From 04f4620cf999846a44089c81720aa920edec6993 Mon Sep 17 00:00:00 2001 From: Viet-Hoa Do Date: Wed, 14 Dec 2022 14:49:56 +0000 Subject: Add multiple output support for dynamic fusion * The dependency graph now can schedule any acyclic graph into a sequential list of operators. This is needed as the output operators now form branches in the graph. * Fix the definition of input, output and intermediate tensors in GpuKernelComponentGroup to support non-linear but sequential list of operators. * Add constraint on GpuOperatorGroup to enforce strictly linear fusion style, but allow output operator as the only form of branch. Resolves: COMPMID-5771 Signed-off-by: Viet-Hoa Do Change-Id: I68de3a31a2456145081f0a397e4e61dd66327682 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8823 Reviewed-by: Gunes Bayir Tested-by: Arm Jenkins Comments-Addressed: Arm Jenkins --- src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp') diff --git a/src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp b/src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp index e4b662b3a8..ef4f2f22a1 100644 --- a/src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp +++ b/src/dynamic_fusion/sketch/gpu/template_writer/cl/ClTemplateStore.cpp @@ -84,9 +84,9 @@ TagLUT ClTemplateStore::get_tag_lut(const GpuKernelVariableTable &vtable, const // Local build options lut["meta_kernel_id"] = id(); lut["DST_TENSOR_TYPE"] = "BUFFER"; - const auto dst_info = comp_group.get_dst_tensors()[0]; - lut["DST_DATA_TYPE"] = dst_info->data_type(); + lut["DST_DATA_TYPE"] = _dst->data_type(); + ARM_COMPUTE_UNUSED(comp_group); return lut; } -- cgit v1.2.1