commit 9198aede8761d6573f2ec3739b48e86f83947b56
parent 3c12def93ed667548bcd9c534c31ec31bfefb8e0
Author: Sebastiano Tronto <sebastiano@tronto.net>
Date: Tue, 15 Oct 2024 20:01:46 +0200
Merge pull request #4 from enricotenuti/master
Neon uint8x8_t corners and minor fixes
Diffstat:
4 files changed, 51 insertions(+), 58 deletions(-)
diff --git a/src/arch/arch.h b/src/arch/arch.h
@@ -15,7 +15,7 @@ typedef __m256i cube_t;
#include <arm_neon.h>
typedef struct {
- uint8x16_t corner;
+ uint8x8_t corner;
uint8x16_t edge;
} cube_t;
diff --git a/src/arch/neon.h b/src/arch/neon.h
@@ -1,24 +1,24 @@
-#define CO2_NEON vdupq_n_u8(0x60)
-#define COCW_NEON vdupq_n_u8(0x20)
-#define CP_NEON vdupq_n_u8(0x07)
+#define CO2_NEON vdup_n_u8(0x60)
+#define COCW_NEON vdup_n_u8(0x20)
+#define CP_NEON vdup_n_u8(0x07)
#define EP_NEON vcombine_u8(vdupq_n_u8(0x0F), vdupq_n_u8(0x0F))
#define EO_NEON vcombine_u8(vdupq_n_u8(0x10), vdupq_n_u8(0x10))
STATIC_INLINE uint8x16_t compose_edges_slim(uint8x16_t, uint8x16_t);
-STATIC_INLINE uint8x16_t compose_corners_slim(uint8x16_t, uint8x16_t);
+STATIC_INLINE uint8x8_t compose_corners_slim(uint8x8_t, uint8x8_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}, \
+ .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, 0, 0, 0, 0}})
// zero cube
#define ZERO_CUBE \
(cube_t) \
{ \
- .corner = vdupq_n_u8(0), \
+ .corner = vdup_n_u8(0), \
.edge = vdupq_n_u8(0) \
}
@@ -30,7 +30,7 @@ STATIC void
pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
{
// First 8 bytes of the corner vector are copied from the c array
- vst1_u8(c, vget_low_u8(cube->corner));
+ vst1_u8(c, cube->corner);
// 12 bytes of the edge vector are copied from the e array
// First 8 bytes
@@ -42,20 +42,17 @@ pieces(cube_t *cube, uint8_t c[static 8], uint8_t e[static 12])
STATIC_INLINE bool
equal(cube_t c1, cube_t c2)
{
- uint8x16_t cmp_corner, cmp_edge;
- uint64x2_t cmp_corner_u64, cmp_edge_u64;
- uint64x2_t cmp_result;
+ uint8x8_t cmp_corner;
+ uint8x16_t cmp_edge;
+ uint64x2_t cmp_corner_u64, cmp_edge_u64, cmp_result;
- // compare the corner vectors
- cmp_corner = vceqq_u8(c1.corner, c2.corner);
- // compare the edge vectors
+ // compare the corner vectors and the edge vectors
+ cmp_corner = vceq_u8(c1.corner, c2.corner);
cmp_edge = vceqq_u8(c1.edge, c2.edge);
- // convert the comparison vectors to 64-bit vectors
- cmp_corner_u64 = vreinterpretq_u64_u8(cmp_corner);
+ // convert the comparison vectors to 64-bit vectors and combine them
+ cmp_corner_u64 = vreinterpretq_u64_u8(vcombine_u64(cmp_corner, cmp_corner));
cmp_edge_u64 = vreinterpretq_u64_u8(cmp_edge);
-
- // combine the comparison vectors
cmp_result = vandq_u64(cmp_corner_u64, cmp_edge_u64);
// check if all the bits are set
@@ -66,15 +63,15 @@ STATIC_INLINE cube_t
invertco(cube_t c)
{
cube_t ret;
- uint8x16_t co, shleft, shright, summed, newco, cleanco;
-
- co = vandq_u8(c.corner, CO2_NEON);
- shleft = vshlq_n_u8(co, 1);
- shright = vshrq_n_u8(co, 1);
- summed = vorrq_u8(shleft, shright);
- newco = vandq_u8(summed, CO2_NEON);
- cleanco = veorq_u8(c.corner, co);
- ret.corner = vorrq_u8(cleanco, newco);
+ uint8x8_t co, shleft, shright, summed, newco, cleanco;
+
+ co = vand_u8(c.corner, CO2_NEON);
+ shleft = vshl_n_u8(co, 1);
+ shright = vshr_n_u8(co, 1);
+ summed = vorr_u8(shleft, shright);
+ newco = vand_u8(summed, CO2_NEON);
+ cleanco = veor_u8(c.corner, co);
+ ret.corner = vorr_u8(cleanco, newco);
ret.edge = c.edge;
return ret;
@@ -120,30 +117,25 @@ compose_edges_slim(uint8x16_t edge1, uint8x16_t edge2)
return ret;
}
-STATIC_INLINE uint8x16_t
-compose_corners_slim(uint8x16_t corner1, uint8x16_t corner2)
+STATIC_INLINE uint8x8_t
+compose_corners_slim(uint8x8_t corner1, uint8x8_t corner2)
{
// Masks
- uint8x16_t p_bits = vdupq_n_u8(PBITS);
- uint8x16_t cobits = vdupq_n_u8(COBITS);
- uint8x16_t cobits2 = vdupq_n_u8(COBITS_2);
- uint8x16_t twist_cw = vdupq_n_u8(CTWIST_CW);
+ uint8x8_t p_bits = vdup_n_u8(PBITS);
+ uint8x8_t cobits = vdup_n_u8(COBITS);
+ uint8x8_t cobits2 = vdup_n_u8(COBITS_2);
+ uint8x8_t twist_cw = vdup_n_u8(CTWIST_CW);
// Find the index and permutation
- uint8x16_t p = vandq_u8(corner2, p_bits);
- uint8x16_t piece1 = vqtbl1q_u8(corner1, p);
+ uint8x8_t p = vand_u8(corner2, p_bits);
+ uint8x8_t piece1 = vtbl1_u8(corner1, p);
// Calculate the orientation
- uint8x16_t aux = vaddq_u8(vandq_u8(corner2, cobits), vandq_u8(piece1, cobits));
- uint8x16_t auy = vshrq_n_u8(vaddq_u8(aux, twist_cw), 2);
- uint8x16_t orien = vandq_u8(vaddq_u8(aux, auy), cobits2);
-
- // Combine the results
- uint8x16_t ret = vorrq_u8(vandq_u8(piece1, p_bits), orien);
+ uint8x8_t aux = vadd_u8(vand_u8(corner2, cobits), vand_u8(piece1, cobits));
+ uint8x8_t auy = vshr_n_u8(vadd_u8(aux, twist_cw), 2);
+ uint8x8_t orien = vand_u8(vadd_u8(aux, auy), cobits2);
- // Mask to clear the last 64 bits of the result
- uint8x16_t mask_last_64 = vsetq_lane_u64(0, vreinterpretq_u64_u8(ret), 1);
- ret = vreinterpretq_u8_u64(mask_last_64);
+ uint8x8_t ret = vorr_u8(vand_u8(piece1, p_bits), orien);
return ret;
}
@@ -167,14 +159,14 @@ inverse(cube_t cube)
// Temp arrays to store the NEON vectors
uint8_t edges[16];
- uint8_t corners[16];
+ uint8_t corners[8];
// Copy the NEON vectors to the arrays
vst1q_u8(edges, cube.edge);
- vst1q_u8(corners, cube.corner);
+ vst1_u8(corners, cube.corner);
uint8_t edge_result[16] = {0};
- uint8_t corner_result[16] = {0};
+ uint8_t corner_result[8] = {0};
// Process the edges
for (i = 0; i < 12; i++)
@@ -194,7 +186,7 @@ inverse(cube_t cube)
// Copy the results back to the NEON vectors
ret.edge = vld1q_u8(edge_result);
- ret.corner = vld1q_u8(corner_result);
+ ret.corner = vld1_u8(corner_result);
return ret;
}
@@ -203,8 +195,8 @@ STATIC_INLINE int64_t
coord_co(cube_t c)
{
// Temp array to store the NEON vector
- uint8_t mem[16];
- vst1q_u8(mem, c.corner);
+ uint8_t mem[8];
+ vst1_u8(mem, c.corner);
int i, p;
int64_t ret;
@@ -219,8 +211,8 @@ STATIC_INLINE int64_t
coord_csep(cube_t c)
{
// Temp array to store the NEON vector
- uint8_t mem[16];
- vst1q_u8(mem, c.corner);
+ uint8_t mem[8];
+ vst1_u8(mem, c.corner);
int64_t ret = 0;
int i, p;
@@ -228,6 +220,7 @@ coord_csep(cube_t c)
ret += p * ((mem[i] & CSEPBIT) >> 2);
return ret;
+ return 0;
}
STATIC_INLINE int64_t
diff --git a/src/solvers/h48/solve.h b/src/solvers/h48/solve.h
@@ -19,7 +19,7 @@ typedef struct {
uint8_t premoves[MAXLEN];
} dfsarg_solveh48_t;
-STATIC uint32_t allowednextmove_h48(uint8_t *, uint8_t, uint32_t);
+STATIC uint32_t allowednextmove_h48(uint8_t *, uint8_t, uint8_t);
STATIC void solve_h48_appendsolution(dfsarg_solveh48_t *);
STATIC_INLINE bool solve_h48_stop(dfsarg_solveh48_t *);
@@ -28,7 +28,7 @@ STATIC int64_t solve_h48(cube_t, int8_t, int8_t,
int8_t, uint64_t, const void *, uint64_t, char *);
STATIC uint32_t
-allowednextmove_h48(uint8_t *moves, uint8_t n, uint32_t h48branch)
+allowednextmove_h48(uint8_t *moves, uint8_t n, uint8_t h48branch)
{
uint32_t result = MM_ALLMOVES;
if (h48branch & MM_NORMALBRANCH)
diff --git a/src/utils/constants.h b/src/utils/constants.h
@@ -93,10 +93,10 @@ STATIC int64_t binomial[12][12] = {
#define TRANS_BDm UINT8_C(46)
#define TRANS_BLm UINT8_C(47)
-#define MM_NORMAL UINT32_C(0x00)
-#define MM_INVERSE UINT32_C(0x01)
-#define MM_INVERSEBRANCH UINT32_C(0x03)
-#define MM_NORMALBRANCH UINT32_C(0x02)
+#define MM_NORMAL UINT8_C(0x00)
+#define MM_INVERSE UINT8_C(0x01)
+#define MM_INVERSEBRANCH UINT8_C(0x03)
+#define MM_NORMALBRANCH UINT8_C(0x02)
#define MM_ALLMOVES UINT32_C(0x3FFFF)
#define MM_NOHALFTURNS UINT32_C(0x2DB6D)
