Merge branch 'master' into dpctl_c_api-constructors

Author: oleksandr-pavlyk

.flake8

@@ -23,6 +23,7 @@ per-file-ignores =
     dpctl/memory/_memory.pyx: E999, E225, E226, E227
     dpctl/program/_program.pyx: E999, E225, E226, E227
     dpctl/tensor/_usmarray.pyx: E999, E225, E226, E227
+    dpctl/tensor/_dlpack.pyx: E999, E225, E226, E227
     dpctl/tensor/numpy_usm_shared.py: F821
     dpctl/tests/_cython_api.pyx: E999, E225, E227, E402
     dpctl/utils/_compute_follows_data.pyx: E999, E225, E227

.github/workflows/conda-package.yml

@@ -215,7 +215,7 @@ jobs:
 
   upload_linux:
     needs: test_linux
-    if: ${{github.ref == 'refs/heads/master' || (startsWith(github.ref, 'refs/heads/release') == true)}}
+    if: ${{github.ref == 'refs/heads/master' || (startsWith(github.ref, 'refs/heads/release') == true) || github.event_name == 'push' && contains(github.ref, 'refs/tags/')}}
     runs-on: ubuntu-latest
     strategy:
       matrix:
@@ -240,7 +240,7 @@ jobs:
 
   upload_windows:
     needs: test_windows
-    if: ${{github.ref == 'refs/heads/master' || (startsWith(github.ref, 'refs/heads/release') == true)}}
+    if: ${{github.ref == 'refs/heads/master' || (startsWith(github.ref, 'refs/heads/release') == true) || github.event_name == 'push' && contains(github.ref, 'refs/tags/')}}
     runs-on: windows-latest
     strategy:
       matrix:

.github/workflows/os-llvm-sycl-build.yml

@@ -10,8 +10,8 @@ jobs:
     runs-on: ubuntu-20.04
 
     env:
-      OCLCPUEXP_FN: oclcpuexp-2021.12.6.0.19_rel.tar.gz
-      FPGAEMU_FN: fpgaemu-2021.12.6.0.19_rel.tar.gz
+      OCLCPUEXP_FN: oclcpuexp-2021.12.9.0.24_rel.tar.gz
+      FPGAEMU_FN: fpgaemu-2021.12.9.0.24_rel.tar.gz
       TBB_FN: oneapi-tbb-2021.4.0-lin.tgz
 
     steps:
@@ -48,8 +48,8 @@ jobs:
               export DOWNLOAD_URL_PREFIX=https://github.com/intel/llvm/releases/download
               rm -rf dpcpp-compiler.tar.gz
               wget ${DOWNLOAD_URL_PREFIX}/${NIGHTLY_TAG}/dpcpp-compiler.tar.gz && echo ${LATEST_LLVM_TAG_SHA} > bundle_id.txt || rm -rf bundle_id.txt
-              [ -f ${OCLCPUEXP_FN} ] || wget ${DOWNLOAD_URL_PREFIX}/2021-07/${OCLCPUEXP_FN} || rm -rf bundle_id.txt
-              [ -f ${FPGAEMU_FN} ] || wget ${DOWNLOAD_URL_PREFIX}/2021-07/${FPGAEMU_FN} || rm -rf bundle_id.txt
+              [ -f ${OCLCPUEXP_FN} ] || wget ${DOWNLOAD_URL_PREFIX}/2021-09/${OCLCPUEXP_FN} || rm -rf bundle_id.txt
+              [ -f ${FPGAEMU_FN} ] || wget ${DOWNLOAD_URL_PREFIX}/2021-09/${FPGAEMU_FN} || rm -rf bundle_id.txt
               [ -f ${TBB_FN} ] || wget https://github.com/oneapi-src/oneTBB/releases/download/v2021.4.0/${TBB_FN} || rm -rf bundle_id.txt
               rm -rf dpcpp_compiler
               tar xf dpcpp-compiler.tar.gz

CHANGELOG.md

@@ -4,10 +4,29 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [Next Release] - TBD
+## [Unreleased]
 
+### Added
+- `dpctl.tensor.asarray`, `dpctl.tensor.empty` implemented (#646).
+- `dpctl.tensor.usm_ndarray` adds support for DLPack protocol. `dpctl.tensor.from_dlpack` implemented (#682).
+
+### Changed
 - dpctl-capi is now renamed to `libsyclinterface` (#666).
 
+## [0.11.3] - 11/30/2021
+
+### Fixed
+
+- Set the last byte in allocated char array to zero [cherry picked from #650] (#699)
+
+## [0.11.2] - 11/29/2021
+
+### Added
+- Extending `dpctl.device_context` with nested contexts (#678)
+
+### Fixed
+- Fixed issue #649 about incorrect behavior of `.T` method on sliced arrays (#653)
+
 ## [0.11.1] - 11/10/2021
 
 ### Changed

MANIFEST.in

@@ -14,5 +14,6 @@ include dpctl/memory/_memory.h
 include dpctl/memory/_memory_api.h
 include dpctl/tensor/_usmarray.h
 include dpctl/tensor/_usmarray_api.h
+recursive-include dpctl/tensor/include *
 include dpctl/tests/input_files/*
 include dpctl/tests/*.pyx

dpctl/__init__.py

@@ -61,6 +61,7 @@
     get_current_queue,
     get_num_activated_queues,
     is_in_device_context,
+    nested_context_factories,
     set_global_queue,
 )
 
@@ -111,6 +112,7 @@
     "get_current_queue",
     "get_num_activated_queues",
     "is_in_device_context",
+    "nested_context_factories",
     "set_global_queue",
 ]
 __all__ += [

dpctl/_sycl_queue_manager.pyx

@@ -19,7 +19,7 @@
 # cython: linetrace=True
 
 import logging
-from contextlib import contextmanager
+from contextlib import ExitStack, contextmanager
 
 from .enum_types import backend_type, device_type
 
@@ -210,6 +210,22 @@ cpdef get_current_backend():
     return _mgr.get_current_backend()
 
 
+nested_context_factories = []
+
+
+def _get_nested_contexts(ctxt):
+    _help_numba_dppy()
+    return (factory(ctxt) for factory in nested_context_factories)
+
+
+def _help_numba_dppy():
+    """Import numba-dppy for registering nested contexts"""
+    try:
+        import numba_dppy
+    except Exception:
+        pass
+
+
 @contextmanager
 def device_context(arg):
     """
@@ -223,6 +239,9 @@ def device_context(arg):
     the context manager's scope. The yielded queue is removed as the currently
     usable queue on exiting the context manager.
 
+    You can register context factory in the list of factories.
+    This context manager uses context factories to create and activate nested contexts.
+
     Args:
         arg : A :class:`dpctl.SyclQueue` object, or a :class:`dpctl.SyclDevice`
               object, or a filter selector string.
@@ -244,11 +263,26 @@ def device_context(arg):
             with dpctl.device_context("level0:gpu:0"):
                 do_something_on_gpu0()
 
+        The following example registers nested context factory:
+
+        .. code-block:: python
+
+            import dctl
+
+            def factory(sycl_queue):
+                ...
+                return context
+
+            dpctl.nested_context_factories.append(factory)
+
     """
     ctxt = None
     try:
         ctxt = _mgr._set_as_current_queue(arg)
-        yield ctxt
+        with ExitStack() as stack:
+            for nested_context in _get_nested_contexts(ctxt):
+                stack.enter_context(nested_context)
+            yield ctxt
     finally:
         # Code to release resource
         if ctxt:

dpctl/tensor/__init__.py

@@ -33,15 +33,20 @@
 from dpctl.tensor._copy_utils import copy_from_numpy as from_numpy
 from dpctl.tensor._copy_utils import copy_to_numpy as asnumpy
 from dpctl.tensor._copy_utils import copy_to_numpy as to_numpy
+from dpctl.tensor._ctors import asarray, empty
+from dpctl.tensor._dlpack import from_dlpack
 from dpctl.tensor._reshape import reshape
 from dpctl.tensor._usmarray import usm_ndarray
 
 __all__ = [
     "usm_ndarray",
+    "asarray",
     "astype",
     "copy",
+    "empty",
     "reshape",
     "from_numpy",
     "to_numpy",
     "asnumpy",
+    "from_dlpack",
 ]

dpctl/tensor/_copy_utils.py

@@ -93,12 +93,12 @@ def copy_to_numpy(ary):
     )
 
 
-def copy_from_numpy(np_ary, usm_type="device", queue=None):
+def copy_from_numpy(np_ary, usm_type="device", sycl_queue=None):
     "Copies numpy array `np_ary` into a new usm_ndarray"
     # This may peform a copy to meet stated requirements
     Xnp = np.require(np_ary, requirements=["A", "O", "C", "E"])
-    if queue:
-        ctor_kwargs = {"queue": queue}
+    if sycl_queue:
+        ctor_kwargs = {"queue": sycl_queue}
     else:
         ctor_kwargs = dict()
     Xusm = dpt.usm_ndarray(