Improve tests code (#57)

* change the test_prange

* fix all tests and examples

* Replace DPPLTestCase with DPPYTestCase

* Fix typos

* To update all the tests in numba_dppy/tests/dppl

* Use explicit device selection in the examples

* Replace dppl with dppy

* Fix int64 to long long conversion on windows

* Fixed test_with_dppy_context_cpu

* Remove unused imports in tests

Co-authored-by: Pokhodenko <sergey.pokhodenko@intel.com>

Author: vlad-perevezentsev

numba_dppy/testing.py

@@ -1,39 +1,13 @@
-from __future__ import print_function, absolute_import, division
-
 import contextlib
 import sys
 
 from numba.core import config
 import unittest
 from numba.tests.support import (
     captured_stdout,
-    SerialMixin,
     redirect_c_stdout,
 )
 
-class DPPYTestCase(SerialMixin, unittest.TestCase):
-    def setUp(self):
-        #init()
-	#TODO
-        pass
-    def tearDown(self):
-        #reset()
-	#TODO
-        pass
-
-class DPPYTextCapture(object):
-    def __init__(self, stream):
-        self._stream = stream
-
-    def getvalue(self):
-        return self._stream.read()
-
-class PythonTextCapture(object):
-    def __init__(self, stream):
-        self._stream = stream
-
-    def getvalue(self):
-        return self._stream.getvalue()
 
 @contextlib.contextmanager
 def captured_dppy_stdout():

numba_dppy/tests/test_arg_accessor.py

@@ -1,10 +1,7 @@
-from __future__ import print_function, division, absolute_import
-
 import numpy as np
 
 import numba_dppy, numba_dppy as dppy
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 import dpctl
 
 
@@ -33,7 +30,7 @@ def call_kernel(global_size, local_size,
 
 
 @unittest.skipUnless(dpctl.has_cpu_queues(), 'test only on CPU system')
-class TestDPPYArgAccessorCPU(DPPYTestCase):
+class TestDPPYArgAccessorCPU(unittest.TestCase):
     def test_arg_with_accessor(self):
         C = np.ones_like(A)
         with dpctl.device_context("opencl:cpu") as cpu_queue:
@@ -50,7 +47,7 @@ def test_arg_without_accessor(self):
 
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
-class TestDPPYArgAccessorGPU(DPPYTestCase):
+class TestDPPYArgAccessorGPU(unittest.TestCase):
     def test_arg_with_accessor(self):
         C = np.ones_like(A)
         with dpctl.device_context("opencl:gpu") as gpu_queue:

numba_dppy/tests/test_arg_types.py

@@ -1,10 +1,7 @@
-from __future__ import print_function, division, absolute_import
-
 import numpy as np
 
 import numba_dppy, numba_dppy as dppy
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 import dpctl
 
 
@@ -24,7 +21,7 @@ def call_mul_device_kernel(global_size, A, B, test):
 
 
 @unittest.skipUnless(dpctl.has_cpu_queues(), 'test only on CPU system')
-class TestDPPYArrayArgCPU(DPPYTestCase):
+class TestDPPYArrayArgCPU(unittest.TestCase):
     def test_integer_arg(self):
         x = np.int32(2)
         with dpctl.device_context("opencl:cpu") as cpu_queue:
@@ -59,7 +56,7 @@ def check_bool_kernel(A, test):
 
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
-class TestDPPYArrayArgGPU(DPPYTestCase):
+class TestDPPYArrayArgGPU(unittest.TestCase):
     def test_integer_arg(self):
         x = np.int32(2)
         with dpctl.device_context("opencl:gpu") as gpu_queue:

numba_dppy/tests/test_atomic_op.py

@@ -1,11 +1,8 @@
-from __future__ import print_function, division, absolute_import
-
 import numpy as np
 
 import numba
 import numba_dppy, numba_dppy as dppy
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 import dpctl
 
 def atomic_add_int32(ary):
@@ -125,7 +122,7 @@ def call_fn_for_datatypes(fn, result, input, global_size):
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
 @unittest.skipUnless(numba_dppy.ocl.atomic_support_present(), 'test only when atomic support is present')
-class TestAtomicOp(DPPYTestCase):
+class TestAtomicOp(unittest.TestCase):
     def test_atomic_add_global(self):
         @dppy.kernel
         def atomic_add(B):

numba_dppy/tests/test_barrier.py

@@ -1,9 +1,6 @@
-from __future__ import print_function, division, absolute_import
-
 import numpy as np
 
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 from numba import float32
 import numba_dppy, numba_dppy as dppy
 import dpctl

numba_dppy/tests/test_black_scholes.py

@@ -1,12 +1,9 @@
-from __future__ import print_function, division, absolute_import
-
 import numpy as np
 import math
 import time
 
 import numba_dppy, numba_dppy as dppy
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 import dpctl
 
 
@@ -49,7 +46,7 @@ def randfloat(rand_var, low, high):
 
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
-class TestDPPYBlackScholes(DPPYTestCase):
+class TestDPPYBlackScholes(unittest.TestCase):
     def test_black_scholes(self):
         OPT_N = 400
         iterations = 2

numba_dppy/tests/test_caching.py

@@ -1,20 +1,16 @@
-from __future__ import print_function
-from timeit import default_timer as time
-
 import sys
 import numpy as np
 import numba_dppy, numba_dppy as dppy
 import dpctl
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 
 
 def data_parallel_sum(a, b, c):
     i = dppy.get_global_id(0)
     c[i] = a[i] + b[i]
 
 
-class TestCaching(DPPYTestCase):
+class TestCaching(unittest.TestCase):
     def test_caching_kernel(self):
         global_size = 10
         N = global_size

numba_dppy/tests/test_device_array_args.py

@@ -1,13 +1,9 @@
 #! /usr/bin/env python
-from __future__ import print_function
-from timeit import default_timer as time
-
 import sys
 import numpy as np
 import numba_dppy, numba_dppy as dppy
 import dpctl
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 
 @dppy.kernel
 def data_parallel_sum(a, b, c):
@@ -24,7 +20,7 @@ def data_parallel_sum(a, b, c):
 
 
 @unittest.skipUnless(dpctl.has_cpu_queues(), 'test only on CPU system')
-class TestDPPYDeviceArrayArgsGPU(DPPYTestCase):
+class TestDPPYDeviceArrayArgsGPU(unittest.TestCase):
     def test_device_array_args_cpu(self):
         c = np.ones_like(a)
 
@@ -35,7 +31,7 @@ def test_device_array_args_cpu(self):
 
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
-class TestDPPYDeviceArrayArgsCPU(DPPYTestCase):
+class TestDPPYDeviceArrayArgsCPU(unittest.TestCase):
     def test_device_array_args_gpu(self):
         c = np.ones_like(a)
 

numba_dppy/tests/test_dpctl_api.py

@@ -1,14 +1,9 @@
-from __future__ import print_function, division, absolute_import
-
-import numpy as np
-
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
 import dpctl
 
 
 @unittest.skipUnless(dpctl.has_gpu_queues(), 'test only on GPU system')
-class TestDPCTLAPI(DPPYTestCase):
+class TestDPCTLAPI(unittest.TestCase):
     def test_dpctl_api(self):
         with dpctl.device_context("opencl:gpu") as gpu_queue:
             dpctl.dump()

numba_dppy/tests/test_dpnp_functions.py

@@ -1,5 +1,4 @@
 #! /usr/bin/env python
-from __future__ import print_function
 from timeit import default_timer as time
 
 import sys
@@ -8,8 +7,8 @@
 import numba_dppy
 import numba_dppy as dppy
 import dpctl
-from numba_dppy.testing import unittest
-from numba_dppy.testing import DPPYTestCase
+import unittest
+
 
 
 def test_for_different_datatypes(fn, test_fn, dims, arg_count, tys, np_all=False, matrix=None):
@@ -83,15 +82,14 @@ def test_for_dimensions(fn, test_fn, dims, tys, np_all=False):
 
 def ensure_dpnp():
     try:
-       # import dpnp
         from numba_dppy.dpnp_glue import dpnp_fptr_interface as dpnp_glue
         return True
     except:
         return False
 
 
 @unittest.skipUnless(ensure_dpnp() and dpctl.has_gpu_queues(), 'test only when dpNP and GPU is available')
-class Testdpnp_functions(DPPYTestCase):
+class Testdpnp_functions(unittest.TestCase):
 
     N = 10