tvm
Public Member Functions | Static Public Member Functions | Public Attributes | List of all members
tvm::relax::CallTIRInplaceAttrs Struct Reference

Attributes used in call_tir_inplace. More...

#include <op.h>

Inheritance diagram for tvm::relax::CallTIRInplaceAttrs:
Collaboration diagram for tvm::relax::CallTIRInplaceAttrs:

Public Member Functions

 TVM_FFI_DECLARE_OBJECT_INFO_FINAL ("relax.attrs.CallTIRInplaceAttrs", CallTIRInplaceAttrs, BaseAttrsNode)
 
- Public Member Functions inherited from tvm::AttrsNodeReflAdapter< CallTIRInplaceAttrs >
void InitByPackedArgs (const ffi::PackedArgs &args, bool allow_unknown) final
 Initialize the attributes by arguments. More...
 
- Public Member Functions inherited from tvm::BaseAttrsNode
virtual ~BaseAttrsNode ()
 virtual destructor More...
 
template<typename... Args>
void InitBySeq (Args &&... args)
 Initialize the attributes by sequence of arguments. More...
 
 TVM_FFI_DECLARE_OBJECT_INFO ("ir.Attrs", BaseAttrsNode, Object)
 

Static Public Member Functions

static void RegisterReflection ()
 

Public Attributes

ffi::Array< Integerinplace_indices
 Indices that describe which input corresponds to which output. More...
 

Additional Inherited Members

- Static Public Attributes inherited from tvm::BaseAttrsNode
static constexpr TVMFFISEqHashKind _type_s_eq_hash_kind = kTVMFFISEqHashKindTreeNode
 

Detailed Description

Attributes used in call_tir_inplace.

Member Function Documentation

◆ RegisterReflection()

static void tvm::relax::CallTIRInplaceAttrs::RegisterReflection ( )
inlinestatic

◆ TVM_FFI_DECLARE_OBJECT_INFO_FINAL()

tvm::relax::CallTIRInplaceAttrs::TVM_FFI_DECLARE_OBJECT_INFO_FINAL ( "relax.attrs.CallTIRInplaceAttrs"  ,
CallTIRInplaceAttrs  ,
BaseAttrsNode   
)

Member Data Documentation

◆ inplace_indices

ffi::Array<Integer> tvm::relax::CallTIRInplaceAttrs::inplace_indices

Indices that describe which input corresponds to which output.

If the ith member has the value k >= 0, then that means that input k should be used to store the ith output. If an element has the value -1, that means a new tensor should be allocated for that output.


The documentation for this struct was generated from the following file: