system/freebsd-src
system/freebsd-src
1
0
Fork 0
mirror of https://github.com/freebsd/freebsd-src synced 2024-07-22 18:56:38 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Update the Arm Optimized Routine library to v23.01

Browse source 
Sponsored by:   Arm Ltd
This commit is contained in:
Andrew Turner 2023-06-09 17:56:02 +01:00
parent 1dd169af71 29866ecb89
commit 072a4ba82a
486 changed files with 20136 additions and 920 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

232
contrib/arm-optimized-routines/LICENSE
View file

@ -1,6 +1,11 @@
MIT License
MIT OR Apache-2.0 WITH LLVM-exception
=====================================
Copyright (c) 1999-2019, Arm Limited.
MIT License
-----------
Copyright (c) 1999-2022, Arm Limited.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
@ -19,3 +24,226 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
Apache-2.0 WITH LLVM-exception
------------------------------
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
--- LLVM Exceptions to the Apache 2.0 License ----
As an exception, if, as a result of your compiling your source code, portions
of this Software are embedded into an Object form of such source code, you
may redistribute such embedded portions in such Object form without complying
with the conditions of Sections 4(a), 4(b) and 4(d) of the License.
In addition, if you combine or link compiled forms of this Software with
software that is licensed under the GPLv2 ("Combined Software") and if a
court of competent jurisdiction determines that the patent provision (Section
3), the indemnity provision (Section 9) or other Section of the License
conflicts with the conditions of the GPLv2, you may retroactively and
prospectively choose to deem waived or otherwise exclude such Section(s) of
the License, but only in their entirety and only with respect to the Combined
Software.

12
contrib/arm-optimized-routines/MAINTAINERS Normal file
View file

@ -0,0 +1,12 @@
/
Szabolcs Nagy <szabolcs.nagy@arm.com>
math/
Szabolcs Nagy <szabolcs.nagy@arm.com>
networking/
Szabolcs Nagy <szabolcs.nagy@arm.com>
pl/
Pierre Blanchard <pierre.blanchard@arm.com>
Joe Ramsay <joe.ramsay@arm.com>
string/
Szabolcs Nagy <szabolcs.nagy@arm.com>
Wilco Dijkstra <wilco.dijkstra@arm.com>

7
contrib/arm-optimized-routines/Makefile
View file

@ -1,7 +1,7 @@
# Makefile - requires GNU make
#
# Copyright (c) 2018-2020, Arm Limited.
# SPDX-License-Identifier: MIT
# Copyright (c) 2018-2022, Arm Limited.
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
srcdir = .
prefix = /usr
@ -11,6 +11,7 @@ includedir = $(prefix)/include
# Configure these in config.mk, do not make changes in this file.
SUBS = math string networking
PLSUBS = math
HOST_CC = cc
HOST_CFLAGS = -std=c99 -O2
HOST_LDFLAGS =
@ -20,6 +21,7 @@ CPPFLAGS =
CFLAGS = -std=c99 -O2
CFLAGS_SHARED = -fPIC
CFLAGS_ALL = -Ibuild/include $(CPPFLAGS) $(CFLAGS)
CFLAGS_PL = -Ibuild/pl/include $(CPPFLAGS) $(CFLAGS) -DPL
LDFLAGS =
LDLIBS =
AR = $(CROSS_COMPILE)ar
@ -51,6 +53,7 @@ $(DIRS):
mkdir -p $@
$(filter %.os,$(ALL_FILES)): CFLAGS_ALL += $(CFLAGS_SHARED)
$(filter %.os,$(ALL_FILES)): CFLAGS_PL += $(CFLAGS_SHARED)
build/%.o: $(srcdir)/%.S
$(CC) $(CFLAGS_ALL) -c -o $@ $<

14
contrib/arm-optimized-routines/README
View file

@ -2,14 +2,17 @@ Arm Optimized Routines
----------------------
This repository contains implementations of library functions
provided by Arm under MIT License (See LICENSE). Contributions
to this project are accepted, but Contributors have to sign an
Assignment Agreement, please follow the instructions in
provided by Arm. The outbound license is available under a dual
license, at the user’s election, as reflected in the LICENSE file.
Contributions to this project are accepted, but Contributors have
to sign an Assignment Agreement, please follow the instructions in
contributor-agreement.pdf. This is needed so upstreaming code
to projects that require copyright assignment is possible.
to projects that require copyright assignment is possible. Further
contribution requirements are documented in README.contributors of
the appropriate subdirectory.
Regular quarterly releases are tagged as vYY.MM, the latest
release is v21.02.
release is v23.01.
Source code layout:
@ -24,6 +27,7 @@ networking/test/ - networking test and benchmark related sources.
string/ - string routines subproject sources.
string/include/ - string library public headers.
string/test/ - string test and benchmark related sources.
pl/... - separately maintained performance library code.
The steps to build the target libraries and run the tests:

44
contrib/arm-optimized-routines/README.contributors Normal file
View file

@ -0,0 +1,44 @@
GENERIC CONTRIBUTION GUIDELINES
===============================
1. Sub-projects are maintained independently and thus have independent
contribution rules. If there exists a README.contributors in the
sub-directory to which the contribution is made, it must be followed.
2. Legal:
- Contributors who are not employed by Arm must sign an Assignment Agreement.
See contributor-agreement.pdf.
- All code must be copyright owned by Arm Limited and the appropriate
copyright notice and license identifier must be present in every source
file.
3. Build:
- Build should only depend on GNU make and posix utilities (shell, awk, sed,
etc) and on a C toolchain.
- Build should pass with the default configuration (see config.mk.dist)
and other supported configurations, with both gcc and clang based
toolchains. (The build should not depend on a recent toolchain, the use
of a new feature should be possible to disable.)
- Currently there is no automated configuration, target specific configuration
should be done via make variables in config.mk. This is the user interface
to the build system, so it should be documented in sufficient detail and
kept reasonably stable.
4. Testing:
- On aarch64 the tests must pass. If the code may behave differently under
some supported configurations (e.g. CFLAGS) those should be tested.
- New symbols are expected to have new associated test code and ideally
benchmark code too.
4. Commits:
- Commit message should be descriptive and should not refer to Arm internal
information (such as Jira tickets, or internal discussions). Non-obvious
decisions should be recorded or explained in the commit message if they are
not explained in source comments.
- Ideally tools and scripts used to write the code should be added to the
repository or at least mentioned in the commit.
- Logically independent changes should not be mixed into the same commit.
5. Style:
- Unless otherwise required differently by the sub-project, follow the
clang-format tool using the style from the gcc contrib/ directory.

24
contrib/arm-optimized-routines/config.mk.dist
View file

@ -1,11 +1,14 @@
# Example config.mk
#
# Copyright (c) 2018-2020, Arm Limited.
# SPDX-License-Identifier: MIT
# Copyright (c) 2018-2022, Arm Limited.
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
# Subprojects to build
SUBS = math string networking
# Subsubprojects to build if subproject pl is built
PLSUBS = math
# Target architecture: aarch64, arm or x86_64
ARCH = aarch64
@ -59,6 +62,23 @@ math-cflags += -ffp-contract=fast -fno-math-errno
# Disable vector math code
#math-cflags += -DWANT_VMATH=0
# Disable/enable SVE vector math code and tests
WANT_SVE_MATH = 0
ifeq ($(WANT_SVE_MATH), 1)
math-cflags += -march=armv8.2-a+sve
endif
math-cflags += -DWANT_SVE_MATH=$(WANT_SVE_MATH)
# If defined to 1, set errno in math functions according to ISO C. Many math
# libraries do not set errno, so this is 0 by default. It may need to be
# set to 1 if math.h has (math_errhandling & MATH_ERRNO) != 0.
WANT_ERRNO = 0
math-cflags += -DWANT_ERRNO=$(WANT_ERRNO)
# If set to 1, set fenv in vector math routines.
WANT_SIMD_EXCEPT = 0
math-cflags += -DWANT_SIMD_EXCEPT=$(WANT_SIMD_EXCEPT)
# Disable fenv checks
#math-ulpflags = -q -f
#math-testflags = -nostatus

15
contrib/arm-optimized-routines/math/Dir.mk
View file

@ -1,7 +1,7 @@
# Makefile fragment - requires GNU make
#
# Copyright (c) 2019, Arm Limited.
# SPDX-License-Identifier: MIT
# Copyright (c) 2019-2022, Arm Limited.
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
S := $(srcdir)/math
B := build/math
@ -15,6 +15,7 @@ math-test-srcs := \
math-test-host-srcs := $(wildcard $(S)/test/rtest/*.[cS])
math-includes := $(patsubst $(S)/%,build/%,$(wildcard $(S)/include/*.h))
math-test-includes := $(patsubst $(S)/%,build/include/%,$(wildcard $(S)/test/*.h))
math-libs := \
build/lib/libmathlib.so \
@ -42,10 +43,11 @@ math-files := \
$(math-tools) \
$(math-host-tools) \
$(math-includes) \
$(math-test-includes) \
all-math: $(math-libs) $(math-tools) $(math-includes)
all-math: $(math-libs) $(math-tools) $(math-includes) $(math-test-includes)
$(math-objs): $(math-includes)
$(math-objs): $(math-includes) $(math-test-includes)
$(math-objs): CFLAGS_ALL += $(math-cflags)
$(B)/test/mathtest.o: CFLAGS_ALL += -fmath-errno
$(math-host-objs): CC = $(HOST_CC)
@ -83,6 +85,9 @@ build/bin/ulp: $(B)/test/ulp.o build/lib/libmathlib.a
build/include/%.h: $(S)/include/%.h
cp $< $@
build/include/test/%.h: $(S)/test/%.h
cp $< $@
build/bin/%.sh: $(S)/test/%.sh
cp $< $@
@ -96,7 +101,7 @@ check-math-rtest: $(math-host-tools) $(math-tools)
cat $(math-rtests) | build/bin/rtest | $(EMULATOR) build/bin/mathtest $(math-testflags)
check-math-ulp: $(math-tools)
ULPFLAGS="$(math-ulpflags)" build/bin/runulp.sh $(EMULATOR)
ULPFLAGS="$(math-ulpflags)" WANT_SIMD_EXCEPT="$(WANT_SIMD_EXCEPT)" build/bin/runulp.sh $(EMULATOR)
check-math: check-math-test check-math-rtest check-math-ulp

78
contrib/arm-optimized-routines/math/README.contributors Normal file
View file

@ -0,0 +1,78 @@
STYLE REQUIREMENTS
==================
1. Most code in this sub-directory is expected to be upstreamed into glibc so
the GNU Coding Standard and glibc specific conventions should be followed
to ease upstreaming.
2. ABI and symbols: the code should be written so it is suitable for inclusion
into a libc with minimal changes. This e.g. means that internal symbols
should be hidden and in the implementation reserved namespace according to
ISO C and POSIX rules. If possible the built shared libraries and static
library archives should be usable to override libc symbols at link time (or
at runtime via LD_PRELOAD). This requires the symbols to follow the glibc ABI
(other than symbol versioning), this cannot be done reliably for static
linking so this is a best effort requirement.
3. API: include headers should be suitable for benchmarking and testing code
and should not conflict with libc headers.
CONTRIBUTION GUIDELINES FOR math SUB-DIRECTORY
==============================================
1. Math functions have quality and performance requirements.
2. Quality:
- Worst-case ULP error should be small in the entire input domain (for most
common double precision scalar functions the target is < 0.66 ULP error,
and < 1 ULP for single precision, even performance optimized function
variant should not have > 5 ULP error if the goal is to be a drop in
replacement for a standard math function), this should be tested
statistically (or on all inputs if possible in reasonable amount of time).
The ulp tool is for this and runulp.sh should be updated for new functions.
- All standard rounding modes need to be supported but in non-default rounding
modes the quality requirement can be relaxed. (Non-nearest rounded
computation can be slow and inaccurate but has to be correct for conformance
reasons.)
- Special cases and error handling need to follow ISO C Annex F requirements,
POSIX requirements, IEEE 754-2008 requirements and Glibc requiremnts:
https://www.gnu.org/software/libc/manual/html_mono/libc.html#Errors-in-Math-Functions
this should be tested by direct tests (glibc test system may be used for it).
- Error handling code should be decoupled from the approximation code as much
as possible. (There are helper functions, these take care of errno as well
as exception raising.)
- Vector math code does not need to work in non-nearest rounding mode and error
handling side effects need not happen (fenv exceptions and errno), but the
result should be correct (within quality requirements, which are lower for
vector code than for scalar code).
- Error bounds of the approximation should be clearly documented.
- The code should build and pass tests on arm, aarch64 and x86_64 GNU linux
systems. (Routines and features can be disabled on specific targets, but
the build must complete). On aarch64, both little- and big-endian targets
are supported as well as valid combinations of architecture extensions.
The configurations that should be tested depend on the contribution.
3. Performance:
- Common math code should be benchmarked on modern aarch64 microarchitectures
over typical inputs.
- Performance improvements should be documented (relative numbers can be
published; it is enough to use the mathbench microbenchmark tool which should
be updated for new functions).
- Attention should be paid to the compilation flags: for aarch64 fma
contraction should be on and math errno turned off so some builtins can be
inlined.
- The code should be reasonably performant on x86_64 too, e.g. some rounding
instructions and fma may not be available on x86_64, such builtins turn into
libc calls with slow code. Such slowdown is not acceptable, a faster fallback
should be present: glibc and bionic use the same code on all targets. (This
does not apply to vector math code).

4
contrib/arm-optimized-routines/math/cosf.c
View file

@ -1,8 +1,8 @@
/*
* Single-precision cos function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2018-2021, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdint.h>

2
contrib/arm-optimized-routines/math/erf.c
View file

@ -2,7 +2,7 @@
* Double-precision erf(x) function.
*
* Copyright (c) 2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/erf_data.c
View file

@ -2,7 +2,7 @@
* Shared data between erf and erfc.
*
* Copyright (c) 2019-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/erff.c
View file

@ -2,7 +2,7 @@
* Single-precision erf(x) function.
*
* Copyright (c) 2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdint.h>

2
contrib/arm-optimized-routines/math/erff_data.c
View file

@ -2,7 +2,7 @@
* Data for approximation of erff.
*
* Copyright (c) 2019-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/exp.c
View file

@ -2,7 +2,7 @@
* Double-precision e^x function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <float.h>

2
contrib/arm-optimized-routines/math/exp2.c
View file

@ -2,7 +2,7 @@
* Double-precision 2^x function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <float.h>

2
contrib/arm-optimized-routines/math/exp2f.c
View file

@ -2,7 +2,7 @@
* Single-precision 2^x function.
*
* Copyright (c) 2017-2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>

2
contrib/arm-optimized-routines/math/exp2f_data.c
View file

@ -2,7 +2,7 @@
* Shared data between expf, exp2f and powf.
*
* Copyright (c) 2017-2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/exp_data.c
View file

@ -2,7 +2,7 @@
* Shared data between exp, exp2 and pow.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/expf.c
View file

@ -2,7 +2,7 @@
* Single-precision e^x function.
*
* Copyright (c) 2017-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>

2
contrib/arm-optimized-routines/math/include/mathlib.h
View file

@ -2,7 +2,7 @@
* Public API.
*
* Copyright (c) 2015-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#ifndef _MATHLIB_H

2
contrib/arm-optimized-routines/math/log.c
View file

@ -2,7 +2,7 @@
* Double-precision log(x) function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <float.h>

2
contrib/arm-optimized-routines/math/log2.c
View file

@ -2,7 +2,7 @@
* Double-precision log2(x) function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <float.h>

2
contrib/arm-optimized-routines/math/log2_data.c
View file

@ -2,7 +2,7 @@
* Data for log2.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/log2f.c
View file

@ -2,7 +2,7 @@
* Single-precision log2 function.
*
* Copyright (c) 2017-2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>

2
contrib/arm-optimized-routines/math/log2f_data.c
View file

@ -2,7 +2,7 @@
* Data definition for log2f.
*
* Copyright (c) 2017-2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/log_data.c
View file

@ -2,7 +2,7 @@
* Data for log.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

6
contrib/arm-optimized-routines/math/logf.c
View file

@ -1,8 +1,8 @@
/*
* Single-precision log function.
*
* Copyright (c) 2017-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2017-2023, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>
@ -57,7 +57,7 @@ logf (float x)
tmp = ix - OFF;
i = (tmp >> (23 - LOGF_TABLE_BITS)) % N;
k = (int32_t) tmp >> 23; /* arithmetic shift */
iz = ix - (tmp & 0x1ff << 23);
iz = ix - (tmp & 0xff800000);
invc = T[i].invc;
logc = T[i].logc;
z = (double_t) asfloat (iz);

2
contrib/arm-optimized-routines/math/logf_data.c
View file

@ -2,7 +2,7 @@
* Data definition for logf.
*
* Copyright (c) 2017-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/math_config.h
View file

@ -2,7 +2,7 @@
* Configuration for math routines.
*
* Copyright (c) 2017-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#ifndef _MATH_CONFIG_H

2
contrib/arm-optimized-routines/math/math_err.c
View file

@ -2,7 +2,7 @@
* Double-precision math error handling.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/math_errf.c
View file

@ -2,7 +2,7 @@
* Single-precision math error handling.
*
* Copyright (c) 2017-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/pow.c
View file

@ -2,7 +2,7 @@
* Double-precision x^y function.
*
* Copyright (c) 2018-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <float.h>

2
contrib/arm-optimized-routines/math/pow_log_data.c
View file

@ -2,7 +2,7 @@
* Data for the log part of pow.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/powf.c
View file

@ -2,7 +2,7 @@
* Single-precision pow function.
*
* Copyright (c) 2017-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>

2
contrib/arm-optimized-routines/math/powf_log2_data.c
View file

@ -2,7 +2,7 @@
* Data definition for powf.
*
* Copyright (c) 2017-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "math_config.h"

2
contrib/arm-optimized-routines/math/s_cos.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_cos.c"

2
contrib/arm-optimized-routines/math/s_cosf.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_cosf.c"

2
contrib/arm-optimized-routines/math/s_exp.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_exp.c"

2
contrib/arm-optimized-routines/math/s_exp2f.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_exp2f.c"

2
contrib/arm-optimized-routines/math/s_exp2f_1u.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_exp2f_1u.c"

2
contrib/arm-optimized-routines/math/s_expf.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_expf.c"

2
contrib/arm-optimized-routines/math/s_expf_1u.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_expf_1u.c"

2
contrib/arm-optimized-routines/math/s_log.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_log.c"

2
contrib/arm-optimized-routines/math/s_logf.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_logf.c"

2
contrib/arm-optimized-routines/math/s_pow.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2020, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_pow.c"

2
contrib/arm-optimized-routines/math/s_powf.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_powf.c"

2
contrib/arm-optimized-routines/math/s_sin.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_sin.c"

2
contrib/arm-optimized-routines/math/s_sinf.c
View file

@ -1,6 +1,6 @@
/*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#define SCALAR 1
#include "v_sinf.c"

4
contrib/arm-optimized-routines/math/sincosf.c
View file

@ -1,8 +1,8 @@
/*
* Single-precision sin/cos function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2018-2021, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdint.h>

4
contrib/arm-optimized-routines/math/sincosf.h
View file

@ -1,8 +1,8 @@
/*
* Header for sinf, cosf and sincosf.
*
* Copyright (c) 2018, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2018-2021, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdint.h>

2
contrib/arm-optimized-routines/math/sincosf_data.c
View file

@ -2,7 +2,7 @@
* Data definition for sinf, cosf and sincosf.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdint.h>

4
contrib/arm-optimized-routines/math/sinf.c
View file

@ -1,8 +1,8 @@
/*
* Single-precision sin function.
*
* Copyright (c) 2018-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2018-2021, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <math.h>

285
contrib/arm-optimized-routines/math/test/mathbench.c
View file

@ -1,8 +1,8 @@
/*
* Microbenchmark for math functions.
*
* Copyright (c) 2018-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2018-2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#undef _GNU_SOURCE
@ -66,6 +66,43 @@ v_float_dup (float x)
{
return (v_float){x, x, x, x};
}
#if WANT_SVE_MATH
#include <arm_sve.h>
typedef svbool_t sv_bool;
typedef svfloat64_t sv_double;
#define sv_double_len() svcntd()
static inline sv_double
sv_double_load (const double *p)
{
svbool_t pg = svptrue_b64();
return svld1(pg, p);
}
static inline sv_double
sv_double_dup (double x)
{
return svdup_n_f64(x);
}
typedef svfloat32_t sv_float;
#define sv_float_len() svcntw()
static inline sv_float
sv_float_load (const float *p)
{
svbool_t pg = svptrue_b32();
return svld1(pg, p);
}
static inline sv_float
sv_float_dup (float x)
{
return svdup_n_f32(x);
}
#endif
#else
/* dummy definitions to make things compile. */
typedef double v_double;
@ -89,7 +126,6 @@ dummyf (float x)
{
return x;
}
#if WANT_VMATH
#if __aarch64__
static v_double
@ -116,101 +152,25 @@ __vn_dummyf (v_float x)
{
return x;
}
__vpcs static v_float
xy__vn_powf (v_float x)
#endif
#if WANT_SVE_MATH
static sv_double
__sv_dummy (sv_double x, sv_bool pg)
{
return __vn_powf (x, x);
return x;
}
__vpcs static v_float
xy_Z_powf (v_float x)
static sv_float
__sv_dummyf (sv_float x, sv_bool pg)
{
return _ZGVnN4vv_powf (x, x);
return x;
}
__vpcs static v_double
xy__vn_pow (v_double x)
{
return __vn_pow (x, x);
}
__vpcs static v_double
xy_Z_pow (v_double x)
{
return _ZGVnN2vv_pow (x, x);
}
#endif
#endif
#endif
static v_float
xy__v_powf (v_float x)
{
return __v_powf (x, x);
}
static v_double
xy__v_pow (v_double x)
{
return __v_pow (x, x);
}
#endif
static float
xy__s_powf (float x)
{
return __s_powf (x, x);
}
static double
xy__s_pow (double x)
{
return __s_pow (x, x);
}
#endif
static double
xypow (double x)
{
return pow (x, x);
}
static float
xypowf (float x)
{
return powf (x, x);
}
static double
xpow (double x)
{
return pow (x, 23.4);
}
static float
xpowf (float x)
{
return powf (x, 23.4f);
}
static double
ypow (double x)
{
return pow (2.34, x);
}
static float
ypowf (float x)
{
return powf (2.34f, x);
}
static float
sincosf_wrap (float x)
{
float s, c;
sincosf (x, &s, &c);
return s + c;
}
#include "test/mathbench_wrappers.h"
static const struct fun
{
@ -228,6 +188,10 @@ static const struct fun
#ifdef __vpcs
__vpcs v_double (*vnd) (v_double);
__vpcs v_float (*vnf) (v_float);
#endif
#if WANT_SVE_MATH
sv_double (*svd) (sv_double, sv_bool);
sv_float (*svf) (sv_float, sv_bool);
#endif
} fun;
} funtab[] = {
@ -237,106 +201,25 @@ static const struct fun
#define VF(func, lo, hi) {#func, 'f', 'v', lo, hi, {.vf = func}},
#define VND(func, lo, hi) {#func, 'd', 'n', lo, hi, {.vnd = func}},
#define VNF(func, lo, hi) {#func, 'f', 'n', lo, hi, {.vnf = func}},
#define SVD(func, lo, hi) {#func, 'd', 's', lo, hi, {.svd = func}},
#define SVF(func, lo, hi) {#func, 'f', 's', lo, hi, {.svf = func}},
D (dummy, 1.0, 2.0)
D (exp, -9.9, 9.9)
D (exp, 0.5, 1.0)
D (exp2, -9.9, 9.9)
D (log, 0.01, 11.1)
D (log, 0.999, 1.001)
D (log2, 0.01, 11.1)
D (log2, 0.999, 1.001)
{"pow", 'd', 0, 0.01, 11.1, {.d = xypow}},
D (xpow, 0.01, 11.1)
D (ypow, -9.9, 9.9)
D (erf, -6.0, 6.0)
F (dummyf, 1.0, 2.0)
F (expf, -9.9, 9.9)
F (exp2f, -9.9, 9.9)
F (logf, 0.01, 11.1)
F (log2f, 0.01, 11.1)
{"powf", 'f', 0, 0.01, 11.1, {.f = xypowf}},
F (xpowf, 0.01, 11.1)
F (ypowf, -9.9, 9.9)
{"sincosf", 'f', 0, 0.1, 0.7, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 0.8, 3.1, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, -3.1, 3.1, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 3.3, 33.3, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 100, 1000, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 1e6, 1e32, {.f = sincosf_wrap}},
F (sinf, 0.1, 0.7)
F (sinf, 0.8, 3.1)
F (sinf, -3.1, 3.1)
F (sinf, 3.3, 33.3)
F (sinf, 100, 1000)
F (sinf, 1e6, 1e32)
F (cosf, 0.1, 0.7)
F (cosf, 0.8, 3.1)
F (cosf, -3.1, 3.1)
F (cosf, 3.3, 33.3)
F (cosf, 100, 1000)
F (cosf, 1e6, 1e32)
F (erff, -4.0, 4.0)
#if WANT_VMATH
D (__s_sin, -3.1, 3.1)
D (__s_cos, -3.1, 3.1)
D (__s_exp, -9.9, 9.9)
D (__s_log, 0.01, 11.1)
{"__s_pow", 'd', 0, 0.01, 11.1, {.d = xy__s_pow}},
F (__s_expf, -9.9, 9.9)
F (__s_expf_1u, -9.9, 9.9)
F (__s_exp2f, -9.9, 9.9)
F (__s_exp2f_1u, -9.9, 9.9)
F (__s_logf, 0.01, 11.1)
{"__s_powf", 'f', 0, 0.01, 11.1, {.f = xy__s_powf}},
F (__s_sinf, -3.1, 3.1)
F (__s_cosf, -3.1, 3.1)
#if __aarch64__
VD (__v_dummy, 1.0, 2.0)
VD (__v_sin, -3.1, 3.1)
VD (__v_cos, -3.1, 3.1)
VD (__v_exp, -9.9, 9.9)
VD (__v_log, 0.01, 11.1)
{"__v_pow", 'd', 'v', 0.01, 11.1, {.vd = xy__v_pow}},
VF (__v_dummyf, 1.0, 2.0)
VF (__v_expf, -9.9, 9.9)
VF (__v_expf_1u, -9.9, 9.9)
VF (__v_exp2f, -9.9, 9.9)
VF (__v_exp2f_1u, -9.9, 9.9)
VF (__v_logf, 0.01, 11.1)
{"__v_powf", 'f', 'v', 0.01, 11.1, {.vf = xy__v_powf}},
VF (__v_sinf, -3.1, 3.1)
VF (__v_cosf, -3.1, 3.1)
#ifdef __vpcs
VND (__vn_dummy, 1.0, 2.0)
VND (__vn_exp, -9.9, 9.9)
VND (_ZGVnN2v_exp, -9.9, 9.9)
VND (__vn_log, 0.01, 11.1)
VND (_ZGVnN2v_log, 0.01, 11.1)
{"__vn_pow", 'd', 'n', 0.01, 11.1, {.vnd = xy__vn_pow}},
{"_ZGVnN2vv_pow", 'd', 'n', 0.01, 11.1, {.vnd = xy_Z_pow}},
VND (__vn_sin, -3.1, 3.1)
VND (_ZGVnN2v_sin, -3.1, 3.1)
VND (__vn_cos, -3.1, 3.1)
VND (_ZGVnN2v_cos, -3.1, 3.1)
VNF (__vn_dummyf, 1.0, 2.0)
VNF (__vn_expf, -9.9, 9.9)
VNF (_ZGVnN4v_expf, -9.9, 9.9)
VNF (__vn_expf_1u, -9.9, 9.9)
VNF (__vn_exp2f, -9.9, 9.9)
VNF (_ZGVnN4v_exp2f, -9.9, 9.9)
VNF (__vn_exp2f_1u, -9.9, 9.9)
VNF (__vn_logf, 0.01, 11.1)
VNF (_ZGVnN4v_logf, 0.01, 11.1)
{"__vn_powf", 'f', 'n', 0.01, 11.1, {.vnf = xy__vn_powf}},
{"_ZGVnN4vv_powf", 'f', 'n', 0.01, 11.1, {.vnf = xy_Z_powf}},
VNF (__vn_sinf, -3.1, 3.1)
VNF (_ZGVnN4v_sinf, -3.1, 3.1)
VNF (__vn_cosf, -3.1, 3.1)
VNF (_ZGVnN4v_cosf, -3.1, 3.1)
#endif
#if WANT_SVE_MATH
SVD (__sv_dummy, 1.0, 2.0)
SVF (__sv_dummyf, 1.0, 2.0)
#endif
#endif
#endif
#include "test/mathbench_funcs.h"
{0},
#undef F
#undef D
@ -344,6 +227,8 @@ VNF (_ZGVnN4v_cosf, -3.1, 3.1)
#undef VD
#undef VNF
#undef VND
#undef SVF
#undef SVD
};
static void
@ -508,6 +393,40 @@ runf_vn_latency (__vpcs v_float f (v_float))
}
#endif
#if WANT_SVE_MATH
static void
run_sv_thruput (sv_double f (sv_double, sv_bool))
{
for (int i = 0; i < N; i += sv_double_len ())
f (sv_double_load (A+i), svptrue_b64 ());
}
static void
runf_sv_thruput (sv_float f (sv_float, sv_bool))
{
for (int i = 0; i < N; i += sv_float_len ())
f (sv_float_load (Af+i), svptrue_b32 ());
}
static void
run_sv_latency (sv_double f (sv_double, sv_bool))
{
sv_double z = sv_double_dup (zero);
sv_double prev = z;
for (int i = 0; i < N; i += sv_double_len ())
prev = f (svmad_f64_x (svptrue_b64 (), prev, z, sv_double_load (A+i)), svptrue_b64 ());
}
static void
runf_sv_latency (sv_float f (sv_float, sv_bool))
{
sv_float z = sv_float_dup (zero);
sv_float prev = z;
for (int i = 0; i < N; i += sv_float_len ())
prev = f (svmad_f32_x (svptrue_b32 (), prev, z, sv_float_load (Af+i)), svptrue_b32 ());
}
#endif
static uint64_t
tic (void)
{
@ -570,6 +489,16 @@ bench1 (const struct fun *f, int type, double lo, double hi)
else if (f->prec == 'f' && type == 'l' && f->vec == 'n')
TIMEIT (runf_vn_latency, f->fun.vnf);
#endif
#if WANT_SVE_MATH
else if (f->prec == 'd' && type == 't' && f->vec == 's')
TIMEIT (run_sv_thruput, f->fun.svd);
else if (f->prec == 'd' && type == 'l' && f->vec == 's')
TIMEIT (run_sv_latency, f->fun.svd);
else if (f->prec == 'f' && type == 't' && f->vec == 's')
TIMEIT (runf_sv_thruput, f->fun.svf);
else if (f->prec == 'f' && type == 'l' && f->vec == 's')
TIMEIT (runf_sv_latency, f->fun.svf);
#endif
if (type == 't')
{

100
contrib/arm-optimized-routines/math/test/mathbench_funcs.h Normal file
View file

@ -0,0 +1,100 @@
/*
* Function entries for mathbench.
*
* Copyright (c) 2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
D (exp, -9.9, 9.9)
D (exp, 0.5, 1.0)
D (exp2, -9.9, 9.9)
D (log, 0.01, 11.1)
D (log, 0.999, 1.001)
D (log2, 0.01, 11.1)
D (log2, 0.999, 1.001)
{"pow", 'd', 0, 0.01, 11.1, {.d = xypow}},
D (xpow, 0.01, 11.1)
D (ypow, -9.9, 9.9)
D (erf, -6.0, 6.0)
F (expf, -9.9, 9.9)
F (exp2f, -9.9, 9.9)
F (logf, 0.01, 11.1)
F (log2f, 0.01, 11.1)
{"powf", 'f', 0, 0.01, 11.1, {.f = xypowf}},
F (xpowf, 0.01, 11.1)
F (ypowf, -9.9, 9.9)
{"sincosf", 'f', 0, 0.1, 0.7, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 0.8, 3.1, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, -3.1, 3.1, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 3.3, 33.3, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 100, 1000, {.f = sincosf_wrap}},
{"sincosf", 'f', 0, 1e6, 1e32, {.f = sincosf_wrap}},
F (sinf, 0.1, 0.7)
F (sinf, 0.8, 3.1)
F (sinf, -3.1, 3.1)
F (sinf, 3.3, 33.3)
F (sinf, 100, 1000)
F (sinf, 1e6, 1e32)
F (cosf, 0.1, 0.7)
F (cosf, 0.8, 3.1)
F (cosf, -3.1, 3.1)
F (cosf, 3.3, 33.3)
F (cosf, 100, 1000)
F (cosf, 1e6, 1e32)
F (erff, -4.0, 4.0)
#if WANT_VMATH
D (__s_sin, -3.1, 3.1)
D (__s_cos, -3.1, 3.1)
D (__s_exp, -9.9, 9.9)
D (__s_log, 0.01, 11.1)
{"__s_pow", 'd', 0, 0.01, 11.1, {.d = xy__s_pow}},
F (__s_expf, -9.9, 9.9)
F (__s_expf_1u, -9.9, 9.9)
F (__s_exp2f, -9.9, 9.9)
F (__s_exp2f_1u, -9.9, 9.9)
F (__s_logf, 0.01, 11.1)
{"__s_powf", 'f', 0, 0.01, 11.1, {.f = xy__s_powf}},
F (__s_sinf, -3.1, 3.1)
F (__s_cosf, -3.1, 3.1)
#if __aarch64__
VD (__v_sin, -3.1, 3.1)
VD (__v_cos, -3.1, 3.1)
VD (__v_exp, -9.9, 9.9)
VD (__v_log, 0.01, 11.1)
{"__v_pow", 'd', 'v', 0.01, 11.1, {.vd = xy__v_pow}},
VF (__v_expf, -9.9, 9.9)
VF (__v_expf_1u, -9.9, 9.9)
VF (__v_exp2f, -9.9, 9.9)
VF (__v_exp2f_1u, -9.9, 9.9)
VF (__v_logf, 0.01, 11.1)
{"__v_powf", 'f', 'v', 0.01, 11.1, {.vf = xy__v_powf}},
VF (__v_sinf, -3.1, 3.1)
VF (__v_cosf, -3.1, 3.1)
#ifdef __vpcs
VND (__vn_exp, -9.9, 9.9)
VND (_ZGVnN2v_exp, -9.9, 9.9)
VND (__vn_log, 0.01, 11.1)
VND (_ZGVnN2v_log, 0.01, 11.1)
{"__vn_pow", 'd', 'n', 0.01, 11.1, {.vnd = xy__vn_pow}},
{"_ZGVnN2vv_pow", 'd', 'n', 0.01, 11.1, {.vnd = xy_Z_pow}},
VND (__vn_sin, -3.1, 3.1)
VND (_ZGVnN2v_sin, -3.1, 3.1)
VND (__vn_cos, -3.1, 3.1)
VND (_ZGVnN2v_cos, -3.1, 3.1)
VNF (__vn_expf, -9.9, 9.9)
VNF (_ZGVnN4v_expf, -9.9, 9.9)
VNF (__vn_expf_1u, -9.9, 9.9)
VNF (__vn_exp2f, -9.9, 9.9)
VNF (_ZGVnN4v_exp2f, -9.9, 9.9)
VNF (__vn_exp2f_1u, -9.9, 9.9)
VNF (__vn_logf, 0.01, 11.1)
VNF (_ZGVnN4v_logf, 0.01, 11.1)
{"__vn_powf", 'f', 'n', 0.01, 11.1, {.vnf = xy__vn_powf}},
{"_ZGVnN4vv_powf", 'f', 'n', 0.01, 11.1, {.vnf = xy_Z_powf}},
VNF (__vn_sinf, -3.1, 3.1)
VNF (_ZGVnN4v_sinf, -3.1, 3.1)
VNF (__vn_cosf, -3.1, 3.1)
VNF (_ZGVnN4v_cosf, -3.1, 3.1)
#endif
#endif
#endif

104
contrib/arm-optimized-routines/math/test/mathbench_wrappers.h Normal file
View file

@ -0,0 +1,104 @@
/*
* Function wrappers for mathbench.
*
* Copyright (c) 2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#if WANT_VMATH
#if __aarch64__
#ifdef __vpcs
__vpcs static v_float
xy__vn_powf (v_float x)
{
return __vn_powf (x, x);
}
__vpcs static v_float
xy_Z_powf (v_float x)
{
return _ZGVnN4vv_powf (x, x);
}
__vpcs static v_double
xy__vn_pow (v_double x)
{
return __vn_pow (x, x);
}
__vpcs static v_double
xy_Z_pow (v_double x)
{
return _ZGVnN2vv_pow (x, x);
}
#endif // __vpcs
static v_float
xy__v_powf (v_float x)
{
return __v_powf (x, x);
}
static v_double
xy__v_pow (v_double x)
{
return __v_pow (x, x);
}
#endif // __aarch64__
static float
xy__s_powf (float x)
{
return __s_powf (x, x);
}
static double
xy__s_pow (double x)
{
return __s_pow (x, x);
}
#endif // WANT_VMATH
static double
xypow (double x)
{
return pow (x, x);
}
static float
xypowf (float x)
{
return powf (x, x);
}
static double
xpow (double x)
{
return pow (x, 23.4);
}
static float
xpowf (float x)
{
return powf (x, 23.4f);
}
static double
ypow (double x)
{
return pow (2.34, x);
}
static float
ypowf (float x)
{
return powf (2.34f, x);
}
static float
sincosf_wrap (float x)
{
float s, c;
sincosf (x, &s, &c);
return s + c;
}

9
contrib/arm-optimized-routines/math/test/mathtest.c
View file

@ -1,8 +1,8 @@
/*
* mathtest.c - test rig for mathlib
*
* Copyright (c) 1998-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 1998-2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <assert.h>
@ -196,9 +196,11 @@ int is_complex_rettype(int rettype) {
#define TFUNCARM(arg,ret,name,tolerance) { t_func, arg, ret, (void*)& ARM_PREFIX(name), m_none, tolerance, #name }
#define MFUNC(arg,ret,name,tolerance) { t_macro, arg, ret, NULL, m_##name, tolerance, #name }
#ifndef PL
/* sincosf wrappers for easier testing. */
static float sincosf_sinf(float x) { float s,c; sincosf(x, &s, &c); return s; }
static float sincosf_cosf(float x) { float s,c; sincosf(x, &s, &c); return c; }
#endif
test_func tfuncs[] = {
/* trigonometric */
@ -218,9 +220,10 @@ test_func tfuncs[] = {
TFUNCARM(at_s,rt_s, tanf, 4*ULPUNIT),
TFUNCARM(at_s,rt_s, sinf, 3*ULPUNIT/4),
TFUNCARM(at_s,rt_s, cosf, 3*ULPUNIT/4),
#ifndef PL
TFUNCARM(at_s,rt_s, sincosf_sinf, 3*ULPUNIT/4),
TFUNCARM(at_s,rt_s, sincosf_cosf, 3*ULPUNIT/4),
#endif
/* hyperbolic */
TFUNC(at_d, rt_d, atanh, 4*ULPUNIT),
TFUNC(at_d, rt_d, asinh, 4*ULPUNIT),

2
contrib/arm-optimized-routines/math/test/rtest/dotest.c
View file

@ -2,7 +2,7 @@
* dotest.c - actually generate mathlib test cases
*
* Copyright (c) 1999-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdio.h>

2
contrib/arm-optimized-routines/math/test/rtest/intern.h
View file

@ -2,7 +2,7 @@
* intern.h
*
* Copyright (c) 1999-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#ifndef mathtest_intern_h

2
contrib/arm-optimized-routines/math/test/rtest/main.c
View file

@ -2,7 +2,7 @@
* main.c
*
* Copyright (c) 1999-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <assert.h>

2
contrib/arm-optimized-routines/math/test/rtest/random.c
View file

@ -2,7 +2,7 @@
* random.c - random number generator for producing mathlib test cases
*
* Copyright (c) 1998-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "types.h"

2
contrib/arm-optimized-routines/math/test/rtest/random.h
View file

@ -2,7 +2,7 @@
* random.h - header for random.c
*
* Copyright (c) 2009-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include "types.h"

2
contrib/arm-optimized-routines/math/test/rtest/semi.c
View file

@ -2,7 +2,7 @@
* semi.c: test implementations of mathlib seminumerical functions
*
* Copyright (c) 1999-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <stdio.h>

2
contrib/arm-optimized-routines/math/test/rtest/semi.h
View file

@ -2,7 +2,7 @@
* semi.h: header for semi.c
*
* Copyright (c) 1999-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#ifndef test_semi_h

2
contrib/arm-optimized-routines/math/test/rtest/types.h
View file

@ -2,7 +2,7 @@
* types.h
*
* Copyright (c) 2005-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#ifndef mathtest_types_h

2
contrib/arm-optimized-routines/math/test/rtest/wrappers.c
View file

@ -2,7 +2,7 @@
* wrappers.c - wrappers to modify output of MPFR/MPC test functions
*
* Copyright (c) 2014-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <assert.h>

2
contrib/arm-optimized-routines/math/test/rtest/wrappers.h
View file

@ -2,7 +2,7 @@
* wrappers.h - wrappers to modify output of MPFR/MPC test functions
*
* Copyright (c) 2014-2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
typedef struct {

47
contrib/arm-optimized-routines/math/test/runulp.sh
View file

@ -2,8 +2,8 @@
# ULP error check script.
#
# Copyright (c) 2019-2020, Arm Limited.
# SPDX-License-Identifier: MIT
# Copyright (c) 2019-2022, Arm Limited.
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
#set -x
set -eu
@ -145,7 +145,7 @@ done
# vector functions
Ldir=0.5
r='n'
flags="${ULPFLAGS:--q} -f"
flags="${ULPFLAGS:--q}"
runs=
check __s_exp 1 && runs=1
runv=
@ -229,7 +229,7 @@ L_sinf=1.4
L_cosf=1.4
L_powf=2.1
while read G F R
while read G F R D
do
[ "$R" = 1 ] || continue
case "$G" in \#*) continue ;; esac
@ -239,7 +239,16 @@ do
do
[ -n "$X" ] || continue
case "$X" in \#*) continue ;; esac
t $F $X
disable_fenv=""
if [ -z "$WANT_SIMD_EXCEPT" ] || [ $WANT_SIMD_EXCEPT -eq 0 ]; then
# If library was built with SIMD exceptions
# disabled, disable fenv checking in ulp
# tool. Otherwise, fenv checking may still be
# disabled by adding -f to the end of the run
# line.
disable_fenv="-f"
fi
t $D $disable_fenv $F $X
done << EOF
$range
EOF
@ -255,10 +264,10 @@ log __v_log $runv
log __vn_log $runvn
log _ZGVnN2v_log $runvn
pow __s_pow $runs
pow __v_pow $runv
pow __vn_pow $runvn
pow _ZGVnN2vv_pow $runvn
pow __s_pow $runs -f
pow __v_pow $runv -f
pow __vn_pow $runvn -f
pow _ZGVnN2vv_pow $runvn -f
sin __s_sin $runs
sin __v_sin $runv
@ -275,18 +284,18 @@ expf __v_expf $runv
expf __vn_expf $runvn
expf _ZGVnN4v_expf $runvn
expf_1u __s_expf_1u $runs
expf_1u __v_expf_1u $runv
expf_1u __vn_expf_1u $runvn
expf_1u __s_expf_1u $runs -f
expf_1u __v_expf_1u $runv -f
expf_1u __vn_expf_1u $runvn -f
exp2f __s_exp2f $runs
exp2f __v_exp2f $runv
exp2f __vn_exp2f $runvn
exp2f _ZGVnN4v_exp2f $runvn
exp2f_1u __s_exp2f_1u $runs
exp2f_1u __v_exp2f_1u $runv
exp2f_1u __vn_exp2f_1u $runvn
exp2f_1u __s_exp2f_1u $runs -f
exp2f_1u __v_exp2f_1u $runv -f
exp2f_1u __vn_exp2f_1u $runvn -f
logf __s_logf $runs
logf __v_logf $runv
@ -303,10 +312,10 @@ cosf __v_cosf $runv
cosf __vn_cosf $runvn
cosf _ZGVnN4v_cosf $runvn
powf __s_powf $runs
powf __v_powf $runv
powf __vn_powf $runvn
powf _ZGVnN4vv_powf $runvn
powf __s_powf $runs -f
powf __v_powf $runv -f
powf __vn_powf $runvn -f
powf _ZGVnN4vv_powf $runvn -f
EOF
[ 0 -eq $FAIL ] || {

2
contrib/arm-optimized-routines/math/test/testcases/directed/cosf.tst
View file

@ -1,7 +1,7 @@
; cosf.tst - Directed test cases for SP cosine
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=cosf op1=7fc00001 result=7fc00001 errno=0
func=cosf op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/erf.tst
View file

@ -1,7 +1,7 @@
; erf.tst - Directed test cases for erf
;
; Copyright (c) 2007-2020, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=erf op1=7ff80000.00000001 result=7ff80000.00000001 errno=0
func=erf op1=fff80000.00000001 result=7ff80000.00000001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/erff.tst
View file

@ -1,7 +1,7 @@
; erff.tst
;
; Copyright (c) 2007-2020, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=erff op1=7fc00001 result=7fc00001 errno=0
func=erff op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/exp.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for exp
;
; Copyright (c) 2018-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=exp op1=7ff80000.00000001 result=7ff80000.00000001 errno=0
func=exp op1=fff80000.00000001 result=7ff80000.00000001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/exp2.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for exp2
;
; Copyright (c) 2018-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=exp2 op1=7ff80000.00000001 result=7ff80000.00000001 errno=0
func=exp2 op1=fff80000.00000001 result=7ff80000.00000001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/exp2f.tst
View file

@ -1,7 +1,7 @@
; exp2f.tst - Directed test cases for exp2f
;
; Copyright (c) 2017-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=exp2f op1=7fc00001 result=7fc00001 errno=0
func=exp2f op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/expf.tst
View file

@ -1,7 +1,7 @@
; expf.tst - Directed test cases for expf
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=expf op1=7fc00001 result=7fc00001 errno=0
func=expf op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/log.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for log
;
; Copyright (c) 2018-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=log op1=7ff80000.00000001 result=7ff80000.00000001 errno=0
func=log op1=fff80000.00000001 result=7ff80000.00000001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/log2.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for log2
;
; Copyright (c) 2018-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=log2 op1=7ff80000.00000001 result=7ff80000.00000001 errno=0
func=log2 op1=fff80000.00000001 result=7ff80000.00000001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/log2f.tst
View file

@ -1,7 +1,7 @@
; log2f.tst - Directed test cases for log2f
;
; Copyright (c) 2017-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=log2f op1=7fc00001 result=7fc00001 errno=0
func=log2f op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/logf.tst
View file

@ -1,7 +1,7 @@
; logf.tst - Directed test cases for logf
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=logf op1=7fc00001 result=7fc00001 errno=0
func=logf op1=ffc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/pow.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for pow
;
; Copyright (c) 2018-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=pow op1=00000000.00000000 op2=00000000.00000000 result=3ff00000.00000000 errno=0
func=pow op1=00000000.00000000 op2=00000000.00000001 result=00000000.00000000 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/powf.tst
View file

@ -1,7 +1,7 @@
; powf.tst - Directed test cases for powf
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=powf op1=7f800001 op2=7f800001 result=7fc00001 errno=0 status=i
func=powf op1=7f800001 op2=ff800001 result=7fc00001 errno=0 status=i

2
contrib/arm-optimized-routines/math/test/testcases/directed/sincosf.tst
View file

@ -1,7 +1,7 @@
; Directed test cases for SP sincos
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=sincosf_sinf op1=7fc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/directed/sinf.tst
View file

@ -1,7 +1,7 @@
; sinf.tst - Directed test cases for SP sine
;
; Copyright (c) 2007-2019, Arm Limited.
; SPDX-License-Identifier: MIT
; SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
func=sinf op1=7fc00001 result=7fc00001 errno=0

2
contrib/arm-optimized-routines/math/test/testcases/random/double.tst
View file

@ -1,7 +1,7 @@
!! double.tst - Random test case specification for DP functions
!!
!! Copyright (c) 1999-2019, Arm Limited.
!! SPDX-License-Identifier: MIT
!! SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
test exp 10000
test exp2 10000

2
contrib/arm-optimized-routines/math/test/testcases/random/float.tst
View file

@ -1,7 +1,7 @@
!! single.tst - Random test case specification for SP functions
!!
!! Copyright (c) 1999-2019, Arm Limited.
!! SPDX-License-Identifier: MIT
!! SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
test sinf 10000
test cosf 10000

248
contrib/arm-optimized-routines/math/test/ulp.c
View file

@ -1,8 +1,8 @@
/*
* ULP error checking tool for math functions.
*
* Copyright (c) 2019-2020, Arm Limited.
* SPDX-License-Identifier: MIT
* Copyright (c) 2019-2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
#include <ctype.h>
@ -214,16 +214,6 @@ struct conf
double errlim;
};
/* Wrappers for sincos. */
static float sincosf_sinf(float x) {(void)cosf(x); return sinf(x);}
static float sincosf_cosf(float x) {(void)sinf(x); return cosf(x);}
static double sincos_sin(double x) {(void)cos(x); return sin(x);}
static double sincos_cos(double x) {(void)sin(x); return cos(x);}
#if USE_MPFR
static int sincos_mpfr_sin(mpfr_t y, const mpfr_t x, mpfr_rnd_t r) { mpfr_cos(y,x,r); return mpfr_sin(y,x,r); }
static int sincos_mpfr_cos(mpfr_t y, const mpfr_t x, mpfr_rnd_t r) { mpfr_sin(y,x,r); return mpfr_cos(y,x,r); }
#endif
/* A bit of a hack: call vector functions twice with the same
input in lane 0 but a different value in other lanes: once
with an in-range value and then with a special case value. */
@ -233,52 +223,81 @@ static int secondcall;
#if __aarch64__ && WANT_VMATH
typedef __f32x4_t v_float;
typedef __f64x2_t v_double;
static const float fv[2] = {1.0f, -INFINITY};
static const double dv[2] = {1.0, -INFINITY};
/* First element of fv and dv may be changed by -c argument. */
static float fv[2] = {1.0f, -INFINITY};
static double dv[2] = {1.0, -INFINITY};
static inline v_float argf(float x) { return (v_float){x,x,x,fv[secondcall]}; }
static inline v_double argd(double x) { return (v_double){x,dv[secondcall]}; }
#if WANT_SVE_MATH
#include <arm_sve.h>
typedef __SVFloat32_t sv_float;
typedef __SVFloat64_t sv_double;
static float v_sinf(float x) { return __v_sinf(argf(x))[0]; }
static float v_cosf(float x) { return __v_cosf(argf(x))[0]; }
static float v_expf_1u(float x) { return __v_expf_1u(argf(x))[0]; }
static float v_expf(float x) { return __v_expf(argf(x))[0]; }
static float v_exp2f_1u(float x) { return __v_exp2f_1u(argf(x))[0]; }
static float v_exp2f(float x) { return __v_exp2f(argf(x))[0]; }
static float v_logf(float x) { return __v_logf(argf(x))[0]; }
static float v_powf(float x, float y) { return __v_powf(argf(x),argf(y))[0]; }
static double v_sin(double x) { return __v_sin(argd(x))[0]; }
static double v_cos(double x) { return __v_cos(argd(x))[0]; }
static double v_exp(double x) { return __v_exp(argd(x))[0]; }
static double v_log(double x) { return __v_log(argd(x))[0]; }
static double v_pow(double x, double y) { return __v_pow(argd(x),argd(y))[0]; }
#ifdef __vpcs
static float vn_sinf(float x) { return __vn_sinf(argf(x))[0]; }
static float vn_cosf(float x) { return __vn_cosf(argf(x))[0]; }
static float vn_expf_1u(float x) { return __vn_expf_1u(argf(x))[0]; }
static float vn_expf(float x) { return __vn_expf(argf(x))[0]; }
static float vn_exp2f_1u(float x) { return __vn_exp2f_1u(argf(x))[0]; }
static float vn_exp2f(float x) { return __vn_exp2f(argf(x))[0]; }
static float vn_logf(float x) { return __vn_logf(argf(x))[0]; }
static float vn_powf(float x, float y) { return __vn_powf(argf(x),argf(y))[0]; }
static double vn_sin(double x) { return __vn_sin(argd(x))[0]; }
static double vn_cos(double x) { return __vn_cos(argd(x))[0]; }
static double vn_exp(double x) { return __vn_exp(argd(x))[0]; }
static double vn_log(double x) { return __vn_log(argd(x))[0]; }
static double vn_pow(double x, double y) { return __vn_pow(argd(x),argd(y))[0]; }
static float Z_sinf(float x) { return _ZGVnN4v_sinf(argf(x))[0]; }
static float Z_cosf(float x) { return _ZGVnN4v_cosf(argf(x))[0]; }
static float Z_expf(float x) { return _ZGVnN4v_expf(argf(x))[0]; }
static float Z_exp2f(float x) { return _ZGVnN4v_exp2f(argf(x))[0]; }
static float Z_logf(float x) { return _ZGVnN4v_logf(argf(x))[0]; }
static float Z_powf(float x, float y) { return _ZGVnN4vv_powf(argf(x),argf(y))[0]; }
static double Z_sin(double x) { return _ZGVnN2v_sin(argd(x))[0]; }
static double Z_cos(double x) { return _ZGVnN2v_cos(argd(x))[0]; }
static double Z_exp(double x) { return _ZGVnN2v_exp(argd(x))[0]; }
static double Z_log(double x) { return _ZGVnN2v_log(argd(x))[0]; }
static double Z_pow(double x, double y) { return _ZGVnN2vv_pow(argd(x),argd(y))[0]; }
static inline sv_float svargf(float x) {
int n = svcntw();
float base[n];
for (int i=0; i<n; i++)
base[i] = (float)x;
base[n-1] = (float) fv[secondcall];
return svld1(svptrue_b32(), base);
}
static inline sv_double svargd(double x) {
int n = svcntd();
double base[n];
for (int i=0; i<n; i++)
base[i] = x;
base[n-1] = dv[secondcall];
return svld1(svptrue_b64(), base);
}
static inline float svretf(sv_float vec) {
int n = svcntw();
float res[n];
svst1(svptrue_b32(), res, vec);
return res[0];
}
static inline double svretd(sv_double vec) {
int n = svcntd();
double res[n];
svst1(svptrue_b64(), res, vec);
return res[0];
}
#endif
#endif
#if WANT_SVE_MATH
long double
dummyl (long double x)
{
return x;
}
double
dummy (double x)
{
return x;
}
static sv_double
__sv_dummy (sv_double x)
{
return x;
}
static sv_float
__sv_dummyf (sv_float x)
{
return x;
}
#endif
#include "test/ulp_wrappers.h"
/* Wrappers for SVE functions. */
#if WANT_SVE_MATH
static double sv_dummy (double x) { return svretd (__sv_dummy (svargd (x))); }
static float sv_dummyf (float x) { return svretf (__sv_dummyf (svargf (x))); }
#endif
struct fun
{
const char *name;
@ -322,83 +341,53 @@ static const struct fun fun[] = {
#define F2(x) F (x##f, x##f, x, mpfr_##x, 2, 1, f2, 0)
#define D1(x) F (x, x, x##l, mpfr_##x, 1, 0, d1, 0)
#define D2(x) F (x, x, x##l, mpfr_##x, 2, 0, d2, 0)
F1 (sin)
F1 (cos)
F (sincosf_sinf, sincosf_sinf, sincos_sin, sincos_mpfr_sin, 1, 1, f1, 0)
F (sincosf_cosf, sincosf_cosf, sincos_cos, sincos_mpfr_cos, 1, 1, f1, 0)
F1 (exp)
F1 (exp2)
F1 (log)
F1 (log2)
F2 (pow)
F1 (erf)
D1 (exp)
D1 (exp2)
D1 (log)
D1 (log2)
D2 (pow)
D1 (erf)
#if WANT_VMATH
F (__s_sinf, __s_sinf, sin, mpfr_sin, 1, 1, f1, 0)
F (__s_cosf, __s_cosf, cos, mpfr_cos, 1, 1, f1, 0)
F (__s_expf_1u, __s_expf_1u, exp, mpfr_exp, 1, 1, f1, 0)
F (__s_expf, __s_expf, exp, mpfr_exp, 1, 1, f1, 0)
F (__s_exp2f_1u, __s_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 0)
F (__s_exp2f, __s_exp2f, exp2, mpfr_exp2, 1, 1, f1, 0)
F (__s_powf, __s_powf, pow, mpfr_pow, 2, 1, f2, 0)
F (__s_logf, __s_logf, log, mpfr_log, 1, 1, f1, 0)
F (__s_sin, __s_sin, sinl, mpfr_sin, 1, 0, d1, 0)
F (__s_cos, __s_cos, cosl, mpfr_cos, 1, 0, d1, 0)
F (__s_exp, __s_exp, expl, mpfr_exp, 1, 0, d1, 0)
F (__s_log, __s_log, logl, mpfr_log, 1, 0, d1, 0)
F (__s_pow, __s_pow, powl, mpfr_pow, 2, 0, d2, 0)
#if __aarch64__
F (__v_sinf, v_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (__v_cosf, v_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (__v_expf_1u, v_expf_1u, exp, mpfr_exp, 1, 1, f1, 1)
F (__v_expf, v_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (__v_exp2f_1u, v_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__v_exp2f, v_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__v_logf, v_logf, log, mpfr_log, 1, 1, f1, 1)
F (__v_powf, v_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (__v_sin, v_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (__v_cos, v_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (__v_exp, v_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (__v_log, v_log, logl, mpfr_log, 1, 0, d1, 1)
F (__v_pow, v_pow, powl, mpfr_pow, 2, 0, d2, 1)
#ifdef __vpcs
F (__vn_sinf, vn_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (__vn_cosf, vn_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (__vn_expf_1u, vn_expf_1u, exp, mpfr_exp, 1, 1, f1, 1)
F (__vn_expf, vn_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (__vn_exp2f_1u, vn_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__vn_exp2f, vn_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__vn_logf, vn_logf, log, mpfr_log, 1, 1, f1, 1)
F (__vn_powf, vn_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (__vn_sin, vn_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (__vn_cos, vn_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (__vn_exp, vn_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (__vn_log, vn_log, logl, mpfr_log, 1, 0, d1, 1)
F (__vn_pow, vn_pow, powl, mpfr_pow, 2, 0, d2, 1)
F (_ZGVnN4v_sinf, Z_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (_ZGVnN4v_cosf, Z_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (_ZGVnN4v_expf, Z_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (_ZGVnN4v_exp2f, Z_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (_ZGVnN4v_logf, Z_logf, log, mpfr_log, 1, 1, f1, 1)
F (_ZGVnN4vv_powf, Z_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (_ZGVnN2v_sin, Z_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (_ZGVnN2v_cos, Z_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (_ZGVnN2v_exp, Z_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (_ZGVnN2v_log, Z_log, logl, mpfr_log, 1, 0, d1, 1)
F (_ZGVnN2vv_pow, Z_pow, powl, mpfr_pow, 2, 0, d2, 1)
#endif
#endif
/* Neon routines. */
#define VF1(x) F (__v_##x##f, v_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define VF2(x) F (__v_##x##f, v_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define VD1(x) F (__v_##x, v_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define VD2(x) F (__v_##x, v_##x, x##l, mpfr_##x, 2, 0, d2, 0)
#define VNF1(x) F (__vn_##x##f, vn_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define VNF2(x) F (__vn_##x##f, vn_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define VND1(x) F (__vn_##x, vn_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define VND2(x) F (__vn_##x, vn_##x, x##l, mpfr_##x, 2, 0, d2, 0)
#define ZVF1(x) F (_ZGVnN4v_##x##f, Z_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define ZVF2(x) F (_ZGVnN4vv_##x##f, Z_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define ZVD1(x) F (_ZGVnN2v_##x, Z_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define ZVD2(x) F (_ZGVnN2vv_##x, Z_##x, x##l, mpfr_##x, 2, 0, d2, 0)
#define ZVNF1(x) VNF1 (x) ZVF1 (x)
#define ZVNF2(x) VNF2 (x) ZVF2 (x)
#define ZVND1(x) VND1 (x) ZVD1 (x)
#define ZVND2(x) VND2 (x) ZVD2 (x)
#define SF1(x) F (__s_##x##f, __s_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define SF2(x) F (__s_##x##f, __s_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define SD1(x) F (__s_##x, __s_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define SD2(x) F (__s_##x, __s_##x, x##l, mpfr_##x, 2, 0, d2, 0)
/* SVE routines. */
#define SVF1(x) F (__sv_##x##f, sv_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define SVF2(x) F (__sv_##x##f, sv_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define SVD1(x) F (__sv_##x, sv_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define SVD2(x) F (__sv_##x, sv_##x, x##l, mpfr_##x, 2, 0, d2, 0)
#define ZSVF1(x) F (_ZGVsMxv_##x##f, Z_sv_##x##f, x, mpfr_##x, 1, 1, f1, 0)
#define ZSVF2(x) F (_ZGVsMxvv_##x##f, Z_sv_##x##f, x, mpfr_##x, 2, 1, f2, 0)
#define ZSVD1(x) F (_ZGVsMxv_##x, Z_sv_##x, x##l, mpfr_##x, 1, 0, d1, 0)
#define ZSVD2(x) F (_ZGVsMxvv_##x, Z_sv_##x, x##l, mpfr_##x, 2, 0, d2, 0)
#include "test/ulp_funcs.h"
#if WANT_SVE_MATH
SVD1 (dummy)
SVF1 (dummy)
#endif
#undef F
#undef F1
#undef F2
#undef D1
#undef D2
#undef SVF1
#undef SVF2
#undef SVD1
#undef SVD2
{0}};
/* Boilerplate for generic calls. */
@ -645,6 +634,11 @@ usage (void)
puts ("-q: quiet.");
puts ("-m: use mpfr even if faster method is available.");
puts ("-f: disable fenv testing (rounding modes and exceptions).");
#if __aarch64__ && WANT_VMATH
puts ("-c: neutral 'control value' to test behaviour when one lane can affect another. \n"
" This should be different from tested input in other lanes, and non-special \n"
" (i.e. should not trigger fenv exceptions). Default is 1.");
#endif
puts ("Supported func:");
for (const struct fun *f = fun; f->name; f++)
printf ("\t%s\n", f->name);
@ -812,6 +806,14 @@ main (int argc, char *argv[])
conf.rc = argv[0][0];
}
break;
#if __aarch64__ && WANT_VMATH
case 'c':
argc--;
argv++;
fv[0] = strtof(argv[0], 0);
dv[0] = strtod(argv[0], 0);
break;
#endif
default:
usage ();
}

2
contrib/arm-optimized-routines/math/test/ulp.h
View file

@ -2,7 +2,7 @@
* Generic functions for ULP error estimation.
*
* Copyright (c) 2019, Arm Limited.
* SPDX-License-Identifier: MIT
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
/* For each different math function type,

78
contrib/arm-optimized-routines/math/test/ulp_funcs.h Normal file
View file

@ -0,0 +1,78 @@
/*
* Function entries for ulp.
*
* Copyright (c) 2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
F1 (sin)
F1 (cos)
F (sincosf_sinf, sincosf_sinf, sincos_sin, sincos_mpfr_sin, 1, 1, f1, 0)
F (sincosf_cosf, sincosf_cosf, sincos_cos, sincos_mpfr_cos, 1, 1, f1, 0)
F1 (exp)
F1 (exp2)
F1 (log)
F1 (log2)
F2 (pow)
F1 (erf)
D1 (exp)
D1 (exp2)
D1 (log)
D1 (log2)
D2 (pow)
D1 (erf)
#if WANT_VMATH
F (__s_sinf, __s_sinf, sin, mpfr_sin, 1, 1, f1, 0)
F (__s_cosf, __s_cosf, cos, mpfr_cos, 1, 1, f1, 0)
F (__s_expf_1u, __s_expf_1u, exp, mpfr_exp, 1, 1, f1, 0)
F (__s_expf, __s_expf, exp, mpfr_exp, 1, 1, f1, 0)
F (__s_exp2f_1u, __s_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 0)
F (__s_exp2f, __s_exp2f, exp2, mpfr_exp2, 1, 1, f1, 0)
F (__s_powf, __s_powf, pow, mpfr_pow, 2, 1, f2, 0)
F (__s_logf, __s_logf, log, mpfr_log, 1, 1, f1, 0)
F (__s_sin, __s_sin, sinl, mpfr_sin, 1, 0, d1, 0)
F (__s_cos, __s_cos, cosl, mpfr_cos, 1, 0, d1, 0)
F (__s_exp, __s_exp, expl, mpfr_exp, 1, 0, d1, 0)
F (__s_log, __s_log, logl, mpfr_log, 1, 0, d1, 0)
F (__s_pow, __s_pow, powl, mpfr_pow, 2, 0, d2, 0)
#if __aarch64__
F (__v_sinf, v_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (__v_cosf, v_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (__v_expf_1u, v_expf_1u, exp, mpfr_exp, 1, 1, f1, 1)
F (__v_expf, v_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (__v_exp2f_1u, v_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__v_exp2f, v_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__v_logf, v_logf, log, mpfr_log, 1, 1, f1, 1)
F (__v_powf, v_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (__v_sin, v_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (__v_cos, v_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (__v_exp, v_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (__v_log, v_log, logl, mpfr_log, 1, 0, d1, 1)
F (__v_pow, v_pow, powl, mpfr_pow, 2, 0, d2, 1)
#ifdef __vpcs
F (__vn_sinf, vn_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (__vn_cosf, vn_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (__vn_expf_1u, vn_expf_1u, exp, mpfr_exp, 1, 1, f1, 1)
F (__vn_expf, vn_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (__vn_exp2f_1u, vn_exp2f_1u, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__vn_exp2f, vn_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (__vn_logf, vn_logf, log, mpfr_log, 1, 1, f1, 1)
F (__vn_powf, vn_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (__vn_sin, vn_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (__vn_cos, vn_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (__vn_exp, vn_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (__vn_log, vn_log, logl, mpfr_log, 1, 0, d1, 1)
F (__vn_pow, vn_pow, powl, mpfr_pow, 2, 0, d2, 1)
F (_ZGVnN4v_sinf, Z_sinf, sin, mpfr_sin, 1, 1, f1, 1)
F (_ZGVnN4v_cosf, Z_cosf, cos, mpfr_cos, 1, 1, f1, 1)
F (_ZGVnN4v_expf, Z_expf, exp, mpfr_exp, 1, 1, f1, 1)
F (_ZGVnN4v_exp2f, Z_exp2f, exp2, mpfr_exp2, 1, 1, f1, 1)
F (_ZGVnN4v_logf, Z_logf, log, mpfr_log, 1, 1, f1, 1)
F (_ZGVnN4vv_powf, Z_powf, pow, mpfr_pow, 2, 1, f2, 1)
F (_ZGVnN2v_sin, Z_sin, sinl, mpfr_sin, 1, 0, d1, 1)
F (_ZGVnN2v_cos, Z_cos, cosl, mpfr_cos, 1, 0, d1, 1)
F (_ZGVnN2v_exp, Z_exp, expl, mpfr_exp, 1, 0, d1, 1)
F (_ZGVnN2v_log, Z_log, logl, mpfr_log, 1, 0, d1, 1)
F (_ZGVnN2vv_pow, Z_pow, powl, mpfr_pow, 2, 0, d2, 1)
#endif
#endif
#endif

59
contrib/arm-optimized-routines/math/test/ulp_wrappers.h Normal file
View file

@ -0,0 +1,59 @@
/*
* Function wrappers for ulp.
*
* Copyright (c) 2022, Arm Limited.
* SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
*/
/* Wrappers for sincos. */
static float sincosf_sinf(float x) {(void)cosf(x); return sinf(x);}
static float sincosf_cosf(float x) {(void)sinf(x); return cosf(x);}
static double sincos_sin(double x) {(void)cos(x); return sin(x);}
static double sincos_cos(double x) {(void)sin(x); return cos(x);}
#if USE_MPFR
static int sincos_mpfr_sin(mpfr_t y, const mpfr_t x, mpfr_rnd_t r) { mpfr_cos(y,x,r); return mpfr_sin(y,x,r); }
static int sincos_mpfr_cos(mpfr_t y, const mpfr_t x, mpfr_rnd_t r) { mpfr_sin(y,x,r); return mpfr_cos(y,x,r); }
#endif
/* Wrappers for vector functions. */
#if __aarch64__ && WANT_VMATH
static float v_sinf(float x) { return __v_sinf(argf(x))[0]; }
static float v_cosf(float x) { return __v_cosf(argf(x))[0]; }
static float v_expf_1u(float x) { return __v_expf_1u(argf(x))[0]; }
static float v_expf(float x) { return __v_expf(argf(x))[0]; }
static float v_exp2f_1u(float x) { return __v_exp2f_1u(argf(x))[0]; }
static float v_exp2f(float x) { return __v_exp2f(argf(x))[0]; }
static float v_logf(float x) { return __v_logf(argf(x))[0]; }
static float v_powf(float x, float y) { return __v_powf(argf(x),argf(y))[0]; }
static double v_sin(double x) { return __v_sin(argd(x))[0]; }
static double v_cos(double x) { return __v_cos(argd(x))[0]; }
static double v_exp(double x) { return __v_exp(argd(x))[0]; }
static double v_log(double x) { return __v_log(argd(x))[0]; }
static double v_pow(double x, double y) { return __v_pow(argd(x),argd(y))[0]; }
#ifdef __vpcs
static float vn_sinf(float x) { return __vn_sinf(argf(x))[0]; }
static float vn_cosf(float x) { return __vn_cosf(argf(x))[0]; }
static float vn_expf_1u(float x) { return __vn_expf_1u(argf(x))[0]; }
static float vn_expf(float x) { return __vn_expf(argf(x))[0]; }
static float vn_exp2f_1u(float x) { return __vn_exp2f_1u(argf(x))[0]; }
static float vn_exp2f(float x) { return __vn_exp2f(argf(x))[0]; }
static float vn_logf(float x) { return __vn_logf(argf(x))[0]; }
static float vn_powf(float x, float y) { return __vn_powf(argf(x),argf(y))[0]; }
static double vn_sin(double x) { return __vn_sin(argd(x))[0]; }
static double vn_cos(double x) { return __vn_cos(argd(x))[0]; }
static double vn_exp(double x) { return __vn_exp(argd(x))[0]; }
static double vn_log(double x) { return __vn_log(argd(x))[0]; }
static double vn_pow(double x, double y) { return __vn_pow(argd(x),argd(y))[0]; }
static float Z_sinf(float x) { return _ZGVnN4v_sinf(argf(x))[0]; }
static float Z_cosf(float x) { return _ZGVnN4v_cosf(argf(x))[0]; }
static float Z_expf(float x) { return _ZGVnN4v_expf(argf(x))[0]; }
static float Z_exp2f(float x) { return _ZGVnN4v_exp2f(argf(x))[0]; }
static float Z_logf(float x) { return _ZGVnN4v_logf(argf(x))[0]; }
static float Z_powf(float x, float y) { return _ZGVnN4vv_powf(argf(x),argf(y))[0]; }
static double Z_sin(double x) { return _ZGVnN2v_sin(argd(x))[0]; }
static double Z_cos(double x) { return _ZGVnN2v_cos(argd(x))[0]; }
static double Z_exp(double x) { return _ZGVnN2v_exp(argd(x))[0]; }
static double Z_log(double x) { return _ZGVnN2v_log(argd(x))[0]; }
static double Z_pow(double x, double y) { return _ZGVnN2vv_pow(argd(x),argd(y))[0]; }
#endif
#endif

2
contrib/arm-optimized-routines/math/tools/cos.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating cos(x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 8; // polynomial degree
a = -pi/4; // interval

2
contrib/arm-optimized-routines/math/tools/exp.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating e^x
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 5; // poly degree
N = 128; // table entries

2
contrib/arm-optimized-routines/math/tools/exp2.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating 2^x
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
// exp2f parameters
deg = 3; // poly degree

2
contrib/arm-optimized-routines/math/tools/log.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating log(1+x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 12; // poly degree
// |log(1+x)| > 0x1p-4 outside the interval

2
contrib/arm-optimized-routines/math/tools/log2.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating log2(1+x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 11; // poly degree
// |log2(1+x)| > 0x1p-4 outside the interval

2
contrib/arm-optimized-routines/math/tools/log2_abs.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating log2(1+x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 7; // poly degree
// interval ~= 1/(2*N), where N is the table entries

2
contrib/arm-optimized-routines/math/tools/log_abs.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating log(1+x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 6; // poly degree
// interval ~= 1/(2*N), where N is the table entries

2
contrib/arm-optimized-routines/math/tools/plot.py
View file

@ -3,7 +3,7 @@
# ULP error plot tool.
#
# Copyright (c) 2019, Arm Limited.
# SPDX-License-Identifier: MIT
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
import numpy as np
import matplotlib.pyplot as plt

2
contrib/arm-optimized-routines/math/tools/remez.jl
View file

@ -4,7 +4,7 @@
# remez.jl - implementation of the Remez algorithm for polynomial approximation
#
# Copyright (c) 2015-2019, Arm Limited.
# SPDX-License-Identifier: MIT
# SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
import Base.\

2
contrib/arm-optimized-routines/math/tools/sin.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating sin(x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 7; // polynomial degree
a = -pi/4; // interval

2
contrib/arm-optimized-routines/math/tools/v_exp.sollya
View file

@ -1,7 +1,7 @@
// polynomial for approximating e^x
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 4; // poly degree
N = 128; // table entries

2
contrib/arm-optimized-routines/math/tools/v_log.sollya
View file

@ -1,7 +1,7 @@
// polynomial used for __v_log(x)
//
// Copyright (c) 2019, Arm Limited.
// SPDX-License-Identifier: MIT
// SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
deg = 6; // poly degree
a = -0x1.fc1p-9;

Some files were not shown because too many files have changed in this diff Show more