CUDAFlags enum
A flag for OrtProvider.cuda. CUDA provides GPU acceleration using NVIDIA GPUs.
Values
- useNone → const CUDAFlags
-
Use default CUDA provider options
const CUDAFlags(0x000) - useArena → const CUDAFlags
-
Use CUDA memory arena for allocations Can improve performance by reducing allocation overhead
const CUDAFlags(0x001) - useCudnnConvAlgoSearch → const CUDAFlags
-
Enable CUDNN convolution algorithm search Finds the fastest convolution algorithm for your model
const CUDAFlags(0x002) - doCopyInDefaultStream → const CUDAFlags
-
Copy inputs/outputs on the same stream as compute Can improve performance in some cases
const CUDAFlags(0x004) - enableCudaGraph → const CUDAFlags
-
Enable CUDA graph capture for better performance Reduces kernel launch overhead
const CUDAFlags(0x008)
Properties
- hashCode → int
-
The hash code for this object.
no setterinherited
- index → int
-
A numeric identifier for the enumerated value.
no setterinherited
- name → String
-
Available on Enum, provided by the EnumName extension
The name of the enum value.no setter - runtimeType → Type
-
A representation of the runtime type of the object.
no setterinherited
- value → int
-
no setteroverride
Methods
-
noSuchMethod(
Invocation invocation) → dynamic -
Invoked when a nonexistent method or property is accessed.
inherited
-
toString(
) → String -
A string representation of this object.
inherited
Operators
-
operator ==(
Object other) → bool -
The equality operator.
inherited