diff options
author | Wilco Dijkstra <wdijkstr@arm.com> | 2018-05-16 15:39:22 +0100 |
---|---|---|
committer | Szabolcs Nagy <szabolcs.nagy@arm.com> | 2018-05-16 15:41:34 +0100 |
commit | 269dc16f8bd323c1e16893ed1481494d4d7bac65 (patch) | |
tree | df2676b8025f08b14a685710080e46dd18e6a19e /math/sincosf_data.c | |
parent | 16e2a571214800410693ff1d7443161547567e8e (diff) | |
download | arm-optimized-routines-269dc16f8bd323c1e16893ed1481494d4d7bac65.tar.gz |
Improve performance of sinf/cosf/sincosf
This patch is a complete rewrite of sinf, cosf and sincosf. The new version
is significantly faster, as well as simple and accurate.
The worst-case ULP is 0.56072, maximum relative error is 0.5303p-23 over all
4 billion inputs. In non-nearest rounding modes the error is 1ULP.
The algorithm uses 3 main cases: small inputs which don't need argument
reduction, small inputs which need a simple range reduction and large inputs
requiring complex range reduction. The code uses approximate integer
comparisons to quickly decide between these cases - on some targets this may
be slow, so this can be configured to use floating point comparisons.
The small range reducer uses a single reduction step to handle values up to
120.0. It is fastest on targets which support inlined round instructions.
The large range reducer uses integer arithmetic for simplicity. It does a
32x96 bit multiply to compute a 64-bit modulo result. This is more than
accurate enough to handle the worst-case cancellation for values close to
an integer multiple of PI/4. It could be further optimized, however it is
already much faster than necessary.
Diffstat (limited to 'math/sincosf_data.c')
-rw-r--r-- | math/sincosf_data.c | 79 |
1 files changed, 79 insertions, 0 deletions
diff --git a/math/sincosf_data.c b/math/sincosf_data.c new file mode 100644 index 0000000..ee6aeb9 --- /dev/null +++ b/math/sincosf_data.c @@ -0,0 +1,79 @@ +/* + * Data definition for sinf, cosf and sincosf. + * + * Copyright (c) 2018, Arm Limited. + * SPDX-License-Identifier: Apache-2.0 + * + * 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. + */ + +#if !WANT_SINGLEPREC + +#include <stdint.h> +#include <math.h> +#include "math_config.h" +#include "sincosf.h" + +/* The constants and polynomials for sine and cosine. The 2nd entry + computes -cos (x) rather than cos (x) to get negation for free. */ +sincos_t sincosf_table[2] = +{ + { + { 1.0, -1.0, -1.0, 1.0 }, +#if HAVE_FAST_ROUND + 0x1.45F306DC9C883p-1, +#else + 0x1.45F306DC9C883p+23, +#endif + 0x1.921FB54442D18p0, + 0x1p0, + -0x1.ffffffd0c621cp-2, + 0x1.55553e1068f19p-5, + -0x1.6c087e89a359dp-10, + 0x1.99343027bf8c3p-16, + -0x1.555545995a603p-3, + 0x1.1107605230bc4p-7, + -0x1.994eb3774cf24p-13 + }, + { + { 1.0, -1.0, -1.0, 1.0 }, +#if HAVE_FAST_ROUND + 0x1.45F306DC9C883p-1, +#else + 0x1.45F306DC9C883p+23, +#endif + 0x1.921FB54442D18p0, + -0x1p0, + 0x1.ffffffd0c621cp-2, + -0x1.55553e1068f19p-5, + 0x1.6c087e89a359dp-10, + -0x1.99343027bf8c3p-16, + -0x1.555545995a603p-3, + 0x1.1107605230bc4p-7, + -0x1.994eb3774cf24p-13 + } +}; + +/* Table with 4/PI to 192 bit precision. To avoid unaligned accesses + only 8 new bits are added per entry, making the table 4 times larger. */ +const uint32_t inv_pio4[24] = +{ + 0xa2, 0xa2f9, 0xa2f983, 0xa2f9836e, + 0xf9836e4e, 0x836e4e44, 0x6e4e4415, 0x4e441529, + 0x441529fc, 0x1529fc27, 0x29fc2757, 0xfc2757d1, + 0x2757d1f5, 0x57d1f534, 0xd1f534dd, 0xf534ddc0, + 0x34ddc0db, 0xddc0db62, 0xc0db6295, 0xdb629599, + 0x6295993c, 0x95993c43, 0x993c4390, 0x3c439041 +}; + +#endif |