File QnnGpuGraph.h

↰ Parent directory (include/QNN/GPU)

A header which defines the QNN GPU specialization of the File QnnGraph.h interface.

Definition (include/QNN/GPU/QnnGpuGraph.h)

Includes

Full File Listing

A header which defines the QNN GPU specialization of the QnnGraph.h interface.

Defines

QNN_GPU_GRAPH_CUSTOM_CONFIG_INIT   {                                                                   \     QNN_GPU_PRECISION_USER_PROVIDED

,   /*precision*/                  \

0u,                                /*disableMemoryOptimizations*/ \

0u,                                /*disableNodeOptimizations*/   \

0u                                 /*disableQueueRecording*/      \

}


QnnGpuGraph_CustomConfig_t initializer macro.

Enums

enum QnnGpu_Precision_t

An enum which defines the different tensor optimization options. A tensor may be optimized to the specified QnnGpu_Precision_t when it is a graph tensor that is not a graph input or a graph output and does not connect two operations from different op packages.

Values:

enumerator QNN_GPU_PRECISION_FP32 = 0

Sets the precision mode to floating point 32-bit (FP32)

enumerator QNN_GPU_PRECISION_FP16 = 1

Sets the precision mode to floating point 16-bit (FP16)

enumerator QNN_GPU_PRECISION_HYBRID = 2

Sets the precision mode to FP16 for storage and FP32 for calculations.

enumerator QNN_GPU_PRECISION_USER_PROVIDED = 3

Uses the tensor data type provided by the user (default)

struct QnnGpuGraph_CustomConfig_t
#include <QnnGpuGraph.h>

A struct which defines the QNN GPU graph custom configuration options. Objects of this type are to be referenced through QnnGraph_CustomConfig_t.

Public Members

QnnGpu_Precision_t precision
uint8_t disableMemoryOptimizations
uint8_t disableNodeOptimizations
uint8_t disableQueueRecording