From 879e8dd2fc8523e4059ba9ced9ea0edb57103778 Mon Sep 17 00:00:00 2001 From: Manuel Bottini Date: Wed, 8 May 2019 17:49:28 +0100 Subject: COMPMID-2115: Fix CLFuseBatchNormalization Change-Id: I0ccb42f90c028f599e318ffc7997a6d922add170 Signed-off-by: Manuel Bottini Reviewed-on: https://review.mlplatform.org/c/1091 Reviewed-by: Gian Marco Iodice Tested-by: Arm Jenkins --- src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp b/src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp index e14b8a3777..150d9b6d1a 100644 --- a/src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp +++ b/src/core/CL/kernels/CLFuseBatchNormalizationKernel.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018 ARM Limited. + * Copyright (c) 2018-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -139,7 +139,7 @@ void CLFuseBatchNormalizationKernel::configure(const ICLTensor *conv_weights, co epsilon)); // Configure kernel window - const unsigned int num_elems_processed_per_iteration_x = 16 / conv_weights->info()->element_size(); + const unsigned int num_elems_processed_per_iteration_x = 4; const int output_width_x = conv_weights->info()->tensor_shape().x(); const bool multi_access_x = (output_width_x / num_elems_processed_per_iteration_x > 0); @@ -216,6 +216,6 @@ void CLFuseBatchNormalizationKernel::run(const arm_compute::Window &window, cl:: { add_1D_tensor_argument(idx, _bn_gamma, vector_slice); } - enqueue(queue, *this, slice, lws_hint()); + enqueue(queue, *this, slice); } } // namespace arm_compute -- cgit v1.2.1