Merge pull request #247 from Hardcode84/numba-mlir-ranbo

Add numba-mlir rambo workloads

Author: ZzEeKkAa

dpbench/benchmarks/rambo/rambo_numba_mlir_k.py

@@ -0,0 +1,32 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+from math import cos, log, pi, sin, sqrt
+
+import numba_mlir.kernel as nb
+
+
+@nb.kernel
+def _rambo(C1, F1, Q1, nout, output):
+    i = nb.get_global_id(0)
+    for j in range(nout):
+        C = 2.0 * C1[i, j] - 1.0
+        S = sqrt(1 - C * C)
+        F = 2.0 * pi * F1[i, j]
+        Q = -log(Q1[i, j])
+
+        output[i, j, 0] = Q
+        output[i, j, 1] = Q * S * sin(F)
+        output[i, j, 2] = Q * S * cos(F)
+        output[i, j, 3] = Q * C
+
+
+def rambo(nevts, nout, C1, F1, Q1, output):
+    _rambo[nevts, ()](
+        C1,
+        F1,
+        Q1,
+        nout,
+        output,
+    )

dpbench/benchmarks/rambo/rambo_numba_mlir_n.py

@@ -0,0 +1,19 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import numba_mlir as nb
+import numpy as np
+
+
+@nb.njit
+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_mlir_p.py

@@ -0,0 +1,22 @@
+# SPDX-FileCopyrightText: 2022 - 2023 Intel Corporation
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import numba
+import numba_mlir as nb
+import numpy as np
+
+
+@nb.njit
+def rambo(nevts, nout, C1, F1, Q1, output):
+    for i in numba.prange(nevts):
+        for j in numba.prange(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