This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
RE: [PATCH v3][aarch64] Implemented reciprocal square root (rsqrt) estimation in -ffast-math
- From: "Kumar, Venkataramanan" <Venkataramanan dot Kumar at amd dot com>
- To: Benedikt Huber <benedikt dot huber at theobroma-systems dot com>, "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
- Cc: "philipp dot tomsich at theobroma-systems dot com" <philipp dot tomsich at theobroma-systems dot com>, "pinskia at gmail dot com" <pinskia at gmail dot com>, "e dot menezes at samsung dot com" <e dot menezes at samsung dot com>, "James Greenhalgh (james dot greenhalgh at arm dot com)" <james dot greenhalgh at arm dot com>, "Marcus Shawcroft (marcus dot shawcroft at arm dot com)" <marcus dot shawcroft at arm dot com>
- Date: Thu, 30 Jul 2015 05:06:43 +0000
- Subject: RE: [PATCH v3][aarch64] Implemented reciprocal square root (rsqrt) estimation in -ffast-math
- Authentication-results: sourceware.org; auth=none
- Authentication-results: spf=none (sender IP is 165.204.84.222) smtp.mailfrom=amd.com; arm.com; dkim=none (message not signed) header.d=none;
- References: <1438192065-29602-1-git-send-email-benedikt dot huber at theobroma-systems dot com>
Hi Benedikt,
I ran SPEC2006 fp with your previous patch (v2) for cortex-a57. Gromacs gains ~5% for -mcpu=cortex-a57 -Ofast and gains ~11% with -mcpu=cortex-a57 -Ofast -mlow-precision-recip-sqrt.
Other FP benchmarks were within noise.
However I will leave it for Aarch64 maintainers to decide on the default tuning.
Regards,
Venkat.
> -----Original Message-----
> From: Benedikt Huber [mailto:benedikt.huber@theobroma-systems.com]
> Sent: Wednesday, July 29, 2015 11:18 PM
> To: gcc-patches@gcc.gnu.org
> Cc: philipp.tomsich@theobroma-systems.com; Kumar, Venkataramanan;
> pinskia@gmail.com; e.menezes@samsung.com; Benedikt Huber
> Subject: [PATCH v3][aarch64] Implemented reciprocal square root (rsqrt)
> estimation in -ffast-math
>
> This third revision of the patch:
> * makes -mrecip default value specified per core.
> * disables rsqrt when -Os is given.
>
> Ok for check in.
>
> Benedikt Huber (1):
> 2015-07-29 Benedikt Huber <benedikt.huber@theobroma-systems.com>
> Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
>
> gcc/ChangeLog | 19 ++++
> gcc/config/aarch64/aarch64-builtins.c | 103 ++++++++++++++++++++
> gcc/config/aarch64/aarch64-opts.h | 7 ++
> gcc/config/aarch64/aarch64-protos.h | 3 +
> gcc/config/aarch64/aarch64-simd.md | 27 ++++++
> gcc/config/aarch64/aarch64.c | 81 ++++++++++++++--
> gcc/config/aarch64/aarch64.md | 3 +
> gcc/config/aarch64/aarch64.opt | 8 ++
> gcc/doc/invoke.texi | 19 ++++
> gcc/testsuite/gcc.target/aarch64/rsqrt-asm-check.c | 63 ++++++++++++
> gcc/testsuite/gcc.target/aarch64/rsqrt.c | 107
> +++++++++++++++++++++
> 11 files changed, 434 insertions(+), 6 deletions(-) create mode 100644
> gcc/testsuite/gcc.target/aarch64/rsqrt-asm-check.c
> create mode 100644 gcc/testsuite/gcc.target/aarch64/rsqrt.c
>
> --
> 1.9.1