Crate yscv_tensor 

yscv-tensor

SIMD-accelerated tensor library. 115 Tensor operations in ops.rs, f32/f16/bf16 support, NumPy-style broadcasting, 32-byte aligned memory.

use yscv_tensor::Tensor;

let a = Tensor::from_vec(vec![2, 3], vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0])?;
let b = Tensor::ones(vec![2, 3]);
let c = (&a + &b)?;

Features

• Data types: f32, f16 (IEEE 754), bf16 (Brain Float)
• Broadcasting: automatic shape expansion following NumPy rules
• SIMD: runtime dispatch for NEON (aarch64) and SSE2/AVX (x86_64)
• Aligned memory: 32-byte aligned allocations for AVX
• Operations: arithmetic, matmul, transpose, reshape, slice, gather, scatter, reduce, clamp, pad, concat, split, topk, sort

Optional Features

[features]
mkl = []      # Intel MKL BLAS backend
armpl = []    # Arm Performance Libraries backend

Tests

207 tests covering shapes, broadcasting, dtypes, edge cases.

Structs

AlignedVec

A Vec-like container that guarantees 32-byte alignment for the data pointer.

Tensor

A compact, contiguous multi-dtype tensor representation.

Enums

DType

Element data type for tensor storage.

Device

Logical device where a tensor resides.

TensorError

Errors returned by tensor construction and math operations.

Constants

CRATE_ID