[libc-commits] [libc] [libc][math][c23] Add cospif16 function (PR #113001)
via libc-commits
libc-commits at lists.llvm.org
Sun Oct 20 09:11:28 PDT 2024
https://github.com/wldfngrs updated https://github.com/llvm/llvm-project/pull/113001
>From 0d59443daf39e994757f372d9aeb30fb4c22418b Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 00:14:59 +0100
Subject: [PATCH 01/11] Add cospif16 function
---
libc/config/linux/aarch64/entrypoints.txt | 1 +
libc/config/linux/x86_64/entrypoints.txt | 1 +
libc/newhdrgen/yaml/math.yaml | 7 +
libc/src/math/CMakeLists.txt | 1 +
libc/src/math/cospif16.h | 21 +++
libc/src/math/generic/CMakeLists.txt | 20 +++
libc/src/math/generic/cospif16.cpp | 141 +++++++++++++++++++++
libc/test/src/math/CMakeLists.txt | 11 ++
libc/test/src/math/cospif16_test.cpp | 38 ++++++
libc/test/src/math/smoke/CMakeLists.txt | 11 ++
libc/test/src/math/smoke/cospif16_test.cpp | 44 +++++++
11 files changed, 296 insertions(+)
create mode 100644 libc/src/math/cospif16.h
create mode 100644 libc/src/math/generic/cospif16.cpp
create mode 100644 libc/test/src/math/cospif16_test.cpp
create mode 100644 libc/test/src/math/smoke/cospif16_test.cpp
diff --git a/libc/config/linux/aarch64/entrypoints.txt b/libc/config/linux/aarch64/entrypoints.txt
index 885827d304efe3..85bb5df358ec3a 100644
--- a/libc/config/linux/aarch64/entrypoints.txt
+++ b/libc/config/linux/aarch64/entrypoints.txt
@@ -607,6 +607,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
libc.src.math.canonicalizef16
libc.src.math.ceilf16
libc.src.math.copysignf16
+ libc.src.math.cospif16
# TODO: aarch64 bug
# Please see https://github.com/llvm/llvm-project/pull/100632#issuecomment-2258772681
# libc.src.math.expf16
diff --git a/libc/config/linux/x86_64/entrypoints.txt b/libc/config/linux/x86_64/entrypoints.txt
index 06ea7bba81f345..f40d752840b85a 100644
--- a/libc/config/linux/x86_64/entrypoints.txt
+++ b/libc/config/linux/x86_64/entrypoints.txt
@@ -611,6 +611,7 @@ if(LIBC_TYPES_HAS_FLOAT16)
libc.src.math.ceilf16
libc.src.math.copysignf16
libc.src.math.coshf16
+ libc.src.math.cospif16
libc.src.math.exp10f16
libc.src.math.exp10m1f16
libc.src.math.exp2f16
diff --git a/libc/newhdrgen/yaml/math.yaml b/libc/newhdrgen/yaml/math.yaml
index 98ea1a0d25fbb7..e0986f00a3b464 100644
--- a/libc/newhdrgen/yaml/math.yaml
+++ b/libc/newhdrgen/yaml/math.yaml
@@ -206,6 +206,13 @@ functions:
return_type: float
arguments:
- type: float
+ - name: cospif16
+ standards:
+ - stdc
+ return_type: _Float16
+ arguments:
+ - type: _Float16
+ guard: LIBC_TYPES_HAS_FLOAT16
- name: ddivl
standards:
- stdc
diff --git a/libc/src/math/CMakeLists.txt b/libc/src/math/CMakeLists.txt
index 516bed499b1941..3836d6562a0748 100644
--- a/libc/src/math/CMakeLists.txt
+++ b/libc/src/math/CMakeLists.txt
@@ -95,6 +95,7 @@ add_math_entrypoint_object(coshf)
add_math_entrypoint_object(coshf16)
add_math_entrypoint_object(cospif)
+add_math_entrypoint_object(cospif16)
add_math_entrypoint_object(daddl)
add_math_entrypoint_object(daddf128)
diff --git a/libc/src/math/cospif16.h b/libc/src/math/cospif16.h
new file mode 100644
index 00000000000000..6779e67cdccae6
--- /dev/null
+++ b/libc/src/math/cospif16.h
@@ -0,0 +1,21 @@
+//===-- Implementation header for cospif16 ---------------------*- 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_COSPIF16_H
+#define LLVM_LIBC_SRC_MATH_COSPIF16_H
+
+#include "src/__support/macros/config.h"
+#include "src/__support/macros/properties/types.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+float16 cospif16(float16 x);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_SINPIF16_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index d7c7a3431d3d95..e535c9e50b9b86 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -422,6 +422,26 @@ add_entrypoint_object(
-O3
)
+
+add_entrypoint_object(
+ cospif16
+ SRCS
+ cospif16.cpp
+ HDRS
+ ../cospif16.h
+ DEPENDS
+ libc.src.__support.common
+ 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.nearest_integer
+ libc.src.__support.FPUtil.polyeval
+ libc.src.__support.macros.properties.types
+ COMPILE_OPTIONS
+ -O3
+)
+
add_entrypoint_object(
sin
SRCS
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
new file mode 100644
index 00000000000000..779aab781b9f07
--- /dev/null
+++ b/libc/src/math/generic/cospif16.cpp
@@ -0,0 +1,141 @@
+//===-- Half-precision cospif 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/cospif16.h"
+#include "src/__support/FPUtil/FEnvImpl.h"
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/FPUtil/PolyEval.h"
+#include "src/__support/FPUtil/cast.h"
+#include "src/__support/FPUtil/multiply_add.h"
+#include "src/__support/FPUtil/nearest_integer.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+// Lookup table for sin(k * pi / 32) with k = 0, ..., 63.
+// Table is generated with Sollya as follows:
+// > display = hexadecimal;
+// > for k from 0 to 63 do { round(sin(k * pi/32), SG, RN); };
+static constexpr float SIN_K_PI_OVER_32[64] = {
+ 0x0.0p0, 0x1.917a6cp-4, 0x1.8f8b84p-3, 0x1.294062p-2,
+ 0x1.87de2ap-2, 0x1.e2b5d4p-2, 0x1.1c73b4p-1, 0x1.44cf32p-1,
+ 0x1.6a09e6p-1, 0x1.8bc806p-1, 0x1.a9b662p-1, 0x1.c38b3p-1,
+ 0x1.d906bcp-1, 0x1.e9f416p-1, 0x1.f6297cp-1, 0x1.fd88dap-1,
+ 0x1p0, 0x1.fd88dap-1, 0x1.f6297cp-1, 0x1.e9f416p-1,
+ 0x1.d906bcp-1, 0x1.c38b3p-1, 0x1.a9b662p-1, 0x1.8bc806p-1,
+ 0x1.6a09e6p-1, 0x1.44cf32p-1, 0x1.1c73b4p-1, 0x1.e2b5d4p-2,
+ 0x1.87de2ap-2, 0x1.294062p-2, 0x1.8f8b84p-3, 0x1.917a6cp-4,
+ 0x0.0p0, -0x1.917a6cp-4, -0x1.8f8b84p-3, -0x1.294062p-2,
+ -0x1.87de2ap-2, -0x1.e2b5d4p-2, -0x1.1c73b4p-1, -0x1.44cf32p-1,
+ -0x1.6a09e6p-1, -0x1.8bc806p-1, -0x1.a9b662p-1, -0x1.c38b3p-1,
+ -0x1.d906bcp-1, -0x1.e9f416p-1, -0x1.f6297ep-1, -0x1.fd88dap-1,
+ -0x1p0, -0x1.fd88dap-1, -0x1.f6297cp-1, -0x1.e9f416p-1,
+ -0x1.d906bcp-1, -0x1.c38b3p-1, -0x1.a9b662p-1, -0x1.8bc806p-1,
+ -0x1.6a09e6p-1, -0x1.44cf32p-1, -0x1.1c73b4p-1, -0x1.e2b5d4p-2,
+ -0x1.87de2ap-2, -0x1.294062p-2, -0x1.8f8b84p-3, -0x1.917a6cp-4};
+
+static LIBC_INLINE int32_t range_reduction(float x, float &y) {
+ float kf = fputil::nearest_integer(x * 32);
+ y = fputil::multiply_add<float>(x, 32.0, -kf);
+
+ return static_cast<int32_t>(kf);
+}
+
+LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
+ using FPBits = typename fputil::FPBits<float16>;
+ FPBits xbits(x);
+
+ uint16_t x_u = xbits.uintval();
+ uint16_t x_abs = x_u & 0x7fff;
+
+ // Range reduction:
+ // For |x| > 1/32, we perform range reduction as follows:
+ // Find k and y such that:
+ // x = (k + y) * 1/32
+ // k is an integer
+ // |y| < 0.5
+ //
+ // This is done by performing:
+ // k = round(x * 32)
+ // y = x * 32 - k
+ //
+ // Once k and y are computed, we then deduce the answer by the sine of sum
+ // formula:
+ // sin(x * pi) = sin((k + y) * pi/32)
+ // = sin(k * pi/32) * cos(y * pi/32) + sin (y * pi/32) * cos (k *
+ // pi/32)
+ // The values of sin(k * pi/32) and cos (k * pi/32) for k = 0...63 are
+ // precomputed and stored using a vector of 64 single precision floats. sin(y
+ // * pi/32) and cos(y * pi/32) are computed using degree-9 chebyshev
+ // polynomials generated by Sollya.
+
+ // For signed zeros
+ if (LIBC_UNLIKELY(x_abs == 0U)) return fputil::cast<float16>(1.0f);
+
+ // Numbers greater or equal to 2^10 are integers, or infinity, or NaN
+ if (LIBC_UNLIKELY(x_abs >= 0x6400)) {
+ if (LIBC_UNLIKELY(x_abs <= 0x67FF)) {
+ return fputil::cast<float16>((x_abs & 0x1) ? -1.0f : 1.0f);
+ }
+
+ // Check for NaN or infintiy values
+ if (LIBC_UNLIKELY(x_abs >= 0x7c00)) {
+ // If value is equal to infinity
+ if (x_abs == 0x7c00) {
+ fputil::set_errno_if_required(EDOM);
+ fputil::raise_except_if_required(FE_INVALID);
+ }
+
+ return x + FPBits::quiet_nan().get_val();
+ }
+
+ return fputil::cast<float16>(1.0f);
+ }
+
+
+ float f32 = x;
+ float y;
+ int32_t k = range_reduction(f32, y);
+
+ float sin_k = SIN_K_PI_OVER_32[k & 63];
+ float cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
+
+ // Recall;
+ // cos(x * pi/32) = cos((k + y) * pi/32)
+ // = cos(y * pi/32) * cos(k * pi/32)
+ // - sin(y * pi/32) * sin(k * pi/32)
+ // Recall, after range reduction, -0.5 <= y <= 0.5. For very small
+ // values of y, calculating sin(y * p/32) can be inaccurate. Generating a
+ // polynomial for sin(y * p/32)/y instead significantly reduces the relative
+ // errors.
+ float ysq = y * y;
+
+ // Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
+ // with:
+ // > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
+ float sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
+ 0x1.a03354p-21f, -0x1.ad02d2p-20f);
+
+ // Note that cosm1_y = cos(y*pi/32) - 1 = cos_y - 1
+ // Derivation:
+ // cos(x * pi) = cos((k + y) * pi/32)
+ // = cos_k * cos_y + sin_k * sin_y
+ // = cos_k * (1 + cos_y - 1) + sin_k * sin_y
+ // Degree-6 minimax even polynomial for cos(y*pi/32) generated by Sollya with:
+ // > P = fpminimax(cos(y*pi/32), [|0, 2, 4, 6|],[|1, SG...|], [0, 0.5]);
+ float cosm1_y = ysq * fputil::polyeval(ysq, -0x1.3bd3ccp-8f, 0x1.03a61ap-18f,
+ 0x1.a6f7a2p-29f);
+
+ if (LIBC_UNLIKELY(sin_y == 0 && cos_k == 0))
+ return fputil::cast<float16>(0.0f);
+
+ // Since, cosm1_y = cos_y - 1, therefore:
+ // cos(x * pi) = cos_k(cosm1_y) + cos_k - sin_k * sin_y
+ return fputil::cast<float16>(fputil::multiply_add(cos_k, cosm1_y, fputil::multiply_add(-sin_k, sin_y, cos_k)));
+}
+} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/math/CMakeLists.txt b/libc/test/src/math/CMakeLists.txt
index 24a5abec898a89..c0209a13287028 100644
--- a/libc/test/src/math/CMakeLists.txt
+++ b/libc/test/src/math/CMakeLists.txt
@@ -45,6 +45,17 @@ add_fp_unittest(
)
+add_fp_unittest(
+ cospif16_test
+ NEED_MPFR
+ SUITE
+ libc-math-unittests
+ SRCS
+ cospif16_test.cpp
+ DEPENDS
+ libc.src.math.cospif16
+)
+
add_fp_unittest(
daddl_test
NEED_MPFR
diff --git a/libc/test/src/math/cospif16_test.cpp b/libc/test/src/math/cospif16_test.cpp
new file mode 100644
index 00000000000000..1ba70c31cc1519
--- /dev/null
+++ b/libc/test/src/math/cospif16_test.cpp
@@ -0,0 +1,38 @@
+//===-- Exhaustive test for cospif16 --------------------------------------===//
+//
+// 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/cospif16.h"
+#include "test/UnitTest/FPMatcher.h"
+#include "test/UnitTest/Test.h"
+#include "utils/MPFRWrapper/MPFRUtils.h"
+
+using LlvmLibcCospif16Test = LIBC_NAMESPACE::testing::FPTest<float16>;
+
+namespace mpfr = LIBC_NAMESPACE::testing::mpfr;
+
+// Range: [0, Inf]
+static constexpr uint16_t POS_START = 0x0000U;
+static constexpr uint16_t POS_STOP = 0x7c00U;
+
+// Range: [-Inf, 0]
+static constexpr uint16_t NEG_START = 0x8000U;
+static constexpr uint16_t NEG_STOP = 0xfc00U;
+
+TEST_F(LlvmLibcCospif16Test, PositiveRange) {
+ for (uint16_t v = POS_START; v <= POS_STOP; ++v) {
+ float16 x = FPBits(v).get_val();
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x, LIBC_NAMESPACE::cospif16(x), 0.5);
+ }
+}
+
+TEST_F(LlvmLibcCospif16Test, NegativeRange) {
+ for (uint16_t v = NEG_START; v <= NEG_STOP; ++v) {
+ float16 x = FPBits(v).get_val();
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x, LIBC_NAMESPACE::cospif16(x), 0.5);
+ }
+}
diff --git a/libc/test/src/math/smoke/CMakeLists.txt b/libc/test/src/math/smoke/CMakeLists.txt
index 3c077240356bd3..8bba7a083da4d7 100644
--- a/libc/test/src/math/smoke/CMakeLists.txt
+++ b/libc/test/src/math/smoke/CMakeLists.txt
@@ -25,6 +25,17 @@ add_fp_unittest(
libc.src.__support.FPUtil.fp_bits
)
+add_fp_unittest(
+ cospif16_test
+ SUITE
+ libc-math-smoke-tests
+ SRCS
+ cospif16_test.cpp
+ DEPENDS
+ libc.src.errno.errno
+ libc.src.math.cospif16
+)
+
add_fp_unittest(
sinf_test
SUITE
diff --git a/libc/test/src/math/smoke/cospif16_test.cpp b/libc/test/src/math/smoke/cospif16_test.cpp
new file mode 100644
index 00000000000000..7daa5d2956c10f
--- /dev/null
+++ b/libc/test/src/math/smoke/cospif16_test.cpp
@@ -0,0 +1,44 @@
+//===-- Unittests for sinpif16 --------------------------------------------===//
+//
+// 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/cospif16.h"
+#include "test/UnitTest/FPMatcher.h"
+#include "test/UnitTest/Test.h"
+
+using LlvmLibcCospif16Test = LIBC_NAMESPACE::testing::FPTest<float16>;
+
+TEST_F(LlvmLibcCospif16Test, SpecialNumbers) {
+ LIBC_NAMESPACE::libc_errno = 0;
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::cospif16(aNaN));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(1.0f, LIBC_NAMESPACE::cospif16(zero));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(1.0f, LIBC_NAMESPACE::cospif16(neg_zero));
+ EXPECT_MATH_ERRNO(0);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::cospif16(inf));
+ EXPECT_MATH_ERRNO(EDOM);
+
+ EXPECT_FP_EQ(aNaN, LIBC_NAMESPACE::cospif16(neg_inf));
+ EXPECT_MATH_ERRNO(EDOM);
+}
+
+TEST_F(LlvmLibcCospif16Test, Integers) {
+ EXPECT_FP_EQ(1.0f, LIBC_NAMESPACE::cospif16(-0x420));
+ EXPECT_FP_EQ(1.0f, LIBC_NAMESPACE::cospif16(-0x1.4p+14));
+ EXPECT_FP_EQ(-1.0f, LIBC_NAMESPACE::cospif16(0x421));
+ EXPECT_FP_EQ(-1.0f, LIBC_NAMESPACE::cospif16(0x333));
+ EXPECT_FP_EQ(zero, LIBC_NAMESPACE::cospif16(-0x1.28p4));
+ EXPECT_FP_EQ(zero, LIBC_NAMESPACE::cospif16(-0x1.ffcp9));
+ EXPECT_FP_EQ(zero, LIBC_NAMESPACE::cospif16(0x1.01p7));
+ EXPECT_FP_EQ(zero, LIBC_NAMESPACE::cospif16(0x1.f6cp9));
+}
>From b6e0457809c4ce8033300f706802e3c79cbe8237 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 00:20:13 +0100
Subject: [PATCH 02/11] clang format
---
libc/docs/math/index.rst | 2 +-
libc/src/math/generic/cospif16.cpp | 23 ++++++++++++-----------
libc/test/src/math/cospif16_test.cpp | 6 ++++--
3 files changed, 17 insertions(+), 14 deletions(-)
diff --git a/libc/docs/math/index.rst b/libc/docs/math/index.rst
index 6591cbbdc15584..84ed0e4135eba6 100644
--- a/libc/docs/math/index.rst
+++ b/libc/docs/math/index.rst
@@ -280,7 +280,7 @@ Higher Math Functions
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
| cosh | |check| | | | |check| | | 7.12.5.4 | F.10.2.4 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
-| cospi | |check| | | | | | 7.12.4.12 | F.10.1.12 |
+| cospi | |check| | | | |check| | | 7.12.4.12 | F.10.1.12 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
| dsqrt | N/A | N/A | |check| | N/A | |check|\* | 7.12.14.6 | F.10.11 |
+-----------+------------------+-----------------+------------------------+----------------------+------------------------+------------------------+----------------------------+
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
index 779aab781b9f07..90deb4a77dad90 100644
--- a/libc/src/math/generic/cospif16.cpp
+++ b/libc/src/math/generic/cospif16.cpp
@@ -75,28 +75,28 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
// polynomials generated by Sollya.
// For signed zeros
- if (LIBC_UNLIKELY(x_abs == 0U)) return fputil::cast<float16>(1.0f);
+ if (LIBC_UNLIKELY(x_abs == 0U))
+ return fputil::cast<float16>(1.0f);
// Numbers greater or equal to 2^10 are integers, or infinity, or NaN
if (LIBC_UNLIKELY(x_abs >= 0x6400)) {
if (LIBC_UNLIKELY(x_abs <= 0x67FF)) {
- return fputil::cast<float16>((x_abs & 0x1) ? -1.0f : 1.0f);
+ return fputil::cast<float16>((x_abs & 0x1) ? -1.0f : 1.0f);
}
-
+
// Check for NaN or infintiy values
if (LIBC_UNLIKELY(x_abs >= 0x7c00)) {
// If value is equal to infinity
if (x_abs == 0x7c00) {
fputil::set_errno_if_required(EDOM);
- fputil::raise_except_if_required(FE_INVALID);
+ fputil::raise_except_if_required(FE_INVALID);
}
return x + FPBits::quiet_nan().get_val();
}
-
- return fputil::cast<float16>(1.0f);
+
+ return fputil::cast<float16>(1.0f);
}
-
float f32 = x;
float y;
@@ -107,8 +107,8 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
// Recall;
// cos(x * pi/32) = cos((k + y) * pi/32)
- // = cos(y * pi/32) * cos(k * pi/32)
- // - sin(y * pi/32) * sin(k * pi/32)
+ // = cos(y * pi/32) * cos(k * pi/32)
+ // - sin(y * pi/32) * sin(k * pi/32)
// Recall, after range reduction, -0.5 <= y <= 0.5. For very small
// values of y, calculating sin(y * p/32) can be inaccurate. Generating a
// polynomial for sin(y * p/32)/y instead significantly reduces the relative
@@ -116,7 +116,7 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
float ysq = y * y;
// Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
- // with:
+ // with:
// > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
float sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
0x1.a03354p-21f, -0x1.ad02d2p-20f);
@@ -136,6 +136,7 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
// Since, cosm1_y = cos_y - 1, therefore:
// cos(x * pi) = cos_k(cosm1_y) + cos_k - sin_k * sin_y
- return fputil::cast<float16>(fputil::multiply_add(cos_k, cosm1_y, fputil::multiply_add(-sin_k, sin_y, cos_k)));
+ return fputil::cast<float16>(fputil::multiply_add(
+ cos_k, cosm1_y, fputil::multiply_add(-sin_k, sin_y, cos_k)));
}
} // namespace LIBC_NAMESPACE_DECL
diff --git a/libc/test/src/math/cospif16_test.cpp b/libc/test/src/math/cospif16_test.cpp
index 1ba70c31cc1519..2c61086ef8735b 100644
--- a/libc/test/src/math/cospif16_test.cpp
+++ b/libc/test/src/math/cospif16_test.cpp
@@ -26,13 +26,15 @@ static constexpr uint16_t NEG_STOP = 0xfc00U;
TEST_F(LlvmLibcCospif16Test, PositiveRange) {
for (uint16_t v = POS_START; v <= POS_STOP; ++v) {
float16 x = FPBits(v).get_val();
- EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x, LIBC_NAMESPACE::cospif16(x), 0.5);
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x,
+ LIBC_NAMESPACE::cospif16(x), 0.5);
}
}
TEST_F(LlvmLibcCospif16Test, NegativeRange) {
for (uint16_t v = NEG_START; v <= NEG_STOP; ++v) {
float16 x = FPBits(v).get_val();
- EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x, LIBC_NAMESPACE::cospif16(x), 0.5);
+ EXPECT_MPFR_MATCH_ALL_ROUNDING(mpfr::Operation::Cospi, x,
+ LIBC_NAMESPACE::cospif16(x), 0.5);
}
}
>From 9a1e08becb07fc6a352c165bebc29825318e60fc Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 00:51:49 +0100
Subject: [PATCH 03/11] doc update
---
libc/src/math/generic/cospif16.cpp | 6 +++---
libc/test/src/math/smoke/cospif16_test.cpp | 2 +-
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
index 90deb4a77dad90..662f1101862b91 100644
--- a/libc/src/math/generic/cospif16.cpp
+++ b/libc/src/math/generic/cospif16.cpp
@@ -66,9 +66,9 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
//
// Once k and y are computed, we then deduce the answer by the sine of sum
// formula:
- // sin(x * pi) = sin((k + y) * pi/32)
- // = sin(k * pi/32) * cos(y * pi/32) + sin (y * pi/32) * cos (k *
- // pi/32)
+ // cos(x * pi) = cos((k + y) * pi/32)
+ // = cos(k * pi/32) * cos(y * pi/32)
+ // + sin(y * pi/32) * sin(k * pi/32)
// The values of sin(k * pi/32) and cos (k * pi/32) for k = 0...63 are
// precomputed and stored using a vector of 64 single precision floats. sin(y
// * pi/32) and cos(y * pi/32) are computed using degree-9 chebyshev
diff --git a/libc/test/src/math/smoke/cospif16_test.cpp b/libc/test/src/math/smoke/cospif16_test.cpp
index 7daa5d2956c10f..f6d7483393191f 100644
--- a/libc/test/src/math/smoke/cospif16_test.cpp
+++ b/libc/test/src/math/smoke/cospif16_test.cpp
@@ -1,4 +1,4 @@
-//===-- Unittests for sinpif16 --------------------------------------------===//
+//===-- Unittests for cospif16 --------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
>From bf93bb460dde909e5a0b13dc2b72ed17c4cb0989 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 00:53:45 +0100
Subject: [PATCH 04/11] clang-format
---
libc/src/math/generic/cospif16.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
index 662f1101862b91..21c5722d244fa7 100644
--- a/libc/src/math/generic/cospif16.cpp
+++ b/libc/src/math/generic/cospif16.cpp
@@ -67,7 +67,7 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
// Once k and y are computed, we then deduce the answer by the sine of sum
// formula:
// cos(x * pi) = cos((k + y) * pi/32)
- // = cos(k * pi/32) * cos(y * pi/32)
+ // = cos(k * pi/32) * cos(y * pi/32)
// + sin(y * pi/32) * sin(k * pi/32)
// The values of sin(k * pi/32) and cos (k * pi/32) for k = 0...63 are
// precomputed and stored using a vector of 64 single precision floats. sin(y
>From 697397e749fd9e05981cfeda4deb5aedc766cf91 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 08:06:43 +0100
Subject: [PATCH 05/11] added sincosf16_utils header, fixed formatting in
existing pr
---
libc/src/math/cospif16.h | 4 +-
libc/src/math/generic/CMakeLists.txt | 27 ++++----
libc/src/math/generic/cospif16.cpp | 69 ++------------------
libc/src/math/generic/sincosf16_utils.h | 83 +++++++++++++++++++++++++
libc/src/math/generic/sinpif16.cpp | 72 ++-------------------
libc/test/src/math/cospif16_test.cpp | 2 +-
6 files changed, 111 insertions(+), 146 deletions(-)
create mode 100644 libc/src/math/generic/sincosf16_utils.h
diff --git a/libc/src/math/cospif16.h b/libc/src/math/cospif16.h
index 6779e67cdccae6..ef9625dfed45f6 100644
--- a/libc/src/math/cospif16.h
+++ b/libc/src/math/cospif16.h
@@ -1,10 +1,10 @@
-//===-- Implementation header for cospif16 ---------------------*- C++ -*-===//
+//===-- Implementation header for cospif16 ----------------------*- 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_COSPIF16_H
#define LLVM_LIBC_SRC_MATH_COSPIF16_H
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index e535c9e50b9b86..f45c88f1af755f 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -351,6 +351,18 @@ add_header_library(
libc.src.__support.common
)
+add_header_library(
+ sincosf16_utils
+ HDRS
+ sincosf16_utils.h
+ DEPENDS
+ .range_reduction
+ libc.src.__support.FPUtil.fp_bits
+ libc.src.__support.FPUtil.polyeval
+ libc.src.__support.FPUtil.nearest_integer
+ libc.src.__support.common
+)
+
add_header_library(
sincos_eval
HDRS
@@ -422,7 +434,6 @@ add_entrypoint_object(
-O3
)
-
add_entrypoint_object(
cospif16
SRCS
@@ -430,14 +441,10 @@ add_entrypoint_object(
HDRS
../cospif16.h
DEPENDS
- libc.src.__support.common
+ .sincosf16_utils
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.nearest_integer
- libc.src.__support.FPUtil.polyeval
- libc.src.__support.macros.properties.types
+ libc.src.__support.FPUtil.fp_bits
COMPILE_OPTIONS
-O3
)
@@ -555,14 +562,10 @@ add_entrypoint_object(
HDRS
../sinpif16.h
DEPENDS
- libc.src.__support.common
+ .sincosf16_utils
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.nearest_integer
- libc.src.__support.FPUtil.polyeval
- libc.src.__support.macros.properties.types
COMPILE_OPTIONS
-O3
)
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
index 21c5722d244fa7..90d781bb830fe6 100644
--- a/libc/src/math/generic/cospif16.cpp
+++ b/libc/src/math/generic/cospif16.cpp
@@ -7,51 +7,20 @@
//===----------------------------------------------------------------------===//
#include "src/math/cospif16.h"
+#include "sincosf16_utils.h"
#include "src/__support/FPUtil/FEnvImpl.h"
#include "src/__support/FPUtil/FPBits.h"
-#include "src/__support/FPUtil/PolyEval.h"
#include "src/__support/FPUtil/cast.h"
#include "src/__support/FPUtil/multiply_add.h"
-#include "src/__support/FPUtil/nearest_integer.h"
-#include "src/__support/common.h"
-#include "src/__support/macros/config.h"
namespace LIBC_NAMESPACE_DECL {
-// Lookup table for sin(k * pi / 32) with k = 0, ..., 63.
-// Table is generated with Sollya as follows:
-// > display = hexadecimal;
-// > for k from 0 to 63 do { round(sin(k * pi/32), SG, RN); };
-static constexpr float SIN_K_PI_OVER_32[64] = {
- 0x0.0p0, 0x1.917a6cp-4, 0x1.8f8b84p-3, 0x1.294062p-2,
- 0x1.87de2ap-2, 0x1.e2b5d4p-2, 0x1.1c73b4p-1, 0x1.44cf32p-1,
- 0x1.6a09e6p-1, 0x1.8bc806p-1, 0x1.a9b662p-1, 0x1.c38b3p-1,
- 0x1.d906bcp-1, 0x1.e9f416p-1, 0x1.f6297cp-1, 0x1.fd88dap-1,
- 0x1p0, 0x1.fd88dap-1, 0x1.f6297cp-1, 0x1.e9f416p-1,
- 0x1.d906bcp-1, 0x1.c38b3p-1, 0x1.a9b662p-1, 0x1.8bc806p-1,
- 0x1.6a09e6p-1, 0x1.44cf32p-1, 0x1.1c73b4p-1, 0x1.e2b5d4p-2,
- 0x1.87de2ap-2, 0x1.294062p-2, 0x1.8f8b84p-3, 0x1.917a6cp-4,
- 0x0.0p0, -0x1.917a6cp-4, -0x1.8f8b84p-3, -0x1.294062p-2,
- -0x1.87de2ap-2, -0x1.e2b5d4p-2, -0x1.1c73b4p-1, -0x1.44cf32p-1,
- -0x1.6a09e6p-1, -0x1.8bc806p-1, -0x1.a9b662p-1, -0x1.c38b3p-1,
- -0x1.d906bcp-1, -0x1.e9f416p-1, -0x1.f6297ep-1, -0x1.fd88dap-1,
- -0x1p0, -0x1.fd88dap-1, -0x1.f6297cp-1, -0x1.e9f416p-1,
- -0x1.d906bcp-1, -0x1.c38b3p-1, -0x1.a9b662p-1, -0x1.8bc806p-1,
- -0x1.6a09e6p-1, -0x1.44cf32p-1, -0x1.1c73b4p-1, -0x1.e2b5d4p-2,
- -0x1.87de2ap-2, -0x1.294062p-2, -0x1.8f8b84p-3, -0x1.917a6cp-4};
-
-static LIBC_INLINE int32_t range_reduction(float x, float &y) {
- float kf = fputil::nearest_integer(x * 32);
- y = fputil::multiply_add<float>(x, 32.0, -kf);
-
- return static_cast<int32_t>(kf);
-}
-
LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
using FPBits = typename fputil::FPBits<float16>;
FPBits xbits(x);
uint16_t x_u = xbits.uintval();
uint16_t x_abs = x_u & 0x7fff;
+ float xf = x;
// Range reduction:
// For |x| > 1/32, we perform range reduction as follows:
@@ -98,38 +67,8 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
return fputil::cast<float16>(1.0f);
}
- float f32 = x;
- float y;
- int32_t k = range_reduction(f32, y);
-
- float sin_k = SIN_K_PI_OVER_32[k & 63];
- float cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
-
- // Recall;
- // cos(x * pi/32) = cos((k + y) * pi/32)
- // = cos(y * pi/32) * cos(k * pi/32)
- // - sin(y * pi/32) * sin(k * pi/32)
- // Recall, after range reduction, -0.5 <= y <= 0.5. For very small
- // values of y, calculating sin(y * p/32) can be inaccurate. Generating a
- // polynomial for sin(y * p/32)/y instead significantly reduces the relative
- // errors.
- float ysq = y * y;
-
- // Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
- // with:
- // > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
- float sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
- 0x1.a03354p-21f, -0x1.ad02d2p-20f);
-
- // Note that cosm1_y = cos(y*pi/32) - 1 = cos_y - 1
- // Derivation:
- // cos(x * pi) = cos((k + y) * pi/32)
- // = cos_k * cos_y + sin_k * sin_y
- // = cos_k * (1 + cos_y - 1) + sin_k * sin_y
- // Degree-6 minimax even polynomial for cos(y*pi/32) generated by Sollya with:
- // > P = fpminimax(cos(y*pi/32), [|0, 2, 4, 6|],[|1, SG...|], [0, 0.5]);
- float cosm1_y = ysq * fputil::polyeval(ysq, -0x1.3bd3ccp-8f, 0x1.03a61ap-18f,
- 0x1.a6f7a2p-29f);
+ float sin_k, cos_k, sin_y, cosm1_y;
+ sincospif16_eval(xf, sin_k, cos_k, sin_y, cosm1_y);
if (LIBC_UNLIKELY(sin_y == 0 && cos_k == 0))
return fputil::cast<float16>(0.0f);
diff --git a/libc/src/math/generic/sincosf16_utils.h b/libc/src/math/generic/sincosf16_utils.h
new file mode 100644
index 00000000000000..456e399e68f79b
--- /dev/null
+++ b/libc/src/math/generic/sincosf16_utils.h
@@ -0,0 +1,83 @@
+//===-- Collection of utils for sinf16/cosf16 ------------------*- 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_GENERIC_SINCOSF16_UTILS_H
+#define LLVM_LIBC_SRC_MATH_GENERIC_SINCOSF16_UTILS_H
+
+#include "src/__support/FPUtil/FPBits.h"
+#include "src/__support/FPUtil/PolyEval.h"
+#include "src/__support/FPUtil/nearest_integer.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+
+namespace LIBC_NAMESPACE_DECL {
+
+// Lookup table for sin(k * pi / 32) with k = 0, ..., 63.
+// Table is generated with Sollya as follows:
+// > display = hexadecimmal;
+// > for k from 0 to 63 do { round(sin(k * pi/32), SG, RN); };
+constexpr float SIN_K_PI_OVER_32[64] = {
+ 0x0.0p0, 0x1.917a6cp-4, 0x1.8f8b84p-3, 0x1.294062p-2,
+ 0x1.87de2ap-2, 0x1.e2b5d4p-2, 0x1.1c73b4p-1, 0x1.44cf32p-1,
+ 0x1.6a09e6p-1, 0x1.8bc806p-1, 0x1.a9b662p-1, 0x1.c38b3p-1,
+ 0x1.d906bcp-1, 0x1.e9f416p-1, 0x1.f6297cp-1, 0x1.fd88dap-1,
+ 0x1p0, 0x1.fd88dap-1, 0x1.f6297cp-1, 0x1.e9f416p-1,
+ 0x1.d906bcp-1, 0x1.c38b3p-1, 0x1.a9b662p-1, 0x1.8bc806p-1,
+ 0x1.6a09e6p-1, 0x1.44cf32p-1, 0x1.1c73b4p-1, 0x1.e2b5d4p-2,
+ 0x1.87de2ap-2, 0x1.294062p-2, 0x1.8f8b84p-3, 0x1.917a6cp-4,
+ 0x0.0p0, -0x1.917a6cp-4, -0x1.8f8b84p-3, -0x1.294062p-2,
+ -0x1.87de2ap-2, -0x1.e2b5d4p-2, -0x1.1c73b4p-1, -0x1.44cf32p-1,
+ -0x1.6a09e6p-1, -0x1.8bc806p-1, -0x1.a9b662p-1, -0x1.c38b3p-1,
+ -0x1.d906bcp-1, -0x1.e9f416p-1, -0x1.f6297ep-1, -0x1.fd88dap-1,
+ -0x1p0, -0x1.fd88dap-1, -0x1.f6297cp-1, -0x1.e9f416p-1,
+ -0x1.d906bcp-1, -0x1.c38b3p-1, -0x1.a9b662p-1, -0x1.8bc806p-1,
+ -0x1.6a09e6p-1, -0x1.44cf32p-1, -0x1.1c73b4p-1, -0x1.e2b5d4p-2,
+ -0x1.87de2ap-2, -0x1.294062p-2, -0x1.8f8b84p-3, -0x1.917a6cp-4};
+
+LIBC_INLINE int32_t range_reduction(float x, float &y) {
+ float kf = fputil::nearest_integer(x * 32);
+ y = fputil::multiply_add<float>(x, 32.0, -kf);
+
+ return static_cast<int32_t>(kf);
+}
+
+LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k, float &sin_y, float &cosm1_y) {
+ float y;
+ int32_t k = range_reduction(xf, y);
+
+ sin_k = SIN_K_PI_OVER_32[k & 63];
+ cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
+
+ // Recall;
+ // sin(x * pi/32) = sin((k + y) * pi/32)
+ // = sin(y * pi/32) * cos(k * pi/32) + cos(y * pi/32) * sin(k *
+ // pi/32) Recall, after range reduction, -0.5 <= y <= 0.5. For very small
+ // values of y, calculating sin(y * p/32) can be inaccurate. Generating a
+ // polynomial for sin(y * p/32)/y instead significantly reduces the relative
+ // errors.
+ float ysq = y * y;
+
+ // Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
+ // with: > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
+ sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
+ 0x1.a03354p-21f, -0x1.ad02d2p-20f);
+
+ // Note that cosm1_y = cos(y*pi/32) - 1 = cos_y - 1
+ // Derivation:
+ // sin(x * pi) = sin((k + y) * pi/32)
+ // = sin_y * cos_k + cos_y * sin_k
+ // = cos_k * sin_y + sin_k * (1 + cos_y - 1)
+ // Degree-6 minimax even polynomial for cos(y*pi/32) generated by Sollya with:
+ // > P = fpminimax(cos(y*pi/32), [|0, 2, 4, 6|],[|1, SG...|], [0, 0.5]);
+ cosm1_y = ysq * fputil::polyeval(ysq, -0x1.3bd3ccp-8f, 0x1.03a61ap-18f,
+ 0x1.a6f7a2p-29f);
+}
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_MATH_GENERIC_SINCOSF16_UTILS_H
diff --git a/libc/src/math/generic/sinpif16.cpp b/libc/src/math/generic/sinpif16.cpp
index 17cca583e0c0ec..2c52d4cd42688d 100644
--- a/libc/src/math/generic/sinpif16.cpp
+++ b/libc/src/math/generic/sinpif16.cpp
@@ -7,52 +7,20 @@
//===----------------------------------------------------------------------===//
#include "src/math/sinpif16.h"
+#include "sincosf16_utils.h"
#include "src/__support/FPUtil/FEnvImpl.h"
#include "src/__support/FPUtil/FPBits.h"
-#include "src/__support/FPUtil/PolyEval.h"
#include "src/__support/FPUtil/cast.h"
#include "src/__support/FPUtil/multiply_add.h"
-#include "src/__support/FPUtil/nearest_integer.h"
-#include "src/__support/common.h"
-#include "src/__support/macros/config.h"
namespace LIBC_NAMESPACE_DECL {
-
-// Lookup table for sin(k * pi / 32) with k = 0, ..., 63.
-// Table is generated with Sollya as follows:
-// > display = hexadecimmal;
-// > for k from 0 to 63 do { round(sin(k * pi/32), SG, RN); };
-static constexpr float SIN_K_PI_OVER_32[64] = {
- 0x0.0p0, 0x1.917a6cp-4, 0x1.8f8b84p-3, 0x1.294062p-2,
- 0x1.87de2ap-2, 0x1.e2b5d4p-2, 0x1.1c73b4p-1, 0x1.44cf32p-1,
- 0x1.6a09e6p-1, 0x1.8bc806p-1, 0x1.a9b662p-1, 0x1.c38b3p-1,
- 0x1.d906bcp-1, 0x1.e9f416p-1, 0x1.f6297cp-1, 0x1.fd88dap-1,
- 0x1p0, 0x1.fd88dap-1, 0x1.f6297cp-1, 0x1.e9f416p-1,
- 0x1.d906bcp-1, 0x1.c38b3p-1, 0x1.a9b662p-1, 0x1.8bc806p-1,
- 0x1.6a09e6p-1, 0x1.44cf32p-1, 0x1.1c73b4p-1, 0x1.e2b5d4p-2,
- 0x1.87de2ap-2, 0x1.294062p-2, 0x1.8f8b84p-3, 0x1.917a6cp-4,
- 0x0.0p0, -0x1.917a6cp-4, -0x1.8f8b84p-3, -0x1.294062p-2,
- -0x1.87de2ap-2, -0x1.e2b5d4p-2, -0x1.1c73b4p-1, -0x1.44cf32p-1,
- -0x1.6a09e6p-1, -0x1.8bc806p-1, -0x1.a9b662p-1, -0x1.c38b3p-1,
- -0x1.d906bcp-1, -0x1.e9f416p-1, -0x1.f6297ep-1, -0x1.fd88dap-1,
- -0x1p0, -0x1.fd88dap-1, -0x1.f6297cp-1, -0x1.e9f416p-1,
- -0x1.d906bcp-1, -0x1.c38b3p-1, -0x1.a9b662p-1, -0x1.8bc806p-1,
- -0x1.6a09e6p-1, -0x1.44cf32p-1, -0x1.1c73b4p-1, -0x1.e2b5d4p-2,
- -0x1.87de2ap-2, -0x1.294062p-2, -0x1.8f8b84p-3, -0x1.917a6cp-4};
-
-static LIBC_INLINE int32_t range_reduction(float x, float &y) {
- float kf = fputil::nearest_integer(x * 32);
- y = fputil::multiply_add<float>(x, 32.0, -kf);
-
- return static_cast<int32_t>(kf);
-}
-
LLVM_LIBC_FUNCTION(float16, sinpif16, (float16 x)) {
using FPBits = typename fputil::FPBits<float16>;
FPBits xbits(x);
uint16_t x_u = xbits.uintval();
uint16_t x_abs = x_u & 0x7fff;
+ float xf = x;
// Range reduction:
// For |x| > 1/32, we perform range reduction as follows:
@@ -68,8 +36,8 @@ LLVM_LIBC_FUNCTION(float16, sinpif16, (float16 x)) {
// Once k and y are computed, we then deduce the answer by the sine of sum
// formula:
// sin(x * pi) = sin((k + y) * pi/32)
- // = sin(k * pi/32) * cos(y * pi/32) + sin (y * pi/32) * cos (k *
- // pi/32)
+ // = sin(k * pi/32) * cos(y * pi/32)
+ // + sin (y * pi/32) * cos (k * pi/32)
// The values of sin(k * pi/32) and cos (k * pi/32) for k = 0...63 are
// precomputed and stored using a vector of 64 single precision floats. sin(y
// * pi/32) and cos(y * pi/32) are computed using degree-9 chebyshev
@@ -94,36 +62,8 @@ LLVM_LIBC_FUNCTION(float16, sinpif16, (float16 x)) {
return FPBits::zero(xbits.sign()).get_val();
}
- float f32 = x;
- float y;
- int32_t k = range_reduction(f32, y);
-
- float sin_k = SIN_K_PI_OVER_32[k & 63];
- float cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
-
- // Recall;
- // sin(x * pi/32) = sin((k + y) * pi/32)
- // = sin(y * pi/32) * cos(k * pi/32) + cos(y * pi/32) * sin(k *
- // pi/32) Recall, after range reduction, -0.5 <= y <= 0.5. For very small
- // values of y, calculating sin(y * p/32) can be inaccurate. Generating a
- // polynomial for sin(y * p/32)/y instead significantly reduces the relative
- // errors.
- float ysq = y * y;
-
- // Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
- // with: > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
- float sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
- 0x1.a03354p-21f, -0x1.ad02d2p-20f);
-
- // Note that cosm1_y = cos(y*pi/32) - 1 = cos_y - 1
- // Derivation:
- // sin(x * pi) = sin((k + y) * pi/32)
- // = sin_y * cos_k + cos_y * sin_k
- // = cos_k * sin_y + sin_k * (1 + cos_y - 1)
- // Degree-6 minimax even polynomial for cos(y*pi/32) generated by Sollya with:
- // > P = fpminimax(cos(y*pi/32), [|0, 2, 4, 6|],[|1, SG...|], [0, 0.5]);
- float cosm1_y = ysq * fputil::polyeval(ysq, -0x1.3bd3ccp-8f, 0x1.03a61ap-18f,
- 0x1.a6f7a2p-29f);
+ float sin_k, cos_k, sin_y, cosm1_y;
+ sincospif16_eval(xf, sin_k, cos_k, sin_y, cosm1_y);
if (LIBC_UNLIKELY(sin_y == 0 && sin_k == 0))
return FPBits::zero(xbits.sign()).get_val();
diff --git a/libc/test/src/math/cospif16_test.cpp b/libc/test/src/math/cospif16_test.cpp
index 2c61086ef8735b..6a32498b0570ac 100644
--- a/libc/test/src/math/cospif16_test.cpp
+++ b/libc/test/src/math/cospif16_test.cpp
@@ -4,7 +4,7 @@
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
-//===---------------------------------------------------------------------===//
+//===----------------------------------------------------------------------===//
#include "src/math/cospif16.h"
#include "test/UnitTest/FPMatcher.h"
>From a7f9f81f11f3c271b7151d2c0955bdb7744222f6 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 08:11:55 +0100
Subject: [PATCH 06/11] formatting
---
libc/src/math/generic/sincosf16_utils.h | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/libc/src/math/generic/sincosf16_utils.h b/libc/src/math/generic/sincosf16_utils.h
index 456e399e68f79b..89f0d931fcf203 100644
--- a/libc/src/math/generic/sincosf16_utils.h
+++ b/libc/src/math/generic/sincosf16_utils.h
@@ -1,10 +1,10 @@
-//===-- Collection of utils for sinf16/cosf16 ------------------*- C++ -*-===//
+//===-- Collection of utils for sinf16/cosf16 -------------------*- 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_GENERIC_SINCOSF16_UTILS_H
#define LLVM_LIBC_SRC_MATH_GENERIC_SINCOSF16_UTILS_H
@@ -14,7 +14,6 @@
#include "src/__support/common.h"
#include "src/__support/macros/config.h"
-
namespace LIBC_NAMESPACE_DECL {
// Lookup table for sin(k * pi / 32) with k = 0, ..., 63.
@@ -46,10 +45,11 @@ LIBC_INLINE int32_t range_reduction(float x, float &y) {
return static_cast<int32_t>(kf);
}
-LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k, float &sin_y, float &cosm1_y) {
+LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k,
+ float &sin_y, float &cosm1_y) {
float y;
int32_t k = range_reduction(xf, y);
-
+
sin_k = SIN_K_PI_OVER_32[k & 63];
cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
@@ -65,7 +65,7 @@ LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k, float &s
// Degree-6 minimax even polynomial for sin(y*pi/32)/y generated by Sollya
// with: > Q = fpminimax(sin(y*pi/32)/y, [|0, 2, 4, 6|], [|SG...|], [0, 0.5]);
sin_y = y * fputil::polyeval(ysq, 0x1.921fb6p-4f, -0x1.4aeabcp-13f,
- 0x1.a03354p-21f, -0x1.ad02d2p-20f);
+ 0x1.a03354p-21f, -0x1.ad02d2p-20f);
// Note that cosm1_y = cos(y*pi/32) - 1 = cos_y - 1
// Derivation:
@@ -75,7 +75,7 @@ LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k, float &s
// Degree-6 minimax even polynomial for cos(y*pi/32) generated by Sollya with:
// > P = fpminimax(cos(y*pi/32), [|0, 2, 4, 6|],[|1, SG...|], [0, 0.5]);
cosm1_y = ysq * fputil::polyeval(ysq, -0x1.3bd3ccp-8f, 0x1.03a61ap-18f,
- 0x1.a6f7a2p-29f);
+ 0x1.a6f7a2p-29f);
}
} // namespace LIBC_NAMESPACE_DECL
>From d63cf622540135ebcd30b970d4bf34c8c3241cf5 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 09:13:58 +0100
Subject: [PATCH 07/11] fixed mpfr_cospi() evaluation for MPFR version < 4.2
---
libc/utils/MPFRWrapper/MPFRUtils.cpp | 18 ++++--------------
1 file changed, 4 insertions(+), 14 deletions(-)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index bd4fbe294a622d..9295280f6b7d07 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -255,19 +255,9 @@ class MPFRNumber {
mpfr_cospi(result.value, value, mpfr_rounding);
return result;
#else
- MPFRNumber value_frac(*this);
- mpfr_frac(value_frac.value, value, MPFR_RNDN);
-
- if (mpfr_cmp_si(value_frac.value, 0.0) == 0) {
- mpz_t integer_part;
- mpz_init(integer_part);
- mpfr_get_z(integer_part, value, MPFR_RNDN);
-
- if (mpz_tstbit(integer_part, 0)) {
- mpfr_set_si(result.value, -1.0, MPFR_RNDN); // odd
- } else {
- mpfr_set_si(result.value, 1.0, MPFR_RNDN); // even
- }
+ if (mpfr_integer_p(value)) {
+ auto d = mpfr_get_si(value, mpfr_rounding);
+ mpfr_set_si(result.value, (d & 1) ? -1 : 1, mpfr_rounding);
return result;
}
@@ -277,7 +267,7 @@ class MPFRNumber {
mpfr_cos(result.value, value_pi.value, mpfr_rounding);
return result;
-#endif
+//#endif
}
MPFRNumber erf() const {
>From 5d620c2ecbf4e29f5198e21abbb21165d5cb9c2a Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 09:41:01 +0100
Subject: [PATCH 08/11] minor fix
---
libc/utils/MPFRWrapper/MPFRUtils.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index 9295280f6b7d07..92cac8c7ffe3c3 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -267,7 +267,7 @@ class MPFRNumber {
mpfr_cos(result.value, value_pi.value, mpfr_rounding);
return result;
-//#endif
+#endif
}
MPFRNumber erf() const {
>From 6b428aab2ac6046074787b712d040fb80cf6a060 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 10:25:05 +0100
Subject: [PATCH 09/11] improved fix for mpfr_cospi() implementation for when
MPFR version < 4.2
---
libc/utils/MPFRWrapper/MPFRUtils.cpp | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index 92cac8c7ffe3c3..3538416698f8a0 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -256,8 +256,11 @@ class MPFRNumber {
return result;
#else
if (mpfr_integer_p(value)) {
- auto d = mpfr_get_si(value, mpfr_rounding);
- mpfr_set_si(result.value, (d & 1) ? -1 : 1, mpfr_rounding);
+ mpz_t integer;
+ mpz_init(integer);
+ mpfr_get_z(integer, value, mpfr_rounding);
+
+ mpfr_set_si(result.value, (mpz_tstbit(integer, 0)) ? -1 : 1, mpfr_rounding);
return result;
}
>From b606662b90951a2fc8e8db1d796ac4cd560e54dc Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sat, 19 Oct 2024 10:54:11 +0100
Subject: [PATCH 10/11] clang format...again
---
libc/utils/MPFRWrapper/MPFRUtils.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libc/utils/MPFRWrapper/MPFRUtils.cpp b/libc/utils/MPFRWrapper/MPFRUtils.cpp
index 3538416698f8a0..d9cfd006707467 100644
--- a/libc/utils/MPFRWrapper/MPFRUtils.cpp
+++ b/libc/utils/MPFRWrapper/MPFRUtils.cpp
@@ -260,7 +260,8 @@ class MPFRNumber {
mpz_init(integer);
mpfr_get_z(integer, value, mpfr_rounding);
- mpfr_set_si(result.value, (mpz_tstbit(integer, 0)) ? -1 : 1, mpfr_rounding);
+ auto d = mpz_tstbit(integer, 0);
+ mpfr_set_si(result.value, d ? -1 : 1, mpfr_rounding);
return result;
}
>From c45bf585f062fd5ad75c233e0a7ac165074205e3 Mon Sep 17 00:00:00 2001
From: wldfngrs <wldfngrs at gmail.com>
Date: Sun, 20 Oct 2024 14:36:31 +0100
Subject: [PATCH 11/11] nit
---
libc/src/math/generic/CMakeLists.txt | 1 -
libc/src/math/generic/cospif16.cpp | 7 +++++--
libc/src/math/generic/sincosf16_utils.h | 5 +++--
3 files changed, 8 insertions(+), 5 deletions(-)
diff --git a/libc/src/math/generic/CMakeLists.txt b/libc/src/math/generic/CMakeLists.txt
index f45c88f1af755f..145c0708796600 100644
--- a/libc/src/math/generic/CMakeLists.txt
+++ b/libc/src/math/generic/CMakeLists.txt
@@ -356,7 +356,6 @@ add_header_library(
HDRS
sincosf16_utils.h
DEPENDS
- .range_reduction
libc.src.__support.FPUtil.fp_bits
libc.src.__support.FPUtil.polyeval
libc.src.__support.FPUtil.nearest_integer
diff --git a/libc/src/math/generic/cospif16.cpp b/libc/src/math/generic/cospif16.cpp
index 90d781bb830fe6..eda1c20f8c17cc 100644
--- a/libc/src/math/generic/cospif16.cpp
+++ b/libc/src/math/generic/cospif16.cpp
@@ -7,13 +7,17 @@
//===----------------------------------------------------------------------===//
#include "src/math/cospif16.h"
+#include "hdr/errno_macros.h"
+#include "hdr/fenv_macros.h"
#include "sincosf16_utils.h"
#include "src/__support/FPUtil/FEnvImpl.h"
#include "src/__support/FPUtil/FPBits.h"
#include "src/__support/FPUtil/cast.h"
#include "src/__support/FPUtil/multiply_add.h"
+#include "src/__support/macros/optimization.h"
namespace LIBC_NAMESPACE_DECL {
+
LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
using FPBits = typename fputil::FPBits<float16>;
FPBits xbits(x);
@@ -49,9 +53,8 @@ LLVM_LIBC_FUNCTION(float16, cospif16, (float16 x)) {
// Numbers greater or equal to 2^10 are integers, or infinity, or NaN
if (LIBC_UNLIKELY(x_abs >= 0x6400)) {
- if (LIBC_UNLIKELY(x_abs <= 0x67FF)) {
+ if (LIBC_UNLIKELY(x_abs <= 0x67FF))
return fputil::cast<float16>((x_abs & 0x1) ? -1.0f : 1.0f);
- }
// Check for NaN or infintiy values
if (LIBC_UNLIKELY(x_abs >= 0x7c00)) {
diff --git a/libc/src/math/generic/sincosf16_utils.h b/libc/src/math/generic/sincosf16_utils.h
index 89f0d931fcf203..39121f58f20240 100644
--- a/libc/src/math/generic/sincosf16_utils.h
+++ b/libc/src/math/generic/sincosf16_utils.h
@@ -5,6 +5,7 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
+
#ifndef LLVM_LIBC_SRC_MATH_GENERIC_SINCOSF16_UTILS_H
#define LLVM_LIBC_SRC_MATH_GENERIC_SINCOSF16_UTILS_H
@@ -38,7 +39,7 @@ constexpr float SIN_K_PI_OVER_32[64] = {
-0x1.6a09e6p-1, -0x1.44cf32p-1, -0x1.1c73b4p-1, -0x1.e2b5d4p-2,
-0x1.87de2ap-2, -0x1.294062p-2, -0x1.8f8b84p-3, -0x1.917a6cp-4};
-LIBC_INLINE int32_t range_reduction(float x, float &y) {
+LIBC_INLINE int32_t range_reduction_sincospif16(float x, float &y) {
float kf = fputil::nearest_integer(x * 32);
y = fputil::multiply_add<float>(x, 32.0, -kf);
@@ -48,7 +49,7 @@ LIBC_INLINE int32_t range_reduction(float x, float &y) {
LIBC_INLINE void sincospif16_eval(float xf, float &sin_k, float &cos_k,
float &sin_y, float &cosm1_y) {
float y;
- int32_t k = range_reduction(xf, y);
+ int32_t k = range_reduction_sincospif16(xf, y);
sin_k = SIN_K_PI_OVER_32[k & 63];
cos_k = SIN_K_PI_OVER_32[(k + 16) & 63];
More information about the libc-commits
mailing list