TOSA: Add support for PAD
Added support for TOSA PAD operator
in line with legacy support
Limitations:
-Rank <= 4
-N = 1 if Rank = 4 for ifms/ofm
-only padding in W and H dimensions
-bool_t not supported
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: I511608202b4c9bf6d86285b559c517fb41741fdf
diff --git a/ethosu/vela/operation.py b/ethosu/vela/operation.py
index 681f498..e9d364e 100644
--- a/ethosu/vela/operation.py
+++ b/ethosu/vela/operation.py
@@ -592,6 +592,9 @@
def get_ifm_ifm2_weights_ofm(self):
return self.ifm, self.ifm2, self.weights, self.ofm
+ def get_ifm_ifm2_ofm(self):
+ return self.ifm, self.ifm2, self.ofm
+
def get_ifm_weights_biases_ofm(self):
return self.ifm, self.weights, self.bias, self.ofm
diff --git a/ethosu/vela/tosa_graph_optimiser.py b/ethosu/vela/tosa_graph_optimiser.py
index 2d1245b..49fc997 100644
--- a/ethosu/vela/tosa_graph_optimiser.py
+++ b/ethosu/vela/tosa_graph_optimiser.py
@@ -35,6 +35,7 @@
from .operation_util import create_avgpool_nop
from .shape4d import Shape4D
from .tensor import create_const_tensor
+from .tensor import create_equivalence_id
def replace_rescale_with_avg_pool(rescale_op):
@@ -417,6 +418,96 @@
return op
+# TODO modified copy of TFLite, solution for TOSA PAD will change so reuse has not been considered
+def convert_pad(op, arch, nng):
+ """
+ Rewrites PAD operator to an add that copies the IFM to the OFM
+ + up to 4 add operators that fill the OFM with zeros at the borders.
+ """
+
+ if op.type != Op.Pad:
+ return op
+
+ # TODO assuming rank <= 4 and N = 1 for rank ==4
+ # This is checked in tosa_supported_operators
+ ifm = op.ifm
+ assert ifm is not None
+ ifm_shape = Shape4D(ifm.shape)
+ ofm = op.ofm
+ assert ofm is not None
+ ofm.ops = []
+ ofm_shape = op.ofm_shapes[0]
+
+ rank = len(ifm.shape)
+ padding = op.inputs[1].values
+ pad_depth = padding[-1]
+ if not (pad_depth == 0).all():
+ print("Warning: For PAD, padding in depth not supported yet")
+ assert False
+
+ top, bottom = 0, 0
+ left, right = 0, 0
+ if rank > 1:
+ left, right = padding[-2][0], padding[-2][1]
+ if rank > 2:
+ top, bottom = padding[-3][0], padding[-3][1]
+ if rank == 4 and not (padding[-4] == 0).all():
+ print("Warning: For PAD, padding not supported in first dimension when rank == 4 yet")
+ assert False
+
+ # Add op that copies IFM to the right place inside the OFM
+ shp0 = Shape4D(0, 0, 0, 0)
+ shp_top = shp0.with_height(top)
+ add_op = create_add_for_concat(op, op.name + "_main", ifm, ifm_shape, shp_top.with_width(left))
+ add_op.activation = op.activation
+
+ quant = ofm.quantization
+ pad_value = ifm.quantization.zero_point
+ # Add operations that fill the borders of the OFM
+ if top > 0:
+ shape = Shape4D(1, top, ofm_shape.width, ofm_shape.depth)
+ zero_tens = create_const_tensor(
+ op.name + "_top",
+ shape.as_list(),
+ ofm.dtype,
+ shape.elements() * [pad_value],
+ np.uint8,
+ quantization=quant, # TODO
+ )
+ # If top/bottom or left/right are equal, the const tensors can be allocated to the same address
+ zero_tens.equivalence_id = create_equivalence_id(tuple(zero_tens.values))
+ create_add_for_concat(op, op.name + "_top", zero_tens, shape, shp0)
+ if bottom > 0:
+ shape = Shape4D(1, bottom, ofm_shape.width, ofm_shape.depth)
+ zero_tens = create_const_tensor(
+ op.name + "_bottom",
+ shape.as_list(),
+ ofm.dtype,
+ shape.elements() * [pad_value],
+ np.uint8,
+ quantization=quant,
+ )
+ zero_tens.equivalence_id = create_equivalence_id(tuple(zero_tens.values))
+ create_add_for_concat(op, op.name + "_bottom", zero_tens, shape, shp0.with_height(ofm_shape.height - bottom))
+ if left > 0:
+ shape = Shape4D(1, ifm_shape.height, left, ofm_shape.depth)
+ zero_tens = create_const_tensor(
+ op.name + "_left", shape.as_list(), ofm.dtype, shape.elements() * [pad_value], np.uint8, quantization=quant
+ )
+ zero_tens.equivalence_id = create_equivalence_id(tuple(zero_tens.values))
+ create_add_for_concat(op, op.name + "_left", zero_tens, shape, shp_top)
+ if right > 0:
+ shape = Shape4D(1, ifm_shape.height, right, ofm_shape.depth)
+ zero_tens = create_const_tensor(
+ op.name + "_right", shape.as_list(), ofm.dtype, shape.elements() * [pad_value], np.uint8, quantization=quant
+ )
+ zero_tens.equivalence_id = create_equivalence_id(tuple(zero_tens.values))
+ create_add_for_concat(op, op.name + "_right", zero_tens, shape, shp_top.with_width(ofm_shape.width - right))
+
+ op.type = Op.ConcatTFLite
+ return add_op
+
+
def fixup_quantization(op, arch, nng):
if op.ifm and op.ifm.quantization.zero_point is None:
op.ifm.quantization.zero_point = 0
@@ -484,7 +575,7 @@
# Post-processing step 1
for idx, sg in enumerate(nng.subgraphs):
nng.subgraphs[idx] = rewrite_graph.rewrite_graph_pre_order(
- nng, sg, arch, [], [rewrite_activation, add_padding_fields],
+ nng, sg, arch, [], [rewrite_activation, convert_pad, add_padding_fields],
)
# Removal of Slice, need to be done after optimisation has been performed,
diff --git a/ethosu/vela/tosa_mapping.py b/ethosu/vela/tosa_mapping.py
index 6efc479..ebbaa0a 100644
--- a/ethosu/vela/tosa_mapping.py
+++ b/ethosu/vela/tosa_mapping.py
@@ -174,7 +174,7 @@
unary_quant_info = QuantSerializer("UnaryQuantInfo", ("input_zp", "output_zp"))
conv_quant_info = QuantSerializer("ConvQuantInfo", ("input_zp", "weight_zp"))
matmul_quant_info = QuantSerializer("MatMulQuantInfo", ("a_zp", "b_zp"))
-pad_quant_info = QuantSerializer("PadQuantInfo", ("input_zp"))
+pad_quant_info = QuantSerializer("PadQuantInfo", ("input_zp",))
unsupported_tosa_operators = {
TosaOp.UNKNOWN,
@@ -218,7 +218,6 @@
TosaOp.REDUCE_MIN,
TosaOp.REDUCE_PRODUCT,
TosaOp.REDUCE_SUM,
- TosaOp.PAD,
TosaOp.REVERSE,
TosaOp.TILE,
TosaOp.GATHER,
@@ -298,12 +297,19 @@
# TODO TosaOp.REDUCE_PRODUCT
# TODO TosaOp.REDUCE_SUM
TosaOp.CONCAT: (Op.Concat, axis_attrs, None, TOSA_CONCAT_INDICES),
- # TODO TosaOp.PAD
+ # TODO Is the padding intended to be dynamic input, TOSA spec state it as attribute
+ # Handled as for TFLite for now
+ TosaOp.PAD: (Op.Pad, None, pad_quant_info, TOSA_IFM_INDICES),
TosaOp.RESHAPE: (Op.Reshape, reshape_attrs, None, TOSA_IFM_INDICES),
# TODO TosaOp.REVERSE
TosaOp.SLICE: (Op.SplitSliceRead, slice_attrs, None, TOSA_IFM_INDICES),
# TODO TosaOp.TILE
- TosaOp.TRANSPOSE: (Op.Transpose, None, None, TOSA_IFM_IFM2_INDICES),
+ TosaOp.TRANSPOSE: (
+ Op.Transpose,
+ None,
+ None,
+ TOSA_IFM_IFM2_INDICES,
+ ), # TODO Is the perms intended to be dynamic input, TOSA spec state it as attribute
# TODO TosaOp.GATHER
# TODO TosaOp.SCATTER
# TODO TosaOp.RESIZE
diff --git a/ethosu/vela/tosa_reader.py b/ethosu/vela/tosa_reader.py
index 94ba350..aadcb0a 100644
--- a/ethosu/vela/tosa_reader.py
+++ b/ethosu/vela/tosa_reader.py
@@ -113,7 +113,7 @@
# Moving permutation to an attribute, to match internal graph representation for now
perms = None
if op_code == TosaOp.TRANSPOSE:
- perms = perms = inputs.pop(1)
+ perms = inputs.pop(1)
indices = TOSA_IFM_INDICES
name = "unknown_op_name"
diff --git a/ethosu/vela/tosa_supported_operators.py b/ethosu/vela/tosa_supported_operators.py
index d368616..a4f822e 100644
--- a/ethosu/vela/tosa_supported_operators.py
+++ b/ethosu/vela/tosa_supported_operators.py
@@ -39,16 +39,15 @@
mac_main_ops = convolution_like_ops | pooling_ops | fc_vector_products
memory_only_ops = set((Op.Reshape, Op.Transpose, Op.Concat, Op.SplitSliceRead,))
-
binary_elem_wise_add_mul_sub = set((Op.Add, Op.Mul, Op.RescaleMul, Op.Sub,))
-
type_conversion_ops = set((Op.Rescale,))
relu_ops = set((Op.Clamp, Op.ReluN,))
activation_ops = relu_ops
+ pad_ops = set((Op.Pad,))
npu_post_ops = activation_ops
supported_operators = (
- mac_main_ops | type_conversion_ops | npu_post_ops | memory_only_ops | binary_elem_wise_add_mul_sub
+ mac_main_ops | type_conversion_ops | npu_post_ops | memory_only_ops | binary_elem_wise_add_mul_sub | pad_ops
)
# Supported data types
@@ -60,12 +59,15 @@
# Setup the generic constraints. Note: the order matters
self.generic_constraints = []
self.generic_constraints.append(TosaSupportedOperators.constraint_tens_dtype)
- self.generic_constraints.append(TosaSupportedOperators.constraint_tens_dimension)
+ self.generic_constraints.append(TosaSupportedOperators.constraint_tens_dimension) # TODO as not supported yet
+ self.generic_constraints.append(TosaSupportedOperators.constraint_rank) # TODO as not supported yet
+ self.generic_constraints.append(TosaSupportedOperators.constraint_batch) # TODO as not supported yet
# Setup specific constraints. Note: the order matters
self.specific_constraints = defaultdict(list)
self.specific_constraints[Op.Transpose].append(TosaSupportedOperators.constraint_ifm_producer)
+ self.specific_constraints[Op.Pad].append(TosaSupportedOperators.constraint_padding_producer)
# Depthwise Conv specific checks:
for op_type in TosaSupportedOperators.depthwise_convolution_ops:
@@ -127,6 +129,38 @@
extra.append(f"Tensor '{tens.name}' has shape: {tens.shape}")
return valid, ", ".join(extra)
+ # TODO This is for a HW limitation, that is to be resolved in SW later on
+ @staticmethod
+ def constraint_rank(op):
+ "Tensor rank must be <= 4"
+ valid = True
+ extra = []
+ tensors = [tens for tens in op.get_ifm_ifm2_weights_ofm() if tens]
+ if not tensors:
+ tensors = [tens for tens in op.inputs if tens]
+ for tens in tensors:
+ rank = len(tens.shape)
+ if not rank <= 4:
+ valid = False
+ extra.append(f"Tensor '{tens.name}' has rank: {rank}")
+ return valid, ", ".join(extra)
+
+ # TODO This is for a HW limitation, that is to be resolved in SW later on
+ @staticmethod
+ def constraint_batch(op):
+ "If Tensor rank is 4 batch of ifms/ofm must be 1"
+ valid = True
+ extra = []
+ tensors = [tens for tens in op.get_ifm_ifm2_ofm() if tens]
+ if not tensors:
+ tensors = [tens for tens in op.inputs if tens]
+ for tens in tensors:
+ rank = len(tens.shape)
+ if rank == 4 and tens.shape[0] != 1:
+ valid = False
+ extra.append(f"Tensor '{tens.name}' has rank: 4 and N: {tens.shape[0]}")
+ return valid, ", ".join(extra)
+
@staticmethod
def constraint_ifm_producer(cls, op):
"Input must be constant data"
@@ -143,6 +177,14 @@
return valid, "Avgpool with pad_top {top} and pad_left {left}"
+ # TODO limit padding to be const data for now.
+ # For TFLite it is assumed to be constant.
+ @staticmethod
+ def constraint_padding_producer(op):
+ "Input must be constant data"
+ valid = op.inputs[1].ops and op.inputs[1].ops[0].type == Op.Const
+ return valid, "PAD Op with non-constant data padding"
+
# TODO duplicates tflite_supported operators, but support for depth multiplier should be added at a later stage
@staticmethod
def constraint_depth_multiplier(op):