summaryrefslogtreecommitdiff
path: root/numpy/core
diff options
context:
space:
mode:
Diffstat (limited to 'numpy/core')
-rw-r--r--numpy/core/src/umath/simd.inc.src76
-rw-r--r--numpy/core/tests/test_umath.py9
2 files changed, 52 insertions, 33 deletions
diff --git a/numpy/core/src/umath/simd.inc.src b/numpy/core/src/umath/simd.inc.src
index 012f7c076..c2fc069c6 100644
--- a/numpy/core/src/umath/simd.inc.src
+++ b/numpy/core/src/umath/simd.inc.src
@@ -2530,6 +2530,7 @@ static NPY_INLINE NPY_GCC_OPT_3 NPY_GCC_TARGET_@ISA@ void
* #vtype = __m256, __m512#
* #vsize = 256, 512#
* #BYTES = 32, 64#
+ * #NUM_LANES = 8, 16#
* #mask = __m256, __mmask16#
* #vsub = , _mask#
* #or_masks =_mm256_or_ps, _mm512_kor#
@@ -2573,7 +2574,7 @@ static NPY_GCC_OPT_3 NPY_GCC_TARGET_@ISA@ void
NPY_TRIG_OP my_trig_op)
{
const npy_intp stride = steps/(npy_intp)sizeof(npy_float);
- const npy_int num_lanes = @BYTES@/(npy_intp)sizeof(npy_float);
+ const npy_int num_lanes = @NUM_LANES@;
npy_float large_number = 71476.0625f;
if (my_trig_op == npy_compute_sin) {
large_number = 117435.992f;
@@ -2622,12 +2623,12 @@ static NPY_GCC_OPT_3 NPY_GCC_TARGET_@ISA@ void
num_lanes);
}
- @vtype@ x;
+ @vtype@ x_in;
if (stride == 1) {
- x = @isa@_masked_load_ps(load_mask, ip);
+ x_in = @isa@_masked_load_ps(load_mask, ip);
}
else {
- x = @isa@_masked_gather_ps(zero_f, ip, vindex, load_mask);
+ x_in = @isa@_masked_gather_ps(zero_f, ip, vindex, load_mask);
}
/*
@@ -2636,10 +2637,10 @@ static NPY_GCC_OPT_3 NPY_GCC_TARGET_@ISA@ void
* these numbers
*/
- glibc_mask = @isa@_in_range_mask(x, large_number,-large_number);
+ glibc_mask = @isa@_in_range_mask(x_in, large_number,-large_number);
glibc_mask = @and_masks@(load_mask, glibc_mask);
- nan_mask = _mm@vsize@_cmp_ps@vsub@(x, x, _CMP_NEQ_UQ);
- x = @isa@_set_masked_lanes_ps(x, zero_f, @or_masks@(nan_mask, glibc_mask));
+ nan_mask = _mm@vsize@_cmp_ps@vsub@(x_in, x_in, _CMP_NEQ_UQ);
+ @vtype@ x = @isa@_set_masked_lanes_ps(x_in, zero_f, @or_masks@(nan_mask, glibc_mask));
npy_int iglibc_mask = @mask_to_int@(glibc_mask);
if (iglibc_mask != @full_mask@) {
@@ -2678,20 +2679,25 @@ static NPY_GCC_OPT_3 NPY_GCC_TARGET_@ISA@ void
}
/* process elements using glibc for large elements */
- if (my_trig_op == npy_compute_cos) {
- for (int ii = 0, jj = 0; iglibc_mask != 0; ii++, jj += stride) {
- if (iglibc_mask & 0x01) {
- op[ii] = npy_cosf(ip[jj]);
+ if (NPY_UNLIKELY(iglibc_mask != 0)) {
+ float NPY_DECL_ALIGNED(@BYTES@) ip_fback[@NUM_LANES@];
+ _mm@vsize@_store_ps(ip_fback, x_in);
+
+ if (my_trig_op == npy_compute_cos) {
+ for (int ii = 0; iglibc_mask != 0; ++ii) {
+ if (iglibc_mask & 0x01) {
+ op[ii] = npy_cosf(ip_fback[ii]);
+ }
+ iglibc_mask = iglibc_mask >> 1;
}
- iglibc_mask = iglibc_mask >> 1;
}
- }
- else {
- for (int ii = 0, jj = 0; iglibc_mask != 0; ii++, jj += stride) {
- if (iglibc_mask & 0x01) {
- op[ii] = npy_sinf(ip[jj]);
+ else {
+ for (int ii = 0; iglibc_mask != 0; ++ii) {
+ if (iglibc_mask & 0x01) {
+ op[ii] = npy_sinf(ip_fback[ii]);
+ }
+ iglibc_mask = iglibc_mask >> 1;
}
- iglibc_mask = iglibc_mask >> 1;
}
}
ip += num_lanes*stride;
@@ -3203,7 +3209,7 @@ AVX512F_log_DOUBLE(npy_double * op,
__mmask8 inf_mask, nan_mask, zero_mask, negx_mask, denormal_mask,
glibc_mask;
- __m512d x;
+ __m512d x_in;
while (num_remaining_elements > 0) {
if (num_remaining_elements < num_lanes) {
load_mask = avx512_get_partial_load_mask_pd(num_remaining_elements,
@@ -3211,27 +3217,27 @@ AVX512F_log_DOUBLE(npy_double * op,
}
if (1 == stride) {
- x = avx512_masked_load_pd(load_mask, ip);
+ x_in = avx512_masked_load_pd(load_mask, ip);
}
else {
- x = avx512_masked_gather_pd(zeros_d, ip, vindex, load_mask);
+ x_in = avx512_masked_gather_pd(zeros_d, ip, vindex, load_mask);
}
/* call glibc when x on [1.0 - 0x1p-4, 1.0 + 0x1.09p-4] */
- __mmask8 m1 = _mm512_cmp_pd_mask(x, mTo_glibc_max, _CMP_LT_OQ);
- __mmask8 m2 = _mm512_cmp_pd_mask(x, mTo_glibc_min, _CMP_GT_OQ);
+ __mmask8 m1 = _mm512_cmp_pd_mask(x_in, mTo_glibc_max, _CMP_LT_OQ);
+ __mmask8 m2 = _mm512_cmp_pd_mask(x_in, mTo_glibc_min, _CMP_GT_OQ);
glibc_mask = m1 & m2;
if (glibc_mask != 0xFF) {
- zero_mask = _mm512_cmp_pd_mask(x, zeros_d, _CMP_EQ_OQ);
- inf_mask = _mm512_cmp_pd_mask(x, mInf, _CMP_EQ_OQ);
- negx_mask = _mm512_cmp_pd_mask(x, zeros_d, _CMP_LT_OQ);
- nan_mask = _mm512_cmp_pd_mask(x, x, _CMP_NEQ_UQ);
+ zero_mask = _mm512_cmp_pd_mask(x_in, zeros_d, _CMP_EQ_OQ);
+ inf_mask = _mm512_cmp_pd_mask(x_in, mInf, _CMP_EQ_OQ);
+ negx_mask = _mm512_cmp_pd_mask(x_in, zeros_d, _CMP_LT_OQ);
+ nan_mask = _mm512_cmp_pd_mask(x_in, x_in, _CMP_NEQ_UQ);
divide_by_zero_mask = divide_by_zero_mask | (zero_mask & load_mask);
invalid_mask = invalid_mask | negx_mask;
- x = avx512_set_masked_lanes_pd(x, zeros_d, negx_mask);
+ __m512d x = avx512_set_masked_lanes_pd(x_in, zeros_d, negx_mask);
__m512i ix = (__m512i)x;
/* Normalize x when it is denormal */
@@ -3303,13 +3309,17 @@ AVX512F_log_DOUBLE(npy_double * op,
}
/* call glibc's log func when x around 1.0f */
- for (int ii = 0, jj = 0; glibc_mask != 0; ii++, jj += stride) {
- if (glibc_mask & 0x01) {
- op[ii] = npy_log(ip[jj]);
+ if (NPY_UNLIKELY(glibc_mask != 0)) {
+ double NPY_DECL_ALIGNED(64) ip_fback[8];
+ _mm512_store_pd(ip_fback, x_in);
+
+ for (int ii = 0; glibc_mask != 0; ++ii) {
+ if (glibc_mask & 0x01) {
+ op[ii] = npy_log(ip_fback[ii]);
+ }
+ glibc_mask = glibc_mask >> 1;
}
- glibc_mask = glibc_mask >> 1;
}
-
ip += num_lanes * stride;
op += num_lanes;
num_remaining_elements -= num_lanes;
diff --git a/numpy/core/tests/test_umath.py b/numpy/core/tests/test_umath.py
index 818b2ad6c..f57493e9c 100644
--- a/numpy/core/tests/test_umath.py
+++ b/numpy/core/tests/test_umath.py
@@ -657,6 +657,11 @@ class TestLog:
yf = np.array(y, dtype=dt)*log2_
assert_almost_equal(np.log(xf), yf)
+ # test aliasing(issue #17761)
+ x = np.array([2, 0.937500, 3, 0.947500, 1.054697])
+ xf = np.log(x)
+ assert_almost_equal(np.log(x, out=x), xf)
+
def test_log_strides(self):
np.random.seed(42)
strides = np.array([-4,-3,-2,-1,1,2,3,4])
@@ -896,6 +901,10 @@ class TestAVXFloat32Transcendental:
x_f64 = np.float64(x_f32)
assert_array_max_ulp(np.sin(x_f32), np.float32(np.sin(x_f64)), maxulp=2)
assert_array_max_ulp(np.cos(x_f32), np.float32(np.cos(x_f64)), maxulp=2)
+ # test aliasing(issue #17761)
+ tx_f32 = x_f32.copy()
+ assert_array_max_ulp(np.sin(x_f32, out=x_f32), np.float32(np.sin(x_f64)), maxulp=2)
+ assert_array_max_ulp(np.cos(tx_f32, out=tx_f32), np.float32(np.cos(x_f64)), maxulp=2)
def test_strided_float32(self):
np.random.seed(42)
View Lorry Controller Status