Table-based transformations (will revert immediately)

This commit changes transformations to being based on a lookup
table intead of a big switch. This is something I had planned for a long
time and the code is nicer with this change. But unfortunately the performance
is noticeably worse, so I'll revert all these changes in the next commit.
I wanted to push this anyway so we can keep track of this.

4 files changed, 115 insertions, 40 deletions

diff --git a/src/arch/avx2.h b/src/arch/avx2.h
index 2c3371e..557ab49 100644
--- a/src/arch/avx2.h
+++ b/src/arch/avx2.h
@@ -6,16 +6,6 @@
 #define EO_AVX2 \
     _mm256_set_epi64x(INT64_C(0x10101010), INT64_C(0x1010101010101010), 0, 0)
 
-#define STATIC_CUBE(c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl, \
-    e_uf, e_ub, e_db, e_df, e_ur, e_ul, e_dl, e_dr, e_fr, e_fl, e_bl, e_br) \
-    _mm256_set_epi8(0, 0, 0, 0, e_br, e_bl, e_fl, e_fr, \
-        e_dr, e_dl, e_ul, e_ur, e_df, e_db, e_ub, e_uf, \
-        0, 0, 0, 0, 0, 0, 0, 0, \
-        c_dbl, c_dfr, c_ubr, c_ufl, c_dbr, c_dfl, c_ubl, c_ufr)
-#define ZERO_CUBE _mm256_set_epi64x(0, 0, 0, 0)
-#define SOLVED_CUBE STATIC_CUBE( \
-    0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
-
 STATIC void
 pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
 {
@@ -26,6 +16,38 @@ pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
         memcpy(e, aux+16, 12);
 }
 
+STATIC_INLINE cube_t
+cubefrompieces(
+        uint8_t c_ufr,
+        uint8_t c_ubl,
+        uint8_t c_dfl,
+        uint8_t c_dbr,
+        uint8_t c_ufl,
+        uint8_t c_ubr,
+        uint8_t c_dfr,
+        uint8_t c_dbl, 
+
+        uint8_t e_uf,
+        uint8_t e_ub,
+        uint8_t e_db,
+        uint8_t e_df,
+        uint8_t e_ur,
+        uint8_t e_ul,
+        uint8_t e_dl,
+        uint8_t e_dr,
+        uint8_t e_fr,
+        uint8_t e_fl,
+        uint8_t e_bl,
+        uint8_t e_br
+) {
+        return _mm256_set_epi8(
+            0, 0, 0, 0, e_br, e_bl, e_fl, e_fr,
+            e_dr, e_dl, e_ul, e_ur, e_df, e_db, e_ub, e_uf,
+            0, 0, 0, 0, 0, 0, 0, 0,
+            c_dbl, c_dfr, c_ubr, c_ufl, c_dbr, c_dfl, c_ubl, c_ufr
+        );
+}
+
 STATIC_INLINE bool
 equal(cube_t c1, cube_t c2)
 {
diff --git a/src/arch/common.h b/src/arch/common.h
index 346403b..1469b30 100644
--- a/src/arch/common.h
+++ b/src/arch/common.h
@@ -1,4 +1,11 @@
+#define ZERO_CUBE cubefrompieces( \
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)
+#define SOLVED_CUBE cubefrompieces( \
+    0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
 STATIC void pieces(cube_t *, uint8_t [static 8], uint8_t [static 12]);
+STATIC_INLINE cube_t cubefrompieces(uint8_t, uint8_t, uint8_t, uint8_t,
+    uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t,
+    uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t, uint8_t);
 STATIC_INLINE bool equal(cube_t, cube_t);
 STATIC_INLINE cube_t invertco(cube_t);
 STATIC_INLINE cube_t compose_epcpeo(cube_t, cube_t);
diff --git a/src/arch/neon.h b/src/arch/neon.h
index acb75c3..9061422 100644
--- a/src/arch/neon.h
+++ b/src/arch/neon.h
@@ -7,25 +7,6 @@
 STATIC_INLINE uint8x16_t compose_edges_slim(uint8x16_t, uint8x16_t);
 STATIC_INLINE uint8x16_t compose_corners_slim(uint8x16_t, uint8x16_t);
 
-// static cube
-#define STATIC_CUBE(c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl, \
-                                        e_uf, e_ub, e_db, e_df, e_ur, e_ul, e_dl, e_dr, e_fr, e_fl, e_bl, e_br) \
-        ((cube_t){ \
-                .corner = {c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl, 0, 0, 0, 0, 0, 0, 0, 0}, \
-                .edge = {e_uf, e_ub, e_db, e_df, e_ur, e_ul, e_dl, e_dr, e_fr, e_fl, e_bl, e_br, 0, 0, 0, 0}})
-
-// zero cube
-#define ZERO_CUBE \
-        (cube_t) \
-        { \
-                .corner = vdupq_n_u8(0), \
-                .edge = vdupq_n_u8(0) \
-        }
-
-// solved cube
-#define SOLVED_CUBE STATIC_CUBE(        \
-        0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
-
 STATIC void
 pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
 {
@@ -39,6 +20,44 @@ pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
         vst1_lane_u32((uint32_t *)(e + 8), vreinterpret_u32_u8(vget_high_u8(cube->edge)), 0);
 }
 
+STATIC_INLINE cube_t
+cubefrompieces(
+        uint8_t c_ufr,
+        uint8_t c_ubl,
+        uint8_t c_dfl,
+        uint8_t c_dbr,
+        uint8_t c_ufl,
+        uint8_t c_ubr,
+        uint8_t c_dfr,
+        uint8_t c_dbl, 
+
+        uint8_t e_uf,
+        uint8_t e_ub,
+        uint8_t e_db,
+        uint8_t e_df,
+        uint8_t e_ur,
+        uint8_t e_ul,
+        uint8_t e_dl,
+        uint8_t e_dr,
+        uint8_t e_fr,
+        uint8_t e_fl,
+        uint8_t e_bl,
+        uint8_t e_br
+) {
+        cube_t ret = {
+                .corner = {
+                        c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl,
+                        0, 0, 0, 0, 0, 0, 0, 0
+                },
+                .edge = {
+                        e_uf, e_ub, e_db, e_df, e_ur, e_ul, e_dl, e_dr, e_fr,
+                        e_fl, e_bl, e_br, 0, 0, 0, 0
+                }
+        };
+
+        return ret;
+}
+
 STATIC_INLINE bool
 equal(cube_t c1, cube_t c2)
 {
diff --git a/src/arch/portable.h b/src/arch/portable.h
index e9a8dd0..3776c90 100644
--- a/src/arch/portable.h
+++ b/src/arch/portable.h
@@ -1,14 +1,3 @@
-#define STATIC_CUBE(c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl, \
-    e_uf, e_ub, e_db, e_df, e_ur, e_ul, e_dl, e_dr, e_fr, e_fl, e_bl, e_br) \
-    ((cube_t) { \
-        .corner = { c_ufr, c_ubl, c_dfl, c_dbr, c_ufl, c_ubr, c_dfr, c_dbl }, \
-        .edge = { e_uf, e_ub, e_db, e_df, e_ur, e_ul, \
-                  e_dl, e_dr, e_fr, e_fl, e_bl, e_br } })
-#define ZERO_CUBE STATIC_CUBE( \
-    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)
-#define SOLVED_CUBE STATIC_CUBE( \
-    0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
-
 STATIC void
 pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
 {
@@ -16,6 +5,44 @@ pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
         memcpy(e, cube->edge, 12);
 }
 
+STATIC_INLINE cube_t
+cubefrompieces(
+        uint8_t c_ufr,
+        uint8_t c_ubl,
+        uint8_t c_dfl,
+        uint8_t c_dbr,
+        uint8_t c_ufl,
+        uint8_t c_ubr,
+        uint8_t c_dfr,
+        uint8_t c_dbl, 
+
+        uint8_t e_uf,
+        uint8_t e_ub,
+        uint8_t e_db,
+        uint8_t e_df,
+        uint8_t e_ur,
+        uint8_t e_ul,
+        uint8_t e_dl,
+        uint8_t e_dr,
+        uint8_t e_fr,
+        uint8_t e_fl,
+        uint8_t e_bl,
+        uint8_t e_br
+) {
+        cube_t ret = {
+                .corner = {
+                        0, 0, 0, 0, 0, 0, 0, 0,
+                        c_dbl, c_dfr, c_ubr, c_ufl, c_dbr, c_dfl, c_ubl, c_ufr
+                },
+                .edge = {
+                        0, 0, 0, 0, e_br, e_bl, e_fl, e_fr,
+                        e_dr, e_dl, e_ul, e_ur, e_df, e_db, e_ub, e_uf
+                }
+        };
+
+        return ret;
+}
+
 STATIC_INLINE bool
 equal(cube_t c1, cube_t c2)
 {