OpenCL Backend

Generate OpenCL kernel code from dew expressions.

Enable

dew-scalar = { version = "0.1", features = ["opencl"] }
dew-linalg = { version = "0.1", features = ["opencl"] }

dew-scalar

Generate Expression

use dew_core::Expr;
use dew_scalar::opencl::emit_opencl;

let expr = Expr::parse("sin(x) + cos(y)").unwrap();
let ocl = emit_opencl(expr.ast()).unwrap();

println!("{}", ocl.code);
// Output: (sin(x) + cos(y))

Generate Kernel Function

use dew_scalar::opencl::emit_opencl_fn;

let expr = Expr::parse("x * x + y * y").unwrap();
let ocl = emit_opencl_fn("distance_squared", expr.ast(), &["x", "y"]).unwrap();

println!("{}", ocl);
// Output:
// float distance_squared(float x, float y) {
//     return ((x * x) + (y * y));
// }

dew-linalg

Generate with Types

use dew_core::Expr;
use dew_linalg::opencl::emit_opencl;
use dew_linalg::Type;
use std::collections::HashMap;

let expr = Expr::parse("dot(a, b)").unwrap();

let mut var_types: HashMap<String, Type> = HashMap::new();
var_types.insert("a".to_string(), Type::Vec3);
var_types.insert("b".to_string(), Type::Vec3);

let result = emit_opencl(expr.ast(), &var_types).unwrap();

println!("{}", result.code);
// Output: dot(a, b)

OpenCL Features

Built-in Vector Types

OpenCL provides native vector types:

dew Type	OpenCL Type
Scalar	float
Vec2	float2
Vec3	float3
Vec4	float4

Built-in Vector Operations

OpenCL has built-in functions for common vector operations:

dew	OpenCL
dot(a, b)	dot(a, b)
cross(a, b)	cross(a, b)
length(v)	length(v)
normalize(v)	normalize(v)
distance(a, b)	distance(a, b)

Vector Constructors

// OpenCL uses cast-style constructors
(float2)(x, y)
(float3)(x, y, z)
(float4)(x, y, z, w)

Function Mapping

dew	OpenCL
sin(x)	sin(x)
cos(x)	cos(x)
sqrt(x)	sqrt(x)
rsqrt(x)	rsqrt(x)
abs(x)	fabs(x)
floor(x)	floor(x)
x ^ y	pow(x, y)
min(a, b)	fmin(a, b)
max(a, b)	fmax(a, b)
pi()	M_PI_F
lerp(a, b, t)	mix(a, b, t)
clamp(x, lo, hi)	clamp(x, lo, hi)

Comparison with CUDA/HIP

Feature	OpenCL	CUDA/HIP
Platform	Cross-platform	NVIDIA/AMD
Vector constructors	(float2)(x, y)	make_float2(x, y)
Built-in dot/cross	Yes	Requires helpers
Math functions	sin()	sinf()
Constants	M_PI_F	M_PI

Use Cases

• Cross-platform GPU compute: Run on NVIDIA, AMD, Intel, and mobile GPUs
• Heterogeneous computing: CPU + GPU on same platform
• FPGA acceleration: OpenCL targets some FPGAs