/* Given a relation "map" between instances of two statements A and B,
* does it relate every instance of A (according to the domain of "src")
* to every instance of B (according to the domain of "dst")?
*/
static isl_bool covers_src_and_dst(__isl_keep isl_map *map,
struct ppcg_grouping_leaf *src, struct ppcg_grouping_leaf *dst)
{
isl_space *space;
isl_set *set1, *set2;
isl_bool is_subset;
space = isl_space_domain(isl_map_get_space(map));
set1 = isl_union_set_extract_set(src->domain, space);
set2 = isl_map_domain(isl_map_copy(map));
is_subset = isl_set_is_subset(set1, set2);
isl_set_free(set1);
isl_set_free(set2);
if (is_subset < 0 || !is_subset)
return is_subset;
space = isl_space_range(isl_map_get_space(map));
set1 = isl_union_set_extract_set(dst->domain, space);
set2 = isl_map_range(isl_map_copy(map));
is_subset = isl_set_is_subset(set1, set2);
isl_set_free(set1);
isl_set_free(set2);
return is_subset;
}
static void
pbb_strip_mine_time_depth (poly_bb_p pbb, int time_depth, int stride)
{
isl_space *d;
isl_constraint *c;
int iter, strip;
/* STRIP is the dimension that iterates with stride STRIDE. */
/* ITER is the dimension that enumerates single iterations inside
one strip that has at most STRIDE iterations. */
strip = time_depth;
iter = strip + 2;
pbb->transformed = isl_map_insert_dims (pbb->transformed, isl_dim_out,
strip, 2);
/* Lower bound of the striped loop. */
d = isl_map_get_space (pbb->transformed);
c = isl_inequality_alloc (isl_local_space_from_space (d));
c = isl_constraint_set_coefficient_si (c, isl_dim_out, strip, -stride);
c = isl_constraint_set_coefficient_si (c, isl_dim_out, iter, 1);
pbb->transformed = isl_map_add_constraint (pbb->transformed, c);
/* Upper bound of the striped loop. */
d = isl_map_get_space (pbb->transformed);
c = isl_inequality_alloc (isl_local_space_from_space (d));
c = isl_constraint_set_coefficient_si (c, isl_dim_out, strip, stride);
c = isl_constraint_set_coefficient_si (c, isl_dim_out, iter, -1);
c = isl_constraint_set_constant_si (c, stride - 1);
pbb->transformed = isl_map_add_constraint (pbb->transformed, c);
/* Static scheduling for ITER level.
This is mandatory to keep the 2d + 1 canonical scheduling format. */
d = isl_map_get_space (pbb->transformed);
c = isl_equality_alloc (isl_local_space_from_space (d));
c = isl_constraint_set_coefficient_si (c, isl_dim_out, strip + 1, 1);
pbb->transformed = isl_map_add_constraint (pbb->transformed, c);
}
static isl_constraint *
build_linearized_memory_access (isl_map *map, poly_dr_p pdr)
{
isl_constraint *res;
isl_local_space *ls = isl_local_space_from_space (isl_map_get_space (map));
unsigned offset, nsubs;
int i;
isl_int size, subsize;
res = isl_equality_alloc (ls);
isl_int_init (size);
isl_int_set_ui (size, 1);
isl_int_init (subsize);
isl_int_set_ui (subsize, 1);
nsubs = isl_set_dim (pdr->extent, isl_dim_set);
/* -1 for the already included L dimension. */
offset = isl_map_dim (map, isl_dim_out) - 1 - nsubs;
res = isl_constraint_set_coefficient_si (res, isl_dim_out, offset + nsubs, -1);
/* Go through all subscripts from last to first. First dimension
is the alias set, ignore it. */
for (i = nsubs - 1; i >= 1; i--)
{
isl_space *dc;
isl_aff *aff;
res = isl_constraint_set_coefficient (res, isl_dim_out, offset + i, size);
dc = isl_set_get_space (pdr->extent);
aff = isl_aff_zero_on_domain (isl_local_space_from_space (dc));
aff = isl_aff_set_coefficient_si (aff, isl_dim_in, i, 1);
isl_set_max (pdr->extent, aff, &subsize);
isl_aff_free (aff);
isl_int_mul (size, size, subsize);
}
isl_int_clear (subsize);
isl_int_clear (size);
return res;
}
static void
pdr_stride_in_loop (mpz_t stride, graphite_dim_t depth, poly_dr_p pdr)
{
poly_bb_p pbb = PDR_PBB (pdr);
isl_map *map;
isl_set *set;
isl_aff *aff;
isl_space *dc;
isl_constraint *lma, *c;
isl_int islstride;
graphite_dim_t time_depth;
unsigned offset, nt;
unsigned i;
/* XXX isl rewrite following comments. */
/* Builds a partial difference equations and inserts them
into pointset powerset polyhedron P. Polyhedron is assumed
to have the format: T|I|T'|I'|G|S|S'|l1|l2.
TIME_DEPTH is the time dimension w.r.t. which we are
differentiating.
OFFSET represents the number of dimensions between
columns t_{time_depth} and t'_{time_depth}.
DIM_SCTR is the number of scattering dimensions. It is
essentially the dimensionality of the T vector.
The following equations are inserted into the polyhedron P:
| t_1 = t_1'
| ...
| t_{time_depth-1} = t'_{time_depth-1}
| t_{time_depth} = t'_{time_depth} + 1
| t_{time_depth+1} = t'_{time_depth + 1}
| ...
| t_{dim_sctr} = t'_{dim_sctr}. */
/* Add the equality: t_{time_depth} = t'_{time_depth} + 1.
This is the core part of this alogrithm, since this
constraint asks for the memory access stride (difference)
between two consecutive points in time dimensions. */
/* Add equalities:
| t1 = t1'
| ...
| t_{time_depth-1} = t'_{time_depth-1}
| t_{time_depth+1} = t'_{time_depth+1}
| ...
| t_{dim_sctr} = t'_{dim_sctr}
This means that all the time dimensions are equal except for
time_depth, where the constraint is t_{depth} = t'_{depth} + 1
step. More to this: we should be careful not to add equalities
to the 'coupled' dimensions, which happens when the one dimension
is stripmined dimension, and the other dimension corresponds
to the point loop inside stripmined dimension. */
/* pdr->accesses: [P1..nb_param,I1..nb_domain]->[a,S1..nb_subscript]
??? [P] not used for PDRs?
pdr->extent: [a,S1..nb_subscript]
pbb->domain: [P1..nb_param,I1..nb_domain]
pbb->transformed: [P1..nb_param,I1..nb_domain]->[T1..Tnb_sctr]
[T] includes local vars (currently unused)
First we create [P,I] -> [T,a,S]. */
map = isl_map_flat_range_product (isl_map_copy (pbb->transformed),
isl_map_copy (pdr->accesses));
/* Add a dimension for L: [P,I] -> [T,a,S,L].*/
map = isl_map_add_dims (map, isl_dim_out, 1);
/* Build a constraint for "lma[S] - L == 0", effectively calculating
L in terms of subscripts. */
lma = build_linearized_memory_access (map, pdr);
/* And add it to the map, so we now have:
[P,I] -> [T,a,S,L] : lma([S]) == L. */
map = isl_map_add_constraint (map, lma);
/* Then we create [P,I,P',I'] -> [T,a,S,L,T',a',S',L']. */
map = isl_map_flat_product (map, isl_map_copy (map));
/* Now add the equality T[time_depth] == T'[time_depth]+1. This will
force L' to be the linear address at T[time_depth] + 1. */
time_depth = psct_dynamic_dim (pbb, depth);
/* Length of [a,S] plus [L] ... */
offset = 1 + isl_map_dim (pdr->accesses, isl_dim_out);
/* ... plus [T]. */
offset += isl_map_dim (pbb->transformed, isl_dim_out);
c = isl_equality_alloc (isl_local_space_from_space (isl_map_get_space (map)));
c = isl_constraint_set_coefficient_si (c, isl_dim_out, time_depth, 1);
c = isl_constraint_set_coefficient_si (c, isl_dim_out,
offset + time_depth, -1);
c = isl_constraint_set_constant_si (c, 1);
map = isl_map_add_constraint (map, c);
/* Now we equate most of the T/T' elements (making PITaSL nearly
the same is (PITaSL)', except for one dimension, namely for 'depth'
(an index into [I]), after translating to index into [T]. Take care
to not produce an empty map, which indicates we wanted to equate
two dimensions that are already coupled via the above time_depth
dimension. Happens with strip mining where several scatter dimension
are interdependend. */
/* Length of [T]. */
nt = pbb_nb_scattering_transform (pbb) + pbb_nb_local_vars (pbb);
for (i = 0; i < nt; i++)
if (i != time_depth)
{
isl_map *temp = isl_map_equate (isl_map_copy (map),
isl_dim_out, i,
isl_dim_out, offset + i);
if (isl_map_is_empty (temp))
isl_map_free (temp);
else
{
isl_map_free (map);
map = temp;
}
}
/* Now maximize the expression L' - L. */
set = isl_map_range (map);
dc = isl_set_get_space (set);
aff = isl_aff_zero_on_domain (isl_local_space_from_space (dc));
aff = isl_aff_set_coefficient_si (aff, isl_dim_in, offset - 1, -1);
aff = isl_aff_set_coefficient_si (aff, isl_dim_in, offset + offset - 1, 1);
isl_int_init (islstride);
isl_set_max (set, aff, &islstride);
isl_int_get_gmp (islstride, stride);
isl_int_clear (islstride);
isl_aff_free (aff);
isl_set_free (set);
if (dump_file && (dump_flags & TDF_DETAILS))
{
char *str;
void (*gmp_free) (void *, size_t);
fprintf (dump_file, "\nStride in BB_%d, DR_%d, depth %d:",
pbb_index (pbb), PDR_ID (pdr), (int) depth);
str = mpz_get_str (0, 10, stride);
fprintf (dump_file, " %s ", str);
mp_get_memory_functions (NULL, NULL, &gmp_free);
(*gmp_free) (str, strlen (str) + 1);
}
}