Dpctl

Added

* Added support of `axis=None` in `dpctl.tensor.concat` [1125](https://github.com/IntelPython/dpctl/pull/1125)
* Added caching for `dpctl.SyclDevice.filter_string` property [1127](https://github.com/IntelPython/dpctl/pull/1127)
* Added `dpctl.tensor.isdtype` from array API [1133](https://github.com/IntelPython/dpctl/pull/1133)
* Added `dpctl.tensor.unstack`, `dpctl.tensor.moveaxis`, `dpctl.tensor.swapaxes` [1137](https://github.com/IntelPython/dpctl/pull/1137), [#1174](https://github.com/IntelPython/dpctl/pull/1174)
* Allow for mutation of `dpctl.tensor.usm_ndarray.flags.writable` [1141](https://github.com/IntelPython/dpctl/pull/1141)
* Added `dpctl.tensor.where` from array API [1147](https://github.com/IntelPython/dpctl/pull/1147)
* Include libtensor headers in `dpctl` installation layout [1185](https://github.com/IntelPython/dpctl/pull/1185)
* Added new properties of `dpctl.tensor.usm_ndarray` object [1199](https://github.com/IntelPython/dpctl/pull/1199)
* Added a list of unary and binary elementwise functions from array API:
- [1203](https://github.com/IntelPython/dpctl/pull/1203): `dpctl.tensor.add`, `dpctl.tensor.divide`, `dpctl.tensor.isnan`, `dpctl.tensor.isinf`, `dpctl.tensor.isfinite`, `dpctl.tensor.cos`, `dpctl.tensor.abs`, `dpctl.tensor.equal`
- [1205](https://github.com/IntelPython/dpctl/pull/1205): `dpctl.tensor.sqrt`
- [1209](https://github.com/IntelPython/dpctl/pull/1209): implements `out` keyword argument
- [1211](https://github.com/IntelPython/dpctl/pull/1211): `dpctl.tensor.multiply`, `dpctl.tensor.subtract`
- [1214](https://github.com/IntelPython/dpctl/pull/1214): `dpctl.tensor.not_equal`
- [1216](https://github.com/IntelPython/dpctl/pull/1216): `dpctl.tensor.exp`, `dpctl.tensor.sin`
- [1217](https://github.com/IntelPython/dpctl/pull/1217): `dpctl.tensor.real`, `dpctl.tensor.imag`, `dpctl.tensor.proj`
- [1218](https://github.com/IntelPython/dpctl/pull/1218): `dpctl.tensor.log`, `dpctl.tensor.log1p`, `dpctl.tensor.expm1`
- [1221](https://github.com/IntelPython/dpctl/pull/1221): `dpctl.tensor.floor_divide`
- [1235](https://github.com/IntelPython/dpctl/pull/1235): `dpctl.tensor.less`
- [1237](https://github.com/IntelPython/dpctl/pull/1237): in-place support for addition, multiplication and subtraction
* Added `dpctl.tensor.all` and `dpctl.tensor.any` [1204](https://github.com/IntelPython/dpctl/pull/1204)
* Added `dpctl.tensor.sum` [1210](https://github.com/IntelPython/dpctl/pull/1210)

Changed

* Updated examples of native Python extensions built using `dpctl` [1108](https://github.com/IntelPython/dpctl/pull/1108)
* Used security flags to compile and link native extensions of `dpctl` [1109](https://github.com/IntelPython/dpctl/pull/1109)
* Changed types of `dpctl.tensor.finfo` and `dpctl.tensor.iinfo` output structure per array API spec [1110](https://github.com/IntelPython/dpctl/pull/1110)
* Consolidated multiple USM temporaries life-time management `host_task`s to improve test suite stability [1111](https://github.com/IntelPython/dpctl/pull/1111)
* MAINT: Improved cmake target dependency tracking [1112](https://github.com/IntelPython/dpctl/pull/1112)
* MAINT: Improved docstrings for existing `dpctl.tensor` functions [1123](https://github.com/IntelPython/dpctl/pull/1123)
* Changed default value of `mode` keyword in `dpctl.tensor.take` and `dpctl.take.put` from `clip` to `wrap` [1132](https://github.com/IntelPython/dpctl/pull/1132)
* Added support for (nested) sequence of `dpctl.tensor.usm_ndarray` objects in `dpctl.tensor.asarray` [1139](https://github.com/IntelPython/dpctl/pull/1139)
* Improved exception handling in `dpctl.tensor.usm_ndarray.__setitem__` special method [1146](https://github.com/IntelPython/dpctl/pull/1146)
* Simplified implementation of copy-and-cast kernels and removed special casing for 2D arrays to conserve binary size [1165](https://github.com/IntelPython/dpctl/pull/1165)
* Improved speed of `dpctl.tensor.usm_ndarray` printing functionality [1187](https://github.com/IntelPython/dpctl/pull/1187)
* Require DPC++ RT 2023.1 to build and run `dpctl` [1195](ttps://github.com/IntelPython/dpctl/pull/1195)
* Compile offloading native extensions with `-fno-sycl-id-queries-fit-in-int` fixing [gh-1184](https://github.com/IntelPython/dpctl/issues/1184), [#1200](https://github.com/IntelPython/dpctl/pull/1200)
* Transition to conda-forge ecosystem [1213](https://github.com/IntelPython/dpctl/pull/1213)



Fixed

* Fix to add empty values check for `dpctl.tensor.place` [1105](https://github.com/IntelPython/dpctl/pull/1105), [#1106](https://github.com/IntelPython/dpctl/pull/1106)
* Fixed [gh-1089](https://github.com/IntelPython/dpctl/issues/1089) by improving `dpctl.tensor.asarray` handling of NumPy arrays viewing into host-accessible USM allocation objects.
* MAINT: Fixed build break with newer GCC and SYCLOS [1118](https://github.com/IntelPython/dpctl/pull/1118)
* Fixed a bug in basic indexing of `dpctl.tensor.usm_ndarray` [1136](https://github.com/IntelPython/dpctl/pull/1136)

Fixed

* Fixed a bug with boolean advanced indexing [1103](https://github.com/IntelPython/dpctl/pull/1103)

Added

* Added `dpctl.SyclDevice.partition_max_sub_devices` property [1005](https://github.com/IntelPython/dpctl/pull/1005)
* Added `dpctl.program.SyclKernel.max_sub_group_size` property [1028](https://github.com/IntelPython/dpctl/pull/1028)
* Implemented printing of `usm_ndarray` [1013](https://github.com/IntelPython/dpctl/pull/1013), [#1043](https://github.com/IntelPython/dpctl/pull/1043), [#1060](https://github.com/IntelPython/dpctl/pull/1060)
* Implemented support for advanced indexing for `dpctl.tensor.usm_ndarray` [1095](https://github.com/IntelPython/dpctl/pull/1095), [#1097](https://github.com/IntelPython/dpctl/pull/1097), [#1099](https://github.com/IntelPython/dpctl/pull/1099), [#1101](https://github.com/IntelPython/dpctl/pull/1101)
* Implemented support for platform listing in `dpctl.__main__` script [1014](https://github.com/IntelPython/dpctl/pull/1014)
* Improved performance of `dpctl.tensor.asnumpy` [1026](https://github.com/IntelPython/dpctl/pull/1026)
* Added `UsmNDArray_Make*` C-API for constructing `dpctl.tensor.usm_ndarray` from native allocations [1050](https://github.com/IntelPython/dpctl/pull/1050), [#1067](https://github.com/IntelPython/dpctl/pull/1067)
* Added support for `dpctl.SyclDevice.native_vector_width_*` device descriptors [1075](https://github.com/IntelPython/dpctl/pull/1075)
* Added ``dpctl::tensor::usm_ndarray::get_shape_vector`` and ``dpctl::tensor::usm_ndarray::get_strides_vector`` methods [1090](https://github.com/IntelPython/dpctl/pull/1090)


Changed

* Removed `dpctl.select_host_device`, `dpctl.has_host_device`, `dpctl.SyclDevice.is_host`, and `dpctl.SyclDevice.has_aspect_host` since support for host device has been removed in DPC++ 2023 and from SYCL 2020 spec [1028](https://github.com/IntelPython/dpctl/pull/1028)
* `usm_ndarray`is made writable by default [1012](https://github.com/IntelPython/dpctl/pull/1012), and writable flag is now checked by `__setitem__`.
* Added convenience signature for C++ utility function in "dpctl4pybind11.hpp" [1016](https://github.com/IntelPython/dpctl/pull/1016)
* Improved error reported when attempting to submit kernel that uses a data-type unsupported by target device [1018](https://github.com/IntelPython/dpctl/pull/1018), [#1040](https://github.com/IntelPython/dpctl/pull/1040)
* Updated C++ code to require DPC++ 2023.0.0 or newer [1028](https://github.com/IntelPython/dpctl/pull/1028), [#1066](https://github.com/IntelPython/dpctl/pull/1066)
* The `dpctl.tensor.Device` class supports `print_device_info` method [1029](https://github.com/IntelPython/dpctl/pull/1029), equality comparison, and hashing [#1048](https://github.com/IntelPython/dpctl/pull/1048)
* Updated version of pybind11 used to 2.10.2 [1031](https://github.com/IntelPython/dpctl/pull/1031)
* Improved internal utility responsible for reduction of iteration space dimensionality [1044](https://github.com/IntelPython/dpctl/pull/1044/), [#1054](https://github.com/IntelPython/dpctl/pull/1054)
* Changed return type of `DCPCTLUSM_GetPointerType` function in SyclInterface library [1061](https://github.com/IntelPython/dpctl/pull/1061), [#1065](https://github.com/IntelPython/dpctl/pull/1065)
* Updated supported version of DLPack to 0.8 [1073](https://github.com/IntelPython/dpctl/pull/1073)
* Implemented queue cache per context/device pair and deployed it in `dpctl.memory`, `dpctl.tensor.from_dlpack` and `dpctl.tensor` array creation functions [1076](https://github.com/IntelPython/dpctl/pull/1076), [#1079](https://github.com/IntelPython/dpctl/pull/1079)

* Maintainance, CI work: [1001](https://github.com/IntelPython/dpctl/pull/1001), [#1009](https://github.com/IntelPython/dpctl/pull/1009), [#1011](https://github.com/IntelPython/dpctl/pull/1011), [#1024](https://github.com/IntelPython/dpctl/pull/1024), [#1030](https://github.com/IntelPython/dpctl/pull/1030), [#1032](https://github.com/IntelPython/dpctl/pull/1032), [#1035](https://github.com/IntelPython/dpctl/pull/1035), [#1037](https://github.com/IntelPython/dpctl/pull/1037), [#1039](https://github.com/IntelPython/dpctl/pull/1039), [#1041](https://github.com/IntelPython/dpctl/pull/1041), [#1045](https://github.com/IntelPython/dpctl/pull/1045), [#1047](https://github.com/IntelPython/dpctl/pull/1047), [#1055](https://github.com/IntelPython/dpctl/pull/1055), [#1057](https://github.com/IntelPython/dpctl/pull/1057), [#1059](https://github.com/IntelPython/dpctl/pull/1059), [#1068](https://github.com/IntelPython/dpctl/pull/1068), [#1070](https://github.com/IntelPython/dpctl/pull/1070), [#1074](https://github.com/IntelPython/dpctl/pull/1074), [#1077](https://github.com/IntelPython/dpctl/pull/1077), [#1078](https://github.com/IntelPython/dpctl/pull/1078), [#1081](https://github.com/IntelPython/dpctl/pull/1081), [#1084](https://github.com/IntelPython/dpctl/pull/1084), [#1085](https://github.com/IntelPython/dpctl/pull/1085), [#1088](https://github.com/IntelPython/dpctl/pull/1088), [#1086](https://github.com/IntelPython/dpctl/pull/1086), [#1092](https://github.com/IntelPython/dpctl/pull/1092), [#1093](https://github.com/IntelPython/dpctl/pull/1093)


Fixed

* Fixed error [gh-998](https://github.com/IntelPython/dpctl/issues/998) in forming Python exception, [#999](https://github.com/IntelPython/dpctl/pull/999).
* A small memory leak fixed, [1000](https://github.com/IntelPython/dpctl/pull/1000)
* Improved dtype support in `dpctl.tensor.full`, PR [1002](https://github.com/IntelPython/dpctl/pull/1002)
* Added missing header file [1008](https://github.com/IntelPython/dpctl/pull/1008) fixing [gh-1007](https://github.com/IntelPython/dpctl/issues/1007)
* Fixed a typo in device-specific dtype mapping [1015](https://github.com/IntelPython/dpctl/pull/1015)
* Fixed default device integer type to align with NumPy's behavior on Windows [1017](https://github.com/IntelPython/dpctl/pull/1017)
* Fixed unexpected overflow in `dpctl.tensor.linspace` when one of the parameters is the largest floating point value [1034](https://github.com/IntelPython/dpctl/pull/1034)
* Constructors `dpctl.tensor.empty`, `dpctl.tensor.zeros`, and `usm_ndarray` constructor itself no longer allow to create array with data-types not supported by targeted device [1042](https://github.com/IntelPython/dpctl/pull/1042)
* Fixed parameter validation in `dpctl.SyclQueue` constructor [1052](https://github.com/IntelPython/dpctl/pull/1052)
* Fixed `usm_type` of the resulting array in `dpctl.tensor.tril` and `dpctl.tensor.triu` functions [1062](https://github.com/IntelPython/dpctl/pull/1062)
* Used DPC++ configuration files to ensure correct use of conda compiler toolchain on Linux [1072](https://github.com/IntelPython/dpctl/pull/1072)
* Fixed issue with empty argument of `dpctl.tensor.meshgrid` function [1080](https://github.com/IntelPython/dpctl/pull/1080/)
* Fixed linking problem on Windows enabling `dpctl` to be functional on Windows for devices not supporting some data types [1083](https://github.com/IntelPython/dpctl/pull/1083)

Added

* Implemented `dpctl.tensor.linspace` function from array-API [875](https://github.com/IntelPython/dpctl/pull/875).
* Implemented `dpctl.tensor.eye` function from array-API [896](https://github.com/IntelPython/dpctl/pull/896).
* Implemented `dpctl.tensor.tril` and `dpctl.tensor.triu` functions from array-API [910](https://github.com/IntelPython/dpctl/pull/910).
* Added data type objects to `dpctl.tensor` namespace, `finfo`, `iinfo`, `can_cast`, and `result_type` functions [913](https://github.com/IntelPython/dpctl/pull/913).
* Implemented `dpctl.tensor.meshgrid` creation function from array-API [920](https://github.com/IntelPython/dpctl/pull/920).
* Implemented convenience class to represent output of `dpctl.tensor.usm_ndarray.flags` property [921](https://github.com/IntelPython/dpctl/pull/921).
* Added new device attributes and kernel's device-specific attributes [894](https://github.com/IntelPython/dpctl/pull/894).
* Added `dpctl.utils.onetrace_enabled` context manager for targeted trace collection [903](https://github.com/IntelPython/dpctl/pull/903).
* Added support for `stream` keyword in `__dlpack__` method, enabling support for sending `usm_ndarray` using mpi4py [906](https://github.com/IntelPython/dpctl/pull/906).
* `dpctl.tensor.asarray` can now transition data between incompatible devices, [951](https://github.com/IntelPython/dpctl/pull/951).
* Introduced `"syclinterface/dpctl_sycl_types_casters.hpp"` header file with declaration of conversion routines between SYCL type pointers and SyclInterface library opaque pointers [960](https://github.com/IntelPython/dpctl/pull/960).
* Added C-API to `dpctl.program.SyclKernel` and `dpctl.program.SyclProgram`. Added type casters for new types to "dpctl4pybind11" and added an example demonstrating its use [970](https://github.com/IntelPython/dpctl/pull/970).
* Introduced "dpctl/sycl.pxd" Cython declaration file to streamline use of SYCL functions from Cython, and added an example demonstrating its use [981](https://github.com/IntelPython/dpctl/pull/981).
* Added experimental support for sharing data allocated on sub-devices via dlpack [984](https://github.com/IntelPython/dpctl/pull/984).
* Added `dpctl.SyclDevice.sub_group_sizes` property to retrieve supported sizes of sub-group by the device [985](https://github.com/IntelPython/dpctl/pull/985).

Changed
* Improved queue compatibility testing in `dpctl.tensor`'s implementation module [900](https://github.com/IntelPython/dpctl/pull/900).
* Added automatic measurement of array-API conformance test suite in CI [901](https://github.com/IntelPython/dpctl/pull/901).
* Improved performance of array metadata transfer from host to device [912](https://github.com/IntelPython/dpctl/pull/912).
* Used `os.add_dll_directory` on Windows to ensure that `DPCTLSyclInterface` library can be found [918](https://github.com/IntelPython/dpctl/pull/918).
* Refactored `dpctl.tensor`'s implementation module [941](https://github.com/IntelPython/dpctl/pull/941) to streamline adding new functionality. Streamlined `dpctl::tensor::usm_ndarray` class implementation.
* Added debugging messaging in case when `DPCTLDynamicLib::getSymbol` encounters errors [956](https://github.com/IntelPython/dpctl/pull/956).
* Updated code base according to changes in DPC++ compiler [952](https://github.com/IntelPython/dpctl/pull/952), [#957](https://github.com/IntelPython/dpctl/pull/957), [#958](https://github.com/IntelPython/dpctl/pull/958).
* Changed `dpctl` to use pybind11 2.10.1 [967](https://github.com/IntelPython/dpctl/pull/967).
* Extended `dpctl.tensor.full` to accept 0d and higher dimensional arrays for fill-value parameter [982](https://github.com/IntelPython/dpctl/pull/982) and [#995](https://github.com/IntelPython/dpctl/pull/995).

Fixed
* Improved SyclDevice constructor error message [893](https://github.com/IntelPython/dpctl/pull/893).
* Fixed issue gh-890 about `dpctl.tensor.reshape` function [915](https://github.com/IntelPython/dpctl/pull/915).
* Fixed unexpected `UnboundLocalError` exception in [922](https://github.com/IntelPython/dpctl/pull/922).
* Fixed bugs in `dpctl.tensor.arange` in [945](https://github.com/IntelPython/dpctl/pull/945).
* Fixed issue with type inferencing in `dpctl.tensor.asarray` in [949](https://github.com/IntelPython/dpctl/pull/949).
* Added missing docstrings for `dpctl.SyclDevice` properties [964](https://github.com/IntelPython/dpctl/pull/964).

Added

* Implemented and deployed dedicated kernels for copying with casting [781](https://github.com/IntelPython/dpctl/781), used in `__setitem__`, implementaion of `asarray`, `dpctl.tensor.copy` functions.
* Implemented dedicated copying kernel for `dpctl.tensor.reshape` function [810](https://github.com/IntelPython/dpctl/810), added support for `copy` keyword [#807](https://github.com/IntelPython/dpctl/807).
* Implemented dedicated kernel to copy with casting from `numpy.ndarray` into `dpctl.tensor.usm_ndarray` [817](https://github.com/IntelPython/dpctl/pull/817).

* Implemented `dpctl.tensor.permute_dims` function from array-API [787](https://github.com/IntelPython/dpctl/pull/787).
* Implemented `dpctl.tensor.expand_dims` function from array-API [788](https://github.com/IntelPython/dpctl/pull/788).
* Implemented `dpctl.tensor.squeeze` function from array-API [790](https://github.com/IntelPython/dpctl/pull/790).
* Implemented `dpctl.tensor.broadcast_to` function from array-API [791](https://github.com/IntelPython/dpctl/791).
* Implemented `dpctl.tensor.broadcast_arrays` function from array-API [798](https://github.com/IntelPython/dpctl/798).
* Implemented `dpctl.tensor.flip` function from array-API [801](https://github.com/IntelPython/dpctl/801).
* Implemented `dpctl.tensor.usm_ndarray.mT` property per array-API [805](https://github.com/IntelPython/dpctl/805).
* Implemented `dpctl.tensor.roll` function from array-API [809](https://github.com/IntelPython/dpctl/809).
* Implemented `dpctl.tensor.arange` function from array-API [814](https://github.com/IntelPython/dpctl/814).
* Implemented `dpctl.tensor.zeros` function from array-API [816](https://github.com/IntelPython/dpctl/816).
* Implemented `dpctl.tensor.zeros` function from array-API [816](https://github.com/IntelPython/dpctl/816).
* Implemented `dpctl.tensor.ones`, `dpctl.tensor.full`, `dpctl.tensor.empty_like`, `dpctl.tensor.zeros_like`, `dpctl.tensor.ones_like`, `dpctl.tensor.full_like` functions from array-API [822](https://github.com/IntelPython/dpctl/pull/822).
* Implemented `DPCTLQueue_Memset` function in SyclInterface library [812](https://github.com/IntelPython/dpctl/812), and exposed it for `dpctl.memory.MemoryUSM*` classes [#815](https://github.com/IntelPython/dpctl/815).
* Implemented `dpctl.utils.get_coerced_usm_type` to deduced usm type of the output array from types of input arrays in compute-follows-data execution model [797](https://github.com/IntelPython/dpctl/pull/797).
* Added `dpctl.SyclDevice.profiling_timer_resolution` property [825](https://github.com/IntelPython/dpctl/pull/825).
* Added `dpctl.SyclDevice.platform` and `dpctl.SyclPlatform.default_context` properties [827](https://github.com/IntelPython/dpctl/pull/827).
* Provided pybind11 example for functions working on `dpctl.tensor.usm_ndarray` container applying oneMKL functions [780](https://github.com/IntelPython/dpctl/pull/780), [#793](https://github.com/IntelPython/dpctl/pull/793), [#819](https://github.com/IntelPython/dpctl/pull/819). The example was expanded to demonstrate implementing iterative linear solvers (Chebyshev solver, and Conjugate-Gradient solver) by asynchronously submitting individual SYCL kernels from Python [#821](https://github.com/IntelPython/dpctl/pull/821), [#833](https://github.com/IntelPython/dpctl/pull/833), [#838](https://github.com/IntelPython/dpctl/pull/838).
* Wrote manual page about working with `dpctl.SyclQueue` [829](https://github.com/IntelPython/dpctl/pull/829).
* Added cmake scripts to dpctl package layout and a way to query the location [853](https://github.com/IntelPython/dpctl/pull/853).
* Implemented `dpctl.tensor.concat` function from array-API [867](https://github.com/IntelPython/dpctl/867).
* Implemented `dpctl.tensor.stack` function from array-API [872](https://github.com/IntelPython/dpctl/872).


Changed

* Enhanced coverage collection for SyclInterface library by also collecting it during pytest run and combining traces with those collected during C-test run [818](https://github.com/IntelPython/dpctl/pull/818). This change also allows to not rebuild SyclInterface library when building C-test executable.
* Exported `keep_args_alive` utility in `dpctl4pybind11.hpp` header [820](https://github.com/IntelPython/dpctl/pull/820). The utility uses `sycl::handler::host_task` to keep given Python arguments alive until eac `sycl::event` from the given vector of events is complete. The host task is scheduled on the SYCL queue provided as the first argument.
* Changed the size of struct underlying `dpctl.SyclEvent` to avoid storing Python object previously used to keep kernel arguments scheduled with `dpctl.SyclQueue.submit` [823](https://github.com/IntelPython/dpctl/pull/823).
* Fixed docstring for `dpctl.SyclTimer` [824](https://github.com/IntelPython/dpctl/pull/824).
* Changed type of exceptions raised on failure to create `dpctl.SyclDevice` from `ValueError` to `dpctl.SyclDeviceCreationError` [826](https://github.com/IntelPython/dpctl/pull/826).
* Improved performance of pybind11 type casters [837](https://github.com/IntelPython/dpctl/pull/837).
* Changed implementation of `dpctl.SyclProgram` from using deprecated `sycl::program` to `sycl::kernel_bundle` [845](https://github.com/IntelPython/dpctl/pull/845).
* Removed deprecated device aspects, added new supported aspects [844](https://github.com/IntelPython/dpctl/pull/844).
* Updated vendored `dlpack.h` to version 0.7 [847](https://github.com/IntelPython/dpctl/pull/847).

Fixed

* Fixed `dpctl.lsplatform()` to work correctly when used from within Jupyter notebook [800](https://github.com/IntelPython/dpctl/pull/800).
* Fixed script to drive debug build [835](https://github.com/IntelPython/dpctl/pull/835) and fixed code to compile in debug mode [#836](https://github.com/IntelPython/dpctl/pull/836).
* Fixed filter selector string produced in outputs of `dpctl.lsplatform(verbosity=2)` and `dpctl.SyclDevice.print_device_info` [866](https://github.com/IntelPython/dpctl/pull/866).
* Fixed issue with slicing reported in gh-870 in [871](https://github.com/IntelPython/dpctl/pull/871).

Added

* Properties added to MemoryUSM* objects. [647](https://github.com/IntelPython/dpctl/pull/647)
* Added `dpctl.tensor.asarray` [646](https://github.com/IntelPython/dpctl/pull/646)
* Implemented DLPack support for usm_ndarray [682](https://github.com/IntelPython/dpctl/pull/682)
* Exported `dpctl.tensor.Device` class [708](https://github.com/IntelPython/dpctl/pull/708) [#718](https://github.com/IntelPython/dpctl/pull/718)
* Added testing of examples in CI [722](https://github.com/IntelPython/dpctl/pull/722)
* Added user manuals to dpctl documentation [712](https://github.com/IntelPython/dpctl/pull/712) [#773](https://github.com/IntelPython/dpctl/pull/773)

Changed

* Folder dpctl-capi/ renamed to libsyclinterface/ in sources and documentation. [666](https://github.com/IntelPython/dpctl/pull/666)
[768](https://github.com/IntelPython/dpctl/pull/768)
* Added workflow to publish rendered documentation on PRs [673](https://github.com/IntelPython/dpctl/pull/673) [#753](https://github.com/IntelPython/dpctl/pull/753) [#726](https://github.com/IntelPython/dpctl/pull/726)
* Synchronization functions and USM allocation functions release GIL [736](https://github.com/IntelPython/dpctl/pull/736) [#766](https://github.com/IntelPython/dpctl/pull/766)
* `dpctl.SyclEvent` destructor is made non-blocking [751](https://github.com/IntelPython/dpctl/pull/751)

Fixed
* Fixed for issue in code of `dpctl.tensor.usm_ndarray.T` [653](https://github.com/IntelPython/dpctl/pull/653)
* Fixed issue with `dpctl.tensor.reshape`'s affect on contiguity flags of usm_ndarray [695](https://github.com/IntelPython/dpctl/pull/695)
* Fixed handling of empty list by `dpctl.tensor.asarray` [694](https://github.com/IntelPython/dpctl/pull/694)
* Fixed type inference with array of empty arrays in `dpctl.tensor.asarray` [697](https://github.com/IntelPython/dpctl/pull/697)
* Fixed issue gh-698 with `dpctl.tensr.asarray` [709](https://github.com/IntelPython/dpctl/pull/709)
* Fixed performance of item assignment from numpy array [724](https://github.com/IntelPython/dpctl/pull/724)
* `DPCTLDeviceMgr_GetNumDevices` should not operate on rejected devices [737](https://github.com/IntelPython/dpctl/pull/737)
* Fixed issue gh-729 for `dpctl.tensor.reshape` applied to 0-element usm_ndarray [756](https://github.com/IntelPython/dpctl/pull/756)
* Fixed issue gh-728 with `dpctl.tensor.astype` [757](https://github.com/IntelPython/dpctl/pull/757)
* Fixed type in memory overlapping test [770](https://github.com/IntelPython/dpctl/pull/770)
* Fixed issue with operator.pos for `dpctl.tensor.usm_ndarray` [783](https://github.com/IntelPython/dpctl/pull/783)
* Only call `PyThread_Ensure` from host_task if the main-thread interpreter is initialized and not finalizing [776](https://github.com/IntelPython/dpctl/pull/776) [#778](https://github.com/IntelPython/dpctl/pull/778) [#721](https://github.com/IntelPython/dpctl/pull/721)

**Full Changelog**: https://github.com/IntelPython/dpctl/compare/0.11.4...0.12.0