From 027947cae6ef1e881e541d68a50d01ad7a559a50 Mon Sep 17 00:00:00 2001
From: Jorijn van der Graaf <jorijnvdgraaf@catcrafts.net>
Date: Mon, 18 May 2026 18:15:12 +0200
Subject: [PATCH] wasm dot

---
 interfaces/Crafter.Math-VectorF32.cppm | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/interfaces/Crafter.Math-VectorF32.cppm b/interfaces/Crafter.Math-VectorF32.cppm
index cc86764..6582d94 100755
--- a/interfaces/Crafter.Math-VectorF32.cppm
+++ b/interfaces/Crafter.Math-VectorF32.cppm
@@ -1578,6 +1578,32 @@ namespace Crafter {
             return sq;
         }
 
+        // Four pairwise dot products packed into one v128. Only the first Len
+        // lanes contribute, so the same routine handles 3- and 4-component
+        // inputs — the 4th lane of Len==3 inputs may be garbage from Cross()
+        // and must not be summed.
+        constexpr static VectorF32<1, 4> Dot(
+            VectorF32<Len, Packing> A0, VectorF32<Len, Packing> A1,
+            VectorF32<Len, Packing> B0, VectorF32<Len, Packing> B1,
+            VectorF32<Len, Packing> C0, VectorF32<Len, Packing> C1,
+            VectorF32<Len, Packing> D0, VectorF32<Len, Packing> D1
+        ) requires((Len == 3 || Len == 4) && Packing == 1) {
+            alignas(16) float a0[4], a1[4], b0[4], b1[4], c0[4], c1[4], d0[4], d1[4];
+            wasm_v128_store(a0, A0.v); wasm_v128_store(a1, A1.v);
+            wasm_v128_store(b0, B0.v); wasm_v128_store(b1, B1.v);
+            wasm_v128_store(c0, C0.v); wasm_v128_store(c1, C1.v);
+            wasm_v128_store(d0, D0.v); wasm_v128_store(d1, D1.v);
+
+            alignas(16) float out[4] = {0,0,0,0};
+            for (std::uint8_t k = 0; k < Len; ++k) {
+                out[0] += a0[k] * a1[k];
+                out[1] += b0[k] * b1[k];
+                out[2] += c0[k] * c1[k];
+                out[3] += d0[k] * d1[k];
+            }
+            return VectorF32<1, 4>(wasm_v128_load(out));
+        }
+
         template<typename... Rest>
             requires((std::is_same_v<Rest, VectorF32<Len, Packing>> && ...))
         constexpr static auto Normalize(VectorF32<Len, Packing> first, Rest... rest) {