diff --git a/include/cglm/simd/arm.h b/include/cglm/simd/arm.h
index 64b2dad..405b9d5 100644
--- a/include/cglm/simd/arm.h
+++ b/include/cglm/simd/arm.h
@@ -79,5 +79,25 @@ glmm_norm_inf(float32x4_t a) {
   return glmm_hmax(glmm_abs(a));
 }
 
+static inline
+float32x4_t
+glmm_fmadd(float32x4_t a, float32x4_t b, float32x4_t c) {
+#if defined(__aarch64__)
+  return vfmaq_f32(a, b, c);
+#else
+  return vmlaq_f32(a, b, c);
+#endif
+}
+
+static inline
+float32x4_t
+glmm_fnmadd(float32x4_t a, float32x4_t b, float32x4_t c) {
+#if defined(__aarch64__)
+  return vfmsq_f32(a, b, c);
+#else
+  return vmlsq_f32(a, b, c);
+#endif
+}
+
 #endif
 #endif /* cglm_simd_arm_h */
diff --git a/include/cglm/simd/x86.h b/include/cglm/simd/x86.h
index bbeccb3..2a5716b 100644
--- a/include/cglm/simd/x86.h
+++ b/include/cglm/simd/x86.h
@@ -197,5 +197,25 @@ glmm_store3(float v[3], __m128 vx) {
   _mm_store_ss(&v[2], glmm_shuff1(vx, 2, 2, 2, 2));
 }
 
+static inline
+__m128
+glmm_fmadd(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return  _mm_fmadd_ps(a, b, c);
+#else
+  return _mm_add_ps(c, _mm_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m128
+glmm_fnmadd(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return  _mm_fnmadd_ps(a, b, c);
+#else
+  return _mm_sub_ps(c, _mm_mul_ps(a, b));
+#endif
+}
+
 #endif
 #endif /* cglm_simd_x86_h */