std/simd

Zen-C provides native SIMD (Single Instruction, Multiple Data) vector types that compile directly to hardware-optimized vector instructions (SSE, AVX, NEON, etc.) supported by the target backend.

Overview

• Native Performance: Leverages LLVM/GCC vector extensions for maximum efficiency.
• Implicitly Portable: Types like f32x4 map to the best available 128-bit hardware specific to the architecture.
• Element-wise arithmetic: Standard operators (+, -, *, /) apply to all lanes simultaneously.
• Broadcasting: Initializing with a single value broadcasts it to all lanes.

Usage

import "std/simd.zc"

fn main() {
    // Initialization (Explicit lanes)
    let a = f32x4 { 1.0, 2.0, 3.0, 4.0 };
    
    // Broadcasting (Single value to all lanes)
    let b = f32x4 { v: 2.0 };
    
    // Element-wise addition
    let c = a + b;   // Result: { 3.0, 4.0, 5.0, 6.0 }
    
    // Lane Access
    let first = c[0];
}

Vector Types

The standard library defines several 128-bit and 256-bit vector types. You can also define custom ones using the @vector(N) attribute.

128-bit Vectors (SSE / NEON)

Type	Element Type	Lanes	Byte Size
f32x4	f32	4	16
f64x2	f64	2	16
i32x4	i32	4	16
u32x4	u32	4	16
i64x2	i64	2	16
u64x2	u64	2	16
i16x8	i16	8	16
u16x8	u16	8	16
i8x16	i8	16	16
u8x16	u8	16	16

256-bit Vectors (AVX / AVX2)

Type	Element Type	Lanes	Byte Size
f32x8	f32	8	32
f64x4	f64	4	32
i32x8	i32	8	32
u32x8	u32	8	32
i64x4	i64	4	32
u64x4	u64	4	32
i16x16	i16	16	32
u16x16	u16	16	32
i8x32	i8	32	32
u8x32	u8	32	32

Operations

Category	Operator	Description
Arithmetic	+, -, *, /	Standard element-wise addition, subtraction, multiplication, and division.
Bitwise	&, |, ^, ~	Bitwise AND, OR, XOR, and NOT across all lanes.
Indexing	[i]	Access or modify individual lanes by index.
Comparison	==, !=, <, >	Returns a boolean mask vector (results vary by backend).