diff --git a/glm/detail/func_geometric_simd.inl b/glm/detail/func_geometric_simd.inl
index 4c7f56bf..809ce70f 100644
--- a/glm/detail/func_geometric_simd.inl
+++ b/glm/detail/func_geometric_simd.inl
@@ -79,6 +79,17 @@ namespace detail
 		}
 	};
 
+	template<>
+	struct compute_normalize<4, float, lowp, true>
+	{
+		GLM_FUNC_QUALIFIER static vec<4, float, lowp> call(vec<4, float, lowp> const& v)
+		{
+			vec<4, float, lowp> Result;
+			Result.data = glm_vec4_fast_normalize(v.data);
+			return Result;
+		}
+	};
+
 	template<qualifier Q>
 	struct compute_faceforward<4, float, Q, true>
 	{
diff --git a/glm/simd/geometric.h b/glm/simd/geometric.h
index 07d7cbcc..680dda49 100644
--- a/glm/simd/geometric.h
+++ b/glm/simd/geometric.h
@@ -75,6 +75,15 @@ GLM_FUNC_QUALIFIER glm_vec4 glm_vec4_cross(glm_vec4 v1, glm_vec4 v2)
 }
 
 GLM_FUNC_QUALIFIER glm_vec4 glm_vec4_normalize(glm_vec4 v)
+{
+	glm_vec4 const dot0 = glm_vec4_dot(v, v);
+	glm_vec4 const sqr0 = _mm_sqrt_ps(dot0);
+	glm_vec4 const isr0 = _mm_div_ps(_mm_set1_ps(1.0f), sqr0);
+	glm_vec4 const mul0 = _mm_mul_ps(v, isr0);
+	return mul0;
+}
+
+GLM_FUNC_QUALIFIER glm_vec4 glm_vec4_fast_normalize(glm_vec4 v)
 {
 	glm_vec4 const dot0 = glm_vec4_dot(v, v);
 	glm_vec4 const isr0 = _mm_rsqrt_ps(dot0);