On Tue, 4 Jul 2023, juzhe.zh...@rivai.ai wrote:
> Hi, Richi.
>
> For GATHER_LOAD which doesn't have len and mask.
>
> Should I keep it as gather_load, then I support both gather_load and
> len_mask_gather_load ?
>
> Or I should normalize it into len_mask_gather_load with length = vf and
> mask = {1,1,1,1,1,...}, then I only need to support len_mask_gather_load
> in RISC-V port?
I think that pattern recog should keep it as GATHER_LOAD or
MASK_GATHER_LOAD depending on what if-conversion did. The later
handling should then use the appropriate vector IFN which could
add a dummy -1 mask or a dummy len and handle partial vectors.
Richard.
> Thanks.
>
>
> juzhe.zh...@rivai.ai
>
> From: Richard Biener
> Date: 2023-07-04 19:17
> To: juzhe.zh...@rivai.ai
> CC: gcc-patches; richard.sandiford
> Subject: Re: Re: [PATCH V3] VECT: Apply LEN_MASK_GATHER_LOAD/SCATTER_STORE
> into vectorizer
> On Tue, 4 Jul 2023, juzhe.zh...@rivai.ai wrote:
>
> > Hi, Richi.
> >
> > >> Eh, it's obvious that we should have the "vectorized" form
> > >> also in the 'scalar' variant. If you think there's no reasonable
> > >> way to add a value for len or bias then instead re-order the
> > >> arguments so 'mask' comes first and the len/bias pair last.
> >
> > I found this patch is buggy when I am testing LEN_MAS_GATHER_LOAD.
> > And reoder 'mask' comes first can not help.
> >
> > Here, we have 'GATHER_LOAD' and 'MASK_GATHER_LOAD',
> > For GATHER_LOAD ((sizetype) src_18(D), _6, 1, 0);
> > I change it into:
> >
> > LEN_MASK_GATHER_LOAD ((sizetype) src_18(D), _6, 1, 0);
> > In this situation, internal_fn_mask_index
> > should return -1.
> >
> > Wheras, MASK_GATHER_LOAD ((sizetype) _56, _8, 1, 0, _33);
> > I change it into LEN_MASK_GATHER_LOAD ((sizetype) _56, _8, 1, 0, _33);
> > it should return the index = 4.
> >
> > I can't differentiate them by only using LEN_MASK_GATHER_LOAD.
> > Could I revise internal_fn_mask_index
> > as follows ?
>
> No, please adjust the gather pattern recognition to produce either
> appropriate LEN_ variant IFNs or simply keep only the unconditional
> and conditional mask variants from patterns but code generate
> the len_ variants. I don't really see what the problem is.
> Maybe you fail to specify the appropriate ifn when you inspect
> the scalar internal fn call?
>
> > int
> > internal_fn_mask_index (internal_fn fn, int nargs)
> > {
> > switch (fn)
> > {
> > case IFN_MASK_LOAD:
> > case IFN_MASK_LOAD_LANES:
> > case IFN_MASK_STORE:
> > case IFN_MASK_STORE_LANES:
> > return 2;
> >
> > case IFN_MASK_GATHER_LOAD:
> > case IFN_MASK_SCATTER_STORE:
> > case IFN_LEN_MASK_LOAD:
> > case IFN_LEN_MASK_STORE:
> > return 4;
> >
> > /* LEN_MASK_GATHER_LOAD/LEN_MASK_SCATTER_STORE are different
> > on arguments before and after vectorized.
> >
> > Before vectorized:
> > LEN_MASK_GATHER_LOAD (ptr, align, offset, mask);
> >
> > After vectorized:
> > LEN_MASK_GATHER_LOAD (ptr, align, offset, len, bias, mask);
> > */
> > case IFN_LEN_MASK_GATHER_LOAD:
> > case IFN_LEN_MASK_SCATTER_STORE:
> > return nargs == 4 ? -1 : nargs == 5 ? 4 : 6;
> >
> > default:
> > return (conditional_internal_fn_code (fn) != ERROR_MARK
> > || get_unconditional_internal_fn (fn) != IFN_LAST ? 0 : -1);
> > }
> > }
> >
> >
> > Thanks.
> >
> >
> > juzhe.zh...@rivai.ai
> >
> > From: Richard Biener
> > Date: 2023-07-04 19:05
> > To: Ju-Zhe Zhong
> > CC: gcc-patches; richard.sandiford
> > Subject: Re: [PATCH V3] VECT: Apply LEN_MASK_GATHER_LOAD/SCATTER_STORE into
> > vectorizer
> > On Tue, 4 Jul 2023, juzhe.zh...@rivai.ai wrote:
> >
> > > From: Ju-Zhe Zhong <juzhe.zh...@rivai.ai>
> > >
> > > Hi, Richard and Richi.
> > >
> > > Address comments from Richard.
> > >
> > > Make gs_info.ifn = LEN_MASK_GATHER_LOAD/LEN_MASK_SCATTER_STORE.
> > >
> > > Since:
> > > /* LEN_MASK_GATHER_LOAD/LEN_MASK_SCATTER_STORE are different
> > > on arguments before and after vectorized.
> > >
> > > Before vectorized:
> > > LEN_MASK_GATHER_LOAD (ptr, align, offset, mask);
> > >
> > > After vectorized:
> > > LEN_MASK_GATHER_LOAD (ptr, align, offset, len, bias, mask);
> > > */
> > >
> > > I add "vectorized_p" default argument into internal_fn_mask_index.
> > > So that we could simplify the codes.
> >
> > Eh, it's obvious that we should have the "vectorized" form
> > also in the 'scalar' variant. If you think there's no reasonable
> > way to add a value for len or bias then instead re-order the
> > arguments so 'mask' comes first and the len/bias pair last.
> >
> > But IMHO "any" len/bias value should do here.
> >
> > The rest looks OK now.
> >
> > Thanks,
> > Richard.
> >
> > > The len_mask_gather_load/len_mask_scatter_store patterns have been added.
> > > Now, this patch applies them into vectorizer.
> > >
> > > Here is the example:
> > >
> > > void
> > > f (int *restrict a,
> > > int *restrict b, int n,
> > > int base, int step,
> > > int *restrict cond)
> > > {
> > > for (int i = 0; i < n; ++i)
> > > {
> > > if (cond[i])
> > > a[i * 4] = b[i];
> > > }
> > > }
> > >
> > > Gimple IR:
> > >
> > > <bb 3> [local count: 105119324]:
> > > _58 = (unsigned long) n_13(D);
> > >
> > > <bb 4> [local count: 630715945]:
> > > # vectp_cond.7_45 = PHI <vectp_cond.7_46(4), cond_14(D)(3)>
> > > # vectp_b.11_51 = PHI <vectp_b.11_52(4), b_15(D)(3)>
> > > # vectp_a.14_55 = PHI <vectp_a.14_56(4), a_16(D)(3)>
> > > # ivtmp_59 = PHI <ivtmp_60(4), _58(3)>
> > > _61 = .SELECT_VL (ivtmp_59, POLY_INT_CST [2, 2]);
> > > ivtmp_44 = _61 * 4;
> > > vect__4.9_47 = .LEN_MASK_LOAD (vectp_cond.7_45, 32B, _61, 0, { -1, ...
> > > });
> > > mask__24.10_49 = vect__4.9_47 != { 0, ... };
> > > vect__8.13_53 = .LEN_MASK_LOAD (vectp_b.11_51, 32B, _61, 0,
> > > mask__24.10_49);
> > > ivtmp_54 = _61 * 16;
> > > .LEN_MASK_SCATTER_STORE (vectp_a.14_55, { 0, 16, 32, ... }, 1,
> > > vect__8.13_53, _61, 0, mask__24.10_49);
> > > vectp_cond.7_46 = vectp_cond.7_45 + ivtmp_44;
> > > vectp_b.11_52 = vectp_b.11_51 + ivtmp_44;
> > > vectp_a.14_56 = vectp_a.14_55 + ivtmp_54;
> > > ivtmp_60 = ivtmp_59 - _61;
> > > if (ivtmp_60 != 0)
> > > goto <bb 4>; [83.33%]
> > > else
> > > goto <bb 5>; [16.67%]
> > >
> > > gcc/ChangeLog:
> > >
> > > * internal-fn.cc (internal_fn_mask_index): Apply
> > > LEN_MASK_GATHER_LOAD/SCATTER_STORE into vectorizer.
> > > * internal-fn.h (internal_fn_mask_index): Ditto.
> > > * optabs-query.cc (supports_vec_gather_load_p): Ditto.
> > > (supports_vec_scatter_store_p): Ditto.
> > > * tree-vect-data-refs.cc (vect_gather_scatter_fn_p): Ditto.
> > > * tree-vect-stmts.cc (exist_non_indexing_operands_for_use_p):
> > > Ditto.
> > > (check_load_store_for_partial_vectors): Ditto.
> > > (vect_get_strided_load_store_ops): Ditto.
> > > (vectorizable_store): Ditto.
> > > (vectorizable_load): Ditto.
> > >
> > > ---
> > > gcc/internal-fn.cc | 16 ++++-
> > > gcc/internal-fn.h | 2 +-
> > > gcc/optabs-query.cc | 2 +
> > > gcc/tree-vect-data-refs.cc | 18 ++++-
> > > gcc/tree-vect-stmts.cc | 135 +++++++++++++++++++++++++++++++------
> > > 5 files changed, 150 insertions(+), 23 deletions(-)
> > >
> > > diff --git a/gcc/internal-fn.cc b/gcc/internal-fn.cc
> > > index 303df102d81..2c78c870de8 100644
> > > --- a/gcc/internal-fn.cc
> > > +++ b/gcc/internal-fn.cc
> > > @@ -4483,7 +4483,7 @@ internal_fn_len_index (internal_fn fn)
> > > otherwise return -1. */
> > >
> > > int
> > > -internal_fn_mask_index (internal_fn fn)
> > > +internal_fn_mask_index (internal_fn fn, bool vectoried_p)
> > > {
> > > switch (fn)
> > > {
> > > @@ -4499,9 +4499,21 @@ internal_fn_mask_index (internal_fn fn)
> > > case IFN_LEN_MASK_STORE:
> > > return 4;
> > >
> > > + /* LEN_MASK_GATHER_LOAD/LEN_MASK_SCATTER_STORE are different
> > > + on arguments before and after vectorized.
> > > +
> > > + Before vectorized:
> > > + LEN_MASK_GATHER_LOAD (ptr, align, offset, mask);
> > > +
> > > + After vectorized:
> > > + LEN_MASK_GATHER_LOAD (ptr, align, offset, len, bias, mask);
> > > + */
> > > case IFN_LEN_MASK_GATHER_LOAD:
> > > case IFN_LEN_MASK_SCATTER_STORE:
> > > - return 6;
> > > + if (vectoried_p)
> > > + return 6;
> > > + else
> > > + return 4;
> > >
> > > default:
> > > return (conditional_internal_fn_code (fn) != ERROR_MARK
> > > diff --git a/gcc/internal-fn.h b/gcc/internal-fn.h
> > > index 4234bbfed87..e9168c16297 100644
> > > --- a/gcc/internal-fn.h
> > > +++ b/gcc/internal-fn.h
> > > @@ -233,7 +233,7 @@ extern bool can_interpret_as_conditional_op_p (gimple
> > > *, tree *,
> > > extern bool internal_load_fn_p (internal_fn);
> > > extern bool internal_store_fn_p (internal_fn);
> > > extern bool internal_gather_scatter_fn_p (internal_fn);
> > > -extern int internal_fn_mask_index (internal_fn);
> > > +extern int internal_fn_mask_index (internal_fn, bool = true);
> > > extern int internal_fn_len_index (internal_fn);
> > > extern int internal_fn_stored_value_index (internal_fn);
> > > extern bool internal_gather_scatter_fn_supported_p (internal_fn, tree,
> > > diff --git a/gcc/optabs-query.cc b/gcc/optabs-query.cc
> > > index 2fdd0d34354..bf1f484e874 100644
> > > --- a/gcc/optabs-query.cc
> > > +++ b/gcc/optabs-query.cc
> > > @@ -676,6 +676,7 @@ supports_vec_gather_load_p (machine_mode mode)
> > > this_fn_optabs->supports_vec_gather_load[mode]
> > > = (supports_vec_convert_optab_p (gather_load_optab, mode)
> > > || supports_vec_convert_optab_p (mask_gather_load_optab, mode)
> > > + || supports_vec_convert_optab_p (len_mask_gather_load_optab, mode)
> > > ? 1 : -1);
> > >
> > > return this_fn_optabs->supports_vec_gather_load[mode] > 0;
> > > @@ -692,6 +693,7 @@ supports_vec_scatter_store_p (machine_mode mode)
> > > this_fn_optabs->supports_vec_scatter_store[mode]
> > > = (supports_vec_convert_optab_p (scatter_store_optab, mode)
> > > || supports_vec_convert_optab_p (mask_scatter_store_optab, mode)
> > > + || supports_vec_convert_optab_p (len_mask_scatter_store_optab, mode)
> > > ? 1 : -1);
> > >
> > > return this_fn_optabs->supports_vec_scatter_store[mode] > 0;
> > > diff --git a/gcc/tree-vect-data-refs.cc b/gcc/tree-vect-data-refs.cc
> > > index ebe93832b1e..ab2af103cb4 100644
> > > --- a/gcc/tree-vect-data-refs.cc
> > > +++ b/gcc/tree-vect-data-refs.cc
> > > @@ -3873,16 +3873,24 @@ vect_gather_scatter_fn_p (vec_info *vinfo, bool
> > > read_p, bool masked_p,
> > > return false;
> > >
> > > /* Work out which function we need. */
> > > - internal_fn ifn, alt_ifn;
> > > + internal_fn ifn, alt_ifn, alt_ifn2;
> > > if (read_p)
> > > {
> > > ifn = masked_p ? IFN_MASK_GATHER_LOAD : IFN_GATHER_LOAD;
> > > alt_ifn = IFN_MASK_GATHER_LOAD;
> > > + /* When target supports LEN_MASK_GATHER_LOAD, we always
> > > + use LEN_MASK_GATHER_LOAD regardless whether len and
> > > + mask are valid or not. */
> > > + alt_ifn2 = IFN_LEN_MASK_GATHER_LOAD;
> > > }
> > > else
> > > {
> > > ifn = masked_p ? IFN_MASK_SCATTER_STORE : IFN_SCATTER_STORE;
> > > alt_ifn = IFN_MASK_SCATTER_STORE;
> > > + /* When target supports LEN_MASK_SCATTER_STORE, we always
> > > + use LEN_MASK_SCATTER_STORE regardless whether len and
> > > + mask are valid or not. */
> > > + alt_ifn2 = IFN_LEN_MASK_SCATTER_STORE;
> > > }
> > >
> > > for (;;)
> > > @@ -3909,6 +3917,14 @@ vect_gather_scatter_fn_p (vec_info *vinfo, bool
> > > read_p, bool masked_p,
> > > *offset_vectype_out = offset_vectype;
> > > return true;
> > > }
> > > + else if (internal_gather_scatter_fn_supported_p (alt_ifn2, vectype,
> > > + memory_type,
> > > + offset_vectype, scale))
> > > + {
> > > + *ifn_out = alt_ifn2;
> > > + *offset_vectype_out = offset_vectype;
> > > + return true;
> > > + }
> > >
> > > if (TYPE_PRECISION (offset_type) >= POINTER_SIZE
> > > && TYPE_PRECISION (offset_type) >= element_bits)
> > > diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
> > > index a0c39268bf0..33ec33f8b8d 100644
> > > --- a/gcc/tree-vect-stmts.cc
> > > +++ b/gcc/tree-vect-stmts.cc
> > > @@ -433,7 +433,7 @@ exist_non_indexing_operands_for_use_p (tree use,
> > > stmt_vec_info stmt_info)
> > > if (call && gimple_call_internal_p (call))
> > > {
> > > internal_fn ifn = gimple_call_internal_fn (call);
> > > - int mask_index = internal_fn_mask_index (ifn);
> > > + int mask_index = internal_fn_mask_index (ifn, false);
> > > if (mask_index >= 0
> > > && use == gimple_call_arg (call, mask_index))
> > > return true;
> > > @@ -1771,6 +1771,18 @@ check_load_store_for_partial_vectors
> > > (loop_vec_info loop_vinfo, tree vectype,
> > > gs_info->offset_vectype,
> > > gs_info->scale))
> > > {
> > > + ifn = (is_load
> > > + ? IFN_LEN_MASK_GATHER_LOAD
> > > + : IFN_LEN_MASK_SCATTER_STORE);
> > > + if (internal_gather_scatter_fn_supported_p (ifn, vectype,
> > > + gs_info->memory_type,
> > > + gs_info->offset_vectype,
> > > + gs_info->scale))
> > > + {
> > > + vec_loop_lens *lens = &LOOP_VINFO_LENS (loop_vinfo);
> > > + vect_record_loop_len (loop_vinfo, lens, nvectors, vectype, 1);
> > > + return;
> > > + }
> > > if (dump_enabled_p ())
> > > dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
> > > "can't operate on partial vectors because"
> > > @@ -3129,16 +3141,39 @@ vect_get_gather_scatter_ops (loop_vec_info
> > > loop_vinfo,
> > > static void
> > > vect_get_strided_load_store_ops (stmt_vec_info stmt_info,
> > > loop_vec_info loop_vinfo,
> > > + gimple_stmt_iterator *gsi,
> > > gather_scatter_info *gs_info,
> > > - tree *dataref_bump, tree *vec_offset)
> > > + tree *dataref_bump, tree *vec_offset,
> > > + vec_loop_lens *loop_lens)
> > > {
> > > struct data_reference *dr = STMT_VINFO_DATA_REF (stmt_info);
> > > tree vectype = STMT_VINFO_VECTYPE (stmt_info);
> > >
> > > - tree bump = size_binop (MULT_EXPR,
> > > - fold_convert (sizetype, unshare_expr (DR_STEP (dr))),
> > > - size_int (TYPE_VECTOR_SUBPARTS (vectype)));
> > > - *dataref_bump = cse_and_gimplify_to_preheader (loop_vinfo, bump);
> > > + if (LOOP_VINFO_USING_SELECT_VL_P (loop_vinfo))
> > > + {
> > > + /* _31 = .SELECT_VL (ivtmp_29, POLY_INT_CST [4, 4]);
> > > + ivtmp_8 = _31 * 16 (step in bytes);
> > > + .LEN_MASK_SCATTER_STORE (vectp_a.9_7, ... );
> > > + vectp_a.9_26 = vectp_a.9_7 + ivtmp_8; */
> > > + tree loop_len
> > > + = vect_get_loop_len (loop_vinfo, gsi, loop_lens, 1, vectype, 0, 0);
> > > + tree tmp
> > > + = fold_build2 (MULT_EXPR, sizetype,
> > > + fold_convert (sizetype, unshare_expr (DR_STEP (dr))),
> > > + loop_len);
> > > + tree bump = make_temp_ssa_name (sizetype, NULL, "ivtmp");
> > > + gassign *assign = gimple_build_assign (bump, tmp);
> > > + gsi_insert_before (gsi, assign, GSI_SAME_STMT);
> > > + *dataref_bump = bump;
> > > + }
> > > + else
> > > + {
> > > + tree bump
> > > + = size_binop (MULT_EXPR,
> > > + fold_convert (sizetype, unshare_expr (DR_STEP (dr))),
> > > + size_int (TYPE_VECTOR_SUBPARTS (vectype)));
> > > + *dataref_bump = cse_and_gimplify_to_preheader (loop_vinfo, bump);
> > > + }
> > >
> > > /* The offset given in GS_INFO can have pointer type, so use the
> > > element
> > > type of the vector instead. */
> > > @@ -8013,7 +8048,7 @@ vectorizable_store (vec_info *vinfo,
> > > return false;
> > > }
> > >
> > > - int mask_index = internal_fn_mask_index (ifn);
> > > + int mask_index = internal_fn_mask_index (ifn, false);
> > > if (mask_index >= 0
> > > && !vect_check_scalar_mask (vinfo, stmt_info, slp_node, mask_index,
> > > &mask, NULL, &mask_dt, &mask_vectype))
> > > @@ -8685,8 +8720,8 @@ vectorizable_store (vec_info *vinfo,
> > > else if (memory_access_type == VMAT_GATHER_SCATTER)
> > > {
> > > aggr_type = elem_type;
> > > - vect_get_strided_load_store_ops (stmt_info, loop_vinfo, &gs_info,
> > > - &bump, &vec_offset);
> > > + vect_get_strided_load_store_ops (stmt_info, loop_vinfo, gsi,
> > > &gs_info,
> > > + &bump, &vec_offset, loop_lens);
> > > }
> > > else
> > > {
> > > @@ -8915,6 +8950,8 @@ vectorizable_store (vec_info *vinfo,
> > > unsigned HOST_WIDE_INT align;
> > >
> > > tree final_mask = NULL_TREE;
> > > + tree final_len = NULL_TREE;
> > > + tree bias = NULL_TREE;
> > > if (loop_masks)
> > > final_mask = vect_get_loop_mask (loop_vinfo, gsi, loop_masks,
> > > vec_num * ncopies,
> > > @@ -8929,8 +8966,41 @@ vectorizable_store (vec_info *vinfo,
> > > if (STMT_VINFO_GATHER_SCATTER_P (stmt_info))
> > > vec_offset = vec_offsets[vec_num * j + i];
> > > tree scale = size_int (gs_info.scale);
> > > +
> > > + if (gs_info.ifn == IFN_LEN_MASK_SCATTER_STORE)
> > > + {
> > > + if (loop_lens)
> > > + {
> > > + final_len
> > > + = vect_get_loop_len (loop_vinfo, gsi, loop_lens,
> > > + vec_num * ncopies, vectype,
> > > + vec_num * j + i, 1);
> > > + }
> > > + else
> > > + {
> > > + tree iv_type = LOOP_VINFO_RGROUP_IV_TYPE (loop_vinfo);
> > > + final_len
> > > + = build_int_cst (iv_type,
> > > + TYPE_VECTOR_SUBPARTS (vectype));
> > > + }
> > > + signed char biasval
> > > + = LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
> > > + bias = build_int_cst (intQI_type_node, biasval);
> > > + if (!final_mask)
> > > + {
> > > + mask_vectype = truth_type_for (vectype);
> > > + final_mask = build_minus_one_cst (mask_vectype);
> > > + }
> > > + }
> > > +
> > > gcall *call;
> > > - if (final_mask)
> > > + if (final_len && final_len)
> > > + call
> > > + = gimple_build_call_internal (IFN_LEN_MASK_SCATTER_STORE,
> > > + 7, dataref_ptr, vec_offset,
> > > + scale, vec_oprnd, final_len,
> > > + bias, final_mask);
> > > + else if (final_mask)
> > > call = gimple_build_call_internal
> > > (IFN_MASK_SCATTER_STORE, 5, dataref_ptr, vec_offset,
> > > scale, vec_oprnd, final_mask);
> > > @@ -9047,9 +9117,6 @@ vectorizable_store (vec_info *vinfo,
> > > machine_mode vmode = TYPE_MODE (vectype);
> > > machine_mode new_vmode = vmode;
> > > internal_fn partial_ifn = IFN_LAST;
> > > - /* Produce 'len' and 'bias' argument. */
> > > - tree final_len = NULL_TREE;
> > > - tree bias = NULL_TREE;
> > > if (loop_lens)
> > > {
> > > opt_machine_mode new_ovmode
> > > @@ -10177,8 +10244,8 @@ vectorizable_load (vec_info *vinfo,
> > > else if (memory_access_type == VMAT_GATHER_SCATTER)
> > > {
> > > aggr_type = elem_type;
> > > - vect_get_strided_load_store_ops (stmt_info, loop_vinfo, &gs_info,
> > > - &bump, &vec_offset);
> > > + vect_get_strided_load_store_ops (stmt_info, loop_vinfo, gsi,
> > > &gs_info,
> > > + &bump, &vec_offset, loop_lens);
> > > }
> > > else
> > > {
> > > @@ -10339,6 +10406,8 @@ vectorizable_load (vec_info *vinfo,
> > > for (i = 0; i < vec_num; i++)
> > > {
> > > tree final_mask = NULL_TREE;
> > > + tree final_len = NULL_TREE;
> > > + tree bias = NULL_TREE;
> > > if (loop_masks
> > > && memory_access_type != VMAT_INVARIANT)
> > > final_mask = vect_get_loop_mask (loop_vinfo, gsi, loop_masks,
> > > @@ -10368,8 +10437,39 @@ vectorizable_load (vec_info *vinfo,
> > > vec_offset = vec_offsets[vec_num * j + i];
> > > tree zero = build_zero_cst (vectype);
> > > tree scale = size_int (gs_info.scale);
> > > +
> > > + if (gs_info.ifn == IFN_LEN_MASK_GATHER_LOAD)
> > > + {
> > > + if (loop_lens)
> > > + {
> > > + final_len = vect_get_loop_len (
> > > + loop_vinfo, gsi, loop_lens, vec_num * ncopies,
> > > + vectype, vec_num * j + i, 1);
> > > + }
> > > + else
> > > + {
> > > + tree iv_type
> > > + = LOOP_VINFO_RGROUP_IV_TYPE (loop_vinfo);
> > > + final_len = build_int_cst (
> > > + iv_type, TYPE_VECTOR_SUBPARTS (vectype));
> > > + }
> > > + signed char biasval
> > > + = LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
> > > + bias = build_int_cst (intQI_type_node, biasval);
> > > + if (!final_mask)
> > > + {
> > > + mask_vectype = truth_type_for (vectype);
> > > + final_mask = build_minus_one_cst (mask_vectype);
> > > + }
> > > + }
> > > +
> > > gcall *call;
> > > - if (final_mask)
> > > + if (final_len && final_mask)
> > > + call = gimple_build_call_internal (
> > > + IFN_LEN_MASK_GATHER_LOAD, 7, dataref_ptr,
> > > + vec_offset, scale, zero, final_len, bias,
> > > + final_mask);
> > > + else if (final_mask)
> > > call = gimple_build_call_internal
> > > (IFN_MASK_GATHER_LOAD, 5, dataref_ptr,
> > > vec_offset, scale, zero, final_mask);
> > > @@ -10462,9 +10562,6 @@ vectorizable_load (vec_info *vinfo,
> > > machine_mode vmode = TYPE_MODE (vectype);
> > > machine_mode new_vmode = vmode;
> > > internal_fn partial_ifn = IFN_LAST;
> > > - /* Produce 'len' and 'bias' argument. */
> > > - tree final_len = NULL_TREE;
> > > - tree bias = NULL_TREE;
> > > if (loop_lens)
> > > {
> > > opt_machine_mode new_ovmode
> > >
> >
> >
>
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH, Frankenstrasse 146, 90461 Nuernberg,
Germany; GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman;
HRB 36809 (AG Nuernberg)
