diff --git a/include/cglm/affine.h b/include/cglm/affine.h
index 869422b..e160bdb 100644
--- a/include/cglm/affine.h
+++ b/include/cglm/affine.h
@@ -55,7 +55,7 @@ glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest);
 CGLM_INLINE
 void
 glm_translate_to(mat4 m, vec3 v, mat4 dest) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat4 t = GLM_MAT4_IDENTITY_INIT;
 
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glmm_store(dest[3],
@@ -82,7 +82,7 @@ glm_translate_to(mat4 m, vec3 v, mat4 dest) {
   glm_vec4_add(v2, t[3], t[3]);
   glm_vec4_add(v3, t[3], t[3]);
 
-  glm__memcpy(float, dest, t, sizeof(mat4));
+  glm_mat4_copy(t, dest);
 #endif
 }
 
@@ -254,7 +254,7 @@ glm_scale(mat4 m, vec3 v) {
 CGLM_INLINE
 void
 glm_scale_uni(mat4 m, float s) {
-  vec3 v = { s, s, s };
+  CGLM_ALIGN(8) vec3 v = { s, s, s };
   glm_scale_to(m, v, m);
 }
 
@@ -269,7 +269,7 @@ glm_scale_uni(mat4 m, float s) {
 CGLM_INLINE
 void
 glm_rotate_x(mat4 m, float angle, mat4 dest) {
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat4 t = GLM_MAT4_IDENTITY_INIT;
   float c, s;
 
   c = cosf(angle);
@@ -294,7 +294,7 @@ glm_rotate_x(mat4 m, float angle, mat4 dest) {
 CGLM_INLINE
 void
 glm_rotate_y(mat4 m, float angle, mat4 dest) {
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat4 t = GLM_MAT4_IDENTITY_INIT;
   float c, s;
 
   c = cosf(angle);
@@ -319,7 +319,7 @@ glm_rotate_y(mat4 m, float angle, mat4 dest) {
 CGLM_INLINE
 void
 glm_rotate_z(mat4 m, float angle, mat4 dest) {
-  mat4  t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat4 t = GLM_MAT4_IDENTITY_INIT;
   float c, s;
 
   c = cosf(angle);
@@ -345,7 +345,7 @@ glm_rotate_z(mat4 m, float angle, mat4 dest) {
 CGLM_INLINE
 void
 glm_rotate_make(mat4 m, float angle, vec3 axis) {
-  vec3  axisn, v, vs;
+  CGLM_ALIGN(8) vec3 axisn, v, vs;
   float c;
 
   c = cosf(angle);
@@ -376,7 +376,7 @@ glm_rotate_make(mat4 m, float angle, vec3 axis) {
 CGLM_INLINE
 void
 glm_rotate(mat4 m, float angle, vec3 axis) {
-  mat4 rot;
+  CGLM_ALIGN(16) mat4 rot;
   glm_rotate_make(rot, angle, axis);
   glm_mul_rot(m, rot, m);
 }
@@ -393,7 +393,7 @@ glm_rotate(mat4 m, float angle, vec3 axis) {
 CGLM_INLINE
 void
 glm_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis) {
-  vec3 pivotInv;
+  CGLM_ALIGN(8) vec3 pivotInv;
 
   glm_vec_inv_to(pivot, pivotInv);
 
@@ -418,12 +418,11 @@ glm_rotate_at(mat4 m, vec3 pivot, float angle, vec3 axis) {
 CGLM_INLINE
 void
 glm_rotate_atm(mat4 m, vec3 pivot, float angle, vec3 axis) {
-  vec3 pivotInv;
+  CGLM_ALIGN(8) vec3 pivotInv;
 
   glm_vec_inv_to(pivot, pivotInv);
 
-  glm_mat4_identity(m);
-  glm_vec_copy(pivot, m[3]);
+  glm_translate_make(m, pivot);
   glm_rotate(m, angle, axis);
   glm_translate(m, pivotInv);
 }
@@ -453,9 +452,8 @@ glm_decompose_scalev(mat4 m, vec3 s) {
 CGLM_INLINE
 bool
 glm_uniscaled(mat4 m) {
-  vec3 s;
+  CGLM_ALIGN(8) vec3 s;
   glm_decompose_scalev(m, s);
-
   return glm_vec_eq_all(s);
 }
 
@@ -470,8 +468,8 @@ glm_uniscaled(mat4 m) {
 CGLM_INLINE
 void
 glm_decompose_rs(mat4 m, mat4 r, vec3 s) {
-  vec4 t = {0.0f, 0.0f, 0.0f, 1.0f};
-  vec3 v;
+  CGLM_ALIGN(16) vec4 t = {0.0f, 0.0f, 0.0f, 1.0f};
+  CGLM_ALIGN(8)  vec3 v;
 
   glm_vec4_copy(m[0], r[0]);
   glm_vec4_copy(m[1], r[1]);
diff --git a/include/cglm/cam.h b/include/cglm/cam.h
index acd573c..28b8a7d 100644
--- a/include/cglm/cam.h
+++ b/include/cglm/cam.h
@@ -332,7 +332,7 @@ glm_lookat(vec3 eye,
            vec3 center,
            vec3 up,
            mat4 dest) {
-  vec3 f, u, s;
+  CGLM_ALIGN(8) vec3 f, u, s;
 
   glm_vec_sub(center, eye, f);
   glm_vec_normalize(f);
@@ -372,7 +372,7 @@ glm_lookat(vec3 eye,
 CGLM_INLINE
 void
 glm_look(vec3 eye, vec3 dir, vec3 up, mat4 dest) {
-  vec3 target;
+  CGLM_ALIGN(8) vec3 target;
   glm_vec_add(eye, dir, target);
   glm_lookat(eye, target, up, dest);
 }
@@ -390,7 +390,7 @@ glm_look(vec3 eye, vec3 dir, vec3 up, mat4 dest) {
 CGLM_INLINE
 void
 glm_look_anyup(vec3 eye, vec3 dir, mat4 dest) {
-  vec3 up;
+  CGLM_ALIGN(8) vec3 up;
   glm_vec_ortho(dir, up);
   glm_look(eye, dir, up, dest);
 }
diff --git a/include/cglm/mat3.h b/include/cglm/mat3.h
index 87bf9b1..1df3a3b 100644
--- a/include/cglm/mat3.h
+++ b/include/cglm/mat3.h
@@ -81,7 +81,7 @@ glm_mat3_copy(mat3 mat, mat3 dest) {
 CGLM_INLINE
 void
 glm_mat3_identity(mat3 mat) {
-  mat3 t = GLM_MAT3_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat3 t = GLM_MAT3_IDENTITY_INIT;
   glm_mat3_copy(t, mat);
 }
 
@@ -155,7 +155,7 @@ glm_mat3_transpose_to(mat3 m, mat3 dest) {
 CGLM_INLINE
 void
 glm_mat3_transpose(mat3 m) {
-  mat3 tmp;
+  CGLM_ALIGN(16) mat3 tmp;
 
   tmp[0][1] = m[1][0];
   tmp[0][2] = m[2][0];
diff --git a/include/cglm/mat4.h b/include/cglm/mat4.h
index f0b6736..6d345f3 100644
--- a/include/cglm/mat4.h
+++ b/include/cglm/mat4.h
@@ -139,7 +139,7 @@ glm_mat4_copy(mat4 mat, mat4 dest) {
 CGLM_INLINE
 void
 glm_mat4_identity(mat4 mat) {
-  mat4 t = GLM_MAT4_IDENTITY_INIT;
+  CGLM_ALIGN(16) mat4 t = GLM_MAT4_IDENTITY_INIT;
   glm_mat4_copy(t, mat);
 }
 
@@ -586,7 +586,7 @@ glm_mat4_inv_fast(mat4 mat, mat4 dest) {
 CGLM_INLINE
 void
 glm_mat4_swap_col(mat4 mat, int col1, int col2) {
-  vec4 tmp;
+  CGLM_ALIGN(16) vec4 tmp;
   glm_vec4_copy(mat[col1], tmp);
   glm_vec4_copy(mat[col2], mat[col1]);
   glm_vec4_copy(tmp, mat[col2]);
@@ -602,7 +602,7 @@ glm_mat4_swap_col(mat4 mat, int col1, int col2) {
 CGLM_INLINE
 void
 glm_mat4_swap_row(mat4 mat, int row1, int row2) {
-  vec4 tmp;
+  CGLM_ALIGN(16) vec4 tmp;
   tmp[0] = mat[0][row1];
   tmp[1] = mat[1][row1];
   tmp[2] = mat[2][row1];
diff --git a/include/cglm/project.h b/include/cglm/project.h
index c71e735..d4295d3 100644
--- a/include/cglm/project.h
+++ b/include/cglm/project.h
@@ -100,7 +100,7 @@ glm_unproject(vec3 pos, mat4 m, vec4 vp, vec3 dest) {
 CGLM_INLINE
 void
 glm_project(vec3 pos, mat4 m, vec4 vp, vec3 dest) {
-  vec4 pos4, vone = GLM_VEC4_ONE_INIT;
+  CGLM_ALIGN(16) vec4 pos4, vone = GLM_VEC4_ONE_INIT;
 
   glm_vec4(pos, 1.0f, pos4);
 
diff --git a/include/cglm/quat.h b/include/cglm/quat.h
index eac853b..fd4f25f 100644
--- a/include/cglm/quat.h
+++ b/include/cglm/quat.h
@@ -99,7 +99,7 @@ glm_translate(mat4 m, vec3 v);
 CGLM_INLINE
 void
 glm_quat_identity(versor q) {
-  versor v = GLM_QUAT_IDENTITY_INIT;
+  CGLM_ALIGN(16) versor v = GLM_QUAT_IDENTITY_INIT;
   glm_vec4_copy(v, q);
 }
 
@@ -131,7 +131,7 @@ glm_quat_init(versor q, float x, float y, float z, float w) {
 CGLM_INLINE
 void
 glm_quatv(versor q, float angle, vec3 axis) {
-  vec3  k;
+  CGLM_ALIGN(8) vec3 k;
   float a, c, s;
 
   a = angle * 0.5f;
@@ -158,7 +158,7 @@ glm_quatv(versor q, float angle, vec3 axis) {
 CGLM_INLINE
 void
 glm_quat(versor q, float angle, float x, float y, float z) {
-  vec3 axis = {x, y, z};
+  CGLM_ALIGN(8) vec3 axis = {x, y, z};
   glm_quatv(q, angle, axis);
 }
 
@@ -267,7 +267,7 @@ glm_quat_conjugate(versor q, versor dest) {
 CGLM_INLINE
 void
 glm_quat_inv(versor q, versor dest) {
-  versor conj;
+  CGLM_ALIGN(8) versor conj;
   glm_quat_conjugate(q, conj);
   glm_vec4_scale(conj, 1.0f / glm_vec4_norm2(q), dest);
 }
@@ -603,7 +603,7 @@ glm_quat_lerp(versor from, versor to, float t, versor dest) {
 CGLM_INLINE
 void
 glm_quat_slerp(versor from, versor to, float t, versor dest) {
-  vec4  q1, q2;
+  CGLM_ALIGN(16) vec4 q1, q2;
   float cosTheta, sinTheta, angle;
 
   cosTheta = glm_quat_dot(from, to);
@@ -646,7 +646,7 @@ glm_quat_slerp(versor from, versor to, float t, versor dest) {
 CGLM_INLINE
 void
 glm_quat_look(vec3 eye, versor ori, mat4 dest) {
-  vec4 t;
+  CGLM_ALIGN(16) vec4 t;
 
   /* orientation */
   glm_quat_mat4t(ori, dest);
@@ -668,7 +668,7 @@ glm_quat_look(vec3 eye, versor ori, mat4 dest) {
 CGLM_INLINE
 void
 glm_quat_for(vec3 dir, vec3 fwd, vec3 up, versor dest) {
-  vec3  axis;
+  CGLM_ALIGN(8) vec3 axis;
   float dot, angle;
 
   dot = glm_vec_dot(dir, fwd);
@@ -702,7 +702,7 @@ glm_quat_for(vec3 dir, vec3 fwd, vec3 up, versor dest) {
 CGLM_INLINE
 void
 glm_quat_forp(vec3 from, vec3 to, vec3 fwd, vec3 up, versor dest) {
-  vec3 dir;
+  CGLM_ALIGN(8) vec3 dir;
   glm_vec_sub(to, from, dir);
   glm_quat_for(dir, fwd, up, dest);
 }
@@ -717,9 +717,9 @@ glm_quat_forp(vec3 from, vec3 to, vec3 fwd, vec3 up, versor dest) {
 CGLM_INLINE
 void
 glm_quat_rotatev(versor q, vec3 v, vec3 dest) {
-  versor p;
-  vec3   u, v1, v2;
-  float  s;
+  CGLM_ALIGN(16) versor p;
+  CGLM_ALIGN(8)  vec3   u, v1, v2;
+  float s;
 
   glm_quat_normalize_to(q, p);
   glm_quat_imag(p, u);
@@ -745,7 +745,7 @@ glm_quat_rotatev(versor q, vec3 v, vec3 dest) {
 CGLM_INLINE
 void
 glm_quat_rotate(mat4 m, versor q, mat4 dest) {
-  mat4 rot;
+  CGLM_ALIGN(16) mat4 rot;
   glm_quat_mat4(q, rot);
   glm_mul_rot(m, rot, dest);
 }
@@ -760,7 +760,7 @@ glm_quat_rotate(mat4 m, versor q, mat4 dest) {
 CGLM_INLINE
 void
 glm_quat_rotate_at(mat4 m, versor q, vec3 pivot) {
-  vec3 pivotInv;
+  CGLM_ALIGN(8) vec3 pivotInv;
 
   glm_vec_inv_to(pivot, pivotInv);
 
@@ -784,12 +784,11 @@ glm_quat_rotate_at(mat4 m, versor q, vec3 pivot) {
 CGLM_INLINE
 void
 glm_quat_rotate_atm(mat4 m, versor q, vec3 pivot) {
-  vec3 pivotInv;
+  CGLM_ALIGN(8) vec3 pivotInv;
 
   glm_vec_inv_to(pivot, pivotInv);
 
-  glm_mat4_identity(m);
-  glm_vec_copy(pivot, m[3]);
+  glm_translate_make(m, pivot);
   glm_quat_rotate(m, q, m);
   glm_translate(m, pivotInv);
 }
diff --git a/include/cglm/vec3.h b/include/cglm/vec3.h
index 50636a1..34a910c 100644
--- a/include/cglm/vec3.h
+++ b/include/cglm/vec3.h
@@ -454,8 +454,7 @@ glm_vec_inv(vec3 v) {
 CGLM_INLINE
 void
 glm_vec_inv_to(vec3 v, vec3 dest) {
-  glm_vec_copy(v, dest);
-  glm_vec_flipsign(dest);
+  glm_vec_flipsign_to(v, dest);
 }
 
 /*!