Crate torsh_core

Crate torsh_core 

Source
Expand description

Core types and traits for the ToRSh deep learning framework

This crate provides fundamental building blocks used throughout ToRSh, including error types, device abstractions, and core traits.

Re-exports§

pub use backend_detection::BackendFeatureDetector;
pub use backend_detection::BackendSummary;
pub use backend_detection::DeviceInfo;
pub use backend_detection::PerformanceTier;
pub use backend_detection::RuntimeFeatures;
pub use backend_detection::WorkloadType;
pub use device::Device;
pub use device::DeviceCapabilities;
pub use device::DeviceType;
pub use dtype::AutoPromote;
pub use dtype::Complex32;
pub use dtype::Complex64;
pub use dtype::ComplexElement;
pub use dtype::DType;
pub use dtype::FloatElement;
pub use dtype::QInt8;
pub use dtype::QUInt8;
pub use dtype::TensorElement;
pub use dtype::TypePromotion;
pub use error::ErrorLocation;
pub use error::Result;
pub use error::TorshError;
pub use ffi::TorshDType;
pub use ffi::TorshDevice;
pub use ffi::TorshErrorCode;
pub use ffi::TorshShape;
pub use interop::ArrowDataType;
pub use interop::ArrowTypeInfo;
pub use interop::ConversionUtils;
pub use interop::FromExternal;
pub use interop::FromExternalZeroCopy;
pub use interop::InteropDocs;
pub use interop::NumpyArrayInfo;
pub use interop::OnnxDataType;
pub use interop::OnnxTensorInfo;
pub use interop::ToExternal;
pub use interop::ToExternalZeroCopy;
pub use memory_debug::detect_memory_leaks;
pub use memory_debug::generate_memory_report;
pub use memory_debug::get_memory_stats;
pub use memory_debug::init_memory_debugger;
pub use memory_debug::init_memory_debugger_with_config;
pub use memory_debug::record_allocation;
pub use memory_debug::record_deallocation;
pub use memory_debug::AllocationInfo;
pub use memory_debug::AllocationPattern;
pub use memory_debug::DebuggingAllocator;
pub use memory_debug::MemoryDebugConfig;
pub use memory_debug::MemoryDebugger;
pub use memory_debug::MemoryLeak;
pub use memory_debug::MemoryReport;
pub use memory_debug::MemoryStats;
pub use memory_debug::SystemDebuggingAllocator;
pub use memory_monitor::AllocationStrategy;
pub use memory_monitor::MemoryMonitorConfig;
pub use memory_monitor::MemoryPressure;
pub use memory_monitor::MemoryPressureThresholds;
pub use memory_monitor::SystemMemoryMonitor;
pub use memory_monitor::SystemMemoryStats;
pub use profiling::get_profiler;
pub use profiling::init_profiler;
pub use profiling::profile_closure;
pub use profiling::OperationContext;
pub use profiling::OperationHandle;
pub use profiling::OperationRecord;
pub use profiling::OperationStats;
pub use profiling::OperationType;
pub use profiling::PerformanceBottleneck;
pub use profiling::PerformanceProfiler;
pub use profiling::ProfilerConfig;
pub use shape::Shape;
pub use simd_arm::ArmSimdOps;
pub use sparse::CompressionStats;
pub use sparse::CooIndices;
pub use sparse::CooStorage;
pub use sparse::CsrIndices;
pub use sparse::CsrStorage;
pub use sparse::SparseFormat;
pub use sparse::SparseMetadata;
pub use sparse::SparseStorage;
pub use storage::allocate_pooled;
pub use storage::clear_pooled_memory;
pub use storage::deallocate_pooled;
pub use storage::pooled_memory_stats;
pub use storage::MemoryFormat;
pub use storage::MemoryPool;
pub use storage::PoolStats;
pub use storage::SharedStorage;
pub use storage::Storage;
pub use storage::StorageView;
pub use scirs2;

Modules§

backend_detection
Backend feature detection system for runtime capability discovery
device
Device abstraction and management system
dtype
error
Error types for ToRSh - Clean Modular Interface
error_recovery
Error recovery mechanisms for graceful degradation
examples
Comprehensive examples for torsh-core modules
ffi
FFI-safe type wrappers for C/C++ integration
inspector
Tensor Inspector
interop
Interoperability traits and utilities for ToRSh
memory_debug
Memory debugging and allocation tracking tools for ToRSh
memory_monitor
Advanced system memory monitoring with platform-specific APIs
prelude
Prelude module for convenient imports
profiling
Performance Profiling Hooks for ToRSh Operations
shape
Shape and stride utilities for tensors
shape_debug
Shape Debugging Utilities
simd_arm
ARM NEON SIMD optimizations for ToRSh core operations
sparse
storage
Storage system for tensor data management

Macros§

impl_custom_tensor_element
Helper macro for easily implementing CustomTensorElement for simple types
index_error
Convenience macro for index errors
profile_operation
Convenience macro for profiling operations
profile_shape_op
Macro for easy shape operation profiling
shape_mismatch_error
Convenience macro for shape mismatch errors
torsh_error_with_location
Convenience macros for error creation with location information

Constants§

VERSION
VERSION_MAJOR
VERSION_MINOR
VERSION_PATCH