feat(simd): update & enable swizzle4_u32_aos()

- fix swizzle impl (byte lane order)
- add/update tests

packages/simd/assembly/index.ts

@@ -2,11 +2,8 @@ export * from "./abs";
 export * from "./add";
 export * from "./addn";
 export * from "./clamp";
-
-// TODO waiting for native impl
 export * from "./div";
 export * from "./divn";
-
 export * from "./dot";
 export * from "./madd";
 export * from "./maddn";
@@ -23,11 +20,8 @@ export * from "./muln";
 export * from "./mulv";
 export * from "./neg";
 export * from "./normalize";
-
-// TODO waiting for native impl
 export * from "./sqrt";
-
 export * from "./sub";
 export * from "./subn";
 export * from "./sum";
-// export * from "./swizzle";
+export * from "./swizzle";

packages/simd/assembly/swizzle.ts

@@ -31,7 +31,38 @@ export function set_lane_f32(
     return res;
 }
 
-export function swizzle4_32(
+/**
+ * Swaps, reorders or replaces vector components in an AOS f32/u32 vec4
+ * buffer. The `x`,`y`,`z`,`w` args indicate the intended lane values
+ * (each 0-3).
+ *
+ * @example
+ * ```ts
+ * simd.f32.set([10, 20, 30, 40], 0)
+ * simd.swizzle4_f32(
+ *   16,         // dest ptr
+ *   0,          // src ptr
+ *   3, 0, 1, 2, // lane IDs
+ *   1,          // num vectors
+ *   4,          // output stride (in f32/u32)
+ *   4           // input stride
+ * )
+ *
+ * simd.f32.slice(4, 8)
+ * // [40, 10, 20, 30]
+ * ```
+ *
+ * @param out -
+ * @param a -
+ * @param x -
+ * @param y -
+ * @param z -
+ * @param w -
+ * @param num -
+ * @param so -
+ * @param sa -
+ */
+export function swizzle4_32_aos(
     out: usize,
     a: usize,
     x: u32,
@@ -46,18 +77,15 @@ export function swizzle4_32(
     so <<= 2;
     sa <<= 2;
     // create swizzle pattern from xyzw
-    // each lane: id * 0x04040404 + 0x00010203
-    // TODO verify order
+    // each lane: id * 0x04040404 + 0x03020100
     let mask = i64x2.replace_lane(
         i64x2.splat(<u64>y * 0x0404040400000000 + <u64>x * 0x0000000004040404),
         1,
         <u64>w * 0x0404040400000000 + <u64>z * 0x0000000004040404
     );
-    mask = i64x2.add(mask, i32x4.splat(0x00010203));
-    v128.store(out, mask);
-
+    mask = i64x2.add(mask, i32x4.splat(0x03020100));
     for (; num-- > 0; ) {
-        // v128.store(out, v128.swizzle(v128.load(a), mask));
+        v128.store(out, v128.swizzle(v128.load(a), mask));
         out += so;
         a += sa;
     }

packages/simd/src/api.ts

@@ -180,6 +180,9 @@ export interface SIMD {
 
     sum4_f32(a: number, num: number, sa: number): number;
 
+    // prettier-ignore
+    swizzle4_32_aos(out: number, a: number, x: number, y: number, z: number, w: number, num: number, so: number, sa: number): number;
+
     /**
      * WASM memory instance given to `init()`.
      */