Merge pull request #196 from IntelPython/rambo_dpnp_numpy

Adding implementations of Rambo using numpy-like vector operations

dpbench/benchmarks/rambo/rambo_dpnp.py

@@ -0,0 +1,17 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import dpnp as np
+
+
+def rambo(nevts, nout, C1, F1, Q1, output):
+    C = 2.0 * C1 - 1.0
+    S = np.sqrt(1 - np.square(C))
+    F = 2.0 * np.pi * F1
+    Q = -np.log(Q1)
+
+    output[:, :, 0] = Q
+    output[:, :, 1] = Q * S * np.sin(F)
+    output[:, :, 2] = Q * S * np.cos(F)
+    output[:, :, 3] = Q * C

dpbench/benchmarks/rambo/rambo_numba_np.py

@@ -0,0 +1,19 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import numba as nb
+import numpy as np
+
+
+@nb.njit(parallel=True, fastmath=True)
+def rambo(nevts, nout, C1, F1, Q1, output):
+    C = 2.0 * C1 - 1.0
+    S = np.sqrt(1 - np.square(C))
+    F = 2.0 * np.pi * F1
+    Q = -np.log(Q1)
+
+    output[:, :, 0] = Q
+    output[:, :, 1] = Q * S * np.sin(F)
+    output[:, :, 2] = Q * S * np.cos(F)
+    output[:, :, 3] = Q * C

dpbench/benchmarks/rambo/rambo_numpy.py

@@ -6,14 +6,12 @@
 
 
 def rambo(nevts, nout, C1, F1, Q1, output):
-    for i in range(nevts):
-        for j in range(nout):
-            C = 2.0 * C1[i, j] - 1.0
-            S = np.sqrt(1 - np.square(C))
-            F = 2.0 * np.pi * F1[i, j]
-            Q = -np.log(Q1[i, j])
+    C = 2.0 * C1 - 1.0
+    S = np.sqrt(1 - np.square(C))
+    F = 2.0 * np.pi * F1
+    Q = -np.log(Q1)
 
-            output[i, j, 0] = Q
-            output[i, j, 1] = Q * S * np.sin(F)
-            output[i, j, 2] = Q * S * np.cos(F)
-            output[i, j, 3] = Q * C
+    output[:, :, 0] = Q
+    output[:, :, 1] = Q * S * np.sin(F)
+    output[:, :, 2] = Q * S * np.cos(F)
+    output[:, :, 3] = Q * C

dpbench/benchmarks/rambo/rambo_python.py

@@ -0,0 +1,19 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import numpy as np
+
+
+def rambo(nevts, nout, C1, F1, Q1, output):
+    for i in range(nevts):
+        for j in range(nout):
+            C = 2.0 * C1[i, j] - 1.0
+            S = np.sqrt(1 - np.square(C))
+            F = 2.0 * np.pi * F1[i, j]
+            Q = -np.log(Q1[i, j])
+
+            output[i, j, 0] = Q
+            output[i, j, 1] = Q * S * np.sin(F)
+            output[i, j, 2] = Q * S * np.cos(F)
+            output[i, j, 3] = Q * C