Update dpnp.isclose with scalar-specific SYCL kernels (#2540)

This PR suggests updating `dpnp.isclose()` function adding a
scalar-specific SYCL kernels for both contiguous and stride cases to
improve performance when `rtol` and `atol` are scalars.
Also extends and updates tests for `dpnp.isclose()`

The new kernel **improves performance** by **up to 10x** compared to the
previous implementation when `rtol` and `atol` are scalars (tested on
PVC).

Author: vlad-perevezentsev

CHANGELOG.md

@@ -38,6 +38,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 * Changed Windows-specific logic in dpnp initialization [#2553](https://github.com/IntelPython/dpnp/pull/2553)
 * Added missing includes to files in ufunc and VM pybind11 extensions [#2571](https://github.com/IntelPython/dpnp/pull/2571)
 * Refactored backend implementation of `dpnp.linalg.solve` to use oneMKL LAPACK `gesv` directly [#2558](https://github.com/IntelPython/dpnp/pull/2558)
+* Improved performance of `dpnp.isclose` function by implementing a dedicated kernel for scalar `rtol` and `atol` arguments [#2540](https://github.com/IntelPython/dpnp/pull/2540)
 
 ### Deprecated
 

dpnp/backend/extensions/common/ext/details/validation_utils_internal.hpp

@@ -114,8 +114,10 @@ inline void check_no_overlap(const array_ptr &input,
     }
 
     const auto &overlap = dpctl::tensor::overlap::MemoryOverlap();
+    const auto &same_logical_tensors =
+        dpctl::tensor::overlap::SameLogicalTensors();
 
-    if (overlap(*input, *output)) {
+    if (overlap(*input, *output) && !same_logical_tensors(*input, *output)) {
         throw py::value_error(name_of(input, names) +
                               " has overlapping memory segments with " +
                               name_of(output, names));

dpnp/backend/extensions/ufunc/CMakeLists.txt

@@ -37,6 +37,7 @@ set(_elementwise_sources
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/heaviside.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/i0.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/interpolate.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/isclose.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/lcm.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/ldexp.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/elementwise_functions/logaddexp2.cpp

dpnp/backend/extensions/ufunc/elementwise_functions/common.cpp

@@ -37,6 +37,7 @@
 #include "heaviside.hpp"
 #include "i0.hpp"
 #include "interpolate.hpp"
+#include "isclose.hpp"
 #include "lcm.hpp"
 #include "ldexp.hpp"
 #include "logaddexp2.hpp"
@@ -66,6 +67,7 @@ void init_elementwise_functions(py::module_ m)
     init_heaviside(m);
     init_i0(m);
     init_interpolate(m);
+    init_isclose(m);
     init_lcm(m);
     init_ldexp(m);
     init_logaddexp2(m);