From: Tamar Christina <Tamar.Christina@arm.com>
To: Richard Biener <rguenther@suse.de>
Cc: Richard Sandiford <Richard.Sandiford@arm.com>, nd <nd@arm.com>,
Tamar Christina via Gcc-patches <gcc-patches@gcc.gnu.org>,
"juzhe.zhong@rivai.ai" <juzhe.zhong@rivai.ai>
Subject: RE: [PATCH]middle-end Add optimized float addsub without needing VEC_PERM_EXPR.
Date: Mon, 31 Oct 2022 11:38:19 +0000 [thread overview]
Message-ID: <VI1PR08MB53252ECDBD8068F7D2E511A1FF379@VI1PR08MB5325.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <nycvar.YFH.7.77.849.2209261103180.6652@jbgna.fhfr.qr>
[-- Attachment #1: Type: text/plain, Size: 7419 bytes --]
Hi All,
This is a respin with all feedback addressed.
Bootstrapped Regtested on aarch64-none-linux-gnu and no issues.
Ok for master?
Thanks,
Tamar
gcc/ChangeLog:
* match.pd: Add fneg/fadd rule.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/simd/addsub_1.c: New test.
* gcc.target/aarch64/sve/addsub_1.c: New test.
--- inline copy of patch ---
diff --git a/gcc/generic-match-head.cc b/gcc/generic-match-head.cc
index cb0fbd32fa6fbea7b3c96462bf54abe891396fd6..aed4dcc8c3d718dcd4cb2da05cf9709a65cdde70 100644
--- a/gcc/generic-match-head.cc
+++ b/gcc/generic-match-head.cc
@@ -39,6 +39,7 @@ along with GCC; see the file COPYING3. If not see
#include "dbgcnt.h"
#include "tm.h"
#include "tree-eh.h"
+#include "langhooks.h"
/* Routine to determine if the types T1 and T2 are effectively
the same for GENERIC. If T1 or T2 is not a type, the test
diff --git a/gcc/gimple-match-head.cc b/gcc/gimple-match-head.cc
index 4c80d77f8ba23b9ce8b67913c2238ff65b291906..9986e3479f903d26446c795113a14c9cc3d4359e 100644
--- a/gcc/gimple-match-head.cc
+++ b/gcc/gimple-match-head.cc
@@ -46,6 +46,7 @@ along with GCC; see the file COPYING3. If not see
#include "dbgcnt.h"
#include "tm.h"
#include "gimple-range.h"
+#include "langhooks.h"
/* Forward declarations of the private auto-generated matchers.
They expect valueized operands in canonical order and do not
diff --git a/gcc/match.pd b/gcc/match.pd
index 1bb936fc4010f98f24bb97671350e8432c55b347..5e747352b413e6595ccbd20d781614d55abc372a 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -7916,6 +7916,65 @@ and,
(simplify (reduc (op @0 VECTOR_CST@1))
(op (reduc:type @0) (reduc:type @1))))
+/* Simplify vector floating point operations of alternating sub/add pairs
+ into using an fneg of a wider element type followed by a normal add.
+ under IEEE 754 the fneg of the wider type will negate every even entry
+ and when doing an add we get a sub of the even and add of every odd
+ elements. */
+(simplify
+ (vec_perm (plus:c @0 @1) (minus @0 @1) VECTOR_CST@2)
+ (if (!VECTOR_INTEGER_TYPE_P (type)
+ && !FLOAT_WORDS_BIG_ENDIAN)
+ (with
+ {
+ /* Build a vector of integers from the tree mask. */
+ vec_perm_builder builder;
+ if (!tree_to_vec_perm_builder (&builder, @2))
+ return NULL_TREE;
+
+ /* Create a vec_perm_indices for the integer vector. */
+ poly_uint64 nelts = TYPE_VECTOR_SUBPARTS (type);
+ vec_perm_indices sel (builder, 2, nelts);
+ }
+ (if (sel.series_p (0, 2, 0, 2))
+ (with
+ {
+ machine_mode vec_mode = TYPE_MODE (type);
+ machine_mode wide_mode;
+ if (!GET_MODE_WIDER_MODE (vec_mode).exists (&wide_mode)
+ || !VECTOR_MODE_P (wide_mode)
+ || (GET_MODE_UNIT_BITSIZE (vec_mode) * 2
+ != GET_MODE_UNIT_BITSIZE (wide_mode)))
+ return NULL_TREE;
+
+ tree stype = lang_hooks.types.type_for_mode (GET_MODE_INNER (wide_mode),
+ TYPE_UNSIGNED (type));
+ if (TYPE_MODE (stype) == BLKmode)
+ return NULL_TREE;
+ tree ntype = build_vector_type_for_mode (stype, wide_mode);
+ if (!VECTOR_TYPE_P (ntype))
+ return NULL_TREE;
+
+ /* The format has to be a non-extended ieee format. */
+ const struct real_format *fmt_old = FLOAT_MODE_FORMAT (vec_mode);
+ const struct real_format *fmt_new = FLOAT_MODE_FORMAT (wide_mode);
+ if (fmt_old == NULL || fmt_new == NULL)
+ return NULL_TREE;
+
+ /* If the target doesn't support v1xx vectors, try using scalar mode xx
+ instead. */
+ if (known_eq (GET_MODE_NUNITS (wide_mode), 1)
+ && !target_supports_op_p (ntype, NEGATE_EXPR, optab_vector))
+ ntype = stype;
+ }
+ (if (fmt_new->signbit_rw
+ == fmt_old->signbit_rw + GET_MODE_UNIT_BITSIZE (vec_mode)
+ && fmt_new->signbit_rw == fmt_new->signbit_ro
+ && targetm.can_change_mode_class (TYPE_MODE (ntype), TYPE_MODE (type), ALL_REGS)
+ && ((optimize_vectors_before_lowering_p () && VECTOR_TYPE_P (ntype))
+ || target_supports_op_p (ntype, NEGATE_EXPR, optab_vector)))
+ (plus (view_convert:type (negate (view_convert:ntype @1))) @0)))))))
+
(simplify
(vec_perm @0 @1 VECTOR_CST@2)
(with
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c b/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..1fb91a34c421bbd2894faa0dbbf1b47ad43310c4
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c
@@ -0,0 +1,56 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_v8_2a_fp16_neon_ok } */
+/* { dg-options "-Ofast" } */
+/* { dg-add-options arm_v8_2a_fp16_neon } */
+/* { dg-final { check-function-bodies "**" "" "" { target { le } } } } */
+
+#pragma GCC target "+nosve"
+
+/*
+** f1:
+** ...
+** fneg v[0-9]+.2d, v[0-9]+.2d
+** fadd v[0-9]+.4s, v[0-9]+.4s, v[0-9]+.4s
+** ...
+*/
+void f1 (float *restrict a, float *restrict b, float *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** d1:
+** ...
+** fneg v[0-9]+.4s, v[0-9]+.4s
+** fadd v[0-9]+.8h, v[0-9]+.8h, v[0-9]+.8h
+** ...
+*/
+void d1 (_Float16 *restrict a, _Float16 *restrict b, _Float16 *res, int n)
+{
+ for (int i = 0; i < (n & -8); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** e1:
+** ...
+** fadd v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d
+** fsub v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d
+** ins v[0-9]+.d\[1\], v[0-9]+.d\[1\]
+** ...
+*/
+void e1 (double *restrict a, double *restrict b, double *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
diff --git a/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c b/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..ea7f9d9db2c8c9a3efe5c7951a314a29b7a7a922
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c
@@ -0,0 +1,52 @@
+/* { dg-do compile } */
+/* { dg-options "-Ofast" } */
+/* { dg-final { check-function-bodies "**" "" "" { target { le } } } } */
+
+/*
+** f1:
+** ...
+** fneg z[0-9]+.d, p[0-9]+/m, z[0-9]+.d
+** fadd z[0-9]+.s, z[0-9]+.s, z[0-9]+.s
+** ...
+*/
+void f1 (float *restrict a, float *restrict b, float *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** d1:
+** ...
+** fneg z[0-9]+.s, p[0-9]+/m, z[0-9]+.s
+** fadd z[0-9]+.h, z[0-9]+.h, z[0-9]+.h
+** ...
+*/
+void d1 (_Float16 *restrict a, _Float16 *restrict b, _Float16 *res, int n)
+{
+ for (int i = 0; i < (n & -8); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** e1:
+** ...
+** fsub z[0-9]+.d, z[0-9]+.d, z[0-9]+.d
+** movprfx z[0-9]+.d, p[0-9]+/m, z[0-9]+.d
+** fadd z[0-9]+.d, p[0-9]+/m, z[0-9]+.d, z[0-9]+.d
+** ...
+*/
+void e1 (double *restrict a, double *restrict b, double *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
[-- Attachment #2: rb15819.patch --]
[-- Type: application/octet-stream, Size: 6827 bytes --]
diff --git a/gcc/generic-match-head.cc b/gcc/generic-match-head.cc
index cb0fbd32fa6fbea7b3c96462bf54abe891396fd6..aed4dcc8c3d718dcd4cb2da05cf9709a65cdde70 100644
--- a/gcc/generic-match-head.cc
+++ b/gcc/generic-match-head.cc
@@ -39,6 +39,7 @@ along with GCC; see the file COPYING3. If not see
#include "dbgcnt.h"
#include "tm.h"
#include "tree-eh.h"
+#include "langhooks.h"
/* Routine to determine if the types T1 and T2 are effectively
the same for GENERIC. If T1 or T2 is not a type, the test
diff --git a/gcc/gimple-match-head.cc b/gcc/gimple-match-head.cc
index 4c80d77f8ba23b9ce8b67913c2238ff65b291906..9986e3479f903d26446c795113a14c9cc3d4359e 100644
--- a/gcc/gimple-match-head.cc
+++ b/gcc/gimple-match-head.cc
@@ -46,6 +46,7 @@ along with GCC; see the file COPYING3. If not see
#include "dbgcnt.h"
#include "tm.h"
#include "gimple-range.h"
+#include "langhooks.h"
/* Forward declarations of the private auto-generated matchers.
They expect valueized operands in canonical order and do not
diff --git a/gcc/match.pd b/gcc/match.pd
index 1bb936fc4010f98f24bb97671350e8432c55b347..5e747352b413e6595ccbd20d781614d55abc372a 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -7916,6 +7916,65 @@ and,
(simplify (reduc (op @0 VECTOR_CST@1))
(op (reduc:type @0) (reduc:type @1))))
+/* Simplify vector floating point operations of alternating sub/add pairs
+ into using an fneg of a wider element type followed by a normal add.
+ under IEEE 754 the fneg of the wider type will negate every even entry
+ and when doing an add we get a sub of the even and add of every odd
+ elements. */
+(simplify
+ (vec_perm (plus:c @0 @1) (minus @0 @1) VECTOR_CST@2)
+ (if (!VECTOR_INTEGER_TYPE_P (type)
+ && !FLOAT_WORDS_BIG_ENDIAN)
+ (with
+ {
+ /* Build a vector of integers from the tree mask. */
+ vec_perm_builder builder;
+ if (!tree_to_vec_perm_builder (&builder, @2))
+ return NULL_TREE;
+
+ /* Create a vec_perm_indices for the integer vector. */
+ poly_uint64 nelts = TYPE_VECTOR_SUBPARTS (type);
+ vec_perm_indices sel (builder, 2, nelts);
+ }
+ (if (sel.series_p (0, 2, 0, 2))
+ (with
+ {
+ machine_mode vec_mode = TYPE_MODE (type);
+ machine_mode wide_mode;
+ if (!GET_MODE_WIDER_MODE (vec_mode).exists (&wide_mode)
+ || !VECTOR_MODE_P (wide_mode)
+ || (GET_MODE_UNIT_BITSIZE (vec_mode) * 2
+ != GET_MODE_UNIT_BITSIZE (wide_mode)))
+ return NULL_TREE;
+
+ tree stype = lang_hooks.types.type_for_mode (GET_MODE_INNER (wide_mode),
+ TYPE_UNSIGNED (type));
+ if (TYPE_MODE (stype) == BLKmode)
+ return NULL_TREE;
+ tree ntype = build_vector_type_for_mode (stype, wide_mode);
+ if (!VECTOR_TYPE_P (ntype))
+ return NULL_TREE;
+
+ /* The format has to be a non-extended ieee format. */
+ const struct real_format *fmt_old = FLOAT_MODE_FORMAT (vec_mode);
+ const struct real_format *fmt_new = FLOAT_MODE_FORMAT (wide_mode);
+ if (fmt_old == NULL || fmt_new == NULL)
+ return NULL_TREE;
+
+ /* If the target doesn't support v1xx vectors, try using scalar mode xx
+ instead. */
+ if (known_eq (GET_MODE_NUNITS (wide_mode), 1)
+ && !target_supports_op_p (ntype, NEGATE_EXPR, optab_vector))
+ ntype = stype;
+ }
+ (if (fmt_new->signbit_rw
+ == fmt_old->signbit_rw + GET_MODE_UNIT_BITSIZE (vec_mode)
+ && fmt_new->signbit_rw == fmt_new->signbit_ro
+ && targetm.can_change_mode_class (TYPE_MODE (ntype), TYPE_MODE (type), ALL_REGS)
+ && ((optimize_vectors_before_lowering_p () && VECTOR_TYPE_P (ntype))
+ || target_supports_op_p (ntype, NEGATE_EXPR, optab_vector)))
+ (plus (view_convert:type (negate (view_convert:ntype @1))) @0)))))))
+
(simplify
(vec_perm @0 @1 VECTOR_CST@2)
(with
diff --git a/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c b/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..1fb91a34c421bbd2894faa0dbbf1b47ad43310c4
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/simd/addsub_1.c
@@ -0,0 +1,56 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_v8_2a_fp16_neon_ok } */
+/* { dg-options "-Ofast" } */
+/* { dg-add-options arm_v8_2a_fp16_neon } */
+/* { dg-final { check-function-bodies "**" "" "" { target { le } } } } */
+
+#pragma GCC target "+nosve"
+
+/*
+** f1:
+** ...
+** fneg v[0-9]+.2d, v[0-9]+.2d
+** fadd v[0-9]+.4s, v[0-9]+.4s, v[0-9]+.4s
+** ...
+*/
+void f1 (float *restrict a, float *restrict b, float *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** d1:
+** ...
+** fneg v[0-9]+.4s, v[0-9]+.4s
+** fadd v[0-9]+.8h, v[0-9]+.8h, v[0-9]+.8h
+** ...
+*/
+void d1 (_Float16 *restrict a, _Float16 *restrict b, _Float16 *res, int n)
+{
+ for (int i = 0; i < (n & -8); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** e1:
+** ...
+** fadd v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d
+** fsub v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d
+** ins v[0-9]+.d\[1\], v[0-9]+.d\[1\]
+** ...
+*/
+void e1 (double *restrict a, double *restrict b, double *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
diff --git a/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c b/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c
new file mode 100644
index 0000000000000000000000000000000000000000..ea7f9d9db2c8c9a3efe5c7951a314a29b7a7a922
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/sve/addsub_1.c
@@ -0,0 +1,52 @@
+/* { dg-do compile } */
+/* { dg-options "-Ofast" } */
+/* { dg-final { check-function-bodies "**" "" "" { target { le } } } } */
+
+/*
+** f1:
+** ...
+** fneg z[0-9]+.d, p[0-9]+/m, z[0-9]+.d
+** fadd z[0-9]+.s, z[0-9]+.s, z[0-9]+.s
+** ...
+*/
+void f1 (float *restrict a, float *restrict b, float *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** d1:
+** ...
+** fneg z[0-9]+.s, p[0-9]+/m, z[0-9]+.s
+** fadd z[0-9]+.h, z[0-9]+.h, z[0-9]+.h
+** ...
+*/
+void d1 (_Float16 *restrict a, _Float16 *restrict b, _Float16 *res, int n)
+{
+ for (int i = 0; i < (n & -8); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
+
+/*
+** e1:
+** ...
+** fsub z[0-9]+.d, z[0-9]+.d, z[0-9]+.d
+** movprfx z[0-9]+.d, p[0-9]+/m, z[0-9]+.d
+** fadd z[0-9]+.d, p[0-9]+/m, z[0-9]+.d, z[0-9]+.d
+** ...
+*/
+void e1 (double *restrict a, double *restrict b, double *res, int n)
+{
+ for (int i = 0; i < (n & -4); i+=2)
+ {
+ res[i+0] = a[i+0] + b[i+0];
+ res[i+1] = a[i+1] - b[i+1];
+ }
+}
next prev parent reply other threads:[~2022-10-31 11:38 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-16 10:58 Tamar Christina
2022-06-17 20:33 ` Andrew Pinski
2022-06-18 10:49 ` Richard Biener
2022-06-20 10:00 ` Tamar Christina
2022-06-20 11:56 ` Richard Biener
2022-06-20 12:05 ` Tamar Christina
2022-06-20 13:10 ` Richard Sandiford
2022-09-23 9:11 ` Tamar Christina
2022-09-23 12:54 ` Richard Biener
2022-09-23 13:07 ` 钟居哲
2022-09-23 13:13 ` Tamar Christina
2022-09-23 13:54 ` Tamar Christina
2022-09-26 11:10 ` Richard Biener
2022-10-31 11:38 ` Tamar Christina [this message]
2022-10-31 15:49 ` Jeff Law
2022-07-01 23:07 ` Jeff Law
2022-07-01 22:57 ` 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=VI1PR08MB53252ECDBD8068F7D2E511A1FF379@VI1PR08MB5325.eurprd08.prod.outlook.com \
--to=tamar.christina@arm.com \
--cc=Richard.Sandiford@arm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=juzhe.zhong@rivai.ai \
--cc=nd@arm.com \
--cc=rguenther@suse.de \
/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).