TESTS PERF add black scholes test (#499)

* TESTS PERF add black scholes test

* TESTS PERF avoid warming up

Author: densmirn

tests/tests_perf/math_tests/test_black_scholes.py

@@ -0,0 +1,57 @@
+import math
+import numpy
+import pytest
+
+from tests.tests_perf.data_generator import *
+from tests.tests_perf.test_perf_base import DPNPTestPerfBase
+
+
+SEED = 7777777
+SL, SH = 10.0, 50.0
+KL, KH = 10.0, 50.0
+TL, TH = 1.0, 2.0
+RISK_FREE = 0.1
+VOLATILITY = 0.2
+
+
+def math_erf(x):
+    result = numpy.empty(x.shape, dtype=x.dtype)
+    for i in range(result.size):
+        result[i] = math.erf(x[i])
+
+    return result
+
+
+# module 'numpy' has no attribute 'erf'
+numpy.erf = math_erf
+
+
+def gen_data(low, high, size):
+    return numpy.random.uniform(low, high, size)
+
+
+def black_scholes_put(lib, S, K, T, r, sigma):
+    d1 = (lib.log(S/K) + (r + sigma*sigma/2.)*T) / sigma*lib.sqrt(T)
+    d2 = d1 - sigma * lib.sqrt(T)
+
+    cdf_d1 = (1 + lib.erf(d1 / lib.sqrt(2))) / 2
+    cdf_d2 = (1 + lib.erf(d2 / lib.sqrt(2))) / 2
+
+    bs_call = S*cdf_d1 - K*lib.exp(-r*T)*cdf_d2
+
+    return K*lib.exp(-r*T) - S + bs_call
+
+
+class TestBlackScholes(DPNPTestPerfBase):
+
+    @pytest.mark.parametrize("dtype", [numpy.float64])
+    @pytest.mark.parametrize("size", [1024, 2048, 4096, 8192])
+    def test_bs_put(self, lib, dtype, size):
+        numpy.random.seed(SEED)
+        S = gen_data(SL, SH, size)
+        K = gen_data(KL, KH, size)
+        T = gen_data(TL, TH, size)
+
+        self.dpnp_benchmark("bs_put", lib, dtype, size,
+                            lib, S, K, T, RISK_FREE, VOLATILITY,
+                            custom_fptr=black_scholes_put)