{% macro flat_index_2d(workgroupSize, name="i", bound="params.count", guardInline=false) %}
{% set wgTerm = workgroupSize ~ "u" if workgroupSize is number else workgroupSize %}
  // 2D-folded flat index: gid.y carries the high bits past the dispatch's
  // per-axis workgroup fold width.
  let {{ name }} = gid.x + gid.y * {{ DISPATCH_FOLD_WIDTH }}u * {{ wgTerm }};
  if ({{ name }} >= {{ bound }}) {
    return;
  }{% endmacro %}
{{ env.wgsl.resourceDeclarations }}

// Rank-2 [N, C] inference vec4 specialization. Vectors run across adjacent channels, so
// scale/bias/mean/var are also bound as vec4<f32> and C must be divisible by 4.
@compute @workgroup_size({{ tunables.WORKGROUP_SIZE }})
fn main(@builtin(global_invocation_id) gid: vec3<u32>) {
{{ flat_index_2d(tunables.WORKGROUP_SIZE, "i", "params.count4") }}
  let channel4 = i % params.channels4;
  let alpha = scale[channel4] * inverseSqrt(input_var[channel4] + vec4<f32>(params.epsilon));
  // Subtract the mean before scaling. The expanded form
  // x * alpha + (bias - mean * alpha) can lose the small residual (x - mean)
  // when |mean| is much larger than |x - mean|.
  y[i] = (x[i] - input_mean[channel4]) * alpha + bias[channel4];
}