24 #ifndef ARM_COMPUTE_ACL_TYPES_H_ 25 #define ARM_COMPUTE_ACL_TYPES_H_ 121 void *(*aligned_alloc)(
void *
user_data,
size_t size,
size_t alignment);
struct AclTensor_ * AclTensor
Opaque Tensor pack object.
struct AclTensorPack_ * AclTensorPack
Opaque Operator object.
Enable bfloat16 data-type support.
Call failed during execution.
Prioritize performance when a single iterations is expected to be performed.
int32_t compute_units
Compute Units that the queue will deploy.
int64_t * strides
Strides on each dimension.
Normal tuning mode, gives a good balance between tuning mode and performance.
void(* free)(void *user_data, void *ptr)
Release a block of size bytes of memory.
Call failed as invalid argument was passed.
Automatic discovery of capabilities.
Native, performant, Compute Library data layout.
Enable SVE optimized paths.
const char * kernel_config_file
Kernel cofiguration file.
int32_t max_compute_units
Max compute units that can be used by a queue created from the context.
Enable paths that use the mmla integer instructions.
Enable NEON optimized paths.
int32_t ndims
Number or dimensions.
AclExecutionMode mode
Execution mode to use.
AclTargetCapabilities capabilities
Target capabilities.
uint64_t AclTargetCapabilities
Error codes returned by the public entry-points.
Call failed as invalid argument was passed.
Call failed as an invalid backend was requested.
Cpu target that uses SIMD extensions.
AclTuningMode mode
Tuning mode.
struct AclContextOptions AclContextOptions
Supported tuning modes.
struct AclOperator_ * AclOperator
int64_t boffset
Offset in terms of bytes for the first element.
Call failed as configuration is unsupported.
Call succeeded, leading to valid state for all involved objects/data.
struct AclQueue_ * AclQueue
Opaque Tensor object.
Fast tuning mode, testing a small portion of the tuning space.
Call failed as requested capability is not implemented.
Enable SVE2 optimized paths.
Data layout where width is the fastest changing dimension.
void(* aligned_free)(void *user_data, void *ptr)
Allocate a block of size bytes of memory.
16-bit brain floating point
AclDataType data_type
Tensor Data type.
AclImportMemoryType
Type of memory to be imported.
struct AclContext_ * AclContext
< Opaque Context object
Call failed due to failure to allocate resources.
bool enable_fast_math
Allow precision loss.
struct AclTensorDescriptor AclTensorDescriptor
Slot type of a tensor.
Enable paths that use the mmla float instructions.
AclCpuCapabilities
Available CPU capabilities.
Exhaustive tuning mode, increased tuning time but with best results.
Prioritize performance when multiple iterations are performed.
Enable float16 data-type support.
AclAllocator * allocator
Allocator to be used by all the memory internally.
Enable paths that use the udot/sdot instructions.
int32_t * shape
Tensor Shape.
Call failed as an object has invalid state.
struct AclAllocator AclAllocator
Context options.
AclExecutionMode
Execution mode types.