@@ -13,16 +13,4 @@ jobs:
         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID}}
         AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY}}
       run: |
-          python3 scripts/oneapi.py spec-venv
-          . spec-venv/bin/activate
-          python scripts/oneapi.py --branch $GITHUB_REF ci
-    - name: Archive site
-      uses: actions/upload-artifact@v1
-      with:
-          name: oneapi-spec-artifacts
-          path: site.zip
-    - name: Archive spellcheck
-      uses: actions/upload-artifact@v1
-      with:
-          name: spell check
-          path: build/spelling/output.txt
+          python3 scripts/oneapi.py --branch $GITHUB_REF ci

source/conf/common_conf.py

@@ -33,8 +33,7 @@ def path_relative_to_repo_root(relative_path):
 ]
 
 env = {
-    'oneapi_version': '0.85',
-    'l0_version': '0.95',
+    'oneapi_version': '0.9',
 }
 
 prolog_template = string.Template("""
@@ -51,7 +50,6 @@ def path_relative_to_repo_root(relative_path):
 .. |dnn_full_name| replace:: oneAPI Deep Neural Network Library
 .. |dnn_version| replace:: $oneapi_version
 .. |l0_full_name| replace:: oneAPI Level Zero
-.. |l0_version| replace:: $l0_version
 .. |tbb_full_name| replace:: oneAPI Threading Building Blocks
 .. |tbb_version| replace:: $oneapi_version
 .. |vpl_full_name| replace:: oneAPI Video Processing Library

source/contributors.rst

@@ -3,11 +3,13 @@ Contributors
 
 .. Alphbetical list of authors, | forces linebreak
 
+| Anastasiya Ageeva, Intel
 | Maksim Banin, Intel
 | Konstantin Boyarinov, Intel
 | Robert Cohn, Intel
 | Max Domeika, Intel
 | Roman Dubtsov, Intel
+| Andrey Fedorov, Intel
 | Evarist Fomenko, Intel
 | Ruslan Israfilov, Intel
 | Alexei Katranov, Intel
@@ -23,6 +25,8 @@ Contributors
 | Todd Rosenquist, Intel
 | Sally Sams, Intel
 | Sanjiv Shah, Intel
+| Mikhail Shiryaev, Intel
+| Timmie Smith, Intel
 | Mikhail Smorkalov, Intel
 | Peng Tu, Intel
 | Zack Waters, Intel

source/elements/dpcpp/source/index.rst

@@ -5,6 +5,6 @@
 .. toctree::
 
    spec/main_objects.rst
-   spec/collective_communication.rst
+   spec/operations.rst
    spec/error_handling.rst
 

source/elements/oneCCL/source/definitions.rst

@@ -1,40 +1,27 @@
 ..
   Copyright 2020 Intel Corporation
 
-==============
+=============
  Introduction
-==============
+=============
 
 The |ccl_full_name| (oneCCL) provides primitives for the
 communication patterns that occur in deep learning applications.
 oneCCL supports both scale-up for platforms with multiple
 oneAPI devices and scale-out for clusters with multiple compute nodes.
 
-oneCCL supports the following communication patterns used in
-deep learning (DL) algorithms:
+oneCCL supports the following communication patterns used in deep learning (DL) algorithms:
 
-- Allreduce
-- Allgatherv
-- Broadcast
-- Reduce
-- Alltoall
+- allgatherv
+- allreduce
+- alltoall
+- alltoallv
+- broadcast
+- reduce
+- reduce_scatter
 
-oneCCL exposes controls over additional optimizations and
-capabilities such as:
+oneCCL exposes controls over additional optimizations and capabilities such as:
 
-* User-defined pre-/post-processing of incoming buffers and reduction
-  operation
 * Prioritization for communication operations
-* Persistent communication operations (enables decoupling one-time
-  initialization and repetitive execution)
-* Fusion of multiple communication operations into the single one
+* Persistent communication operations (enables decoupling one-time initialization and repetitive execution)
 * Unordered communication operations
-* :code:`Allreduce` on sparse data
-
-Intel has published an `open source implementation`_ with the Apache
-license. The `open source implementation`_ includes a comprehensive
-test suite.  Consult the `README`_ for directions.
-
-.. _`open source implementation`: https://github.com/oneapi-src/oneCCL
-.. _`README`: https://github.com/oneapi-src/oneCCL/blob/master/README.md
-

source/elements/oneCCL/source/introduction.rst

@@ -5,5 +5,5 @@
 .. toctree::
 
    spec/generic_workflow.rst
-   spec/gpu_support.rst
-   spec/cpu_support.rst   
+   spec/host_communication.rst
+   spec/device_communication.rst