From: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
To: Andrew Pinski <pinskia@gmail.com>,
James Greenhalgh <James.Greenhalgh@arm.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>, nd <nd@arm.com>
Subject: Re: [PATCH][AArch64] Remove '*' from movsi/di/ti patterns
Date: Wed, 13 Sep 2017 11:46:00 -0000 [thread overview]
Message-ID: <DB6PR0801MB20535FB82DD9B09571509DF7836E0@DB6PR0801MB2053.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <CA+=Sn1mWtGBV7g=0L=MPkcL-phdtiFxSL7ar4NcMEk8jN9OVTg@mail.gmail.com>
Andrew Pinski wrote:
>
> Note this caused a few testsuite failures:
Confirmed. The diffs show the new sequence is always better. I've
committed this:
Update vmov_n_1.c now we are generating better code for dup:
ldr s0, [x0]
dup v0.2s, v0.s[0]
ret
Instead of:
ldr w0, [x0]
dup v0.2s, w0
ret
ChangeLog:
2017-09-13 Wilco Dijkstra <wdijkstr@arm.com>
* gcc.target/aarch64/vmov_n_1.c: Update dup scan-assembler.
--
diff --git a/gcc/testsuite/gcc.target/aarch64/vmov_n_1.c b/gcc/testsuite/gcc.target/aarch64/vmov_n_1.c
index 485a1a970bbcd30cb45a2f69bbd9f62f8258d3df..d0c284296a5e256be5cf5f859b113cdd62f929ba 100644
--- a/gcc/testsuite/gcc.target/aarch64/vmov_n_1.c
+++ b/gcc/testsuite/gcc.target/aarch64/vmov_n_1.c
@@ -190,8 +190,9 @@ TESTFUNC_NAME (reg_len, data_type, data_len) () \
OBSCURE_FUNC (64, 32, f)
TESTFUNC (64, 32, f)
-/* "dup Vd.2s, Rn" is less preferable then "dup Vd.2s, Vn.s[lane]". */
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 1 } } */
+/* "dup Vd.2s, Rn" is less preferable than "dup Vd.2s, Vn.s[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.2s, w\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (64, 64, f)
TESTFUNC (64, 64, f)
@@ -216,7 +217,9 @@ TESTFUNC (64, 16, s)
OBSCURE_FUNC (64, 32, s)
TESTFUNC (64, 32, s)
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2s, w\[0-9\]+" 2 } } */
+/* "dup Vd.2s, Rn" is less preferable than "dup Vd.2s, Vn.s[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.2s, w\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (64, 64, s)
TESTFUNC (64, 64, s)
@@ -242,13 +245,15 @@ TESTFUNC (64, 64, u)
OBSCURE_FUNC (128, 32, f)
TESTFUNC (128, 32, f)
-/* "dup Vd.4s, Rn" is less preferable then "dup Vd.4s, Vn.s[lane]". */
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.4s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 1 } } */
+/* "dup Vd.4s, Rn" is less preferable than "dup Vd.4s, Vn.s[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.4s, w\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.4s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (128, 64, f)
TESTFUNC (128, 64, f)
-/* "dup Vd.2d, Rn" is less preferable then "dup Vd.2d, Vn.d[lane]". */
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2d, v\[0-9\]+\.d\\\[\[0-9\]+\\\]" 1 } } */
+/* "dup Vd.2d, Rn" is less preferable than "dup Vd.2d, Vn.d[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.2d, x\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2d, v\[0-9\]+\.d\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (128, 8, p)
TESTFUNC (128, 8, p)
@@ -268,11 +273,15 @@ TESTFUNC (128, 16, s)
OBSCURE_FUNC (128, 32, s)
TESTFUNC (128, 32, s)
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.4s, w\[0-9\]+" 2 } } */
+/* "dup Vd.4s, Rn" is less preferable than "dup Vd.4s, Vn.s[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.4s, w\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.4s, v\[0-9\]+\.s\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (128, 64, s)
TESTFUNC (128, 64, s)
-/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2d, x\[0-9\]+" 2 } } */
+/* "dup Vd.2d, Rn" is less preferable than "dup Vd.2d, Vn.d[lane]". */
+/* { dg-final { scan-assembler-not "dup\\tv\[0-9\]+\.2d, x\[0-9\]+" } } */
+/* { dg-final { scan-assembler-times "dup\\tv\[0-9\]+\.2d, v\[0-9\]+\.d\\\[\[0-9\]+\\\]" 3 } } */
OBSCURE_FUNC (128, 8, u)
TESTFUNC (128, 8, u)
prev parent reply other threads:[~2017-09-13 11:46 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-26 13:46 Wilco Dijkstra
2017-08-01 10:18 ` Wilco Dijkstra
2017-08-15 16:27 ` Wilco Dijkstra
2017-09-09 6:58 ` Andrew Pinski
2017-09-12 16:11 ` James Greenhalgh
2017-09-13 2:10 ` Andrew Pinski
2017-09-13 11:46 ` Wilco Dijkstra [this message]
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=DB6PR0801MB20535FB82DD9B09571509DF7836E0@DB6PR0801MB2053.eurprd08.prod.outlook.com \
--to=wilco.dijkstra@arm.com \
--cc=James.Greenhalgh@arm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=nd@arm.com \
--cc=pinskia@gmail.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).