From: Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>
To: Andrea Corallo <Andrea.Corallo@arm.com>,
"gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>
Cc: Richard Earnshaw <Richard.Earnshaw@arm.com>,
Andrea Corallo <Andrea.Corallo@arm.com>
Subject: RE: [PATCH 08/23] arm: improve tests for vcmlaq*
Date: Fri, 20 Jan 2023 17:53:10 +0000 [thread overview]
Message-ID: <PAXPR08MB6926A6AE8102832892025F6193C59@PAXPR08MB6926.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <20230120163948.752531-9-andrea.corallo@arm.com>
> -----Original Message-----
> From: Andrea Corallo <andrea.corallo@arm.com>
> Sent: Friday, January 20, 2023 4:40 PM
> To: gcc-patches@gcc.gnu.org
> Cc: Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>; Richard Earnshaw
> <Richard.Earnshaw@arm.com>; Andrea Corallo <Andrea.Corallo@arm.com>
> Subject: [PATCH 08/23] arm: improve tests for vcmlaq*
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/arm/mve/intrinsics/vcmlaq_f16.c: Improve test.
> * gcc.target/arm/mve/intrinsics/vcmlaq_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_m_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_m_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f32.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f16.c: Likewise.
> * gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f32.c: Likewise.
Ok.
Thanks,
Kyrill
> ---
> .../arm/mve/intrinsics/vcmlaq_f16.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_f32.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_m_f16.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_m_f32.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot180_f16.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot180_f32.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot180_m_f16.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot180_m_f32.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot270_f16.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot270_f32.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot270_m_f16.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot270_m_f32.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot90_f16.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot90_f32.c | 24 +++++++++++--
> .../arm/mve/intrinsics/vcmlaq_rot90_m_f16.c | 34 ++++++++++++++++---
> .../arm/mve/intrinsics/vcmlaq_rot90_m_f32.c | 34 ++++++++++++++++---
> 16 files changed, 416 insertions(+), 48 deletions(-)
>
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f16.c
> index fa7d0c05e8c..bb8a99790a0 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f16.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_f16 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f32.c
> index 166bf421f14..71ec4b8479c 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_f32.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_f32 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f16.c
> index 0929f5a0a89..3db345d0791 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f16.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_m_f16 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f32.c
> index 1f4ba453cbc..dcbd2dccce5 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_m_f32.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_m_f32 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #0(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f16.c
> index fc6ba30b36a..f76ae2383a2 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f16.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot180_f16 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot180 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f32.c
> index dbe3f26b3b9..c97d0d0d852 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_f32.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot180_f32 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot180 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f16.c
> index 84a3bd81ad9..132cdf9954f 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f16.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot180_m_f16 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot180_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f32.c
> index 61f5716f0b7..99e96ebe3a9 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot180_m_f32.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot180_m_f32 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #180(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot180_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f16.c
> index 1b0bef91f37..fae85105feb 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f16.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot270_f16 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot270 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f32.c
> index 83e15025e44..54a9b662772 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_f32.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot270_f32 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot270 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f16.c
> index 6e033b1d7ab..e34f83165c3 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f16.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot270_m_f16 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot270_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f32.c
> index 4928341076d..cdba91b8e8c 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot270_m_f32.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot270_m_f32 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #270(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot270_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f16.c
> index 16744a4582c..f767b2b6e6b 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f16.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot90_f16 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f16 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c)
> {
> return vcmlaq_rot90 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f32.c
> index f1f19a87ba7..6c9b24f271d 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_f32.c
> @@ -1,21 +1,41 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot90_f32 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vcmla.f32 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c)
> {
> return vcmlaq_rot90 (a, b, c);
> }
>
> -/* { dg-final { scan-assembler "vcmla.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f16.c
> index 7133ddc7ad5..9141c9e6f90 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f16.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot90_m_f16 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f16 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float16x8_t
> foo1 (float16x8_t a, float16x8_t b, float16x8_t c, mve_pred16_t p)
> {
> return vcmlaq_rot90_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f16" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> diff --git
> a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f32.c
> index 6022e3be538..f317d411806 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vcmlaq_rot90_m_f32.c
> @@ -1,23 +1,49 @@
> /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
> /* { dg-add-options arm_v8_1m_mve_fp } */
> /* { dg-additional-options "-O2" } */
> +/* { dg-final { check-function-bodies "**" "" } } */
>
> #include "arm_mve.h"
>
> +#ifdef __cplusplus
> +extern "C" {
> +#endif
> +
> +/*
> +**foo:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot90_m_f32 (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
>
> +/*
> +**foo1:
> +** ...
> +** vmsr p0, (?:ip|fp|r[0-9]+)(?: @.*|)
> +** ...
> +** vpst(?: @.*|)
> +** ...
> +** vcmlat.f32 q[0-9]+, q[0-9]+, q[0-9]+, #90(?: @.*|)
> +** ...
> +*/
> float32x4_t
> foo1 (float32x4_t a, float32x4_t b, float32x4_t c, mve_pred16_t p)
> {
> return vcmlaq_rot90_m (a, b, c, p);
> }
>
> -/* { dg-final { scan-assembler "vpst" } } */
> -/* { dg-final { scan-assembler "vcmlat.f32" } } */
> +#ifdef __cplusplus
> +}
> +#endif
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> \ No newline at end of file
> --
> 2.25.1
next prev parent reply other threads:[~2023-01-20 17:53 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-20 16:39 [PATCH 00/23] arm: rework MVE testsuite and rework backend where necessary (3rd chunck) Andrea Corallo
2023-01-20 16:39 ` [PATCH 01/23] arm: improve tests and fix vclsq* Andrea Corallo
2023-01-20 17:44 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 02/23] arm: improve tests and fix vclzq* Andrea Corallo
2023-01-20 17:46 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 03/23] arm: improve tests and fix vnegq* Andrea Corallo
2023-01-20 17:47 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 04/23] arm: improve tests for vmulhq* Andrea Corallo
2023-01-20 17:48 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 05/23] arm: improve tests for vmullbq* Andrea Corallo
2023-01-20 17:51 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 06/23] arm: improve tests for vmulltq* Andrea Corallo
2023-01-20 17:52 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 07/23] arm: improve tests for vcaddq* Andrea Corallo
2023-01-20 17:52 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 08/23] arm: improve tests for vcmlaq* Andrea Corallo
2023-01-20 17:53 ` Kyrylo Tkachov [this message]
2023-01-20 16:39 ` [PATCH 09/23] arm: improve tests for vcmulq* Andrea Corallo
2023-01-20 17:59 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 10/23] arm: improve tests and fix vqabsq* Andrea Corallo
2023-01-20 18:01 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 11/23] arm: improve tests for vqdmladhq* Andrea Corallo
2023-01-20 18:03 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 12/23] arm: improve tests for vqdmladhxq* Andrea Corallo
2023-01-20 18:04 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 13/23] arm: improve tests for vqrdmladhq* Andrea Corallo
2023-01-20 18:04 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 14/23] arm: improve tests for vqrdmladhxq* Andrea Corallo
2023-01-20 18:05 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 15/23] arm: improve tests for vqrdmlashq* Andrea Corallo
2023-01-20 18:06 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 16/23] arm: improve tests for vqdmlsdhq* Andrea Corallo
2023-01-20 18:06 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 17/23] arm: improve tests for vqdmlsdhxq* Andrea Corallo
2023-01-20 18:07 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 18/23] arm: improve tests for vqrdmlsdhq* Andrea Corallo
2023-01-20 18:07 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 19/23] arm: improve tests for vqrdmlsdhxq* Andrea Corallo
2023-01-20 18:08 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 20/23] arm: improve tests for vqrdmulhq* Andrea Corallo
2023-01-20 18:08 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 21/23] arm: improve tests and fix vqnegq* Andrea Corallo
2023-01-20 18:08 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 22/23] arm: improve tests for vld2q* Andrea Corallo
2023-01-20 18:09 ` Kyrylo Tkachov
2023-01-20 16:39 ` [PATCH 23/23] arm: fix missing extern "C" in MVE tests Andrea Corallo
2023-01-20 18:10 ` Kyrylo Tkachov
2023-01-25 13:53 ` Andrea Corallo
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=PAXPR08MB6926A6AE8102832892025F6193C59@PAXPR08MB6926.eurprd08.prod.outlook.com \
--to=kyrylo.tkachov@arm.com \
--cc=Andrea.Corallo@arm.com \
--cc=Richard.Earnshaw@arm.com \
--cc=gcc-patches@gcc.gnu.org \
/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).