Skip to content
Snippets Groups Projects
Commit bccd8cb3 authored by James Willis's avatar James Willis
Browse files

Removed unused functions.

parent ad6842f9
No related branches found
No related tags found
1 merge request!406Doself2 vectorisation
Hide whitespace changes
Inline Side-by-side
src/runner_doiact_vec.c
+ 0
− 498
View file @ bccd8cb3
...@@ -858,504 +858,6 @@ __attribute__((always_inline)) INLINE void runner_doself1_density_vec( ...@@ -858,504 +858,6 @@ __attribute__((always_inline)) INLINE void runner_doself1_density_vec(
__attribute__((always_inline)) INLINE void runner_doself2_force_vec( __attribute__((always_inline)) INLINE void runner_doself2_force_vec(
struct runner *r, struct cell *restrict c) { struct runner *r, struct cell *restrict c) {
#ifdef WITH_VECTORIZATION
//static int intCount = 0;
const struct engine *e = r->e;
int doi_mask;
struct part *restrict pi;
int count_align;
int num_vec_proc = 1;//NUM_VEC_PROC;
struct part *piq[VEC_SIZE], *pjq[VEC_SIZE];
struct part *restrict parts = c->parts;
const int count = c->count;
vector v_hi, v_hig2, v_r2;
//TIMER_TIC
if (!cell_is_active(c, e)) return;
if (!cell_is_drifted(c, e)) cell_drift_particles(c, e);
/* Get the particle cache from the runner and re-allocate
* the cache if it is not big enough for the cell. */
struct cache *restrict cell_cache = &r->ci_cache;
if (cell_cache->count < count) {
cache_init(cell_cache, count);
}
/* Read the particles from the cell and store them locally in the cache. */
cache_read_particles(c, cell_cache);
/* Create secondary cache to store particle interactions. */
//struct c2_cache int_cache;
//int icount = 0, icount_align = 0;
/* Loop over the particles in the cell. */
for (int pid = 0; pid < count; pid++) {
/* Get a pointer to the ith particle. */
pi = &parts[pid];
/* Is the ith particle active? */
if (!part_is_active(pi, e)) continue;
vector pix, piy, piz;
const float hi = cell_cache->h[pid];
/* Fill particle pi vectors. */
pix.v = vec_set1(cell_cache->x[pid]);
piy.v = vec_set1(cell_cache->y[pid]);
piz.v = vec_set1(cell_cache->z[pid]);
v_hi.v = vec_set1(hi);
const float hig2 = hi * hi * kernel_gamma2;
v_hig2.v = vec_set1(hig2);
/* Reset cumulative sums of update vectors. */
vector a_hydro_xSum, a_hydro_ySum, a_hydro_zSum, h_dtSum, v_sigSum, entropy_dtSum;
/* Get the inverse of hi. */
vector v_hi_inv;
v_hi_inv = vec_reciprocal(v_hi);
a_hydro_xSum.v = vec_setzero();
a_hydro_ySum.v = vec_setzero();
a_hydro_zSum.v = vec_setzero();
h_dtSum.v = vec_setzero();
v_sigSum.v = vec_set1(pi->force.v_sig);
entropy_dtSum.v = vec_setzero();
/* Pad cache if there is a serial remainder. */
count_align = count;
int rem = count % (num_vec_proc * VEC_SIZE);
if (rem != 0) {
int pad = (num_vec_proc * VEC_SIZE) - rem;
count_align += pad;
/* Set positions to the same as particle pi so when the r2 > 0 mask is
* applied these extra contributions are masked out.*/
for (int i = count; i < count_align; i++) {
cell_cache->x[i] = pix.f[0];
cell_cache->y[i] = piy.f[0];
cell_cache->z[i] = piz.f[0];
}
}
vector pjx, pjy, pjz;
//vector pjvx, pjvy, pjvz, mj;
vector hj, hjg2;
//vector pjx2, pjy2, pjz2;
//vector pjvx2, pjvy2, pjvz2, mj2, hj_2, hjg2_2;
for(int k=0; k<VEC_SIZE; k++)
piq[k] = pi;
/* Find all of particle pi's interacions and store needed values in the
* secondary cache.*/
for (int pjd = 0; pjd < count_align; pjd += (num_vec_proc * VEC_SIZE)) {
/* Load 2 sets of vectors from the particle cache. */
pjx.v = vec_load(&cell_cache->x[pjd]);
pjy.v = vec_load(&cell_cache->y[pjd]);
pjz.v = vec_load(&cell_cache->z[pjd]);
//pjvx.v = vec_load(&cell_cache->vx[pjd]);
//pjvy.v = vec_load(&cell_cache->vy[pjd]);
//pjvz.v = vec_load(&cell_cache->vz[pjd]);
//mj.v = vec_load(&cell_cache->m[pjd]);
hj.v = vec_load(&cell_cache->h[pjd]);
hjg2.v = vec_mul(vec_mul(hj.v,hj.v), kernel_gamma2_vec.v);
//pjx2.v = vec_load(&cell_cache->x[pjd + VEC_SIZE]);
//pjy2.v = vec_load(&cell_cache->y[pjd + VEC_SIZE]);
//pjz2.v = vec_load(&cell_cache->z[pjd + VEC_SIZE]);
//pjvx2.v = vec_load(&cell_cache->vx[pjd + VEC_SIZE]);
//pjvy2.v = vec_load(&cell_cache->vy[pjd + VEC_SIZE]);
//pjvz2.v = vec_load(&cell_cache->vz[pjd + VEC_SIZE]);
//mj2.v = vec_load(&cell_cache->m[pjd + VEC_SIZE]);
//hj_2.v = vec_load(&cell_cache->h[pjd + VEC_SIZE]);
//hjg2_2.v = vec_mul(vec_mul(hj_2.v,hj_2.v), kernel_gamma2_vec.v);
vector v_hj_inv;
v_hj_inv = vec_reciprocal(hj);
/* Compute the pairwise distance. */
vector v_dx_tmp, v_dy_tmp, v_dz_tmp;
//vector v_dx_tmp2, v_dy_tmp2, v_dz_tmp2, v_r2_2;
v_dx_tmp.v = vec_sub(pix.v, pjx.v);
//v_dx_tmp2.v = vec_sub(pix.v, pjx2.v);
v_dy_tmp.v = vec_sub(piy.v, pjy.v);
//v_dy_tmp2.v = vec_sub(piy.v, pjy2.v);
v_dz_tmp.v = vec_sub(piz.v, pjz.v);
//v_dz_tmp2.v = vec_sub(piz.v, pjz2.v);
v_r2.v = vec_mul(v_dx_tmp.v, v_dx_tmp.v);
//v_r2_2.v = vec_mul(v_dx_tmp2.v, v_dx_tmp2.v);
v_r2.v = vec_fma(v_dy_tmp.v, v_dy_tmp.v, v_r2.v);
//v_r2_2.v = vec_fma(v_dy_tmp2.v, v_dy_tmp2.v, v_r2_2.v);
v_r2.v = vec_fma(v_dz_tmp.v, v_dz_tmp.v, v_r2.v);
//v_r2_2.v = vec_fma(v_dz_tmp2.v, v_dz_tmp2.v, v_r2_2.v);
/* Form a mask from r2 < hig2 and r2 > 0.*/
#ifdef HAVE_AVX512_F
// KNL_MASK_16 doi_mask, doi_mask_check, doi_mask2, doi_mask2_check;
KNL_MASK_16 doi_mask_check, doi_mask2, doi_mask2_check;
doi_mask_check = vec_cmp_gt(v_r2.v, vec_setzero());
doi_mask = vec_cmp_lt(v_r2.v, v_hig2.v);
doi_mask2_check = vec_cmp_gt(v_r2_2.v, vec_setzero());
doi_mask2 = vec_cmp_lt(v_r2_2.v, v_hig2.v);
doi_mask = doi_mask & doi_mask_check;
doi_mask2 = doi_mask2 & doi_mask2_check;
#else
vector v_doi_mask, v_doi_mask_check, v_doi_N3_mask;
//vector v_doi_mask2, v_doi_mask2_check, v_doi_N3_mask2;
//int doi_mask2;
/* Form r2 > 0 mask, r2 < hig2 mask and r2 < hjg2 mask. */
v_doi_mask_check.v = vec_cmp_gt(v_r2.v, vec_setzero());
v_doi_mask.v = vec_cmp_lt(v_r2.v, v_hig2.v);
v_doi_N3_mask.v = vec_cmp_lt(v_r2.v, hjg2.v);
/* Form r2 > 0 mask and r2 < hig2 mask. */
//v_doi_mask2_check.v = vec_cmp_gt(v_r2_2.v, vec_setzero());
//v_doi_mask2.v = vec_cmp_lt(v_r2_2.v, v_hig2.v);
//v_doi_N3_mask2.v = vec_cmp_lt(v_r2_2.v, v_hjg2_2.v);
v_doi_mask.v = vec_and(vec_add(v_doi_mask.v, v_doi_N3_mask.v), v_doi_mask_check.v);
/* Combine two masks and form integer mask. */
doi_mask = vec_cmp_result(v_doi_mask.v);
//doi_mask2 = vec_cmp_result(vec_add(vec_and(v_doi_mask2.v, v_doi_mask2_check.v), v_doi_N3_mask2.v));
#endif /* HAVE_AVX512_F */
for(int k=0; k<VEC_SIZE; k++)
pjq[k] = &parts[pjd + k];
/* If there are any interactions left pack interaction values into c2
* cache. */
if (doi_mask) {
for(int k=0; k<VEC_SIZE; k++) {
if( v_r2.f[k] == 0.f) v_r2.f[k] = 1.f;
}
//intCount += __builtin_popcount(doi_mask);
runner_iact_nonsym_1_vec_force(&v_r2, &v_dx_tmp, &v_dy_tmp, &v_dz_tmp,
v_hi_inv, v_hj_inv, piq, pjq,
&a_hydro_xSum, &a_hydro_ySum, &a_hydro_zSum,
&h_dtSum, &v_sigSum, &entropy_dtSum, v_doi_mask);
}
}
VEC_HADD(a_hydro_xSum, pi->a_hydro[0]);
VEC_HADD(a_hydro_ySum, pi->a_hydro[1]);
VEC_HADD(a_hydro_zSum, pi->a_hydro[2]);
VEC_HADD(h_dtSum, pi->force.h_dt);
for(int k=0; k<VEC_SIZE; k++)
pi->force.v_sig = max(pi->force.v_sig, v_sigSum.f[k]);
VEC_HADD(entropy_dtSum, pi->entropy_dt);
} /* loop over all particles. */
//TIMER_TOC(timer_doself_force);
#endif /* WITH_VECTORIZATION */
}
/**
* @brief Compute the cell self-interaction (non-symmetric) using vector
* intrinsics with one particle pi at a time.
*
* @param r The #runner.
* @param c The #cell.
*/
__attribute__((always_inline)) INLINE void runner_doself2_force_vec_2(
struct runner *r, struct cell *restrict c) {
#ifdef WITH_VECTORIZATION
//static int intCount = 0;
const struct engine *e = r->e;
int doi_mask;
struct part *restrict pi;
int count_align;
int num_vec_proc = 2;//NUM_VEC_PROC;
struct part *restrict parts = c->parts;
const int count = c->count;
vector v_hi, v_vix, v_viy, v_viz, v_hig2, v_r2;
vector v_rhoi, v_grad_hi, v_pOrhoi2, v_balsara_i, v_ci;
//TIMER_TIC
if (!cell_is_active(c, e)) return;
if (!cell_is_drifted(c, e)) cell_drift_particles(c, e);
/* Get the particle cache from the runner and re-allocate
* the cache if it is not big enough for the cell. */
struct cache *restrict cell_cache = &r->ci_cache;
if (cell_cache->count < count) {
cache_init(cell_cache, count);
}
/* Read the particles from the cell and store them locally in the cache. */
cache_read_particles(c, cell_cache);
/* Create secondary cache to store particle interactions. */
//struct c2_cache int_cache;
//int icount = 0, icount_align = 0;
/* Loop over the particles in the cell. */
for (int pid = 0; pid < count; pid++) {
/* Get a pointer to the ith particle. */
pi = &parts[pid];
/* Is the ith particle active? */
if (!part_is_active(pi, e)) continue;
vector pix, piy, piz;
const float hi = cell_cache->h[pid];
/* Fill particle pi vectors. */
pix.v = vec_set1(cell_cache->x[pid]);
piy.v = vec_set1(cell_cache->y[pid]);
piz.v = vec_set1(cell_cache->z[pid]);
v_hi.v = vec_set1(hi);
v_vix.v = vec_set1(cell_cache->vx[pid]);
v_viy.v = vec_set1(cell_cache->vy[pid]);
v_viz.v = vec_set1(cell_cache->vz[pid]);
v_rhoi.v = vec_set1(cell_cache->rho[pid]);
v_grad_hi.v = vec_set1(cell_cache->grad_h[pid]);
v_pOrhoi2.v = vec_set1(cell_cache->pOrho2[pid]);
v_balsara_i.v = vec_set1(cell_cache->balsara[pid]);
v_ci.v = vec_set1(cell_cache->soundspeed[pid]);
const float hig2 = hi * hi * kernel_gamma2;
v_hig2.v = vec_set1(hig2);
/* Reset cumulative sums of update vectors. */
vector a_hydro_xSum, a_hydro_ySum, a_hydro_zSum, h_dtSum, v_sigSum, entropy_dtSum;
/* Get the inverse of hi. */
vector v_hi_inv;
v_hi_inv = vec_reciprocal(v_hi);
a_hydro_xSum.v = vec_setzero();
a_hydro_ySum.v = vec_setzero();
a_hydro_zSum.v = vec_setzero();
h_dtSum.v = vec_setzero();
v_sigSum.v = vec_set1(pi->force.v_sig);
entropy_dtSum.v = vec_setzero();
/* Pad cache if there is a serial remainder. */
count_align = count;
int rem = count % (num_vec_proc * VEC_SIZE);
if (rem != 0) {
int pad = (num_vec_proc * VEC_SIZE) - rem;
count_align += pad;
/* Set positions to the same as particle pi so when the r2 > 0 mask is
* applied these extra contributions are masked out.*/
for (int i = count; i < count_align; i++) {
cell_cache->x[i] = pix.f[0];
cell_cache->y[i] = piy.f[0];
cell_cache->z[i] = piz.f[0];
cell_cache->h[i] = 1.f;
}
}
vector pjx, pjy, pjz;
vector hj, hjg2;
vector pjx2, pjy2, pjz2;
vector hj_2, hjg2_2;
/* Find all of particle pi's interacions and store needed values in the
* secondary cache.*/
for (int pjd = 0; pjd < count_align; pjd += (num_vec_proc * VEC_SIZE)) {
/* Load 2 sets of vectors from the particle cache. */
pjx.v = vec_load(&cell_cache->x[pjd]);
pjy.v = vec_load(&cell_cache->y[pjd]);
pjz.v = vec_load(&cell_cache->z[pjd]);
hj.v = vec_load(&cell_cache->h[pjd]);
hjg2.v = vec_mul(vec_mul(hj.v,hj.v), kernel_gamma2_vec.v);
pjx2.v = vec_load(&cell_cache->x[pjd + VEC_SIZE]);
pjy2.v = vec_load(&cell_cache->y[pjd + VEC_SIZE]);
pjz2.v = vec_load(&cell_cache->z[pjd + VEC_SIZE]);
//pjvx2.v = vec_load(&cell_cache->vx[pjd + VEC_SIZE]);
//pjvy2.v = vec_load(&cell_cache->vy[pjd + VEC_SIZE]);
//pjvz2.v = vec_load(&cell_cache->vz[pjd + VEC_SIZE]);
//mj2.v = vec_load(&cell_cache->m[pjd + VEC_SIZE]);
hj_2.v = vec_load(&cell_cache->h[pjd + VEC_SIZE]);
hjg2_2.v = vec_mul(vec_mul(hj_2.v,hj_2.v), kernel_gamma2_vec.v);
/* Compute the pairwise distance. */
vector v_dx_tmp, v_dy_tmp, v_dz_tmp;
vector v_dx_tmp2, v_dy_tmp2, v_dz_tmp2, v_r2_2;
v_dx_tmp.v = vec_sub(pix.v, pjx.v);
v_dx_tmp2.v = vec_sub(pix.v, pjx2.v);
v_dy_tmp.v = vec_sub(piy.v, pjy.v);
v_dy_tmp2.v = vec_sub(piy.v, pjy2.v);
v_dz_tmp.v = vec_sub(piz.v, pjz.v);
v_dz_tmp2.v = vec_sub(piz.v, pjz2.v);
v_r2.v = vec_mul(v_dx_tmp.v, v_dx_tmp.v);
v_r2_2.v = vec_mul(v_dx_tmp2.v, v_dx_tmp2.v);
v_r2.v = vec_fma(v_dy_tmp.v, v_dy_tmp.v, v_r2.v);
v_r2_2.v = vec_fma(v_dy_tmp2.v, v_dy_tmp2.v, v_r2_2.v);
v_r2.v = vec_fma(v_dz_tmp.v, v_dz_tmp.v, v_r2.v);
v_r2_2.v = vec_fma(v_dz_tmp2.v, v_dz_tmp2.v, v_r2_2.v);
/* Form a mask from r2 < hig2 and r2 > 0.*/
#ifdef HAVE_AVX512_F
// KNL_MASK_16 doi_mask, doi_mask_check, doi_mask2, doi_mask2_check;
KNL_MASK_16 doi_mask_check, doi_mask2, doi_mask2_check;
doi_mask_check = vec_cmp_gt(v_r2.v, vec_setzero());
doi_mask = vec_cmp_lt(v_r2.v, v_hig2.v);
doi_mask2_check = vec_cmp_gt(v_r2_2.v, vec_setzero());
doi_mask2 = vec_cmp_lt(v_r2_2.v, v_hig2.v);
doi_mask = doi_mask & doi_mask_check;
doi_mask2 = doi_mask2 & doi_mask2_check;
#else
vector v_doi_mask, v_doi_mask_check, v_doi_N3_mask;
vector v_doi_mask2, v_doi_mask2_check, v_doi_N3_mask2;
int doi_mask2;
/* Form r2 > 0 mask, r2 < hig2 mask and r2 < hjg2 mask. */
v_doi_mask_check.v = vec_cmp_gt(v_r2.v, vec_setzero());
v_doi_mask.v = vec_cmp_lt(v_r2.v, v_hig2.v);
v_doi_N3_mask.v = vec_cmp_lt(v_r2.v, hjg2.v);
/* Form r2 > 0 mask and r2 < hig2 mask. */
v_doi_mask2_check.v = vec_cmp_gt(v_r2_2.v, vec_setzero());
v_doi_mask2.v = vec_cmp_lt(v_r2_2.v, v_hig2.v);
v_doi_N3_mask2.v = vec_cmp_lt(v_r2_2.v, hjg2_2.v);
v_doi_mask.v = vec_and(vec_add(v_doi_mask.v, v_doi_N3_mask.v), v_doi_mask_check.v);
v_doi_mask2.v = vec_and(vec_add(v_doi_mask2.v, v_doi_N3_mask2.v), v_doi_mask2_check.v);
/* Combine two masks and form integer mask. */
doi_mask = vec_cmp_result(v_doi_mask.v);
doi_mask2 = vec_cmp_result(v_doi_mask2.v);
#endif /* HAVE_AVX512_F */
/* If there are any interactions left pack interaction values into c2
* cache. */
if (doi_mask) {
for(int k=0; k<VEC_SIZE; k++) {
if( v_r2.f[k] == 0.f) v_r2.f[k] = 1.f;
}
//intCount += __builtin_popcount(doi_mask);
vector pjvx, pjvy, pjvz, mj, v_hj_inv;
vector v_rhoj, v_grad_hj, v_pOrhoj2, v_balsara_j, v_cj;
v_hj_inv = vec_reciprocal(hj);
mj.v = vec_load(&cell_cache->m[pjd]);
pjvx.v = vec_load(&cell_cache->vx[pjd]);
pjvy.v = vec_load(&cell_cache->vy[pjd]);
pjvz.v = vec_load(&cell_cache->vz[pjd]);
v_rhoj.v = vec_load(&cell_cache->rho[pjd]);
v_grad_hj.v = vec_load(&cell_cache->grad_h[pjd]);
v_pOrhoj2.v = vec_load(&cell_cache->pOrho2[pjd]);
v_balsara_j.v = vec_load(&cell_cache->balsara[pjd]);
v_cj.v = vec_load(&cell_cache->soundspeed[pjd]);
runner_iact_nonsym_1_vec_force_2(&v_r2, &v_dx_tmp, &v_dy_tmp, &v_dz_tmp,
&v_vix, &v_viy, &v_viz, &v_rhoi, &v_grad_hi, &v_pOrhoi2, &v_balsara_i, &v_ci,
&pjvx, &pjvy, &pjvz, &v_rhoj, &v_grad_hj, &v_pOrhoj2, &v_balsara_j, &v_cj, &mj,
v_hi_inv, v_hj_inv,
&a_hydro_xSum, &a_hydro_ySum, &a_hydro_zSum,
&h_dtSum, &v_sigSum, &entropy_dtSum, v_doi_mask);
}
if (doi_mask2) {
for(int k=0; k<VEC_SIZE; k++) {
if( v_r2_2.f[k] == 0.f) v_r2_2.f[k] = 1.f;
}
vector pjvx, pjvy, pjvz, mj, v_hj_inv;
vector v_rhoj, v_grad_hj, v_pOrhoj2, v_balsara_j, v_cj;
v_hj_inv = vec_reciprocal(hj_2);
mj.v = vec_load(&cell_cache->m[pjd + VEC_SIZE]);
pjvx.v = vec_load(&cell_cache->vx[pjd + VEC_SIZE]);
pjvy.v = vec_load(&cell_cache->vy[pjd + VEC_SIZE]);
pjvz.v = vec_load(&cell_cache->vz[pjd + VEC_SIZE]);
v_rhoj.v = vec_load(&cell_cache->rho[pjd + VEC_SIZE]);
v_grad_hj.v = vec_load(&cell_cache->grad_h[pjd + VEC_SIZE]);
v_pOrhoj2.v = vec_load(&cell_cache->pOrho2[pjd + VEC_SIZE]);
v_balsara_j.v = vec_load(&cell_cache->balsara[pjd + VEC_SIZE]);
v_cj.v = vec_load(&cell_cache->soundspeed[pjd + VEC_SIZE]);
runner_iact_nonsym_1_vec_force_2(&v_r2_2, &v_dx_tmp2, &v_dy_tmp2, &v_dz_tmp2,
&v_vix, &v_viy, &v_viz, &v_rhoi, &v_grad_hi, &v_pOrhoi2, &v_balsara_i, &v_ci,
&pjvx, &pjvy, &pjvz, &v_rhoj, &v_grad_hj, &v_pOrhoj2, &v_balsara_j, &v_cj, &mj,
v_hi_inv, v_hj_inv,
&a_hydro_xSum, &a_hydro_ySum, &a_hydro_zSum,
&h_dtSum, &v_sigSum, &entropy_dtSum, v_doi_mask2);
}
}
VEC_HADD(a_hydro_xSum, pi->a_hydro[0]);
VEC_HADD(a_hydro_ySum, pi->a_hydro[1]);
VEC_HADD(a_hydro_zSum, pi->a_hydro[2]);
VEC_HADD(h_dtSum, pi->force.h_dt);
for(int k=0; k<VEC_SIZE; k++)
pi->force.v_sig = max(pi->force.v_sig, v_sigSum.f[k]);
VEC_HADD(entropy_dtSum, pi->entropy_dt);
} /* loop over all particles. */
//message("No. of force interactions: %d", intCount);
//TIMER_TOC(timer_doself_force);
#endif /* WITH_VECTORIZATION */
}
/**
* @brief Compute the cell self-interaction (non-symmetric) using vector
* intrinsics with one particle pi at a time.
*
* @param r The #runner.
* @param c The #cell.
*/
__attribute__((always_inline)) INLINE void runner_doself2_force_vec_3(
struct runner *r, struct cell *restrict c) {
#ifdef WITH_VECTORIZATION #ifdef WITH_VECTORIZATION
const struct engine *e = r->e; const struct engine *e = r->e;
struct part *restrict pi; struct part *restrict pi;
......
src/runner_doiact_vec.h
+ 0
− 3
View file @ bccd8cb3
...@@ -36,9 +36,6 @@ ...@@ -36,9 +36,6 @@
/* Function prototypes. */ /* Function prototypes. */
void runner_doself1_density_vec(struct runner *r, struct cell *restrict c); void runner_doself1_density_vec(struct runner *r, struct cell *restrict c);
void runner_doself2_force_vec(struct runner *r, struct cell *restrict c); void runner_doself2_force_vec(struct runner *r, struct cell *restrict c);
void runner_doself2_force_vec_2(struct runner *r, struct cell *restrict c);
void runner_doself2_force_vec_3(struct runner *r, struct cell *restrict c);
void runner_doself1_density_vec_2(struct runner *r, struct cell *restrict c);
void runner_dopair1_density_vec(struct runner *r, struct cell *restrict ci, void runner_dopair1_density_vec(struct runner *r, struct cell *restrict ci,
struct cell *restrict cj); struct cell *restrict cj);
......
tests/test125cells.c
+ 2
− 4
View file @ bccd8cb3
...@@ -32,9 +32,9 @@ ...@@ -32,9 +32,9 @@
#include "swift.h" #include "swift.h"
#if defined(WITH_VECTORIZATION) #if defined(WITH_VECTORIZATION)
#define DOSELF2 runner_doself2_force_vec_3 #define DOSELF2 runner_doself2_force_vec
#define DOPAIR2 runner_dopair2_density_vec #define DOPAIR2 runner_dopair2_density_vec
#define DOSELF2_NAME "runner_doself2_force_vec_3" #define DOSELF2_NAME "runner_doself2_force_vec"
#define DOPAIR2_NAME "runner_dopair2_force_vec" #define DOPAIR2_NAME "runner_dopair2_force_vec"
#endif #endif
...@@ -441,8 +441,6 @@ void runner_doself1_density(struct runner *r, struct cell *ci); ...@@ -441,8 +441,6 @@ void runner_doself1_density(struct runner *r, struct cell *ci);
void runner_dopair2_force(struct runner *r, struct cell *ci, struct cell *cj); void runner_dopair2_force(struct runner *r, struct cell *ci, struct cell *cj);
void runner_doself2_force(struct runner *r, struct cell *ci); void runner_doself2_force(struct runner *r, struct cell *ci);
void runner_doself2_force_vec(struct runner *r, struct cell *ci); void runner_doself2_force_vec(struct runner *r, struct cell *ci);
void runner_doself2_force_vec_2(struct runner *r, struct cell *ci);
void runner_doself2_force_vec_3(struct runner *r, struct cell *ci);
/* And go... */ /* And go... */
int main(int argc, char *argv[]) { int main(int argc, char *argv[]) {
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment