`dot()`

Function in Builtins

dot(a, b): dot product, vector to scalar.

import { dot } from 'typeshade'

The signature, the description and the examples come from the compiler's own source at commit 26de7be8.

Syntax

function dot<K extends `vec${number}<f64>`>(
  a: ReadonlyNode<K>,
  b: NoInfer<ReadonlyNode<K>>,
): Node<'f64'>;
function dot<K extends `vec${number}<i32>`>(
  a: ReadonlyNode<K>,
  b: NoInfer<ReadonlyNode<K>>,
): Node<'i32'>;
function dot<K extends `vec${number}<u32>`>(
  a: ReadonlyNode<K>,
  b: NoInfer<ReadonlyNode<K>>,
): Node<'u32'>;
function dot<K extends `vec${number}<f32>`>(
  a: ReadonlyNode<K>,
  b: NoInfer<ReadonlyNode<K>>,
): Node<'f32'>;

Parameters

a ReadonlyNode<K>
b NoInfer<ReadonlyNode<K>>

Return value

Node<'f64'>

Description

Like length, the return precision follows the operand key: f32 vectors return f32, emulated-double vec${N}<f64> vectors return f64. Both operands share one key, so dot(v2, v3) is a tsc error.

An integer dot returns that integer kind and takes the dotI / dotU id, because GLSL ES 3.00 has no integer dot at all and a WebGL2 driver refuses the call; the ids carry a GLSL column that expands to the multiply-add the spelling is missing. The choice is divergentIntegerId, shared with the "use typeshade" front end.

Examples

Example

import { fn, dot, vec3, f32T } from 'typeshade'

const luma = fn('luma', { x: f32T }, ({ x }) => dot(vec3(x, x, x), vec3(0.2, 0.7, 0.1)))

Targets

Target	Support	Notes
WGSL (WebGPU)	Supported	`dot(a, b)`. Over `f64` operands the fp64 pass rewrites it into its `df64` helper.
GLSL ES 3.00 (WebGL2)	Supported	`dot(a, b)`. Over `f64` operands the fp64 pass rewrites it into its `df64` helper.
CPU oracle	Supported	Evaluated in f64 by the oracle.

In the guide

fp64

Source

src/core/ir/node.ts, line 1821, at commit 26de7be8

Edit this page Report a problem

dot()