Crate slsl
Source - add
- div
- mul
- sub
- arrayn
- Creates an
ArrayN from a list of values at compile time. - impl_asum_half
- Generate L1 norm (sum of absolute values) operations for half precision types
- impl_asum_signed
- Generate L1 norm (sum of absolute values) operations for signed numeric types
- impl_asum_unsigned
- Generate L1 norm (sum of absolute values) operations for unsigned numeric types
- impl_dot
- Generate dot product implementations for integer types that return larger types to avoid overflow
- impl_gemm
- Generate GEMM implementations for other numeric types (fallback to simple implementation)
- impl_gemm_half
- Generate general matrix multiplication for half precision types using unified gemm library
- impl_gemm_sd
- Generate general matrix multiplication for all numeric types
Generate optimized GEMM implementations for f32 and f64 using unified gemm library
- impl_max_i
- Generate vectorized maximum index for all numeric types
- impl_max_i_half
- Generate vectorized maximum index for half precision types
- impl_max_v
- Generate vectorized maximum value for all numeric types
- impl_max_v_half
- Generate vectorized maximum value for half precision types
- impl_max_vi
- Generate vectorized maximum value with index for all numeric types
- impl_max_vi_half
- Generate vectorized maximum value with index for half precision types
- impl_mean_float
- Generate mean operations for floating point types
- impl_mean_half
- Generate mean operations for half precision types
- impl_mean_int
- Generate mean operations for integer types
- impl_min_i
- Generate vectorized minimum index for all numeric types
- impl_min_i_half
- Generate vectorized minimum index for half precision types
- impl_min_max_i
- Generate vectorized min and max indices for all numeric types
- impl_min_max_i_half
- Generate vectorized min and max indices for half precision types
- impl_min_max_v
- Generate vectorized min and max values for all numeric types
- impl_min_max_v_half
- Generate vectorized min and max values for half precision types
- impl_min_max_vi
- Generate vectorized min and max values with indices for all numeric types
- impl_min_max_vi_half
- Generate vectorized min and max values with indices for half precision types
- impl_min_v
- Generate vectorized minimum value for all numeric types
- impl_min_v_half
- Generate vectorized minimum value for half precision types
- impl_min_vi
- Generate vectorized minimum value with index for all numeric types
- impl_min_vi_half
- Generate vectorized minimum value with index for half precision types
- impl_relu
- Generate vectorized ReLU operation for floating point types
- impl_relu_half
- Generate vectorized ReLU operation for half precision types
- impl_relu_int
- Generate vectorized ReLU operation for integer types
- impl_relu_uint
- Generate vectorized ReLU operation for unsigned integer types
- impl_scal
- Generate scale operations for all numeric types
- impl_sum_int
- Generate sum operations for integer types
- impl_v_abs
- Generate vectorized absolute value for all numeric types
- impl_v_abs_half
- Generate vectorized absolute value for f16 and bf16
- impl_v_add
- Generate vectorized element-wise addition for all numeric types
- impl_v_add_half
- Generate vectorized element-wise addition for half precision types
- impl_v_add_scalar
- Generate scalar addition operations for all numeric types
- impl_v_add_scalar_half
- Generate scalar addition operations for half precision types
- impl_v_ceil
- Generate vectorized ceiling function for all numeric types
- impl_v_ceil_half
- Generate vectorized ceiling function for f16 and bf16
- impl_v_cos
- Generate vectorized cosine function for all numeric types
- impl_v_cos_half
- Generate vectorized cosine function for f16 and bf16
- impl_v_div
- Generate vectorized element-wise division for all numeric types
- impl_v_div_half
- Generate vectorized element-wise division for half precision types
- impl_v_div_scalar
- Generate scalar division operations for all numeric types
- impl_v_div_scalar_half
- Generate scalar division operations for half precision types
- impl_v_exp
- Generate vectorized exponential function for all numeric types
- impl_v_exp_half
- Generate vectorized exponential function for f16 and bf16
- impl_v_floor
- Generate vectorized floor function for all numeric types
- impl_v_floor_half
- Generate vectorized floor function for f16 and bf16
- impl_v_log
- Generate vectorized natural logarithm for all numeric types
- impl_v_log_half
- Generate vectorized natural logarithm for f16 and bf16
- impl_v_mul
- Generate vectorized element-wise multiplication for all numeric types
- impl_v_mul_half
- Generate vectorized element-wise multiplication for half precision types
- impl_v_mul_scalar
- Generate scalar multiplication operations for all numeric types
- impl_v_mul_scalar_half
- Generate scalar multiplication operations for half precision types
- impl_v_neg
- Generate vectorized negation for signed numeric types
- impl_v_neg_half
- Generate vectorized negation for f16 and bf16
- impl_v_pow
- Generate vectorized power function for all numeric types
- impl_v_recip
- Generate vectorized reciprocal function for all numeric types
- impl_v_recip_half
- Generate vectorized reciprocal function for f16 and bf16
- impl_v_round
- Generate vectorized round function for all numeric types
- impl_v_round_half
- Generate vectorized round function for f16 and bf16
- impl_v_sin
- Generate vectorized sine function for all numeric types
- impl_v_sin_half
- Generate vectorized sine for f16 and bf16
- impl_v_sqr
- Generate vectorized square (element-wise) for all numeric types
- impl_v_sqr_half
- Generate vectorized square (element-wise) for f16 and bf16
- impl_v_sqrt
- Generate vectorized square root for all numeric types
- impl_v_sqrt_half
- Generate vectorized square root for f16 and bf16
- impl_v_sub
- Generate vectorized element-wise subtraction for all numeric types
- impl_v_sub_half
- Generate vectorized element-wise subtraction for half precision types
- impl_v_sub_scalar
- Generate scalar subtraction operations for all numeric types
- impl_v_sub_scalar_half
- Generate scalar subtraction operations for half precision types
- impl_v_tan
- Generate vectorized tangent function for all numeric types
- impl_v_tan_half
- Generate vectorized tangent function for f16 and bf16
- impl_v_tanh
- Generate vectorized hyperbolic tangent for all numeric types
- impl_v_tanh_half
- Generate vectorized hyperbolic tangent for f16 and bf16
- s
- Creates slice specifications for tensor slicing operations.
- ArrayN
- A fixed-capacity array with dynamic length.
- DimIter
- High-performance iterator over tensor dimensions
- Storage
- Thread-safe reference-counted storage for tensor data.
- TensorBase
- Core tensor structure that can work with different storage backends.
- TensorTypeIter
- Index-based iterator - mirrors the for_at() approach
- TensorTypeMetaItem
- A typed iterator yielding indices, byte offset and value reference
- TensorTypeMetaIter
- Typed iterator with metadata - thin wrapper around TensorTypeIter
- UninitVec
- A safe wrapper for uninitialized Vec operations
- DType
- Enumeration of all supported data types for tensor elements.
- SliceElem
- MAX_DIM
- Dim
- Trait for types that can be converted to a single dimension index.
- Dims
- Trait for types that can be converted to multiple dimension indices.
- IntoSliceElem
- OpsTrait
- OpsTrait trait for accelerated tensor operations
- StorageTrait
- Trait for types that can provide storage functionality.
- TensorElement
- Trait for types that can be used as tensor elements
- WideSimd
- get_num_threads
- global_backend
- Get a reference to the global backend instance
- reduce_shape_stride
- ArrayUsize8
- Type alias for 8-element array with usize elements.
- Backend
- Shape
- Type alias for representing tensor shapes.
- SliceSpecs
- Stride
- Type alias for representing memory strides.
- Tensor
- Type alias for owned tensors with
Storage backend. - TensorView
- Type alias for tensor views that borrow from existing storage.