aboutsummaryrefslogtreecommitdiff
path: root/src/core/CL/kernels
diff options
context:
space:
mode:
authorMatthew Jackson <matthew.jackson@arm.com>2019-08-30 15:19:42 +0100
committerMatthew Jackson <matthew.jackson@arm.com>2019-09-02 13:54:07 +0000
commitc2a60593436387d20ff142a619f4c3955a5cd41b (patch)
tree4b42d0e876e40b8439d826c75e57ca72303aaae6 /src/core/CL/kernels
parentc5f48adafede995cae6fcb2f44471c9bbcc8a125 (diff)
downloadComputeLibrary-c2a60593436387d20ff142a619f4c3955a5cd41b.tar.gz
COMPMID-2639: CLPadLayer support for 4D padding
Add support for 4D padding to CLPadLayerKernel. Add validation tests with 4D padding. Change-Id: I5579cc441a155c03fa1d14c6e77ba8ec693a806d Signed-off-by: Matthew Jackson <matthew.jackson@arm.com> Reviewed-on: https://review.mlplatform.org/c/1847 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/kernels')
-rw-r--r--src/core/CL/kernels/CLPadLayerKernel.cpp8
1 files changed, 8 insertions, 0 deletions
diff --git a/src/core/CL/kernels/CLPadLayerKernel.cpp b/src/core/CL/kernels/CLPadLayerKernel.cpp
index 5270e31bbe..52b65c39b1 100644
--- a/src/core/CL/kernels/CLPadLayerKernel.cpp
+++ b/src/core/CL/kernels/CLPadLayerKernel.cpp
@@ -99,6 +99,12 @@ void CLPadLayerKernel::configure(const ICLTensor *input, ICLTensor *output, cons
{
build_opts.add_option("-DPAD_NEAR=" + support::cpp11::to_string(padding.at(2).first));
build_opts.add_option("-DSRC_DEPTH=" + support::cpp11::to_string(input->info()->dimension(2)));
+
+ if(padding.size() > 3)
+ {
+ build_opts.add_option("-DPAD_BTOP=" + support::cpp11::to_string(padding.at(3).first));
+ build_opts.add_option("-DSRC_BATCH=" + support::cpp11::to_string(input->info()->dimension(3)));
+ }
}
}
@@ -130,11 +136,13 @@ void CLPadLayerKernel::run(const Window &window, cl::CommandQueue &queue)
Window slice_out = window.first_slice_window_3D();
Window slice_in = win_in.first_slice_window_3D();
+ unsigned int batch = 0;
do
{
unsigned int idx = 0;
add_3D_tensor_argument(idx, _input, slice_in);
add_3D_tensor_argument(idx, _output, slice_out);
+ add_argument<unsigned int>(idx, batch++);
enqueue(queue, *this, slice_out, lws_hint());
}