Skip to content

GitLab

Commit ac92c086 authored by peastman's avatar peastman
Browse files

Eliminated uses of __builtin_shuffle() in AltiVec code, since it produced bugs

parent de1d245a
Hide whitespace changes
Inline Side-by-side
Showing with 17 additions and 12 deletions
openmmapi/include/openmm/internal/vectorize_ppc.h
View file @ ac92c086
...@@ -249,25 +249,30 @@ static inline float dot3(const fvec4& v1, const fvec4& v2) { ...@@ -249,25 +249,30 @@ static inline float dot3(const fvec4& v1, const fvec4& v2) {
static inline float dot4(const fvec4& v1, const fvec4& v2) { static inline float dot4(const fvec4& v1, const fvec4& v2) {
fvec4 r = v1*v2; fvec4 r = v1*v2;
fvec4 temp = __builtin_shuffle(r.val, r.val, (__m128i) {0, 1, -1, -1})+__builtin_shuffle(r.val, r.val, (__m128i) {2, 3, -1, -1}); fvec4 temp = r + vec_sld(r.val, r.val, 8);
return temp[0]+temp[1]; return temp[0]+temp[1];
} }
static inline fvec4 cross(const fvec4& v1, const fvec4& v2) { static inline fvec4 cross(const fvec4& v1, const fvec4& v2) {
__m128 temp = v2.val*__builtin_shuffle(v1.val, v1.val, (__m128i) {2, 0, 1, 3}) - vector unsigned char perm = (vector unsigned char) {8, 9, 10, 11, 0, 1, 2, 3, 4, 5, 6, 7, 12, 13, 14, 15};
v1.val*__builtin_shuffle(v2.val, v2.val, (__m128i) {2, 0, 1, 3}); __m128 temp = v2.val*vec_perm(v1.val, v1.val, perm) -
return __builtin_shuffle(temp, temp, (__m128i) {2, 0, 1, 3}); v1.val*vec_perm(v2.val, v2.val, perm);
return vec_perm(temp, temp, perm);
} }
static inline void transpose(fvec4& v1, fvec4& v2, fvec4& v3, fvec4& v4) { static inline void transpose(fvec4& v1, fvec4& v2, fvec4& v3, fvec4& v4) {
__m128 a1 = __builtin_shuffle(v1.val, v2.val, (__m128i) {0, 4, 2, 6}); vector unsigned char perm1 = (vector unsigned char) {0, 1, 2, 3, 16, 17, 18, 19, 8, 9, 10, 11, 24, 25, 26, 27};
__m128 a2 = __builtin_shuffle(v1.val, v2.val, (__m128i) {1, 5, 3, 7}); vector unsigned char perm2 = (vector unsigned char) {4, 5, 6, 7, 20, 21, 22, 23, 12, 13, 14, 15, 28, 29, 30, 31};
__m128 a3 = __builtin_shuffle(v3.val, v4.val, (__m128i) {0, 4, 2, 6}); __m128 a1 = vec_perm(v1.val, v2.val, perm1);
__m128 a4 = __builtin_shuffle(v3.val, v4.val, (__m128i) {1, 5, 3, 7}); __m128 a2 = vec_perm(v1.val, v2.val, perm2);
v1 = __builtin_shuffle(a1, a3, (__m128i) {0, 1, 4, 5}); __m128 a3 = vec_perm(v3.val, v4.val, perm1);
v2 = __builtin_shuffle(a2, a4, (__m128i) {0, 1, 4, 5}); __m128 a4 = vec_perm(v3.val, v4.val, perm2);
v3 = __builtin_shuffle(a1, a3, (__m128i) {2, 3, 6, 7}); vector unsigned char perm3 = (vector unsigned char) {0, 1, 2, 3, 4, 5, 6, 7, 16, 17, 18, 19, 20, 21, 22, 23};
v4 = __builtin_shuffle(a2, a4, (__m128i) {2, 3, 6, 7}); vector unsigned char perm4 = (vector unsigned char) {8, 9, 10, 11, 12, 13, 14, 15, 24, 25, 26, 27, 28, 29, 30, 31};
v1 = vec_perm(a1, a3, perm3);
v2 = vec_perm(a2, a4, perm3);
v3 = vec_perm(a1, a3, perm4);
v4 = vec_perm(a2, a4, perm4);
} }
// Functions that operate on ivec4s. // Functions that operate on ivec4s.
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment