aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON/kernels
diff options
context:
space:
mode:
authorAnthony Barbier <anthony.barbier@arm.com>2017-07-03 17:39:37 +0100
committerAnthony Barbier <anthony.barbier@arm.com>2018-09-17 14:15:39 +0100
commitac69aa137e360340fe9f148f019d93af6c3d8336 (patch)
tree88c2fb6ea8693c69b19d3b7b38fe26cd916303c8 /src/core/NEON/kernels
parent05da6dd102c988081c7d5eccb227f559f740ceef (diff)
downloadComputeLibrary-ac69aa137e360340fe9f148f019d93af6c3d8336.tar.gz
COMPMID-418 Add check and fix comments after preprocessor conditions
Change-Id: I1353fd652ee180e3931e58b4ce13d651a48c7e2c Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79567 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
Diffstat (limited to 'src/core/NEON/kernels')
-rw-r--r--src/core/NEON/kernels/NEAccumulateKernel.cpp2
-rw-r--r--src/core/NEON/kernels/NEBox3x3Kernel.cpp2
-rw-r--r--src/core/NEON/kernels/NECannyEdgeKernel.cpp2
-rw-r--r--src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp4
-rw-r--r--src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp18
-rw-r--r--src/core/NEON/kernels/NEHarrisCornersKernel.cpp2
-rw-r--r--src/core/NEON/kernels/NEIm2ColKernel.cpp4
-rw-r--r--src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp4
-rw-r--r--src/core/NEON/kernels/NEMagnitudePhaseKernel.cpp2
-rw-r--r--src/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.cpp2
10 files changed, 21 insertions, 21 deletions
diff --git a/src/core/NEON/kernels/NEAccumulateKernel.cpp b/src/core/NEON/kernels/NEAccumulateKernel.cpp
index e5b933a781..6e54dd64a3 100644
--- a/src/core/NEON/kernels/NEAccumulateKernel.cpp
+++ b/src/core/NEON/kernels/NEAccumulateKernel.cpp
@@ -131,7 +131,7 @@ void NEAccumulateWeightedFP16Kernel::run(const Window &window)
},
input, accum);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
namespace
{
diff --git a/src/core/NEON/kernels/NEBox3x3Kernel.cpp b/src/core/NEON/kernels/NEBox3x3Kernel.cpp
index d7e6d73cd7..551c903dd9 100644
--- a/src/core/NEON/kernels/NEBox3x3Kernel.cpp
+++ b/src/core/NEON/kernels/NEBox3x3Kernel.cpp
@@ -103,7 +103,7 @@ void NEBox3x3FP16Kernel::run(const Window &window)
},
input, output);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
BorderSize NEBox3x3Kernel::border_size() const
{
diff --git a/src/core/NEON/kernels/NECannyEdgeKernel.cpp b/src/core/NEON/kernels/NECannyEdgeKernel.cpp
index 85a2cd5855..26df6f6b8b 100644
--- a/src/core/NEON/kernels/NECannyEdgeKernel.cpp
+++ b/src/core/NEON/kernels/NECannyEdgeKernel.cpp
@@ -787,7 +787,7 @@ void NEGradientFP16Kernel::configure(const ITensor *gx, const ITensor *gy, ITens
INEKernel::configure(win);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
namespace
{
diff --git a/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp b/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp
index 7d659ab2e6..57d2807b8a 100644
--- a/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp
+++ b/src/core/NEON/kernels/NEGEMMMatrixAdditionKernel.cpp
@@ -89,7 +89,7 @@ void matrix_addition_f16(const ITensor *input, ITensor *output, const Window &wi
},
in, out);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
void matrix_addition_qs8(const ITensor *input, ITensor *output, const Window &window, float beta)
{
@@ -142,7 +142,7 @@ void NEGEMMMatrixAdditionKernel::configure(const ITensor *input, ITensor *output
#ifdef ARM_COMPUTE_ENABLE_FP16
_func = &matrix_addition_f16;
break;
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
default:
ARM_COMPUTE_ERROR("Data type not supported");
break;
diff --git a/src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp b/src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp
index 101c5c8132..bff16ec329 100644
--- a/src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp
+++ b/src/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.cpp
@@ -244,7 +244,7 @@ void vector_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(vec_a)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + in_b_stride)));
-#endif
+#endif /* __arm__ */
auto vec_a_end_addr = vec_a + num_elems_vec_a;
for(; vec_a <= (vec_a_end_addr - 4);)
@@ -267,7 +267,7 @@ void vector_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 2 * in_b_stride)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 3 * in_b_stride)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 4 * in_b_stride)));
-#endif
+#endif /* __arm__ */
acc0 = vmlaq_lane_f32(acc0, b00, a0l, 0);
acc1 = vmlaq_lane_f32(acc1, b01, a0l, 0);
@@ -527,7 +527,7 @@ void matrix_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_a0)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b0)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b1)));
-#endif
+#endif /* __arm__ */
auto mtx_b0_end_addr = mtx_b0 + num_elems_matrix_b_x;
for(; mtx_b0 <= (mtx_b0_end_addr - 32);)
@@ -546,7 +546,7 @@ void matrix_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_a0)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b0)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b1)));
-#endif
+#endif /* __arm__ */
// 4x4 block 0
acc00 = vmlaq_f32(acc00, b00, a0);
@@ -637,7 +637,7 @@ void matrix_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_a0)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b0)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b1)));
-#endif
+#endif /* __arm__ */
// 4x4 block 0
acc00 = vmlaq_f32(acc00, b00, a0);
@@ -728,7 +728,7 @@ void matrix_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_a0)));
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b0)));
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b1)));
-#endif
+#endif /* __arm__ */
// 4x4 block 0
acc00 = vmlaq_f32(acc00, b00, a0);
acc10 = vmlaq_f32(acc10, b00, a1);
@@ -915,9 +915,9 @@ void matrix_matrix_multiply_f16(const ITensor *input0, const ITensor *input1, IT
vst1q_f16(mtx_out + 3 * out_stride, c.val[3]);
},
ina, inb, out);
-#else
+#else /* ARM_COMPUTE_ENABLE_FP16 */
ARM_COMPUTE_ERROR("Not implemented");
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
}
template <bool multiply_alpha>
@@ -1028,7 +1028,7 @@ void matrix_matrix_multiply_qs8(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_a0)));
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b0)));
asm volatile("PLD [%0, #128*2]" ::"r"(reinterpret_cast<const uint8_t *>(mtx_b1)));
-#endif
+#endif /* __arm__ */
// Second accumulation
acc00_qs16 = vqmlal_qs8(acc00_qs16, b02, a4, fixed_point_position);
diff --git a/src/core/NEON/kernels/NEHarrisCornersKernel.cpp b/src/core/NEON/kernels/NEHarrisCornersKernel.cpp
index 585676bb87..d44f4ce3b8 100644
--- a/src/core/NEON/kernels/NEHarrisCornersKernel.cpp
+++ b/src/core/NEON/kernels/NEHarrisCornersKernel.cpp
@@ -360,7 +360,7 @@ void NEHarrisScoreFP16Kernel<block_size>::configure(const IImage *input1, const
INEKernel::configure(win);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
template class arm_compute::NEHarrisScoreKernel<3>;
template class arm_compute::NEHarrisScoreKernel<5>;
diff --git a/src/core/NEON/kernels/NEIm2ColKernel.cpp b/src/core/NEON/kernels/NEIm2ColKernel.cpp
index 99daa2e5e7..8c9d12c57c 100644
--- a/src/core/NEON/kernels/NEIm2ColKernel.cpp
+++ b/src/core/NEON/kernels/NEIm2ColKernel.cpp
@@ -305,7 +305,7 @@ void NEIm2ColKernel::configure(const ITensor *input, ITensor *output, const Size
case DataType::F16:
_func = &NEIm2ColKernel::run_reduced<float16_t>;
break;
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
case DataType::QS8:
_func = &NEIm2ColKernel::run_reduced<qint8_t>;
break;
@@ -325,7 +325,7 @@ void NEIm2ColKernel::configure(const ITensor *input, ITensor *output, const Size
case DataType::F16:
_func = ((pad_x == 0) && (pad_y == 0)) ? &NEIm2ColKernel::run_generic<float16_t, false> : &NEIm2ColKernel::run_generic<float16_t, true>;
break;
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
case DataType::QS8:
_func = ((pad_x == 0) && (pad_y == 0)) ? &NEIm2ColKernel::run_generic<qint8_t, false> : &NEIm2ColKernel::run_generic<qint8_t, true>;
break;
diff --git a/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp b/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp
index ab84efbf23..895799c6ca 100644
--- a/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp
+++ b/src/core/NEON/kernels/NELocallyConnectedMatrixMultiplyKernel.cpp
@@ -89,7 +89,7 @@ void vector_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(vec_a)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b)));
asm volatile("PLD [%0, #128*4]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + in_b_stride)));
-#endif
+#endif /* __arm__ */
const float *vec_a_end_addr = vec_a + num_elems_vec_a;
@@ -113,7 +113,7 @@ void vector_matrix_multiply_f32(const ITensor *input0, const ITensor *input1, IT
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 2 * in_b_stride)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 3 * in_b_stride)));
asm volatile("PLD [%0, #128*1]" ::"r"(reinterpret_cast<const uint8_t *>(matrix_b + 4 * in_b_stride)));
-#endif
+#endif /* __arm __ */
acc0 = vmlaq_lane_f32(acc0, b00, a0l, 0);
acc1 = vmlaq_lane_f32(acc1, b01, a0l, 0);
diff --git a/src/core/NEON/kernels/NEMagnitudePhaseKernel.cpp b/src/core/NEON/kernels/NEMagnitudePhaseKernel.cpp
index a874d219d7..599dad6c70 100644
--- a/src/core/NEON/kernels/NEMagnitudePhaseKernel.cpp
+++ b/src/core/NEON/kernels/NEMagnitudePhaseKernel.cpp
@@ -428,7 +428,7 @@ template class arm_compute::NEMagnitudePhaseFP16Kernel<MagnitudeType::L1NORM, Ph
template class arm_compute::NEMagnitudePhaseFP16Kernel<MagnitudeType::L2NORM, PhaseType::SIGNED>;
template class arm_compute::NEMagnitudePhaseFP16Kernel<MagnitudeType::L1NORM, PhaseType::UNSIGNED>;
template class arm_compute::NEMagnitudePhaseFP16Kernel<MagnitudeType::L2NORM, PhaseType::UNSIGNED>;
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
namespace
{
diff --git a/src/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.cpp b/src/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.cpp
index 1826c474f7..ece7b9a10f 100644
--- a/src/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.cpp
+++ b/src/core/NEON/kernels/NENonMaximaSuppression3x3Kernel.cpp
@@ -224,7 +224,7 @@ void NENonMaximaSuppression3x3FP16Kernel::configure(const ITensor *input, ITenso
INEKernel::configure(win);
}
-#endif
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
namespace
{