aboutsummaryrefslogtreecommitdiff
path: root/src/core/cpu/kernels/pooling/neon/fp16.cpp
diff options
context:
space:
mode:
authorManuel Bottini <manuel.bottini@arm.com>2021-03-23 11:50:34 +0000
committerManuel Bottini <manuel.bottini@arm.com>2021-04-06 11:28:16 +0000
commitca62c6f53eb7244e6fed9f7e932608aa2496d9eb (patch)
treee5c7630c40d9f009e9baef4e849c6c7cc6ca90a7 /src/core/cpu/kernels/pooling/neon/fp16.cpp
parent4ed7b39dbbe8ccc6267a9eacefca51717c3b3e10 (diff)
downloadComputeLibrary-ca62c6f53eb7244e6fed9f7e932608aa2496d9eb.tar.gz
Mixed data-layout testing on high priority operators
Change data layouts after the configure in validation tests for: - Scale - Pooling - FullyConnected - DepthwiseConvolution - DirectConvolution - FFTConvolution - WinogradConvolution - GEMMConvolution (Indirect GEMM included) Extending fixtures Fixes for new mixed data layout tests Resolves: COMPMID-4162 Change-Id: I2f2eb2075f7e24ab3872249d88cadb57b82c5dde Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5326 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Diffstat (limited to 'src/core/cpu/kernels/pooling/neon/fp16.cpp')
-rw-r--r--src/core/cpu/kernels/pooling/neon/fp16.cpp4
1 files changed, 2 insertions, 2 deletions
diff --git a/src/core/cpu/kernels/pooling/neon/fp16.cpp b/src/core/cpu/kernels/pooling/neon/fp16.cpp
index 314be3704e..1ecceafe86 100644
--- a/src/core/cpu/kernels/pooling/neon/fp16.cpp
+++ b/src/core/cpu/kernels/pooling/neon/fp16.cpp
@@ -93,7 +93,7 @@ void pooling2_f16_maxpool_indices(const ITensor *src, ITensor *dst0, ITensor *ds
// Store result
vst1q_f16(reinterpret_cast<float16_t *>(out.ptr()) + x_off, vres);
- const uint32_t offset_base = offset_no_padding<float16_t>(in.offset(), id, *src->info(), pool_stride_x, pool_stride_y);
+ const uint32_t offset_base = offset_no_padding<float16_t>(in.offset(), id, *src->info(), pool_stride_x, pool_stride_y, DataLayout::NHWC);
const uint32_t offset_x0 = (uint32_t)offset_base / sizeof(float16_t) + x_off;
const uint32_t offset_x1 = (uint32_t)offset_x0 + in_stride_y / sizeof(float16_t) - pad_right;
const uint32_t offset_x2 = (uint32_t)offset_x0 + in_stride_z / sizeof(float16_t) - pad_right * src->info()->tensor_shape()[1];
@@ -132,7 +132,7 @@ void pooling2_f16_maxpool_indices(const ITensor *src, ITensor *dst0, ITensor *ds
// Store result
*(reinterpret_cast<float16_t *>(out.ptr()) + x_off) = res;
- const uint32_t offset_base = offset_no_padding<float16_t>(in.offset(), id, *src->info(), pool_stride_x, pool_stride_y);
+ const uint32_t offset_base = offset_no_padding<float16_t>(in.offset(), id, *src->info(), pool_stride_x, pool_stride_y, DataLayout::NHWC);
const uint32_t offset_x0 = (uint32_t)offset_base / sizeof(float16_t) + x_off;
const uint32_t offset_x1 = (uint32_t)offset_x0 + in_stride_y / sizeof(float16_t) - pad_right;
const uint32_t offset_x2 = (uint32_t)offset_x0 + in_stride_z / sizeof(float16_t) - pad_right * src->info()->tensor_shape()[1];