[libc][math][c23] Add acoshf16 C23 math function. #130588
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
harrisonGPU
force-pushed
the
libcAcoshf16
branch
2 times, most recently
from
5113bfe to
0dea69f
Compare
overmighty
requested changes
Mar 11, 2025
overmighty
reviewed
Mar 11, 2025
harrisonGPU
force-pushed
the
libcAcoshf16
branch
from
9af9063 to
2c14051
Compare
52 tasks
Member
|
@llvm/pr-subscribers-libc Author: Harrison Hao (harrisonGPU) ChangesImplementation of acoshf16 function for 16-bit inputs. Full diff: https://github.com/llvm/llvm-project/pull/130588.diff 11 Files Affected:
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 12dc87bf945fd..b705d81fff79c 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -653,6 +653,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
# math.h C23 _Float16 entrypoints
libc.src.math.asinf16
libc.src.math.acosf16
+ libc.src.math.acoshf16
libc.src.math.canonicalizef16
libc.src.math.ceilf16
libc.src.math.copysignf16
diff --git a/libc/docs/headers/math/index.rst b/libc/docs/headers/math/index.rst
index 5b855ce4881c3..8dc7c8c1a0966 100644
--- a/libc/docs/headers/math/index.rst
+++ b/libc/docs/headers/math/index.rst
@@ -251,7 +251,7 @@ Higher Math Functions
+===========+==================+=================+========================+======================+========================+========================+============================+
| acos | |check| | | | |check| | | 7.12.4.1 | F.10.1.1 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| acosh | |check| | | | | | 7.12.5.1 | F.10.2.1 |
+| acosh | |check| | | | |check| | | 7.12.5.1 | F.10.2.1 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
| acospi | | | | | | 7.12.4.8 | F.10.1.8 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
diff --git a/libc/include/math.yaml b/libc/include/math.yaml
index a66f981030864..7fd73445cc74e 100644
--- a/libc/include/math.yaml
+++ b/libc/include/math.yaml
@@ -27,6 +27,13 @@ functions:
return_type: float
arguments:
- type: float
+ name: acoshf16
+ standards:
+ - stdc
+ return_type: _Float16
+ arguments:
+ - type: _Float16
+ guard: LIBC_TYPES_HAS_FLOAT16
- name: asin
standards:
- stdc
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index f18a73d46f9aa..bdf8b923416d8 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -46,6 +46,7 @@ add_math_entrypoint_object(acosf16)
add_math_entrypoint_object(acosh)
add_math_entrypoint_object(acoshf)
+add_math_entrypoint_object(acoshf16)
add_math_entrypoint_object(asin)
add_math_entrypoint_object(asinf)
diff --git a/libc/src/math/acoshf16.h b/libc/src/math/acoshf16.h
new file mode 100644
index 0000000000000..f471ecfeb6154
--- /dev/null
+++ b/libc/src/math/acoshf16.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for acoshf16 ----------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_MATH_ACOSHF16_H
+#define LLVM_LIBC_SRC_MATH_ACOSHF16_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+float16 acoshf16(float16 x);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_ACOSHF16_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index 3114289bad486..fd5ecd72b765f 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -3944,6 +3944,27 @@ add_entrypoint_object(
libc.src.__support.macros.optimization
)
+add_entrypoint_object(
+ acoshf16
+ SRCS
+ acoshf16.cpp
+ HDRS
+ ../acoshf16.h
+ DEPENDS
+ .common_constants
+ .explogxf
+ libc.hdr.errno_macros
+ libc.hdr.fenv_macros
+ libc.src.__support.FPUtil.cast
+ libc.src.__support.FPUtil.fenv_impl
+ libc.src.__support.FPUtil.fp_bits
+ libc.src.__support.FPUtil.multiply_add
+ libc.src.__support.FPUtil.polyeval
+ libc.src.__support.FPUtil.sqrt
+ libc.src.__support.macros.optimization
+ libc.src.__support.macros.properties.types
+)
+
add_entrypoint_object(
asinhf
SRCS
diff --git a/libc/src/math/generic/acoshf16.cpp b/libc/src/math/generic/acoshf16.cpp
new file mode 100644
index 0000000000000..1a4f68987ee6b
--- /dev/null
+++ b/libc/src/math/generic/acoshf16.cpp
@@ -0,0 +1,88 @@
+//===-- Half-precision acoshf16 function ----------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/math/acoshf16.h"
+#include "src/__support/FPUtil/FEnvImpl.h"
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/FPUtil/cast.h"
+#include "src/__support/FPUtil/except_value_utils.h"
+#include "src/__support/FPUtil/generic/sqrt.h"
+#include "src/__support/FPUtil/multiply_add.h"
+#include "src/__support/macros/optimization.h"
+#include "src/math/generic/explogxf.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+static constexpr size_t N_EXCEPTS = 2;
+static constexpr fputil::ExceptValues<float16, N_EXCEPTS> ACOSHF16_EXCEPTS{
+ {// (input, RZ output, RU offset, RD offset, RN offset)
+ {0x41B7, 0x3ED8, 1, 0, 0},
+ {0x3CE4, 0x393E, 1, 0, 1}}};
+
+LLVM_LIBC_FUNCTION(float16, acoshf16, (float16 x)) {
+ using FPBits = fputil::FPBits<float16>;
+ FPBits xbits(x);
+ uint16_t x_u = xbits.uintval();
+ uint16_t x_abs = x_u & 0x7fff;
+
+ // Check for NaN input first.
+ if (LIBC_UNLIKELY(xbits.is_nan())) {
+ if (xbits.is_signaling_nan()) {
+ fputil::raise_except_if_required(FE_INVALID);
+ return FPBits::quiet_nan().get_val();
+ }
+ return x;
+ }
+
+ // Check for infinite inputs.
+ if (LIBC_UNLIKELY(xbits.is_inf())) {
+ if (xbits.is_neg()) {
+ fputil::set_errno_if_required(EDOM);
+ fputil::raise_except_if_required(FE_INVALID);
+ return FPBits::quiet_nan().get_val();
+ }
+ return x;
+ }
+
+ // Domain error for inputs less than 1.0.
+ if (LIBC_UNLIKELY(x_abs < 0x3c00U)) {
+ fputil::set_errno_if_required(EDOM);
+ fputil::raise_except_if_required(FE_INVALID);
+ return FPBits::quiet_nan().get_val();
+ }
+
+ // acosh(1.0) exactly equals 0.0
+ if (LIBC_UNLIKELY(x_u == 0x3c00U))
+ return float16(0.0f);
+
+ if (auto r = ACOSHF16_EXCEPTS.lookup(xbits.uintval());
+ LIBC_UNLIKELY(r.has_value()))
+ return r.value();
+
+ float xf32 = x;
+ // High precision for inputs very close to 1.0
+ if (LIBC_UNLIKELY(xf32 < 1.25f)) {
+ float delta = xf32 - 1.0f;
+ float sqrt_2_delta = fputil::sqrt<float>(2.0 * delta);
+ float x2 = delta;
+ float pe = fputil::polyeval(x2, 0x1.0000000000000p+0f,
+ -0x1.55551a83a9472p-4f, 0x1.331601c4b8ecfp-6f,
+ -0x1.6890f49eb0acbp-8f, 0x1.8f3a617040a6ap-10f);
+ float approx = sqrt_2_delta * pe;
+ return fputil::cast<float16>(approx);
+ }
+
+ // Standard computation for general case.
+ float sqrt_term =
+ fputil::sqrt<float>(fputil::multiply_add(xf32, xf32, -1.0f));
+ float result = static_cast<float>(log_eval(xf32 + sqrt_term));
+
+ return fputil::cast<float16>(result);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/src/math/CMakeLists.txt
index 53ddd301900c0..311d56a2a33ab 100644
--- a/libc/test/src/math/CMakeLists.txt
+++ b/libc/test/src/math/CMakeLists.txt
@@ -2173,6 +2173,17 @@ add_fp_unittest(
libc.src.__support.FPUtil.fp_bits
)
+add_fp_unittest(
+ acoshf16_test
+ NEED_MPFR
+ SUITE
+ libc-math-unittests
+ SRCS
+ acoshf16_test.cpp
+ DEPENDS
+ libc.src.math.acoshf16
+)
+
add_fp_unittest(
asinf_test
NEED_MPFR
diff --git a/libc/test/src/math/acoshf16_test.cpp b/libc/test/src/math/acoshf16_test.cpp
new file mode 100644
index 0000000000000..a779d3110f329
--- /dev/null
+++ b/libc/test/src/math/acoshf16_test.cpp
@@ -0,0 +1,76 @@
+//===-- Unittests for acoshf16 --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/FPUtil/cast.h"
+#include "src/errno/libc_errno.h"
+#include "src/math/acoshf16.h"
+#include "test/UnitTest/FPMatcher.h"
+#include "utils/MPFRWrapper/MPFRUtils.h"
+#include <stdint.h>
+
+using LlvmLibcAcoshf16Test = LIBC_NAMESPACE::testing::FPTest<float16>;
+namespace mpfr = LIBC_NAMESPACE::testing::mpfr;
+
+static constexpr uint16_t START = 0x3c00U;
+static constexpr uint16_t STOP = 0x7bffU;
+
+TEST_F(LlvmLibcAcoshf16Test, SpecialNumbers) {
+ LIBC_NAMESPACE::libc_errno = 0;
+
+ EXPECT_FP_EQ_ALL_ROUNDING(aNaN, LIBC_NAMESPACE::acoshf16(aNaN));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ_WITH_EXCEPTION(aNaN, LIBC_NAMESPACE::acoshf16(sNaN), FE_INVALID);
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ_ALL_ROUNDING(float16(0.0f),
+ LIBC_NAMESPACE::acoshf16(float16(1.0f)));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ_ALL_ROUNDING(aNaN, LIBC_NAMESPACE::acoshf16(float16(0.5f)));
+ EXPECT_MATH_ERRNO(EDOM);
+
+ EXPECT_FP_EQ_ALL_ROUNDING(inf, LIBC_NAMESPACE::acoshf16(inf));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ_ALL_ROUNDING(aNaN, LIBC_NAMESPACE::acoshf16(neg_inf));
+ EXPECT_MATH_ERRNO(EDOM);
+}
+
+TEST_F(LlvmLibcAcoshf16Test, PositiveRange) {
+ for (uint16_t v = START; v <= STOP; ++v) {
+ float16 x = FPBits(v).get_val();
+
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Acosh, x,
+ LIBC_NAMESPACE::acoshf16(x), 0.5);
+ }
+}
+
+TEST_F(LlvmLibcAcoshf16Test, SpecificBitPatterns) {
+ constexpr int N = 12;
+ constexpr uint16_t INPUTS[N] = {
+ 0x3C00, // x = 1.0
+ 0x3C01, // x = just above 1.0 (minimally larger than 1)
+ 0x3E00, // x = 1.5
+ 0x4200, // x = 3.0
+ 0x4500, // x = 5.0
+ 0x4900, // x = 10.0
+ 0x51FF, // x = ~47.94
+ 0x5CB0, // x = ~300.0
+ 0x643F, // x = ~1087.6
+ 0x77FF, // x = just below next exponent interval (max for exponent 0x1D)
+ 0x7801, // x = just above previous value (min for exponent 0x1E)
+ 0x7BFF // x = 65504.0 (max finite half)
+ };
+ for (int i = 0; i < N; ++i) {
+ float16 x = FPBits(INPUTS[i]).get_val();
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Acosh, x,
+ LIBC_NAMESPACE::acoshf16(x), 0.5);
+ }
+}
diff --git a/libc/test/src/math/smoke/CMakeLists.txt b/libc/test/src/math/smoke/CMakeLists.txt
index 6f94440d826d9..4303afd242e22 100644
--- a/libc/test/src/math/smoke/CMakeLists.txt
+++ b/libc/test/src/math/smoke/CMakeLists.txt
@@ -3945,6 +3945,17 @@ add_fp_unittest(
libc.src.__support.FPUtil.fp_bits
)
+add_fp_unittest(
+ acoshf16_test
+ SUITE
+ libc-math-smoke-tests
+ SRCS
+ acoshf16_test.cpp
+ DEPENDS
+ libc.src.errno.errno
+ libc.src.math.acoshf16
+)
+
add_fp_unittest(
asinf_test
SUITE
diff --git a/libc/test/src/math/smoke/acoshf16_test.cpp b/libc/test/src/math/smoke/acoshf16_test.cpp
new file mode 100644
index 0000000000000..b8d13b9fe88c3
--- /dev/null
+++ b/libc/test/src/math/smoke/acoshf16_test.cpp
@@ -0,0 +1,41 @@
+//===-- Unittests for acoshf16 --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/errno/libc_errno.h"
+#include "src/math/acoshf16.h"
+#include "test/UnitTest/FPMatcher.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcAcoshf16Test = LIBC_NAMESPACE::testing::FPTest<float16>;
+
+TEST_F(LlvmLibcAcoshf16Test, SpecialNumbers) {
+ LIBC_NAMESPACE::libc_errno = 0;
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::acoshf16(aNaN));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ_WITH_EXCEPTION(aNaN, LIBC_NAMESPACE::acoshf16(sNaN), FE_INVALID);
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::acoshf16(zero));
+ EXPECT_MATH_ERRNO(EDOM);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::acoshf16(neg_zero));
+ EXPECT_MATH_ERRNO(EDOM);
+
+ EXPECT_FP_EQ(inf, LIBC_NAMESPACE::acoshf16(inf));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::acoshf16(neg_inf));
+ EXPECT_MATH_ERRNO(EDOM);
+
+ EXPECT_FP_EQ(float16(0.0f), LIBC_NAMESPACE::acoshf16(float16(1.0f)));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::acoshf16(float16(0.5f)));
+ EXPECT_MATH_ERRNO(EDOM);
+}
|
lntue
reviewed
Mar 14, 2025
harrisonGPU
force-pushed
the
libcAcoshf16
branch
from
3922002 to
6d9a926
Compare
overmighty
requested changes
Mar 16, 2025
harrisonGPU
force-pushed
the
libcAcoshf16
branch
from
6d9a926 to
6c54d4b
Compare
Member
|
You don't need to force-push if you didn't rebase btw. |
Contributor
Author
Yes, I did rebase. |
overmighty
requested changes
Mar 17, 2025
harrisonGPU
force-pushed
the
libcAcoshf16
branch
from
c7f1f4f to
c8f5517
Compare
Contributor
Author
|
ping @overmighty |
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 20, 2025
lntue
reviewed
Mar 21, 2025
lntue
approved these changes
Mar 22, 2025
Contributor
lntue
left a comment
lntue
left a comment
There was a problem hiding this comment.
LGTM. Please wait to see if @overmighty has any other comments.
overmighty
requested changes
Mar 22, 2025
overmighty
approved these changes
Mar 22, 2025
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/71/builds/18786 Here is the relevant piece of the build log for the reference |
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/104/builds/18785 Here is the relevant piece of the build log for the reference |
Contributor
|
@harrisonGPU aarch64 failure: Can you investigate to see if it can be fixed quickly, or temporarily disable the assertion statement if you need more time? |
Contributor
Author
I will try to investigate this failing test, and for now, I have temporarily disabled it. #132580 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implementation of acoshf16 function for 16-bit inputs.