24 #ifndef TVM_RELAY_ATTRS_MEMORY_H_
25 #define TVM_RELAY_ATTRS_MEMORY_H_
50 .describe(
"The dtype of the tensor to allocate.")
66 .describe(
"The dtype of the tensor to allocate.")
71 "The shape to cast the return type of the allocation to, "
72 "used to specify the shape obtained via further analysis.");
The base class of the all the Use "curiously recurring template pattern".
Definition: attrs.h:870
Managed reference to RelayExprNode.
Definition: expr.h:442
Managed reference to TypeNode.
Definition: type.h:93
Managed reference class to VirtualDeviceNode.
Definition: virtual_device.h:271
static VirtualDevice FullyUnconstrained()
Returns the unique fully unconstrained VirtualDevice.
Array, container representing a contiguous sequence of ObjectRefs.
Definition: array.h:289
Runtime primitive data type.
Definition: data_type.h:43
static DataType Float(int bits, int lanes=1)
Construct an float type.
Definition: data_type.h:236
Helpers for attribute objects.
#define TVM_ATTR_FIELD(FieldName)
Declare an attribute field.
Definition: attrs.h:76
std::vector< TensorType > FlattenTupleType(const Type &type)
Expr ToTupleType(const Type &t, const std::vector< Expr > &exprs)
std::vector< Expr > FromTupleType(const Type &type, const Expr &expr)
runtime implementation for LibTorch/TorchScript.
Definition: analyzer.h:36
Relay expression language.
Options for allocating storage.
Definition: memory.h:44
DataType dtype
Definition: memory.h:45
TVM_DECLARE_ATTRS(AllocStorageAttrs, "relay.attrs.AllocStorageAttrs")
Definition: memory.h:48
VirtualDevice virtual_device
Definition: memory.h:46
Options for allocating tensors.
Definition: memory.h:59
Constant const_shape
Definition: memory.h:60
Array< IndexExpr > assert_shape
Definition: memory.h:61
TVM_DECLARE_ATTRS(AllocTensorAttrs, "relay.attrs.AllocTensorAttrs")
Definition: memory.h:64
DataType dtype
Definition: memory.h:62
A compile time representation for where data is to be stored at runtime, and how to compile code to c...