Re: Re: [PATCH V3] VECT: Apply LEN_MASK_{LOAD,STORE} into vectorizer

["juzhe.zhong@rivai.ai" <juzhe.zhong@rivai.ai>]

[-- Attachment #1: Type: text/plain, Size: 19806 bytes --]

Hi, Richi. Thanks so much for the review and comments.

>> Can you instead adjust get_len_load_store_mode and
>>can_vec_mask_load_store_p to provide the optab they matched on
>>via the corresponding IFN code as additional output (add a
>>pointer argument, you can default it to nullptr and only
>>fill in the detail in the context that need it)?
>>Like above the _len case can then simply
>>take precedence.

Do you mean I remove partial_or_mask_vector_ifn
then use can_vec_mask_load_store_p or get_len_load_store_mode to calculate IFN instead?

I tried to add pointer argument internal_fn * into get_len_load_store_mode but it fail to compile:

../../../riscv-gcc/gcc/optabs-query.h:191:63: error: ‘internal_fn’ has not been declared
 opt_machine_mode get_len_load_store_mode (machine_mode, bool, internal_fn* = nullptr);

I am not sure whether I am on the same page with you.
Could you help me with that?

Thanks.


juzhe.zhong@rivai.ai
 
From: Richard Biener
Date: 2023-06-21 16:53
To: Ju-Zhe Zhong
CC: gcc-patches; richard.sandiford
Subject: Re: [PATCH V3] VECT: Apply LEN_MASK_{LOAD,STORE} into vectorizer
On Tue, 20 Jun 2023, juzhe.zhong@rivai.ai wrote:
 
> From: Ju-Zhe Zhong <juzhe.zhong@rivai.ai>
> 
> gcc/ChangeLog:
> 
>         * internal-fn.cc (expand_partial_store_optab_fn): Add LEN_MASK_{LOAD,STORE} vectorizer support.
>         (internal_load_fn_p): Ditto.
>         (internal_store_fn_p): Ditto.
>         (internal_fn_mask_index): Ditto.
>         (internal_fn_stored_value_index): Ditto.
>         (internal_len_load_store_bias): Ditto.
>         * optabs-query.cc (can_vec_mask_load_store_p): Ditto.
>         (get_len_load_store_mode): Ditto.
>         * tree-vect-stmts.cc (check_load_store_for_partial_vectors): Ditto.
>         (get_all_ones_mask): New function.
>         (vectorizable_store): Add LEN_MASK_{LOAD,STORE} vectorizer support.
>         (vectorizable_load): Ditto.
> D
> ---
>  gcc/internal-fn.cc     |  35 +++++-
>  gcc/optabs-query.cc    |  25 ++++-
>  gcc/tree-vect-stmts.cc | 234 ++++++++++++++++++++++++++++++-----------
>  3 files changed, 227 insertions(+), 67 deletions(-)
> 
> diff --git a/gcc/internal-fn.cc b/gcc/internal-fn.cc
> index c911ae790cb..e10c21de5f1 100644
> --- a/gcc/internal-fn.cc
> +++ b/gcc/internal-fn.cc
> @@ -2949,7 +2949,7 @@ expand_partial_load_optab_fn (internal_fn, gcall *stmt, convert_optab optab)
>   * OPTAB.  */
>  
>  static void
> -expand_partial_store_optab_fn (internal_fn, gcall *stmt, convert_optab optab)
> +expand_partial_store_optab_fn (internal_fn ifn, gcall *stmt, convert_optab optab)
>  {
>    class expand_operand ops[5];
>    tree type, lhs, rhs, maskt, biast;
> @@ -2957,7 +2957,7 @@ expand_partial_store_optab_fn (internal_fn, gcall *stmt, convert_optab optab)
>    insn_code icode;
>  
>    maskt = gimple_call_arg (stmt, 2);
> -  rhs = gimple_call_arg (stmt, 3);
> +  rhs = gimple_call_arg (stmt, internal_fn_stored_value_index (ifn));
>    type = TREE_TYPE (rhs);
>    lhs = expand_call_mem_ref (type, stmt, 0);
>  
> @@ -4435,6 +4435,7 @@ internal_load_fn_p (internal_fn fn)
>      case IFN_GATHER_LOAD:
>      case IFN_MASK_GATHER_LOAD:
>      case IFN_LEN_LOAD:
> +    case IFN_LEN_MASK_LOAD:
>        return true;
>  
>      default:
> @@ -4455,6 +4456,7 @@ internal_store_fn_p (internal_fn fn)
>      case IFN_SCATTER_STORE:
>      case IFN_MASK_SCATTER_STORE:
>      case IFN_LEN_STORE:
> +    case IFN_LEN_MASK_STORE:
>        return true;
>  
>      default:
> @@ -4494,6 +4496,10 @@ internal_fn_mask_index (internal_fn fn)
>      case IFN_MASK_STORE_LANES:
>        return 2;
>  
> +    case IFN_LEN_MASK_LOAD:
> +    case IFN_LEN_MASK_STORE:
> +      return 3;
> +
>      case IFN_MASK_GATHER_LOAD:
>      case IFN_MASK_SCATTER_STORE:
>        return 4;
> @@ -4519,6 +4525,9 @@ internal_fn_stored_value_index (internal_fn fn)
>      case IFN_LEN_STORE:
>        return 3;
>  
> +    case IFN_LEN_MASK_STORE:
> +      return 4;
> +
>      default:
>        return -1;
>      }
> @@ -4583,13 +4592,31 @@ internal_len_load_store_bias (internal_fn ifn, machine_mode mode)
>  {
>    optab optab = direct_internal_fn_optab (ifn);
>    insn_code icode = direct_optab_handler (optab, mode);
> +  int bias_argno = 3;
> +  if (icode == CODE_FOR_nothing)
> +    {
> +      machine_mode mask_mode
> + = targetm.vectorize.get_mask_mode (mode).require ();
> +      if (ifn == IFN_LEN_LOAD)
> + {
> +   /* Try LEN_MASK_LOAD.  */
> +   optab = direct_internal_fn_optab (IFN_LEN_MASK_LOAD);
> + }
> +      else
> + {
> +   /* Try LEN_MASK_STORE.  */
> +   optab = direct_internal_fn_optab (IFN_LEN_MASK_STORE);
> + }
> +      icode = convert_optab_handler (optab, mode, mask_mode);
> +      bias_argno = 4;
> +    }
>  
>    if (icode != CODE_FOR_nothing)
>      {
>        /* For now we only support biases of 0 or -1.  Try both of them.  */
> -      if (insn_operand_matches (icode, 3, GEN_INT (0)))
> +      if (insn_operand_matches (icode, bias_argno, GEN_INT (0)))
>  return 0;
> -      if (insn_operand_matches (icode, 3, GEN_INT (-1)))
> +      if (insn_operand_matches (icode, bias_argno, GEN_INT (-1)))
>  return -1;
>      }
>  
> diff --git a/gcc/optabs-query.cc b/gcc/optabs-query.cc
> index 276f8408dd7..4394d391200 100644
> --- a/gcc/optabs-query.cc
> +++ b/gcc/optabs-query.cc
> @@ -566,11 +566,14 @@ can_vec_mask_load_store_p (machine_mode mode,
>     bool is_load)
>  {
>    optab op = is_load ? maskload_optab : maskstore_optab;
> +  optab len_op = is_load ? len_maskload_optab : len_maskstore_optab;
>    machine_mode vmode;
>  
>    /* If mode is vector mode, check it directly.  */
>    if (VECTOR_MODE_P (mode))
> -    return convert_optab_handler (op, mode, mask_mode) != CODE_FOR_nothing;
> +    return convert_optab_handler (op, mode, mask_mode) != CODE_FOR_nothing
> +    || convert_optab_handler (len_op, mode, mask_mode)
> + != CODE_FOR_nothing;
>  
>    /* Otherwise, return true if there is some vector mode with
>       the mask load/store supported.  */
> @@ -584,7 +587,9 @@ can_vec_mask_load_store_p (machine_mode mode,
>    vmode = targetm.vectorize.preferred_simd_mode (smode);
>    if (VECTOR_MODE_P (vmode)
>        && targetm.vectorize.get_mask_mode (vmode).exists (&mask_mode)
> -      && convert_optab_handler (op, vmode, mask_mode) != CODE_FOR_nothing)
> +      && (convert_optab_handler (op, vmode, mask_mode) != CODE_FOR_nothing
> +   || convert_optab_handler (len_op, vmode, mask_mode)
> +        != CODE_FOR_nothing))
>      return true;
>  
>    auto_vector_modes vector_modes;
> @@ -592,7 +597,9 @@ can_vec_mask_load_store_p (machine_mode mode,
>    for (machine_mode base_mode : vector_modes)
>      if (related_vector_mode (base_mode, smode).exists (&vmode)
>  && targetm.vectorize.get_mask_mode (vmode).exists (&mask_mode)
> - && convert_optab_handler (op, vmode, mask_mode) != CODE_FOR_nothing)
> + && (convert_optab_handler (op, vmode, mask_mode) != CODE_FOR_nothing
> +     || convert_optab_handler (len_op, vmode, mask_mode)
> + != CODE_FOR_nothing))
>        return true;
>    return false;
>  }
> @@ -608,17 +615,27 @@ opt_machine_mode
>  get_len_load_store_mode (machine_mode mode, bool is_load)
>  {
>    optab op = is_load ? len_load_optab : len_store_optab;
> +  optab masked_op = is_load ? len_maskload_optab : len_maskstore_optab;
>    gcc_assert (VECTOR_MODE_P (mode));
>  
>    /* Check if length in lanes supported for this mode directly.  */
>    if (direct_optab_handler (op, mode))
>      return mode;
>  
> +  /* Check if length in lanes supported by len_maskload/store.  */
> +  machine_mode mask_mode;
> +  if (targetm.vectorize.get_mask_mode (mode).exists (&mask_mode)
> +      && convert_optab_handler (masked_op, mode, mask_mode) != CODE_FOR_nothing)
> +    return mode;
> +
>    /* Check if length in bytes supported for same vector size VnQI.  */
>    machine_mode vmode;
>    poly_uint64 nunits = GET_MODE_SIZE (mode);
>    if (related_vector_mode (mode, QImode, nunits).exists (&vmode)
> -      && direct_optab_handler (op, vmode))
> +      && (direct_optab_handler (op, vmode)
> +   || (targetm.vectorize.get_mask_mode (vmode).exists (&mask_mode)
> +       && convert_optab_handler (masked_op, vmode, mask_mode)
> +    != CODE_FOR_nothing)))
>      return vmode;
>  
>    return opt_machine_mode ();
> diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
> index 056a0ecb2be..416fde9a9a6 100644
> --- a/gcc/tree-vect-stmts.cc
> +++ b/gcc/tree-vect-stmts.cc
> @@ -1819,16 +1819,8 @@ check_load_store_for_partial_vectors (loop_vec_info loop_vinfo, tree vectype,
>    poly_uint64 nunits = TYPE_VECTOR_SUBPARTS (vectype);
>    poly_uint64 vf = LOOP_VINFO_VECT_FACTOR (loop_vinfo);
>    machine_mode mask_mode;
> -  bool using_partial_vectors_p = false;
> -  if (targetm.vectorize.get_mask_mode (vecmode).exists (&mask_mode)
> -      && can_vec_mask_load_store_p (vecmode, mask_mode, is_load))
> -    {
> -      nvectors = group_memory_nvectors (group_size * vf, nunits);
> -      vect_record_loop_mask (loop_vinfo, masks, nvectors, vectype, scalar_mask);
> -      using_partial_vectors_p = true;
> -    }
> -
>    machine_mode vmode;
> +  bool using_partial_vectors_p = false;
>    if (get_len_load_store_mode (vecmode, is_load).exists (&vmode))
>      {
>        nvectors = group_memory_nvectors (group_size * vf, nunits);
> @@ -1837,6 +1829,13 @@ check_load_store_for_partial_vectors (loop_vec_info loop_vinfo, tree vectype,
>        vect_record_loop_len (loop_vinfo, lens, nvectors, vectype, factor);
>        using_partial_vectors_p = true;
>      }
> +  else if (targetm.vectorize.get_mask_mode (vecmode).exists (&mask_mode)
> +    && can_vec_mask_load_store_p (vecmode, mask_mode, is_load))
> +    {
> +      nvectors = group_memory_nvectors (group_size * vf, nunits);
> +      vect_record_loop_mask (loop_vinfo, masks, nvectors, vectype, scalar_mask);
> +      using_partial_vectors_p = true;
> +    }
>  
>    if (!using_partial_vectors_p)
>      {
> @@ -2809,6 +2808,58 @@ vect_build_zero_merge_argument (vec_info *vinfo,
>    return vect_init_vector (vinfo, stmt_info, merge, vectype, NULL);
>  }
>  
> +/* Get all-ones vector mask for corresponding vectype.  */
> +
> +static tree
> +get_all_ones_mask (machine_mode vmode)
> +{
> +  machine_mode maskmode = targetm.vectorize.get_mask_mode (vmode).require ();
> +  poly_uint64 nunits = GET_MODE_NUNITS (maskmode);
> +  tree masktype = build_truth_vector_type_for_mode (nunits, maskmode);
> +  return constant_boolean_node (true, masktype);
> +}
> +
> +/* Get the partial vector IFN that the target supports.
> +
> +   For partial contiguous load, we could return IFN_LEN_LOAD, IFN_MASK_LOAD
> +   or IFN_LEN_MASK_LOAD.
> +
> +   For partial contiguous load, we could return IFN_LEN_STORE, IFN_MASK_STORE
> +   or IFN_LEN_MASK_STORE.
> +*/
> +
> +static internal_fn
> +partial_or_mask_vector_ifn (machine_mode vecmode, bool is_load)
> +{
> +  machine_mode maskmode;
> +  machine_mode vmode;
 
Can you instead adjust get_len_load_store_mode and
can_vec_mask_load_store_p to provide the optab they matched on
via the corresponding IFN code as additional output (add a
pointer argument, you can default it to nullptr and only
fill in the detail in the context that need it)?
Like above the _len case can then simply
take precedence.
 
The rest looks good now.
 
Thanks,
Richard.
 
> +  if (get_len_load_store_mode (vecmode, is_load).exists (&vmode))
> +    {
> +      if (targetm.vectorize.get_mask_mode (vecmode).exists (&maskmode)
> +   && can_vec_mask_load_store_p (vecmode, maskmode, is_load))
> + {
> +   if (is_load)
> +     return IFN_LEN_MASK_LOAD;
> +   else
> +     return IFN_LEN_MASK_STORE;
> + }
> +      if (is_load)
> + return IFN_LEN_LOAD;
> +      else
> + return IFN_LEN_STORE;
> +    }
> +  else if (targetm.vectorize.get_mask_mode (vecmode).exists (&maskmode)
> +    && can_vec_mask_load_store_p (vecmode, maskmode, is_load))
> +    {
> +      if (is_load)
> + return IFN_MASK_LOAD;
> +      else
> + return IFN_MASK_STORE;
> +    }
> +  return IFN_LAST;
> +}
> +
>  /* Build a gather load call while vectorizing STMT_INFO.  Insert new
>     instructions before GSI and add them to VEC_STMT.  GS_INFO describes
>     the gather load operation.  If the load is conditional, MASK is the
> @@ -8945,30 +8996,46 @@ vectorizable_store (vec_info *vinfo,
>  }
>  
>        /* Arguments are ready.  Create the new vector stmt.  */
> -       if (final_mask)
> - {
> -   tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
> -   gcall *call
> -     = gimple_build_call_internal (IFN_MASK_STORE, 4,
> -   dataref_ptr, ptr,
> -   final_mask, vec_oprnd);
> -   gimple_call_set_nothrow (call, true);
> -   vect_finish_stmt_generation (vinfo, stmt_info, call, gsi);
> -   new_stmt = call;
> - }
> -       else if (loop_lens)
> +       internal_fn partial_ifn
> + = partial_or_mask_vector_ifn (TYPE_MODE (vectype), false);
> +       tree final_len = NULL_TREE;
> +       machine_mode vmode = TYPE_MODE (vectype);
> +       machine_mode new_vmode;
> +
> +       /* Produce 'len' argument.  */
> +       if (loop_lens)
>  {
> -   machine_mode vmode = TYPE_MODE (vectype);
>    opt_machine_mode new_ovmode
>      = get_len_load_store_mode (vmode, false);
> -   machine_mode new_vmode = new_ovmode.require ();
> +   new_vmode = new_ovmode.require ();
>    unsigned factor
>      = (new_ovmode == vmode) ? 1 : GET_MODE_UNIT_SIZE (vmode);
> -   tree final_len
> -     = vect_get_loop_len (loop_vinfo, gsi, loop_lens,
> - vec_num * ncopies, vectype,
> - vec_num * j + i, factor);
> -   tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
> +   final_len = vect_get_loop_len (loop_vinfo, gsi, loop_lens,
> + vec_num * ncopies, vectype,
> + vec_num * j + i, factor);
> + }
> +       if (partial_ifn == IFN_LEN_MASK_STORE)
> + {
> +   if (!final_len)
> +     {
> +       /* Pass VF value to 'len' argument of LEN_MASK_STORE if
> +        * LOOP_LENS is invalid.  */
> +       tree iv_type = LOOP_VINFO_RGROUP_IV_TYPE (loop_vinfo);
> +       final_len
> + = build_int_cst (iv_type,
> + TYPE_VECTOR_SUBPARTS (vectype));
> +     }
> +   if (!final_mask)
> +     {
> +       /* Pass all ones value to 'mask' argument of
> +        * LEN_MASK_STORE if final_mask is invalid.  */
> +       final_mask = get_all_ones_mask (vmode);
> +     }
> + }
> +
> +       tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
> +       if (final_len)
> + {
>    /* Need conversion if it's wrapped with VnQI.  */
>    if (vmode != new_vmode)
>      {
> @@ -8987,14 +9054,32 @@ vectorizable_store (vec_info *vinfo,
>        vec_oprnd = var;
>      }
>  
> -   signed char biasval =
> -     LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
> +   signed char biasval
> +     = LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
>  
>    tree bias = build_int_cst (intQI_type_node, biasval);
> +   gcall *call;
> +
> +   if (final_mask)
> +     call = gimple_build_call_internal (IFN_LEN_MASK_STORE, 6,
> +        dataref_ptr, ptr,
> +        final_len, final_mask,
> +        vec_oprnd, bias);
> +   else
> +     call
> +       = gimple_build_call_internal (IFN_LEN_STORE, 5,
> +     dataref_ptr, ptr, final_len,
> +     vec_oprnd, bias);
> +   gimple_call_set_nothrow (call, true);
> +   vect_finish_stmt_generation (vinfo, stmt_info, call, gsi);
> +   new_stmt = call;
> + }
> +       else if (final_mask)
> + {
>    gcall *call
> -     = gimple_build_call_internal (IFN_LEN_STORE, 5, dataref_ptr,
> -   ptr, final_len, vec_oprnd,
> -   bias);
> +     = gimple_build_call_internal (IFN_MASK_STORE, 4,
> +   dataref_ptr, ptr,
> +   final_mask, vec_oprnd);
>    gimple_call_set_nothrow (call, true);
>    vect_finish_stmt_generation (vinfo, stmt_info, call, gsi);
>    new_stmt = call;
> @@ -10304,45 +10389,66 @@ vectorizable_load (vec_info *vinfo,
>        align, misalign);
>      align = least_bit_hwi (misalign | align);
>  
> -     if (final_mask)
> -       {
> - tree ptr = build_int_cst (ref_type,
> -   align * BITS_PER_UNIT);
> - gcall *call
> -   = gimple_build_call_internal (IFN_MASK_LOAD, 3,
> - dataref_ptr, ptr,
> - final_mask);
> - gimple_call_set_nothrow (call, true);
> - new_stmt = call;
> - data_ref = NULL_TREE;
> -       }
> -     else if (loop_lens && memory_access_type != VMAT_INVARIANT)
> +     internal_fn partial_ifn
> +       = partial_or_mask_vector_ifn (TYPE_MODE (vectype), true);
> +     tree final_len = NULL_TREE;
> +     machine_mode vmode = TYPE_MODE (vectype);
> +     machine_mode new_vmode;
> +
> +     /* Produce 'len' argument.  */
> +     if (loop_lens)
>        {
> - machine_mode vmode = TYPE_MODE (vectype);
>  opt_machine_mode new_ovmode
> -   = get_len_load_store_mode (vmode, true);
> - machine_mode new_vmode = new_ovmode.require ();
> +   = get_len_load_store_mode (vmode, false);
> + new_vmode = new_ovmode.require ();
>  unsigned factor = (new_ovmode == vmode)
>      ? 1
>      : GET_MODE_UNIT_SIZE (vmode);
> - tree final_len
> + final_len
>    = vect_get_loop_len (loop_vinfo, gsi, loop_lens,
>         vec_num * ncopies, vectype,
>         vec_num * j + i, factor);
> - tree ptr
> -   = build_int_cst (ref_type, align * BITS_PER_UNIT);
> +       }
> +     if (partial_ifn == IFN_LEN_MASK_LOAD)
> +       {
> + if (!final_len)
> +   {
> +     /* Pass VF value to 'len' argument of LEN_MASK_STORE
> +      * if LOOP_LENS is invalid.  */
> +     tree iv_type
> +       = LOOP_VINFO_RGROUP_IV_TYPE (loop_vinfo);
> +     final_len
> +       = build_int_cst (iv_type,
> +        TYPE_VECTOR_SUBPARTS (vectype));
> +   }
> + if (!final_mask)
> +   {
> +     /* Pass all ones value to 'mask' argument of
> +      * LEN_MASK_STORE if final_mask is invalid.  */
> +     final_mask = get_all_ones_mask (vmode);
> +   }
> +       }
> +
> +     tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
> +     if (final_len && memory_access_type != VMAT_INVARIANT)
> +       {
> + gcall *call;
>  
>  tree qi_type = unsigned_intQI_type_node;
>  
> - signed char biasval =
> -   LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
> + signed char biasval
> +   = LOOP_VINFO_PARTIAL_LOAD_STORE_BIAS (loop_vinfo);
>  
>  tree bias = build_int_cst (intQI_type_node, biasval);
> -
> - gcall *call
> -   = gimple_build_call_internal (IFN_LEN_LOAD, 4,
> - dataref_ptr, ptr,
> - final_len, bias);
> + if (final_mask)
> +   call = gimple_build_call_internal (IFN_LEN_MASK_LOAD,
> +      5, dataref_ptr,
> +      ptr, final_len,
> +      final_mask, bias);
> + else
> +   call = gimple_build_call_internal (IFN_LEN_LOAD, 4,
> +      dataref_ptr, ptr,
> +      final_len, bias);
>  gimple_call_set_nothrow (call, true);
>  new_stmt = call;
>  data_ref = NULL_TREE;
> @@ -10363,6 +10469,16 @@ vectorizable_load (vec_info *vinfo,
>       VIEW_CONVERT_EXPR, op);
>    }
>        }
> +     else if (final_mask)
> +       {
> + gcall *call
> +   = gimple_build_call_internal (IFN_MASK_LOAD, 3,
> + dataref_ptr, ptr,
> + final_mask);
> + gimple_call_set_nothrow (call, true);
> + new_stmt = call;
> + data_ref = NULL_TREE;
> +       }
>      else
>        {
>  tree ltype = vectype;
> 
 
-- 
Richard Biener <rguenther@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)