Update reinterpret tensor as 1D for CPU add

* Use the same implementation as other layers.

Resolves: COMPMID-5108
Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Change-Id: I5a50259b398b71ca1f61b5ee8daa539bf8263fac
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8501
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

5 files changed, 0 insertions, 66 deletions

diff --git a/src/cpu/kernels/add/generic/neon/fp16.cpp b/src/cpu/kernels/add/generic/neon/fp16.cpp
index bb6636af1e..1e3bc3c986 100644
--- a/src/cpu/kernels/add/generic/neon/fp16.cpp
+++ b/src/cpu/kernels/add/generic/neon/fp16.cpp
@@ -33,11 +33,6 @@ void add_fp16_neon(const ITensor *src0, const ITensor *src1, ITensor *dst, const
 {
     return add_same_neon<float16_t>(src0, src1, dst, policy, window);
 }
-
-void add_fp16_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    return add_same_neon_as_1d_array<float16_t>(src0, src1, dst, policy, window);
-}
 }
 } // namespace arm_compute
 #endif /* (__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) */
diff --git a/src/cpu/kernels/add/generic/neon/fp32.cpp b/src/cpu/kernels/add/generic/neon/fp32.cpp
index 1d313a191d..1f599b1968 100644
--- a/src/cpu/kernels/add/generic/neon/fp32.cpp
+++ b/src/cpu/kernels/add/generic/neon/fp32.cpp
@@ -32,10 +32,5 @@ void add_fp32_neon(const ITensor *src0, const ITensor *src1, ITensor *dst, const
 {
     return add_same_neon<float>(src0, src1, dst, policy, window);
 }
-
-void add_fp32_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    return add_same_neon_as_1d_array<float>(src0, src1, dst, policy, window);
-}
 }
 } // namespace arm_compute
diff --git a/src/cpu/kernels/add/generic/neon/impl.cpp b/src/cpu/kernels/add/generic/neon/impl.cpp
index 0f7b31c754..1a0b44fa8c 100644
--- a/src/cpu/kernels/add/generic/neon/impl.cpp
+++ b/src/cpu/kernels/add/generic/neon/impl.cpp
@@ -128,35 +128,6 @@ void add_same_neon(const ITensor *src0, const ITensor *src1, ITensor *dst, const
     }
 }
 
-template <typename ScalarType>
-void add_same_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    const ScalarType *src0_ptr = reinterpret_cast<const ScalarType *>(src0->buffer());
-    const ScalarType *src1_ptr = reinterpret_cast<const ScalarType *>(src1->buffer());
-    ScalarType       *dst_ptr  = reinterpret_cast<ScalarType *>(dst->buffer());
-
-    constexpr int window_step_x  = 16 / sizeof(ScalarType);
-    const auto    window_start_x = static_cast<int>(window.x().start());
-    const auto    window_end_x   = static_cast<int>(window.x().end());
-
-    int x = window_start_x;
-    for(; x <= (window_end_x - window_step_x); x += window_step_x)
-    {
-        const auto val1 = wrapper::vloadq(src0_ptr + x);
-        const auto val2 = wrapper::vloadq(src1_ptr + x);
-        const auto res  = (policy == ConvertPolicy::SATURATE) ? wrapper::vqadd(val1, val2) : wrapper::vadd(val1, val2);
-        wrapper::vstore(dst_ptr + x, res);
-    }
-
-    // Compute left-over elements
-    for(; x < window_end_x; ++x)
-    {
-        const auto val1 = *(src0_ptr + x);
-        const auto val2 = *(src1_ptr + x);
-        *(dst_ptr + x)  = (policy == ConvertPolicy::SATURATE) ? wrapper::add_sat(val1, val2) : val1 + val2;
-    }
-}
-
 bool add_q8_neon_fixedpoint_possible(const ITensorInfo *src0, const ITensorInfo *src1, const ITensorInfo *dst)
 {
     const auto iq0 = src0->quantization_info().uniform();
@@ -383,15 +354,6 @@ template void add_same_neon<int16_t>(const ITensor *src0, const ITensor *src1, I
 template void add_same_neon<float16_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
 #endif /* (__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) */
 
-template void add_same_neon_as_1d_array<float>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-template void add_same_neon_as_1d_array<uint8_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-template void add_same_neon_as_1d_array<int32_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-template void add_same_neon_as_1d_array<int16_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-
-#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS)
-template void add_same_neon_as_1d_array<float16_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-#endif /* (__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) */
-
 template void add_q8_neon_fixedpoint<int8_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
 template void add_q8_neon_fixedpoint<uint8_t>(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
 
diff --git a/src/cpu/kernels/add/generic/neon/impl.h b/src/cpu/kernels/add/generic/neon/impl.h
index e6a12fb4c0..91f347ff9c 100644
--- a/src/cpu/kernels/add/generic/neon/impl.h
+++ b/src/cpu/kernels/add/generic/neon/impl.h
@@ -33,9 +33,6 @@ namespace cpu
 template <typename ScalarType>
 void add_same_neon(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
 
-template <typename ScalarType>
-void add_same_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window);
-
 bool add_q8_neon_fixedpoint_possible(const ITensorInfo *src0, const ITensorInfo *src1, const ITensorInfo *dst);
 
 template <typename ScalarType>
diff --git a/src/cpu/kernels/add/generic/neon/integer.cpp b/src/cpu/kernels/add/generic/neon/integer.cpp
index ffead03474..5698d6d552 100644
--- a/src/cpu/kernels/add/generic/neon/integer.cpp
+++ b/src/cpu/kernels/add/generic/neon/integer.cpp
@@ -42,20 +42,5 @@ void add_s32_neon(const ITensor *src0, const ITensor *src1, ITensor *dst, const
 {
     return add_same_neon<int32_t>(src0, src1, dst, policy, window);
 }
-
-void add_u8_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    return add_same_neon_as_1d_array<uint8_t>(src0, src1, dst, policy, window);
-}
-
-void add_s16_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    return add_same_neon_as_1d_array<int16_t>(src0, src1, dst, policy, window);
-}
-
-void add_s32_neon_as_1d_array(const ITensor *src0, const ITensor *src1, ITensor *dst, const ConvertPolicy &policy, const Window &window)
-{
-    return add_same_neon_as_1d_array<int32_t>(src0, src1, dst, policy, window);
-}
 }
 } // namespace arm_compute