COMPMID-926 Add depth multiplier support to NEON/CL/GLES depthwise convolution
Change-Id: I03f32c62350e5ea43e77bb15fc5a832d83719e3b
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/126657
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/src/core/GLES_COMPUTE/cs_shaders/depthwise_convolution3x3.cs b/src/core/GLES_COMPUTE/cs_shaders/depthwise_convolution3x3.cs
index adfc126..134cc10 100644
--- a/src/core/GLES_COMPUTE/cs_shaders/depthwise_convolution3x3.cs
+++ b/src/core/GLES_COMPUTE/cs_shaders/depthwise_convolution3x3.cs
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017 ARM Limited.
+ * Copyright (c) 2017-2018 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -108,6 +108,8 @@
uint z_index = gl_GlobalInvocationID.z;
TENSOR_ITERATOR_ADVANCE_IN_BYTES(weights_iter, z_index * weights_attrs.stride_z);
+ src_iter.current_offset_in_bytes -= int((z_index - z_index / uint(DEPTH_MULTIPLIER)) * src_attrs.step_z);
+
vec4 w[3];
w[0] = LOAD_UNPACK4_CURRENT_ITEM_HALF(weights_ptr, weights_iter);
w[1] = LOAD_UNPACK4_HALF(weights_ptr, TENSOR3D_OFFSET(weights_iter, 0, 1, 0));
@@ -263,6 +265,8 @@
uint z_index = gl_GlobalInvocationID.z;
TENSOR_ITERATOR_ADVANCE_IN_BYTES(weights_iter, z_index * weights_attrs.stride_z);
+ src_iter.current_offset_in_bytes -= int((z_index - z_index / uint(DEPTH_MULTIPLIER)) * src_attrs.step_z);
+
vec4 w[3];
w[0] = LOAD_UNPACK4_CURRENT_ITEM_HALF(weights_ptr, weights_iter);
w[1] = LOAD_UNPACK4_HALF(weights_ptr, TENSOR3D_OFFSET(weights_iter, 0, 1, 0));