def apply(self, graph: SDFGState, sdfg: SDFG):
tasklet = self.tasklet
map_exit = self.map_exit
outer_map_exit = self.outer_map_exit
memlet = None
edge = None
for e in graph.out_edges(map_exit):
memlet = e.data
# TODO: What if there's more than one?
if e.dst == outer_map_exit and isinstance(sdfg.arrays[memlet.data],
data.Stream):
edge = e
break
tasklet_memlet = None
for e in graph.out_edges(tasklet):
tasklet_memlet = e.data
if tasklet_memlet.data == memlet.data:
break
bbox = map_exit.map.range.bounding_box_size()
bbox_approx = [symbolic.overapproximate(dim) for dim in bbox]
dataname = memlet.data
# Create the new node: Temporary stream and an access node
newname, _ = sdfg.add_stream('trans_' + dataname,
sdfg.arrays[memlet.data].dtype,
bbox_approx[0],
storage=sdfg.arrays[memlet.data].storage,
transient=True,
find_new_name=True)
snode = graph.add_access(newname)
to_stream_mm = copy.deepcopy(memlet)
to_stream_mm.data = snode.data
tasklet_memlet.data = snode.data
if self.with_buffer:
newname_arr, _ = sdfg.add_transient('strans_' + dataname,
[bbox_approx[0]],
sdfg.arrays[memlet.data].dtype,
find_new_name=True)
anode = graph.add_access(newname_arr)
to_array_mm = copy.deepcopy(memlet)
to_array_mm.data = anode.data
graph.add_edge(snode, None, anode, None, to_array_mm)
else:
anode = snode
# Reconnect, assuming one edge to the stream
graph.remove_edge(edge)
graph.add_edge(map_exit, edge.src_conn, snode, None, to_stream_mm)
graph.add_edge(anode, None, outer_map_exit, edge.dst_conn, memlet)
return
def apply(self, graph: SDFGState, sdfg: SDFG):
import dace.libraries.blas as blas
transpose_a = self.transpose_a
_at = self.at
transpose_b = self.transpose_b
_bt = self.bt
a_times_b = self.a_times_b
for src, src_conn, _, _, memlet in graph.in_edges(transpose_a):
graph.add_edge(src, src_conn, a_times_b, '_b', memlet)
graph.remove_node(transpose_a)
for src, src_conn, _, _, memlet in graph.in_edges(transpose_b):
graph.add_edge(src, src_conn, a_times_b, '_a', memlet)
graph.remove_node(transpose_b)
graph.remove_node(_at)
graph.remove_node(_bt)
for _, _, dst, dst_conn, memlet in graph.out_edges(a_times_b):
subset = dcpy(memlet.subset)
subset.squeeze()
size = subset.size()
shape = [size[1], size[0]]
break
tmp_name, tmp_arr = sdfg.add_temp_transient(shape, a_times_b.dtype)
tmp_acc = graph.add_access(tmp_name)
transpose_c = blas.Transpose('_Transpose_', a_times_b.dtype)
for edge in graph.out_edges(a_times_b):
_, _, dst, dst_conn, memlet = edge
graph.remove_edge(edge)
graph.add_edge(transpose_c, '_out', dst, dst_conn, memlet)
graph.add_edge(a_times_b, '_c', tmp_acc, None, dace.Memlet.from_array(tmp_name, tmp_arr))
graph.add_edge(tmp_acc, None, transpose_c, '_inp', dace.Memlet.from_array(tmp_name, tmp_arr))
