From: Philipp Tomsich <philipp.tomsich@vrull.eu>
To: gcc-patches@gcc.gnu.org
Cc: Kito Cheng <kito.cheng@gmail.com>,
Christoph Muellner <christoph.muellner@vrull.eu>,
Palmer Dabbelt <palmer@rivosinc.com>,
Andrew Waterman <andrew@sifive.com>,
Vineet Gupta <vineetg@rivosinc.com>,
Philipp Tomsich <philipp.tomsich@vrull.eu>
Subject: [RFC PATCH v1 08/10] ifcvt: add if-conversion to conditional-zero instructions
Date: Fri, 10 Feb 2023 23:41:48 +0100 [thread overview]
Message-ID: <20230210224150.2801962-9-philipp.tomsich@vrull.eu> (raw)
In-Reply-To: <20230210224150.2801962-1-philipp.tomsich@vrull.eu>
Some architectures, as it the case on RISC-V with the proposed
ZiCondOps and the vendor-defined XVentanaCondOps, define a
conditional-zero instruction that is equivalent to:
- the positive form: rd = (rc != 0) ? rs : 0
- the negated form: rd = (rc == 0) ? rs : 0
While noce_try_store_flag_mask will somewhat work for this case, it
will generate a number of atomic RTX that will misdirect the cost
calculation and may be too long (i.e., 4 RTX and more) to successfully
merge at combine-time.
Instead, we add two new transforms that attempt to build up what we
define as the canonical form of a conditional-zero expression:
(set (match_operand 0 "register_operand" "=r")
(and (neg (eq_or_ne (match_operand 1 "register_operand" "r")
(const_int 0)))
(match_operand 2 "register_operand" "r")))
Architectures that provide a conditional-zero are thus expected to
define an instruction matching this pattern in their backend.
Based on this, we support the following cases:
- noce_try_condzero:
a ? a : b
a ? b : 0 (and then/else swapped)
!a ? b : 0 (and then/else swapped)
- noce_try_condzero_arith:
conditional-plus, conditional-minus, conditional-and,
conditional-or, conditional-xor, conditional-shift,
conditional-and
Given that this is hooked into the CE passes, it is less powerful than
a tree-pass (e.g., it can not transform cases where an extension, such
as for uint16_t operations is in either the then or else-branch
together with the arithmetic) but already covers a good array of cases
and triggers across SPEC CPU 2017.
Adding transformations in a tree pass should come in a future
improvement.
gcc/ChangeLog:
* ifcvt.cc (noce_emit_insn): Add prototype.
(noce_emit_condzero): Helper for noce_try_condzero and
noce_try_condzero_arith transforms.
(noce_try_condzero): New transform.
(noce_try_condzero_arith): New transform for conditional
arithmetic that can be built up by exploiting that the
conditional-zero instruction will inject 0, which acts
as the neutral element for operations.
(noce_process_if_block): Call noce_try_condzero and
noce_try_condzero_arith.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/xventanacondops-and-01.c: New test.
* gcc.target/riscv/xventanacondops-and-02.c: New test.
* gcc.target/riscv/xventanacondops-eq-01.c: New test.
* gcc.target/riscv/xventanacondops-eq-02.c: New test.
* gcc.target/riscv/xventanacondops-lt-01.c: New test.
* gcc.target/riscv/xventanacondops-ne-01.c: New test.
* gcc.target/riscv/xventanacondops-xor-01.c: New test.
Signed-off-by: Philipp Tomsich <philipp.tomsich@vrull.eu>
---
gcc/ifcvt.cc | 216 ++++++++++++++++++
.../gcc.target/riscv/zicond-and-01.c | 16 ++
.../gcc.target/riscv/zicond-and-02.c | 15 ++
gcc/testsuite/gcc.target/riscv/zicond-eq-01.c | 11 +
gcc/testsuite/gcc.target/riscv/zicond-eq-02.c | 14 ++
gcc/testsuite/gcc.target/riscv/zicond-lt-01.c | 16 ++
gcc/testsuite/gcc.target/riscv/zicond-ne-01.c | 10 +
.../gcc.target/riscv/zicond-xor-01.c | 14 ++
8 files changed, 312 insertions(+)
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-and-01.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-and-02.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-eq-01.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-eq-02.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-lt-01.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-ne-01.c
create mode 100644 gcc/testsuite/gcc.target/riscv/zicond-xor-01.c
diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
index 008796838f7..7ac3bd8f18e 100644
--- a/gcc/ifcvt.cc
+++ b/gcc/ifcvt.cc
@@ -97,6 +97,7 @@ static int find_if_case_2 (basic_block, edge, edge);
static int dead_or_predicable (basic_block, basic_block, basic_block,
edge, int);
static void noce_emit_move_insn (rtx, rtx);
+static rtx_insn *noce_emit_insn (rtx);
static rtx_insn *block_has_only_trap (basic_block);
static void need_cmov_or_rewire (basic_block, hash_set<rtx_insn *> *,
hash_map<rtx_insn *, int> *);
@@ -787,6 +788,9 @@ static rtx noce_get_alt_condition (struct noce_if_info *, rtx, rtx_insn **);
static int noce_try_minmax (struct noce_if_info *);
static int noce_try_abs (struct noce_if_info *);
static int noce_try_sign_mask (struct noce_if_info *);
+static rtx noce_emit_condzero (struct noce_if_info *, rtx, bool = false);
+static int noce_try_condzero (struct noce_if_info *);
+static int noce_try_condzero_arith (struct noce_if_info *);
/* Return the comparison code for reversed condition for IF_INFO,
or UNKNOWN if reversing the condition is not possible. */
@@ -1664,6 +1668,214 @@ noce_try_addcc (struct noce_if_info *if_info)
return FALSE;
}
+/* Helper to noce_try_condzero: cond ? a : 0. */
+static rtx
+noce_emit_condzero (struct noce_if_info *if_info, rtx a, bool reverse)
+{
+ /* The canonical form for a conditional-zero-or-value is:
+ (set (match_operand 0 "register_operand" "=r")
+ (and (neg (eq_or_ne (match_operand 1 "register_operand" "r")
+ (const_int 0)))
+ (match_operand 2 "register_operand" "r")))
+ */
+
+ machine_mode opmode = GET_MODE (if_info->x);
+ enum rtx_code code = GET_CODE (if_info->cond);
+ rtx cond;
+ rtx op_a = XEXP (if_info->cond, 0);
+ rtx op_b = XEXP (if_info->cond, 1);
+
+ /* If it is not a EQ/NE comparison against const0_rtx, canonicalize
+ by first synthesizing a truth-value and then building a NE
+ condition around it. */
+ if ((code != EQ && code != NE) || XEXP (if_info->cond, 1) != const0_rtx)
+ {
+ rtx tmp = gen_reg_rtx (opmode);
+
+ start_sequence ();
+ cond = gen_rtx_fmt_ee (code, opmode, op_a, op_b);
+ if (!noce_emit_insn (gen_rtx_SET (tmp, cond)))
+ {
+ end_sequence ();
+
+ /* If we can't emit this pattern, try to reverse it and
+ invert the polarity of the second test. */
+ start_sequence ();
+ cond = gen_rtx_fmt_ee (reverse_condition (code), opmode, op_a, op_b);
+ if (!noce_emit_insn (gen_rtx_SET (tmp, cond))) {
+ end_sequence ();
+ return NULL_RTX;
+ }
+
+ /* We have recovered by reversing the first comparison,
+ so we need change the second one around as well... */
+ reverse = !reverse;
+ }
+ rtx_insn *seq = get_insns ();
+ end_sequence ();
+ emit_insn (seq);
+
+ /* Set up the second comparison that will be embedded in the
+ canonical conditional-zero-or-value RTX. */
+ code = NE;
+ op_a = tmp;
+ op_b = const0_rtx;
+ }
+
+ cond = gen_rtx_fmt_ee (reverse ? reverse_condition (code) : code,
+ opmode, op_a, op_b);
+
+ /* Build (and (neg (eq_or_ne ... const0_rtx)) (reg <a>)) */
+ rtx target = gen_reg_rtx (opmode);
+ rtx czero = gen_rtx_AND (opmode, gen_rtx_NEG (opmode, cond), a);
+ noce_emit_move_insn (target, czero);
+
+ return target;
+}
+
+/* Use a conditional-zero instruction for "if (test) x = 0;", if available. */
+static int
+noce_try_condzero (struct noce_if_info *if_info)
+{
+ rtx target;
+ rtx_insn *seq;
+ int reversep = 0;
+ /* Keep local copies of the constituent elements of if_info, as we
+ may be changing them. We are not allowed to modify if_info
+ though, as we may fail in this function and can't leave different
+ semantics behind for the next functions. */
+ rtx a = if_info->a;
+ rtx b = if_info->b;
+ rtx x = if_info->x;
+ rtx cond = if_info->cond;
+ enum rtx_code code = GET_CODE (cond);
+ rtx cond_arg0 = XEXP (cond, 0);
+ rtx cond_arg1 = XEXP (cond, 1);
+ rtx orig_b = NULL_RTX;
+
+ if (!noce_simple_bbs (if_info))
+ return FALSE;
+
+ /* We may encounter the form "(b != 0) ? b : a", which can be
+ simplified to "b | ((b != 0) ? 0 : a)". */
+ if (code == NE && cond_arg1 == const0_rtx &&
+ REG_P (b) && rtx_equal_p (b, cond_arg0))
+ {
+ orig_b = b;
+ b = const0_rtx;
+ }
+
+ /* We may encounter the form "(b == 0) ? b : a", which can be
+ simplied to "(b == 0) ? 0 : a". */
+ if (code == EQ && cond_arg1 == const0_rtx &&
+ REG_P (b) && rtx_equal_p (b, cond_arg0))
+ {
+ b = const0_rtx;
+ }
+
+ start_sequence ();
+
+ if ((a == const0_rtx && (REG_P (b) || rtx_equal_p (b, x)))
+ || ((reversep = (noce_reversed_cond_code (if_info) != UNKNOWN))
+ && b == const0_rtx && (REG_P (a) || rtx_equal_p (a, x))))
+ {
+ target = noce_emit_condzero(if_info, reversep ? a : b, reversep);
+
+ /* Handle the case where we replace b in "(b != 0) ? b : a" with
+ with const0_rtx to then emit "b | ((b != 0) ? 0 : a)". */
+ if (orig_b && target)
+ target = expand_simple_binop (GET_MODE (x), IOR, orig_b,
+ target, x, 0, OPTAB_WIDEN);
+
+ if (target)
+ {
+ if (target != if_info->x)
+ noce_emit_move_insn (if_info->x, target);
+
+ seq = end_ifcvt_sequence (if_info);
+ if (!seq || !targetm.noce_conversion_profitable_p (seq, if_info))
+ return FALSE;
+
+ emit_insn_before_setloc (seq, if_info->jump,
+ INSN_LOCATION (if_info->insn_a));
+ if_info->transform_name = "noce_try_condzero";
+
+ return TRUE;
+ }
+ }
+
+ end_sequence ();
+
+ return FALSE;
+}
+
+/* Convert "if (test) x op= a;" to a branchless sequence using the
+ canonical form for a conditional-zero. */
+static int
+noce_try_condzero_arith (struct noce_if_info *if_info)
+{
+ rtx target;
+ rtx_insn *seq;
+ rtx_code op = GET_CODE (if_info->a);
+ const rtx arg0 = XEXP (if_info->a, 0);
+ const rtx arg1 = XEXP (if_info->a, 1);
+
+ if (!noce_simple_bbs (if_info))
+ return FALSE;
+
+ /* Check for no else condition. */
+ if (!rtx_equal_p (if_info->x, if_info->b))
+ return FALSE;
+
+ if (op != PLUS && op != MINUS && op != IOR && op != XOR &&
+ op != ASHIFT && op != ASHIFTRT && op != LSHIFTRT && op != AND)
+ return FALSE;
+
+ if (!rtx_equal_p (if_info->x, arg0))
+ return FALSE;
+
+ start_sequence ();
+
+ target = noce_emit_condzero(if_info, arg1, op != AND ? true : false);
+
+ if (target)
+ {
+ rtx op1 = if_info->x;
+
+ if (op == AND)
+ {
+ /* Emit "tmp = x & val;" followed by "tmp |= !cond ? x : 0;" */
+ op1 = expand_simple_binop (GET_MODE (if_info->x), AND, op1,
+ arg1, NULL_RTX, 0, OPTAB_WIDEN);
+ op = IOR;
+ }
+
+ if (op1)
+ target = expand_simple_binop (GET_MODE (if_info->x), op, op1,
+ target, if_info->x, 0, OPTAB_WIDEN);
+ }
+
+ if (target)
+ {
+ if (target != if_info->x)
+ noce_emit_move_insn (if_info->x, target);
+
+ seq = end_ifcvt_sequence (if_info);
+ if (!seq || !targetm.noce_conversion_profitable_p (seq, if_info))
+ return FALSE;
+
+ emit_insn_before_setloc(seq, if_info->jump,
+ INSN_LOCATION(if_info->insn_a));
+ if_info->transform_name = "noce_try_condzero_arith";
+
+ return TRUE;
+ }
+
+ end_sequence ();
+
+ return FALSE;
+}
+
/* Convert "if (test) x = 0;" to "x &= -(test == 0);" */
static int
@@ -3967,8 +4179,12 @@ noce_process_if_block (struct noce_if_info *if_info)
{
if (noce_try_addcc (if_info))
goto success;
+ if (noce_try_condzero (if_info))
+ goto success;
if (noce_try_store_flag_mask (if_info))
goto success;
+ if (noce_try_condzero_arith (if_info))
+ goto success;
if (HAVE_conditional_move
&& noce_try_cmove_arith (if_info))
goto success;
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-and-01.c b/gcc/testsuite/gcc.target/riscv/zicond-and-01.c
new file mode 100644
index 00000000000..d9b0ff00756
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-and-01.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64 -mbranch-cost=4" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" "-Os" "-Oz" } } */
+
+long and1(long a, long b, long c, long d)
+{
+ if (c < d)
+ a &= b;
+
+ return a;
+}
+
+/* { dg-final { scan-assembler-times "and\t" 1 } } */
+/* { dg-final { scan-assembler-times "slt" 1 } } */
+/* { dg-final { scan-assembler-times "czero.nez" 1 } } */
+/* { dg-final { scan-assembler-times "or\t" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-and-02.c b/gcc/testsuite/gcc.target/riscv/zicond-and-02.c
new file mode 100644
index 00000000000..80f417cfb54
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-and-02.c
@@ -0,0 +1,15 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64 -mbranch-cost=4" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" "-Os" "-Oz" } } */
+
+int and2(int a, int b, long c)
+{
+ if (c)
+ a &= b;
+
+ return a;
+}
+
+/* { dg-final { scan-assembler-times "and\t" 1 } } */
+/* { dg-final { scan-assembler-times "czero.nez" 1 } } */
+/* { dg-final { scan-assembler-times "or\t" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-eq-01.c b/gcc/testsuite/gcc.target/riscv/zicond-eq-01.c
new file mode 100644
index 00000000000..4f933c1db60
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-eq-01.c
@@ -0,0 +1,11 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" } } */
+
+long
+eq1 (long a, long b)
+{
+ return (a == 0) ? b : 0;
+}
+
+/* { dg-final { scan-assembler-times "czero.nez" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-eq-02.c b/gcc/testsuite/gcc.target/riscv/zicond-eq-02.c
new file mode 100644
index 00000000000..a7bc747ab1d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-eq-02.c
@@ -0,0 +1,14 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" } } */
+
+long
+eq2 (long a, long b)
+{
+ if (a == 0)
+ return b;
+
+ return 0;
+}
+
+/* { dg-final { scan-assembler-times "czero.nez" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-lt-01.c b/gcc/testsuite/gcc.target/riscv/zicond-lt-01.c
new file mode 100644
index 00000000000..830bfc6449f
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-lt-01.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" "-Os" "-Oz" } } */
+
+long long sink (long long);
+
+long long lt3 (long long a, long long b)
+{
+ if (a < b)
+ b = 0;
+
+ return sink(b);
+}
+
+/* { dg-final { scan-assembler-times "czero.nez\t" 1 } } */
+/* { dg-final { scan-assembler-times "slt\t" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-ne-01.c b/gcc/testsuite/gcc.target/riscv/zicond-ne-01.c
new file mode 100644
index 00000000000..f25e601ae3c
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-ne-01.c
@@ -0,0 +1,10 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" } } */
+
+long long ne1(long long a, long long b)
+{
+ return (a != 0) ? b : 0;
+}
+
+/* { dg-final { scan-assembler-times "czero.eqz" 1 } } */
diff --git a/gcc/testsuite/gcc.target/riscv/zicond-xor-01.c b/gcc/testsuite/gcc.target/riscv/zicond-xor-01.c
new file mode 100644
index 00000000000..c45a3be2680
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/zicond-xor-01.c
@@ -0,0 +1,14 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gc_zicond -mabi=lp64" } */
+/* { dg-skip-if "" { *-*-* } { "-O0" "-Og" "-Os" "-Oz" } } */
+
+long xor1(long crc, long poly)
+{
+ if (crc & 1)
+ crc ^= poly;
+
+ return crc;
+}
+
+/* { dg-final { scan-assembler-times "czero.eqz" 1 } } */
+/* { dg-final { scan-assembler-times "xor\t" 1 } } */
--
2.34.1
next prev parent reply other threads:[~2023-02-10 22:42 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-10 22:41 [RFC PATCH v1 00/10] RISC-V: Support the Zicond (conditional-operations) extension Philipp Tomsich
2023-02-10 22:41 ` [RFC PATCH v1 01/10] docs: Document a canonical RTL for a conditional-zero insns Philipp Tomsich
2023-02-10 23:18 ` Andrew Pinski
2023-02-10 22:41 ` [RFC PATCH v1 02/10] RISC-V: Recognize Zicond (conditional operations) extension Philipp Tomsich
2023-04-20 17:44 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 03/10] RISC-V: Generate czero.eqz/nez on noce_try_store_flag_mask if-conversion Philipp Tomsich
2023-04-20 17:53 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 04/10] RISC-V: Support immediates in Zicond Philipp Tomsich
2023-04-20 18:00 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 05/10] RISC-V: Support noce_try_store_flag_mask as czero.eqz/czero.nez Philipp Tomsich
2023-04-21 19:31 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 06/10] RISC-V: Recognize sign-extract + and cases for czero.eqz/nez Philipp Tomsich
2023-04-21 19:40 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 07/10] RISC-V: Recognize bexti in negated if-conversion Philipp Tomsich
2023-04-21 19:56 ` Jeff Law
2023-02-10 22:41 ` Philipp Tomsich [this message]
2023-02-10 23:07 ` [RFC PATCH v1 08/10] ifcvt: add if-conversion to conditional-zero instructions Andrew Pinski
2023-02-13 17:32 ` Richard Sandiford
2023-02-13 18:43 ` Jeff Law
2023-02-13 18:53 ` Andrew Pinski
2023-02-13 7:31 ` Jeff Law
2023-02-28 16:42 ` Maciej W. Rozycki
2023-03-11 15:50 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 09/10] RISC-V: Recognize xventanacondops extension Philipp Tomsich
2023-04-21 19:57 ` Jeff Law
2023-04-25 9:53 ` Kito Cheng
2023-04-25 10:15 ` Philipp Tomsich
2023-04-25 10:43 ` Kito Cheng
2023-04-26 2:28 ` Jeff Law
2023-02-10 22:41 ` [RFC PATCH v1 10/10] RISC-V: Support XVentanaCondOps extension Philipp Tomsich
2023-04-21 19:58 ` Jeff Law
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=20230210224150.2801962-9-philipp.tomsich@vrull.eu \
--to=philipp.tomsich@vrull.eu \
--cc=andrew@sifive.com \
--cc=christoph.muellner@vrull.eu \
--cc=gcc-patches@gcc.gnu.org \
--cc=kito.cheng@gmail.com \
--cc=palmer@rivosinc.com \
--cc=vineetg@rivosinc.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).