On Tue, 4 Jun 2024, Richard Sandiford wrote:
> Richard Biener <rguent...@suse.de> writes:
> > The following emulates classical interleaving for SLP load permutes
> > that we are unlikely handling natively. This is to handle cases
> > where interleaving (or load/store-lanes) is the optimal choice for
> > vectorizing even when we are doing that within SLP. An example
> > would be
> >
> > void foo (int * __restrict a, int * b)
> > {
> > for (int i = 0; i < 16; ++i)
> > {
> > a[4*i + 0] = b[4*i + 0] * 3;
> > a[4*i + 1] = b[4*i + 1] + 3;
> > a[4*i + 2] = (b[4*i + 2] * 3 + 3);
> > a[4*i + 3] = b[4*i + 3] * 3;
> > }
> > }
> >
> > where currently the SLP store is merging four single-lane SLP
> > sub-graphs but none of the loads in it can be code-generated
> > with V4SImode vectors and a VF of four as the permutes would need
> > three vectors.
>
> Nice!
>
> > The patch introduces a lowering phase after SLP discovery but
> > before SLP pattern recognition or permute optimization that
> > analyzes all loads from the same dataref group and creates an
> > interleaving scheme starting from an unpermuted load.
> >
> > What can be handled is quite restrictive, matching only a subset
> > of the non-SLP interleaving cases (the power-of-two group size
> > ones, in addition only cases without gaps). The interleaving
> > vectorization in addition can handle size 3 and 5 - but I am not
> > sure if it's possible to do that in a VL agnostic way. It
> > should be still possible to set up the SLP graph in a way that
> > a load-lane could be matched from SLP pattern recognition.
>
> Yeah, I don't think it would be possible to decompose a 3- or
> 5-lane grouped load into a series of VLA 2-input permutes.
> But (as I think you're saying) it seems like a load-3-lanes would just
> be a load with a LANE_PERMUTATION of N, N+3, N+6, N+9, ... for lane N.
> Is that right?
Yes, that's how it looks without this patch. I think we'd need
a load node loading N, N+1, N+2, ... and then permute nodes
with N, N+3, ... and N+1, N+4, ... and N+2, N+5 ... so we generate
one .LOAD_LANES from the load node and the permutes pick up the
correct vector defs? I'm not sure yet how classification and
code generation would work for this.
The store side is already on trunk with the single SLP store node
getting lanes via permutes.
It might be we want a load/store node with N inputs/outputs as the
best representation and use lane_permutation to indicate the
input (for stores) and output (for loads) "permute".
> > As said gaps are currently not handled - for SLP we have a
> > representational issue that SLP_TREE_SCALAR_STMTS for "gap lanes"
> > would need to be filled in some way (even if we just push NULL).
> >
> > The patch misses multi-level even/odd handling as well as CSEing
> > intermediate generated permutes. Both is quite straight-forward
> > to add, but eventually there's a better or more general strategy
> > for lowering? The main goal of the patch is to avoid falling
> > back to non-SLP for cases the interleaving code handles.
>
> Does the multi-level thing including examples like:
>
> int a[2 * 16];
> int b[8 * 16];
> void f()
> {
> for (int i = 0; i < 16; ++i)
> {
> a[i * 2 + 0] += b[i * 8 + 0] + b[i * 8 + 1] + b[i * 8 + 2] + b[i * 8 +
> 3];
> a[i * 2 + 1] += b[i * 8 + 4] + b[i * 8 + 5] + b[i * 8 + 6] + b[i * 8 +
> 7];
> }
> }
>
> ? For that we generate:
>
> _45 = VEC_PERM_EXPR <vect__4.9_63, vect__4.10_61, { 1, 3, 5, 7 }>;
> _44 = VEC_PERM_EXPR <vect__4.11_59, vect__4.12_57, { 1, 3, 5, 7 }>;
> _43 = VEC_PERM_EXPR <_45, _44, { 1, 3, 5, 7 }>;
> _49 = VEC_PERM_EXPR <vect__4.9_63, vect__4.10_61, { 0, 2, 4, 6 }>;
> _48 = VEC_PERM_EXPR <vect__4.11_59, vect__4.12_57, { 0, 2, 4, 6 }>;
> _47 = VEC_PERM_EXPR <_49, _48, { 1, 3, 5, 7 }>;
> _53 = VEC_PERM_EXPR <vect__4.9_63, vect__4.10_61, { 0, 1, 4, 5 }>;
> _52 = VEC_PERM_EXPR <vect__4.11_59, vect__4.12_57, { 0, 1, 4, 5 }>;
> _51 = VEC_PERM_EXPR <_53, _52, { 1, 3, 5, 7 }>;
> _54 = VEC_PERM_EXPR <_49, _48, { 0, 2, 4, 6 }>;
>
> (two even level 1, one even level 2, one odd level 1), whereas
> preferring 2xeven + 2xodd would avoid the third set of first-level
> permutes:
>
> _45 = VEC_PERM_EXPR <vect__4.9_63, vect__4.10_61, { 1, 3, 5, 7 }>;
> _44 = VEC_PERM_EXPR <vect__4.11_59, vect__4.12_57, { 1, 3, 5, 7 }>;
> _43 = VEC_PERM_EXPR <_45, _44, { 1, 3, 5, 7 }>;
> _49 = VEC_PERM_EXPR <vect__4.9_63, vect__4.10_61, { 0, 2, 4, 6 }>;
> _48 = VEC_PERM_EXPR <vect__4.11_59, vect__4.12_57, { 0, 2, 4, 6 }>;
> _47 = VEC_PERM_EXPR <_49, _48, { 1, 3, 5, 7 }>;
> _51 = VEC_PERM_EXPR <_45, _44, { 0, 2, 4, 6 }>;
> _54 = VEC_PERM_EXPR <_49, _48, { 0, 2, 4, 6 }>;
The multi-level issue is more when a single reduction to N/2
inputs still doesn't get you to the point where you can do
the permute with two inputs. I think the above is more because
each load is handled individually, not taking into account
redundancies across loads when you have freedom in the
even/odd, level combinations (which you usually have).
I suppose it should be possible to handle this to some extent,
not sure what the best strategy is when trying to avoid brute-force
searching for an optimal set (esp. when multi-level interleaving
will be involved).
> > Comments and suggestions welcome, esp. what representation
> > you'd think is suitable for SLP pattern matching to
> > load/store-lane and how to represent that? Maybe this lowering
> > should happen directly in vect_lower_load_permutations?
>
> If the load-lanes representation is as simple as above, it sounds like
> it could be deferred to pattern matching. Not sure what the result
> would look like though. It would be nice if (at least for costing
> purposes) we could have a single node for all lanes of the load-lanes,
> rather than create a separate node for each lane and rely on later CSE.
> (Or do we already have a good representation for this? It's been too
> long, sorry.)
Yeah, as said above having a load-lane node with multiple outputs
would be the best match, similar for store-lane. It's probably
easiest to generate those right from this lowering until we
re-implement SLP discovery from scratch.
> Bit of trivia below:
>
> > Thanks,
> > Richard.
> >
> > * tree-vect-slp.cc (vllp_cmp): New function.
> > (vect_lower_load_permutations): Likewise.
> > (vect_analyze_slp): Call it.
> > ---
> > gcc/tree-vect-slp.cc | 279 +++++++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 279 insertions(+)
> >
> > diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
> > index 7e3d0107b4e..766b773452f 100644
> > --- a/gcc/tree-vect-slp.cc
> > +++ b/gcc/tree-vect-slp.cc
> > @@ -3839,6 +3839,279 @@ vect_analyze_slp_instance (vec_info *vinfo,
> > return res;
> > }
> >
> > +/* qsort comparator ordering SLP load nodes. */
> > +
> > +static int
> > +vllp_cmp (const void *a_, const void *b_)
> > +{
> > + const slp_tree a = *(const slp_tree *)a_;
> > + const slp_tree b = *(const slp_tree *)b_;
> > + stmt_vec_info a0 = SLP_TREE_SCALAR_STMTS (a)[0];
> > + stmt_vec_info b0 = SLP_TREE_SCALAR_STMTS (b)[0];
> > + if (STMT_VINFO_GROUPED_ACCESS (a0)
> > + && STMT_VINFO_GROUPED_ACCESS (b0)
> > + && DR_GROUP_FIRST_ELEMENT (a0) == DR_GROUP_FIRST_ELEMENT (b0))
> > + {
> > + /* Same group, order after lanes used. */
> > + if (SLP_TREE_LANES (a) < SLP_TREE_LANES (b))
> > + return 1;
> > + else if (SLP_TREE_LANES (a) > SLP_TREE_LANES (b))
> > + return -1;
> > + else
> > + {
> > + /* Try to order loads using the same lanes together, breaking
> > + the tie with the lane number that first differs. */
> > + if (!SLP_TREE_LOAD_PERMUTATION (a).exists ()
> > + && !SLP_TREE_LOAD_PERMUTATION (b).exists ())
> > + return 0;
>
> Does the comparison need to be "stable", with a further tie-breaker
> when a != b? Or does our qsort not rely on that?
It doesn't, there's stable_sort if we'd rely on a specific order
on the consumer side.
> > + else if (SLP_TREE_LOAD_PERMUTATION (a).exists ()
> > + && !SLP_TREE_LOAD_PERMUTATION (b).exists ())
> > + return 1;
> > + else if (!SLP_TREE_LOAD_PERMUTATION (a).exists ()
> > + && SLP_TREE_LOAD_PERMUTATION (b).exists ())
> > + return -1;
> > + else
> > + {
> > + for (unsigned i = 0; i < SLP_TREE_LANES (a); ++i)
> > + if (SLP_TREE_LOAD_PERMUTATION (a)[i]
> > + != SLP_TREE_LOAD_PERMUTATION (b)[i])
> > + {
> > + /* In-order lane first, that's what the above case for
> > + no permutation does. */
> > + if (SLP_TREE_LOAD_PERMUTATION (a)[i] == i)
> > + return -1;
> > + else if (SLP_TREE_LOAD_PERMUTATION (b)[i] == i)
> > + return 1;
> > + else if (SLP_TREE_LOAD_PERMUTATION (a)[i]
> > + < SLP_TREE_LOAD_PERMUTATION (b)[i])
> > + return -1;
> > + else
> > + return 1;
> > + }
> > + return 0;
> > + }
> > + }
> > + }
> > + else /* Different groups or non-groups. */
> > + {
> > + /* Order groups as their first element to keep them together. */
> > + if (STMT_VINFO_GROUPED_ACCESS (a0))
> > + a0 = DR_GROUP_FIRST_ELEMENT (a0);
> > + if (STMT_VINFO_GROUPED_ACCESS (b0))
> > + b0 = DR_GROUP_FIRST_ELEMENT (b0);
> > + if (a0 == b0)
> > + return 0;
> > + /* Tie using UID. */
> > + else if (gimple_uid (STMT_VINFO_STMT (a0))
> > + < gimple_uid (STMT_VINFO_STMT (b0)))
> > + return -1;
> > + else
> > + {
> > + gcc_assert (gimple_uid (STMT_VINFO_STMT (a0))
> > + != gimple_uid (STMT_VINFO_STMT (b0)));
> > + return 1;
> > + }
> > + }
> > +}
> > +
> > +/* Process the set of LOADS that are all from the same dataref group. */
> > +
> > +static void
> > +vect_lower_load_permutations (loop_vec_info loop_vinfo,
> > + scalar_stmts_to_slp_tree_map_t *bst_map,
> > + const array_slice<slp_tree> &loads)
> > +{
> > + /* We at this point want to lower without a fixed VF or vector
> > + size in mind which means we cannot actually compute whether we
> > + need three or more vectors for a load permutation yet. So always
> > + lower. */
> > + stmt_vec_info first
> > + = DR_GROUP_FIRST_ELEMENT (SLP_TREE_SCALAR_STMTS (loads[0])[0]);
> > +
> > + /* ??? In principle we have to consider a gap up to the next full
> > + vector, but we have to actually represent a scalar stmt for the
> > + gaps value so delay handling this. The same is true for
> > + inbetween gaps which the load places in the load-permutation
> > + represent. It's probably not worth trying an intermediate packing
> > + to vectors without gap even if that might handle some more cases.
> > + Instead get the gap case correct in some way. */
> > + unsigned group_lanes = 0;
> > + for (stmt_vec_info s = first; s; s = DR_GROUP_NEXT_ELEMENT (s))
> > + {
> > + if ((s == first && DR_GROUP_GAP (s) != 0)
> > + || (s != first && DR_GROUP_GAP (s) != 1))
> > + return;
> > + group_lanes++;
> > + }
> > + /* Only a power-of-two number of lanes matches interleaving. */
> > + if (exact_log2 (group_lanes) == -1)
> > + return;
> > +
> > + for (slp_tree load : loads)
> > + {
> > + /* Leave masked or gather loads alone for now. */
> > + if (!SLP_TREE_CHILDREN (load).is_empty ())
> > + continue;
> > +
> > + /* We need to lower only loads of less than half of the groups
> > + lanes, including duplicate lanes. */
> > + if (SLP_TREE_LANES (load) >= group_lanes / 2)
> > + continue;
> > +
> > + /* Lower by reducing the group to half its size using an
> > + interleaving scheme. For this try to compute whether all
> > + elements needed for this loads are in even or odd elements of
>
> this load (or these loads)
>
> > + a even/odd decomposition with N consecutive elements.
>
> an even/odd
>
> > + Thus { e, e, o, o, e, e, o, o } woud be an even/odd decomposition
> > + with N == 2. */
> > + unsigned even = (1 << ceil_log2 (DR_GROUP_SIZE (first))) - 1;
>
> Is this DR_GROUP_SIZE (first) conceptually different from group_lanes
> above? If not, I think it'd be a bit easier to follow if this line reused
> the exact_log2 result above.
Once we look at groups with gaps it's different - the load permutation
lane indices have gaps represented, so a a[0], a[2], a[3] group
would have a load permutation of { 0, 2, 3 }. group_lanes is the
number of lanes in the output of the load which has unused/gap lanes
stripped.
I've short-cut handling of groups with intermediate gaps and also with
gaps at the end for simplicity as I have to decide what to put into
SLP_TREE_SCALAR_STMTS for the unpermuted SLP load node which would
have those gaps "represented" (I'm quite sure a NULL ICEs left and
right, duplicating the previous lane sounds appealing even though
it's wrong ...). As said, this is more a proof-of-concept ;)
For the next iteration I'm going to add some test coverage, esp.
also for the multi-level case and will see to handle gaps.
> > + unsigned odd = even;
> > + for (unsigned l : SLP_TREE_LOAD_PERMUTATION (load))
> > + {
> > + even &= ~l;
> > + odd &= l;
> > + }
> > + /* Give up when this doesn't match up with an interleaving scheme.
> > */
> > + if (!even && !odd)
> > + continue;
> > +
> > + /* First build (and possibly re-use) a load node for the
> > + unpermuted group. */
> > + vec<stmt_vec_info> stmts;
> > + stmts.create (group_lanes);
> > + for (stmt_vec_info s = first; s; s = DR_GROUP_NEXT_ELEMENT (s))
> > + stmts.quick_push (s);
> > + poly_uint64 max_nunits;
> > + bool *matches = XALLOCAVEC (bool, group_lanes);
> > + unsigned limit = 1;
> > + unsigned tree_size = 0;
> > + slp_tree l0 = vect_build_slp_tree (loop_vinfo, stmts,
> > + group_lanes,
> > + &max_nunits, matches, &limit,
> > + &tree_size, bst_map);
> > +
> > + /* Build the permute to get the original load permutation order. */
> > + lane_permutation_t final_perm;
> > + final_perm.create (SLP_TREE_LANES (load));
> > + for (unsigned i = 0; i < SLP_TREE_LANES (load); ++i)
> > + final_perm.quick_push
> > + (std::make_pair (0, SLP_TREE_LOAD_PERMUTATION (load)[i]));
> > +
> > + /* Now build a even or odd extraction from the unpermuted load. */
>
> an even
Thanks,
Richard.
> Thanks,
> Richard
>
> > + lane_permutation_t perm;
> > + perm.create (group_lanes / 2);
> > + unsigned level;
> > + if (even
> > + && ((level = 1 << ctz_hwi (even)), true)
> > + && group_lanes % (2 * level) == 0)
> > + {
> > + /* { 0, 1, ... 4, 5 ..., } */
> > + unsigned level = 1 << ctz_hwi (even);
> > + for (unsigned i = 0; i < group_lanes / 2 / level; ++i)
> > + for (unsigned j = 0; j < level; ++j)
> > + perm.quick_push (std::make_pair (0, 2 * i * level + j));
> > + }
> > + else if (odd)
> > + {
> > + /* { ..., 2, 3, ... 6, 7 } */
> > + unsigned level = 1 << ctz_hwi (odd);
> > + gcc_assert (group_lanes % (2 * level) == 0);
> > + for (unsigned i = 0; i < group_lanes / 2 / level; ++i)
> > + for (unsigned j = 0; j < level; ++j)
> > + perm.quick_push (std::make_pair (0, (2 * i + 1) * level + j));
> > + }
> > + else
> > + gcc_unreachable ();
> > +
> > + /* And update final_perm. */
> > + for (unsigned i = 0; i < SLP_TREE_LANES (load); ++i)
> > + {
> > + unsigned l = final_perm[i].second;
> > + unsigned j;
> > + for (j = 0; j < perm.length (); ++j)
> > + if (perm[j].second == l)
> > + {
> > + final_perm[i].second = j;
> > + break;
> > + }
> > + gcc_assert (j < perm.length ());
> > + }
> > +
> > + slp_tree p = vect_create_new_slp_node (1, VEC_PERM_EXPR);
> > + SLP_TREE_CHILDREN (p).quick_push (l0);
> > + SLP_TREE_LANE_PERMUTATION (p) = perm;
> > + SLP_TREE_VECTYPE (p) = SLP_TREE_VECTYPE (load);
> > + SLP_TREE_LANES (p) = perm.length ();
> > + SLP_TREE_REPRESENTATIVE (p) = SLP_TREE_REPRESENTATIVE (load);
> > + /* ??? We should have scalar stmts for this and use bst_map
> > + to CSE. But we do not want to pick up original SLP load
> > + nodes with a load-permutation here. */
> > + /* ??? We need to iterate if group_lanes / 2 is still too large. */
> > + /* ??? Ideally pick the best even/odd scheme usable for
> > + most of the loads. -> do a multi-step scheme? */
> > +
> > + /* And finally from the ordered reduction node create the
> > + permute to shuffle the lanes into the original load-permutation
> > + order. We replace the original load node with this. */
> > + SLP_TREE_CODE (load) = VEC_PERM_EXPR;
> > + SLP_TREE_LOAD_PERMUTATION (load).release ();
> > + SLP_TREE_LANE_PERMUTATION (load) = final_perm;
> > + SLP_TREE_CHILDREN (load).create (1);
> > + SLP_TREE_CHILDREN (load).quick_push (p);
> > + }
> > +}
> > +
> > +/* Transform SLP loads in the SLP graph created by SLP discovery to
> > + group loads from the same group and lower load permutations that
> > + are unlikely to be supported into a series of permutes.
> > + In the degenerate case of having only single-lane SLP instances
> > + this should result in a series of permute nodes emulating an
> > + interleaving scheme. */
> > +
> > +static void
> > +vect_lower_load_permutations (loop_vec_info loop_vinfo,
> > + scalar_stmts_to_slp_tree_map_t *bst_map)
> > +{
> > + /* Gather and sort loads across all instances. */
> > + hash_set<slp_tree> visited;
> > + auto_vec<slp_tree> loads;
> > + for (auto inst : loop_vinfo->slp_instances)
> > + vect_gather_slp_loads (loads, SLP_INSTANCE_TREE (inst), visited);
> > + if (loads.is_empty ())
> > + return;
> > + loads.qsort (vllp_cmp);
> > +
> > + /* Now process each dataref group separately. */
> > + unsigned firsti = 0;
> > + for (unsigned i = 1; i < loads.length (); ++i)
> > + {
> > + slp_tree first = loads[firsti];
> > + slp_tree next = loads[i];
> > + stmt_vec_info a0 = SLP_TREE_SCALAR_STMTS (first)[0];
> > + stmt_vec_info b0 = SLP_TREE_SCALAR_STMTS (next)[0];
> > + if (STMT_VINFO_GROUPED_ACCESS (a0)
> > + && STMT_VINFO_GROUPED_ACCESS (b0)
> > + && DR_GROUP_FIRST_ELEMENT (a0) == DR_GROUP_FIRST_ELEMENT (b0))
> > + continue;
> > + /* Just one SLP load of a possible group, leave those alone. */
> > + if (i == firsti + 1)
> > + {
> > + firsti = i;
> > + continue;
> > + }
> > + /* Now we have multiple SLP loads of the same group from
> > + firsti to i - 1. */
> > + vect_lower_load_permutations (loop_vinfo, bst_map,
> > + make_array_slice (&loads[firsti],
> > + i - firsti));
> > + firsti = i;
> > + }
> > + if (firsti < loads.length () - 1)
> > + vect_lower_load_permutations (loop_vinfo, bst_map,
> > + make_array_slice (&loads[firsti],
> > + loads.length () - firsti));
> > +}
> > +
> > /* Check if there are stmts in the loop can be vectorized using SLP.
> > Build SLP
> > trees of packed scalar stmts if SLP is possible. */
> >
> > @@ -3982,6 +4255,12 @@ vect_analyze_slp (vec_info *vinfo, unsigned
> > max_tree_size)
> > }
> > }
> >
> > + /* When we end up with load permutations that we cannot possibly handle,
> > + like those requiring three vector inputs, lower them using
> > interleaving
> > + like schemes. */
> > + if (loop_vec_info loop_vinfo = dyn_cast <loop_vec_info> (vinfo))
> > + vect_lower_load_permutations (loop_vinfo, bst_map);
> > +
> > hash_set<slp_tree> visited_patterns;
> > slp_tree_to_load_perm_map_t perm_cache;
> > slp_compat_nodes_map_t compat_cache;
>
