From: "Li, Pan2" <pan2.li@intel.com>
To: Richard Biener <richard.guenther@gmail.com>
Cc: "gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>,
"juzhe.zhong@rivai.ai" <juzhe.zhong@rivai.ai>,
"kito.cheng@gmail.com" <kito.cheng@gmail.com>,
"tamar.christina@arm.com" <tamar.christina@arm.com>,
"Liu, Hongtao" <hongtao.liu@intel.com>
Subject: RE: [PATCH v4 1/3] Internal-fn: Support new IFN SAT_ADD for unsigned scalar int
Date: Tue, 14 May 2024 14:14:49 +0000 [thread overview]
Message-ID: <MW5PR11MB59086674B904F7079865D65DA9E32@MW5PR11MB5908.namprd11.prod.outlook.com> (raw)
In-Reply-To: <CAFiYyc35L1ec9vqXM-pCix44jV9o1R3RMC_NvPm3i_x4b_NRPQ@mail.gmail.com>
Thanks Richard for comments.
> If you require a gassign please statically type your function
> argument as gassign * instead and remove this assert.
Sure
> As addition to Tamars good comments why do you set *cfg_changed_p to
> true? You are
> not changing the CFG afer all?
Yes, we can add it back in future if we really changed cfg, will update in v5 (include vect patch 2/3) after all test passed.
Pan
-----Original Message-----
From: Richard Biener <richard.guenther@gmail.com>
Sent: Tuesday, May 14, 2024 9:18 PM
To: Li, Pan2 <pan2.li@intel.com>
Cc: gcc-patches@gcc.gnu.org; juzhe.zhong@rivai.ai; kito.cheng@gmail.com; tamar.christina@arm.com; Liu, Hongtao <hongtao.liu@intel.com>
Subject: Re: [PATCH v4 1/3] Internal-fn: Support new IFN SAT_ADD for unsigned scalar int
On Mon, May 6, 2024 at 4:48 PM <pan2.li@intel.com> wrote:
>
> From: Pan Li <pan2.li@intel.com>
>
> This patch would like to add the middle-end presentation for the
> saturation add. Aka set the result of add to the max when overflow.
> It will take the pattern similar as below.
>
> SAT_ADD (x, y) => (x + y) | (-(TYPE)((TYPE)(x + y) < x))
>
> Take uint8_t as example, we will have:
>
> * SAT_ADD (1, 254) => 255.
> * SAT_ADD (1, 255) => 255.
> * SAT_ADD (2, 255) => 255.
> * SAT_ADD (255, 255) => 255.
>
> Given below example for the unsigned scalar integer uint64_t:
>
> uint64_t sat_add_u64 (uint64_t x, uint64_t y)
> {
> return (x + y) | (- (uint64_t)((uint64_t)(x + y) < x));
> }
>
> Before this patch:
> uint64_t sat_add_uint64_t (uint64_t x, uint64_t y)
> {
> long unsigned int _1;
> _Bool _2;
> long unsigned int _3;
> long unsigned int _4;
> uint64_t _7;
> long unsigned int _10;
> __complex__ long unsigned int _11;
>
> ;; basic block 2, loop depth 0
> ;; pred: ENTRY
> _11 = .ADD_OVERFLOW (x_5(D), y_6(D));
> _1 = REALPART_EXPR <_11>;
> _10 = IMAGPART_EXPR <_11>;
> _2 = _10 != 0;
> _3 = (long unsigned int) _2;
> _4 = -_3;
> _7 = _1 | _4;
> return _7;
> ;; succ: EXIT
>
> }
>
> After this patch:
> uint64_t sat_add_uint64_t (uint64_t x, uint64_t y)
> {
> uint64_t _7;
>
> ;; basic block 2, loop depth 0
> ;; pred: ENTRY
> _7 = .SAT_ADD (x_5(D), y_6(D)); [tail call]
> return _7;
> ;; succ: EXIT
> }
>
> We perform the tranform during widen_mult because that the sub-expr of
> SAT_ADD will be optimized to .ADD_OVERFLOW. We need to try the .SAT_ADD
> pattern first and then .ADD_OVERFLOW, or we may never catch the pattern
> .SAT_ADD. Meanwhile, the isel pass is after widen_mult and then we
> cannot perform the .SAT_ADD pattern match as the sub-expr will be
> optmized to .ADD_OVERFLOW first.
>
> The below tests are passed for this patch:
> 1. The riscv fully regression tests.
> 2. The aarch64 fully regression tests.
> 3. The x86 bootstrap tests.
> 4. The x86 fully regression tests.
>
> PR target/51492
> PR target/112600
>
> gcc/ChangeLog:
>
> * internal-fn.cc (commutative_binary_fn_p): Add type IFN_SAT_ADD
> to the return true switch case(s).
> * internal-fn.def (SAT_ADD): Add new signed optab SAT_ADD.
> * match.pd: Add unsigned SAT_ADD match.
> * optabs.def (OPTAB_NL): Remove fixed-point limitation for us/ssadd.
> * tree-ssa-math-opts.cc (gimple_unsigned_integer_sat_add): New extern
> func decl generated in match.pd match.
> (match_saturation_arith): New func impl to match the saturation arith.
> (math_opts_dom_walker::after_dom_children): Try match saturation
> arith.
>
> Signed-off-by: Pan Li <pan2.li@intel.com>
> ---
> gcc/internal-fn.cc | 1 +
> gcc/internal-fn.def | 2 ++
> gcc/match.pd | 28 ++++++++++++++++++++++++
> gcc/optabs.def | 4 ++--
> gcc/tree-ssa-math-opts.cc | 46 +++++++++++++++++++++++++++++++++++++++
> 5 files changed, 79 insertions(+), 2 deletions(-)
>
> diff --git a/gcc/internal-fn.cc b/gcc/internal-fn.cc
> index 0a7053c2286..73045ca8c8c 100644
> --- a/gcc/internal-fn.cc
> +++ b/gcc/internal-fn.cc
> @@ -4202,6 +4202,7 @@ commutative_binary_fn_p (internal_fn fn)
> case IFN_UBSAN_CHECK_MUL:
> case IFN_ADD_OVERFLOW:
> case IFN_MUL_OVERFLOW:
> + case IFN_SAT_ADD:
> case IFN_VEC_WIDEN_PLUS:
> case IFN_VEC_WIDEN_PLUS_LO:
> case IFN_VEC_WIDEN_PLUS_HI:
> diff --git a/gcc/internal-fn.def b/gcc/internal-fn.def
> index 848bb9dbff3..25badbb86e5 100644
> --- a/gcc/internal-fn.def
> +++ b/gcc/internal-fn.def
> @@ -275,6 +275,8 @@ DEF_INTERNAL_SIGNED_OPTAB_FN (MULHS, ECF_CONST | ECF_NOTHROW, first,
> DEF_INTERNAL_SIGNED_OPTAB_FN (MULHRS, ECF_CONST | ECF_NOTHROW, first,
> smulhrs, umulhrs, binary)
>
> +DEF_INTERNAL_SIGNED_OPTAB_FN (SAT_ADD, ECF_CONST, first, ssadd, usadd, binary)
> +
> DEF_INTERNAL_COND_FN (ADD, ECF_CONST, add, binary)
> DEF_INTERNAL_COND_FN (SUB, ECF_CONST, sub, binary)
> DEF_INTERNAL_COND_FN (MUL, ECF_CONST, smul, binary)
> diff --git a/gcc/match.pd b/gcc/match.pd
> index d401e7503e6..7058e4cbe29 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -3043,6 +3043,34 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
> || POINTER_TYPE_P (itype))
> && wi::eq_p (wi::to_wide (int_cst), wi::max_value (itype))))))
>
> +/* Unsigned Saturation Add */
> +(match (usadd_left_part @0 @1)
> + (plus:c @0 @1)
> + (if (INTEGRAL_TYPE_P (type)
> + && TYPE_UNSIGNED (TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@1)))))
> +
> +(match (usadd_right_part @0 @1)
> + (negate (convert (lt (plus:c @0 @1) @0)))
> + (if (INTEGRAL_TYPE_P (type)
> + && TYPE_UNSIGNED (TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@1)))))
> +
> +(match (usadd_right_part @0 @1)
> + (negate (convert (gt @0 (plus:c @0 @1))))
> + (if (INTEGRAL_TYPE_P (type)
> + && TYPE_UNSIGNED (TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@0))
> + && types_match (type, TREE_TYPE (@1)))))
> +
> +/* Unsigned saturation add, case 1 (branchless):
> + SAT_U_ADD = (X + Y) | - ((X + Y) < X) or
> + SAT_U_ADD = (X + Y) | - (X > (X + Y)). */
> +(match (unsigned_integer_sat_add @0 @1)
> + (bit_ior:c (usadd_left_part @0 @1) (usadd_right_part @0 @1)))
> +
> /* x > y && x != XXX_MIN --> x > y
> x > y && x == XXX_MIN --> false . */
> (for eqne (eq ne)
> diff --git a/gcc/optabs.def b/gcc/optabs.def
> index ad14f9328b9..3f2cb46aff8 100644
> --- a/gcc/optabs.def
> +++ b/gcc/optabs.def
> @@ -111,8 +111,8 @@ OPTAB_NX(add_optab, "add$F$a3")
> OPTAB_NX(add_optab, "add$Q$a3")
> OPTAB_VL(addv_optab, "addv$I$a3", PLUS, "add", '3', gen_intv_fp_libfunc)
> OPTAB_VX(addv_optab, "add$F$a3")
> -OPTAB_NL(ssadd_optab, "ssadd$Q$a3", SS_PLUS, "ssadd", '3', gen_signed_fixed_libfunc)
> -OPTAB_NL(usadd_optab, "usadd$Q$a3", US_PLUS, "usadd", '3', gen_unsigned_fixed_libfunc)
> +OPTAB_NL(ssadd_optab, "ssadd$a3", SS_PLUS, "ssadd", '3', gen_signed_fixed_libfunc)
> +OPTAB_NL(usadd_optab, "usadd$a3", US_PLUS, "usadd", '3', gen_unsigned_fixed_libfunc)
> OPTAB_NL(sub_optab, "sub$P$a3", MINUS, "sub", '3', gen_int_fp_fixed_libfunc)
> OPTAB_NX(sub_optab, "sub$F$a3")
> OPTAB_NX(sub_optab, "sub$Q$a3")
> diff --git a/gcc/tree-ssa-math-opts.cc b/gcc/tree-ssa-math-opts.cc
> index 705f4a4695a..35a46edc9f6 100644
> --- a/gcc/tree-ssa-math-opts.cc
> +++ b/gcc/tree-ssa-math-opts.cc
> @@ -4026,6 +4026,44 @@ arith_overflow_check_p (gimple *stmt, gimple *cast_stmt, gimple *&use_stmt,
> return 0;
> }
>
> +extern bool gimple_unsigned_integer_sat_add (tree, tree*, tree (*)(tree));
> +
> +/*
> + * Try to match saturation arith pattern(s).
> + * 1. SAT_ADD (unsigned)
> + * _7 = _4 + _6;
> + * _8 = _4 > _7;
> + * _9 = (long unsigned int) _8;
> + * _10 = -_9;
> + * _12 = _7 | _10;
> + * =>
> + * _12 = .SAT_ADD (_4, _6); */
> +static bool
> +match_saturation_arith (gimple_stmt_iterator *gsi, gimple *stmt,
> + bool *cfg_changed_p)
> +{
> + gcall *call = NULL;
> + bool changed_p = false;
> +
> + gcc_assert (is_gimple_assign (stmt));
If you require a gassign please statically type your function
argument as gassign * instead and remove this assert.
> +
> + tree ops[2];
> + tree lhs = gimple_assign_lhs (stmt);
> +
> + if (gimple_unsigned_integer_sat_add (lhs, ops, NULL)
> + && direct_internal_fn_supported_p (IFN_SAT_ADD, TREE_TYPE (lhs),
> + OPTIMIZE_FOR_SPEED))
> + {
> + call = gimple_build_call_internal (IFN_SAT_ADD, 2, ops[0], ops[1]);
> + gimple_call_set_lhs (call, lhs);
> + gsi_replace (gsi, call, true);
> + changed_p = true;
> + *cfg_changed_p = changed_p;
As addition to Tamars good comments why do you set *cfg_changed_p to
true? You are
not changing the CFG afer all?
> + }
> +
> + return changed_p;
> +}
> +
> /* Recognize for unsigned x
> x = y - z;
> if (x > y)
> @@ -5886,6 +5924,14 @@ math_opts_dom_walker::after_dom_children (basic_block bb)
>
> fma_deferring_state fma_state (param_avoid_fma_max_bits > 0);
>
> + for (gsi = gsi_after_labels (bb); !gsi_end_p (gsi); gsi_next (&gsi))
> + {
> + gimple *stmt = gsi_stmt (gsi);
> +
> + if (is_gimple_assign (stmt))
> + match_saturation_arith (&gsi, stmt, m_cfg_changed_p);
> + }
> +
> for (gsi = gsi_after_labels (bb); !gsi_end_p (gsi);)
> {
> gimple *stmt = gsi_stmt (gsi);
> --
> 2.34.1
>
next prev parent reply other threads:[~2024-05-14 14:15 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-06 12:07 [PATCH v1] Internal-fn: Introduce new internal function SAT_ADD pan2.li
2024-04-07 7:03 ` [PATCH v2] " pan2.li
2024-04-28 12:10 ` Li, Pan2
2024-04-29 7:53 ` [PATCH v3] " pan2.li
2024-05-01 17:06 ` Tamar Christina
2024-05-02 3:10 ` Li, Pan2
2024-05-02 3:25 ` Tamar Christina
2024-05-02 10:57 ` Li, Pan2
2024-05-02 12:57 ` Tamar Christina
2024-05-03 1:45 ` Li, Pan2
2024-05-06 14:48 ` [PATCH v4 1/3] Internal-fn: Support new IFN SAT_ADD for unsigned scalar int pan2.li
2024-05-13 9:09 ` Tamar Christina
2024-05-13 13:36 ` Li, Pan2
2024-05-13 15:03 ` Tamar Christina
2024-05-14 1:50 ` Li, Pan2
2024-05-14 13:18 ` Richard Biener
2024-05-14 14:14 ` Li, Pan2 [this message]
2024-05-06 14:49 ` [PATCH v4 2/3] VECT: Support new IFN SAT_ADD for unsigned vector int pan2.li
2024-05-13 9:10 ` Tamar Christina
2024-05-14 13:21 ` Richard Biener
2024-05-06 14:50 ` [PATCH v4 3/3] RISC-V: Implement IFN SAT_ADD for both the scalar and vector pan2.li
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=MW5PR11MB59086674B904F7079865D65DA9E32@MW5PR11MB5908.namprd11.prod.outlook.com \
--to=pan2.li@intel.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hongtao.liu@intel.com \
--cc=juzhe.zhong@rivai.ai \
--cc=kito.cheng@gmail.com \
--cc=richard.guenther@gmail.com \
--cc=tamar.christina@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).