/* * // Copyright (c) Radzivon Bartoshyk 7/2025. All rights reserved. * // * // Redistribution and use in source and binary forms, with or without modification, * // are permitted provided that the following conditions are met: * // * // 1. Redistributions of source code must retain the above copyright notice, this * // list of conditions and the following disclaimer. * // * // 2. Redistributions in binary form must reproduce the above copyright notice, * // this list of conditions and the following disclaimer in the documentation * // and/or other materials provided with the distribution. * // * // 3. Neither the name of the copyright holder nor the names of its * // contributors may be used to endorse or promote products derived from * // this software without specific prior written permission. * // * // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * // DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE * // FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * // DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * // SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER * // CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * // OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ use crate::common::{f_fmla, f_fmlaf}; use crate::double_double::DoubleDouble; use crate::dyadic_float::DyadicFloat128; use std::ops::Mul; pub(crate) trait PolyevalMla { fn polyeval_mla(a: Self, b: Self, c: Self) -> Self; } impl PolyevalMla for f64 { #[inline(always)] fn polyeval_mla(a: Self, b: Self, c: Self) -> Self { f_fmla(a, b, c) } } impl PolyevalMla for f32 { #[inline(always)] fn polyeval_mla(a: Self, b: Self, c: Self) -> Self { f_fmlaf(a, b, c) } } impl PolyevalMla for DoubleDouble { #[inline(always)] fn polyeval_mla(a: Self, b: Self, c: Self) -> Self { DoubleDouble::mul_add(a, b, c) } } impl PolyevalMla for DyadicFloat128 { #[inline(always)] fn polyeval_mla(a: Self, b: Self, c: Self) -> Self { c.quick_add(&a.quick_mul(&b)) } } // impl PolyevalMla for DyadicFloat256 { // #[inline(always)] // fn polyeval_mla(a: Self, b: Self, c: Self) -> Self { // c.quick_add(&a.quick_mul(&b)) // } // } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval6>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, ) -> T { let x2 = x * x; let u0 = T::polyeval_mla(x, a5, a4); let u1 = T::polyeval_mla(x, a3, a2); let u2 = T::polyeval_mla(x, a1, a0); let v0 = T::polyeval_mla(x2, u0, u1); T::polyeval_mla(x2, v0, u2) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn dd_quick_polyeval6( x: DoubleDouble, a0: DoubleDouble, a1: DoubleDouble, a2: DoubleDouble, a3: DoubleDouble, a4: DoubleDouble, a5: DoubleDouble, ) -> DoubleDouble { let x2 = DoubleDouble::quick_mult(x, x); let u0 = DoubleDouble::quick_mul_add(x, a5, a4); let u1 = DoubleDouble::quick_mul_add(x, a3, a2); let u2 = DoubleDouble::quick_mul_add(x, a1, a0); let v0 = DoubleDouble::quick_mul_add(x2, u0, u1); DoubleDouble::quick_mul_add(x2, v0, u2) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn dd_quick_polyeval6_fma( x: DoubleDouble, a0: DoubleDouble, a1: DoubleDouble, a2: DoubleDouble, a3: DoubleDouble, a4: DoubleDouble, a5: DoubleDouble, ) -> DoubleDouble { let x2 = DoubleDouble::quick_mult_fma(x, x); let u0 = DoubleDouble::quick_mul_add_fma(x, a5, a4); let u1 = DoubleDouble::quick_mul_add_fma(x, a3, a2); let u2 = DoubleDouble::quick_mul_add_fma(x, a1, a0); let v0 = DoubleDouble::quick_mul_add_fma(x2, u0, u1); DoubleDouble::quick_mul_add_fma(x2, v0, u2) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn d_polyeval6(x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64, a5: f64) -> f64 { let x2 = x * x; let u0 = f64::mul_add(x, a5, a4); let u1 = f64::mul_add(x, a3, a2); let u2 = f64::mul_add(x, a1, a0); let v0 = f64::mul_add(x2, u0, u1); f64::mul_add(x2, v0, u2) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval9>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, ) -> T { let mut acc = a8; acc = T::polyeval_mla(x, acc, a7); acc = T::polyeval_mla(x, acc, a6); acc = T::polyeval_mla(x, acc, a5); acc = T::polyeval_mla(x, acc, a4); acc = T::polyeval_mla(x, acc, a3); acc = T::polyeval_mla(x, acc, a2); acc = T::polyeval_mla(x, acc, a1); T::polyeval_mla(x, acc, a0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_estrin_polyeval9>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let p0 = T::polyeval_mla(x, a1, a0); let p1 = T::polyeval_mla(x, a3, a2); let p2 = T::polyeval_mla(x, a5, a4); let p3 = T::polyeval_mla(x, a7, a6); let q0 = T::polyeval_mla(x2, p1, p0); let q1 = T::polyeval_mla(x2, p3, p2); let r0 = T::polyeval_mla(x4, q1, q0); T::polyeval_mla(x8, a8, r0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval10>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let p0 = T::polyeval_mla(x, a1, a0); let p1 = T::polyeval_mla(x, a3, a2); let p2 = T::polyeval_mla(x, a5, a4); let p3 = T::polyeval_mla(x, a7, a6); let p4 = T::polyeval_mla(x, a9, a8); let q0 = T::polyeval_mla(x2, p1, p0); let q1 = T::polyeval_mla(x2, p3, p2); let r0 = T::polyeval_mla(x4, q1, q0); T::polyeval_mla(x8, p4, r0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn dd_quick_polyeval10( x: DoubleDouble, a0: DoubleDouble, a1: DoubleDouble, a2: DoubleDouble, a3: DoubleDouble, a4: DoubleDouble, a5: DoubleDouble, a6: DoubleDouble, a7: DoubleDouble, a8: DoubleDouble, a9: DoubleDouble, ) -> DoubleDouble { let x2 = DoubleDouble::quick_mult(x, x); let x4 = DoubleDouble::quick_mult(x2, x2); let x8 = DoubleDouble::quick_mult(x4, x4); let p0 = DoubleDouble::quick_mul_add(x, a1, a0); let p1 = DoubleDouble::quick_mul_add(x, a3, a2); let p2 = DoubleDouble::quick_mul_add(x, a5, a4); let p3 = DoubleDouble::quick_mul_add(x, a7, a6); let p4 = DoubleDouble::quick_mul_add(x, a9, a8); let q0 = DoubleDouble::quick_mul_add(x2, p1, p0); let q1 = DoubleDouble::quick_mul_add(x2, p3, p2); let r0 = DoubleDouble::quick_mul_add(x4, q1, q0); DoubleDouble::quick_mul_add(x8, p4, r0) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn dd_quick_polyeval10_fma( x: DoubleDouble, a0: DoubleDouble, a1: DoubleDouble, a2: DoubleDouble, a3: DoubleDouble, a4: DoubleDouble, a5: DoubleDouble, a6: DoubleDouble, a7: DoubleDouble, a8: DoubleDouble, a9: DoubleDouble, ) -> DoubleDouble { let x2 = DoubleDouble::quick_mult_fma(x, x); let x4 = DoubleDouble::quick_mult_fma(x2, x2); let x8 = DoubleDouble::quick_mult_fma(x4, x4); let p0 = DoubleDouble::quick_mul_add_fma(x, a1, a0); let p1 = DoubleDouble::quick_mul_add_fma(x, a3, a2); let p2 = DoubleDouble::quick_mul_add_fma(x, a5, a4); let p3 = DoubleDouble::quick_mul_add_fma(x, a7, a6); let p4 = DoubleDouble::quick_mul_add_fma(x, a9, a8); let q0 = DoubleDouble::quick_mul_add_fma(x2, p1, p0); let q1 = DoubleDouble::quick_mul_add_fma(x2, p3, p2); let r0 = DoubleDouble::quick_mul_add_fma(x4, q1, q0); DoubleDouble::quick_mul_add_fma(x8, p4, r0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval11>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let q0 = T::polyeval_mla(x, a1, a0); let q1 = T::polyeval_mla(x, a3, a2); let q2 = T::polyeval_mla(x, a5, a4); let q3 = T::polyeval_mla(x, a7, a6); let q4 = T::polyeval_mla(x, a9, a8); let r0 = T::polyeval_mla(x2, q1, q0); let r1 = T::polyeval_mla(x2, q3, q2); let s0 = T::polyeval_mla(x4, r1, r0); let s1 = T::polyeval_mla(x2, a10, q4); T::polyeval_mla(x8, s1, s0) } #[inline(always)] pub(crate) fn f_polyeval3(x: T, a0: T, a1: T, a2: T) -> T { T::polyeval_mla(x, T::polyeval_mla(x, a2, a1), a0) } #[inline(always)] #[allow(unused)] pub(crate) fn d_polyeval3(x: f64, a0: f64, a1: f64, a2: f64) -> f64 { f64::mul_add(x, f64::mul_add(x, a2, a1), a0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval4(x: T, a0: T, a1: T, a2: T, a3: T) -> T { let t2 = T::polyeval_mla(x, a3, a2); let t5 = T::polyeval_mla(x, t2, a1); T::polyeval_mla(x, t5, a0) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn d_polyeval4(x: f64, a0: f64, a1: f64, a2: f64, a3: f64) -> f64 { let t2 = f64::mul_add(x, a3, a2); let t5 = f64::mul_add(x, t2, a1); f64::mul_add(x, t5, a0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_estrin_polyeval4>( x: T, a0: T, a1: T, a2: T, a3: T, ) -> T { let x2 = x * x; let p01 = T::polyeval_mla(x, a1, a0); let p23 = T::polyeval_mla(x, a3, a2); T::polyeval_mla(x2, p23, p01) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval13>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let t0 = T::polyeval_mla(x, a3, a2); let t1 = T::polyeval_mla(x, a1, a0); let t2 = T::polyeval_mla(x, a7, a6); let t3 = T::polyeval_mla(x, a5, a4); let t4 = T::polyeval_mla(x, a11, a10); let t5 = T::polyeval_mla(x, a9, a8); let q0 = T::polyeval_mla(x2, t0, t1); let q1 = T::polyeval_mla(x2, t2, t3); let q2 = T::polyeval_mla(x2, t4, t5); let q3 = a12; let r0 = T::polyeval_mla(x4, q1, q0); let r1 = T::polyeval_mla(x4, q3, q2); T::polyeval_mla(x8, r1, r0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval12>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let e0 = T::polyeval_mla(x, a1, a0); let e1 = T::polyeval_mla(x, a3, a2); let e2 = T::polyeval_mla(x, a5, a4); let e3 = T::polyeval_mla(x, a7, a6); let e4 = T::polyeval_mla(x, a9, a8); let e5 = T::polyeval_mla(x, a11, a10); let f0 = T::polyeval_mla(x2, e1, e0); let f1 = T::polyeval_mla(x2, e3, e2); let f2 = T::polyeval_mla(x2, e5, e4); let g0 = T::polyeval_mla(x4, f1, f0); T::polyeval_mla(x8, f2, g0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval14>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let g0 = T::polyeval_mla(x, a1, a0); let g1 = T::polyeval_mla(x, a3, a2); let g2 = T::polyeval_mla(x, a5, a4); let g3 = T::polyeval_mla(x, a7, a6); let g4 = T::polyeval_mla(x, a9, a8); let g5 = T::polyeval_mla(x, a11, a10); let g6 = T::polyeval_mla(x, a13, a12); let h0 = T::polyeval_mla(x2, g1, g0); let h1 = T::polyeval_mla(x2, g3, g2); let h2 = T::polyeval_mla(x2, g5, g4); let q0 = T::polyeval_mla(x4, h1, h0); let q1 = T::polyeval_mla(x4, g6, h2); T::polyeval_mla(x8, q1, q0) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn d_polyeval14( x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64, a5: f64, a6: f64, a7: f64, a8: f64, a9: f64, a10: f64, a11: f64, a12: f64, a13: f64, ) -> f64 { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let g0 = f64::mul_add(x, a1, a0); let g1 = f64::mul_add(x, a3, a2); let g2 = f64::mul_add(x, a5, a4); let g3 = f64::mul_add(x, a7, a6); let g4 = f64::mul_add(x, a9, a8); let g5 = f64::mul_add(x, a11, a10); let g6 = f64::mul_add(x, a13, a12); let h0 = f64::mul_add(x2, g1, g0); let h1 = f64::mul_add(x2, g3, g2); let h2 = f64::mul_add(x2, g5, g4); let q0 = f64::mul_add(x4, h1, h0); let q1 = f64::mul_add(x4, g6, h2); f64::mul_add(x8, q1, q0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval7( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, ) -> T { let t1 = T::polyeval_mla(x, a6, a5); let t2 = T::polyeval_mla(x, t1, a4); let t3 = T::polyeval_mla(x, t2, a3); let t4 = T::polyeval_mla(x, t3, a2); let t5 = T::polyeval_mla(x, t4, a1); T::polyeval_mla(x, t5, a0) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn d_polyeval7( x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64, a5: f64, a6: f64, ) -> f64 { let t1 = f64::mul_add(x, a6, a5); let t2 = f64::mul_add(x, t1, a4); let t3 = f64::mul_add(x, t2, a3); let t4 = f64::mul_add(x, t3, a2); let t5 = f64::mul_add(x, t4, a1); f64::mul_add(x, t5, a0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_estrin_polyeval7>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let b0 = T::polyeval_mla(x, a1, a0); let b1 = T::polyeval_mla(x, a3, a2); let b2 = T::polyeval_mla(x, a5, a4); let c0 = T::polyeval_mla(x2, b1, b0); let c1 = T::polyeval_mla(x2, a6, b2); T::polyeval_mla(x4, c1, c0) } #[inline(always)] #[allow(unused)] #[allow(clippy::too_many_arguments)] pub(crate) fn d_estrin_polyeval7( x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64, a5: f64, a6: f64, ) -> f64 { let x2 = x * x; let x4 = x2 * x2; let b0 = f64::mul_add(x, a1, a0); let b1 = f64::mul_add(x, a3, a2); let b2 = f64::mul_add(x, a5, a4); let c0 = f64::mul_add(x2, b1, b0); let c1 = f64::mul_add(x2, a6, b2); f64::mul_add(x4, c1, c0) } #[allow(clippy::too_many_arguments)] #[inline(always)] pub(crate) fn f_polyeval5(x: T, a0: T, a1: T, a2: T, a3: T, a4: T) -> T { let mut acc = a4; acc = T::polyeval_mla(x, acc, a3); acc = T::polyeval_mla(x, acc, a2); acc = T::polyeval_mla(x, acc, a1); T::polyeval_mla(x, acc, a0) } #[allow(clippy::too_many_arguments)] #[inline(always)] #[allow(unused)] pub(crate) fn d_polyeval5(x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64) -> f64 { let mut acc = a4; acc = f64::mul_add(x, acc, a3); acc = f64::mul_add(x, acc, a2); acc = f64::mul_add(x, acc, a1); f64::mul_add(x, acc, a0) } #[allow(clippy::too_many_arguments)] #[inline(always)] pub(crate) fn f_estrin_polyeval5>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, ) -> T { let x2 = x * x; let p01 = T::polyeval_mla(x, a1, a0); let p23 = T::polyeval_mla(x, a3, a2); let t = T::polyeval_mla(x2, a4, p23); T::polyeval_mla(x2, t, p01) } #[allow(clippy::too_many_arguments)] #[inline(always)] #[allow(unused)] pub(crate) fn d_estrin_polyeval5(x: f64, a0: f64, a1: f64, a2: f64, a3: f64, a4: f64) -> f64 { let x2 = x * x; let p01 = f64::mul_add(x, a1, a0); let p23 = f64::mul_add(x, a3, a2); let t = f64::mul_add(x2, a4, p23); f64::mul_add(x2, t, p01) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval8( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, ) -> T { let z0 = T::polyeval_mla(x, a7, a6); let t1 = T::polyeval_mla(x, z0, a5); let t2 = T::polyeval_mla(x, t1, a4); let t3 = T::polyeval_mla(x, t2, a3); let t4 = T::polyeval_mla(x, t3, a2); let t5 = T::polyeval_mla(x, t4, a1); T::polyeval_mla(x, t5, a0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_estrin_polyeval8>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let p0 = T::polyeval_mla(x, a1, a0); let p1 = T::polyeval_mla(x, a3, a2); let p2 = T::polyeval_mla(x, a5, a4); let p3 = T::polyeval_mla(x, a7, a6); let q0 = T::polyeval_mla(x2, p1, p0); let q1 = T::polyeval_mla(x2, p3, p2); T::polyeval_mla(x4, q1, q0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval16>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, a15: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let q0 = T::polyeval_mla(x, a1, a0); let q1 = T::polyeval_mla(x, a3, a2); let q2 = T::polyeval_mla(x, a5, a4); let q3 = T::polyeval_mla(x, a7, a6); let q4 = T::polyeval_mla(x, a9, a8); let q5 = T::polyeval_mla(x, a11, a10); let q6 = T::polyeval_mla(x, a13, a12); let q7 = T::polyeval_mla(x, a15, a14); let r0 = T::polyeval_mla(x2, q1, q0); let r1 = T::polyeval_mla(x2, q3, q2); let r2 = T::polyeval_mla(x2, q5, q4); let r3 = T::polyeval_mla(x2, q7, q6); let s0 = T::polyeval_mla(x4, r1, r0); let s1 = T::polyeval_mla(x4, r3, r2); T::polyeval_mla(x8, s1, s0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval15>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let e0 = T::polyeval_mla(x, a1, a0); let e1 = T::polyeval_mla(x, a3, a2); let e2 = T::polyeval_mla(x, a5, a4); let e3 = T::polyeval_mla(x, a7, a6); let e4 = T::polyeval_mla(x, a9, a8); let e5 = T::polyeval_mla(x, a11, a10); let e6 = T::polyeval_mla(x, a13, a12); // Level 2 let f0 = T::polyeval_mla(x2, e1, e0); let f1 = T::polyeval_mla(x2, e3, e2); let f2 = T::polyeval_mla(x2, e5, e4); let f3 = T::polyeval_mla(x2, a14, e6); // Level 3 let g0 = T::polyeval_mla(x4, f1, f0); let g1 = T::polyeval_mla(x4, f3, f2); // Final T::polyeval_mla(x8, g1, g0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval18>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, a15: T, a16: T, a17: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let x16 = x8 * x8; let q0 = T::polyeval_mla(x, a1, a0); let q1 = T::polyeval_mla(x, a3, a2); let q2 = T::polyeval_mla(x, a5, a4); let q3 = T::polyeval_mla(x, a7, a6); let q4 = T::polyeval_mla(x, a9, a8); let q5 = T::polyeval_mla(x, a11, a10); let q6 = T::polyeval_mla(x, a13, a12); let q7 = T::polyeval_mla(x, a15, a14); let q8 = T::polyeval_mla(x, a17, a16); let r0 = T::polyeval_mla(x2, q1, q0); let r1 = T::polyeval_mla(x2, q3, q2); let r2 = T::polyeval_mla(x2, q5, q4); let r3 = T::polyeval_mla(x2, q7, q6); let s0 = T::polyeval_mla(x4, r1, r0); let s1 = T::polyeval_mla(x4, r3, r2); let t0 = T::polyeval_mla(x8, s1, s0); T::polyeval_mla(x16, q8, t0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval19>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, a15: T, a16: T, a17: T, a18: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let x16 = x8 * x8; // Level 0: pairs let e0 = T::polyeval_mla(x, a1, a0); // a0 + a1·x let e1 = T::polyeval_mla(x, a3, a2); // a2 + a3·x let e2 = T::polyeval_mla(x, a5, a4); let e3 = T::polyeval_mla(x, a7, a6); let e4 = T::polyeval_mla(x, a9, a8); let e5 = T::polyeval_mla(x, a11, a10); let e6 = T::polyeval_mla(x, a13, a12); let e7 = T::polyeval_mla(x, a15, a14); let e8 = T::polyeval_mla(x, a17, a16); // Level 1: combine with x² let f0 = T::polyeval_mla(x2, e1, e0); let f1 = T::polyeval_mla(x2, e3, e2); let f2 = T::polyeval_mla(x2, e5, e4); let f3 = T::polyeval_mla(x2, e7, e6); // Level 2: combine with x⁴ let g0 = T::polyeval_mla(x4, f1, f0); let g1 = T::polyeval_mla(x4, f3, f2); // Level 3: combine with x⁸ let h0 = T::polyeval_mla(x8, g1, g0); // Final: combine with x¹⁶ let final_poly = T::polyeval_mla(x16, e8, h0); // Degree 18: Add a18·x¹⁸ // This assumes `x18 = x16 * x2`, since x² already computed let x18 = x16 * x2; T::polyeval_mla(x18, a18, final_poly) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval22>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, a15: T, a16: T, a17: T, a18: T, a19: T, a20: T, a21: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let x16 = x8 * x8; let p0 = T::polyeval_mla(x, a1, a0); // a1·x + a0 let p1 = T::polyeval_mla(x, a3, a2); // a3·x + a2 let p2 = T::polyeval_mla(x, a5, a4); let p3 = T::polyeval_mla(x, a7, a6); let p4 = T::polyeval_mla(x, a9, a8); let p5 = T::polyeval_mla(x, a11, a10); let p6 = T::polyeval_mla(x, a13, a12); let p7 = T::polyeval_mla(x, a15, a14); let p8 = T::polyeval_mla(x, a17, a16); let p9 = T::polyeval_mla(x, a19, a18); let p10 = T::polyeval_mla(x, a21, a20); let q0 = T::polyeval_mla(x2, p1, p0); // (a3·x + a2)·x² + (a1·x + a0) let q1 = T::polyeval_mla(x2, p3, p2); let q2 = T::polyeval_mla(x2, p5, p4); let q3 = T::polyeval_mla(x2, p7, p6); let q4 = T::polyeval_mla(x2, p9, p8); let r0 = T::polyeval_mla(x4, q1, q0); // q1·x⁴ + q0 let r1 = T::polyeval_mla(x4, q3, q2); let s0 = T::polyeval_mla(x8, r1, r0); // r1·x⁸ + r0 let r2 = T::polyeval_mla(x4, p10, q4); // p10·x⁴ + q4 T::polyeval_mla(x16, r2, s0) } #[inline(always)] #[allow(clippy::too_many_arguments)] pub(crate) fn f_polyeval24>( x: T, a0: T, a1: T, a2: T, a3: T, a4: T, a5: T, a6: T, a7: T, a8: T, a9: T, a10: T, a11: T, a12: T, a13: T, a14: T, a15: T, a16: T, a17: T, a18: T, a19: T, a20: T, a21: T, a22: T, a23: T, ) -> T { let x2 = x * x; let x4 = x2 * x2; let x8 = x4 * x4; let x16 = x8 * x8; // Group degree 0–1 let e0 = T::polyeval_mla(x, a1, a0); // Group degree 2–3 let e1 = T::polyeval_mla(x, a3, a2); // Group degree 4–5 let e2 = T::polyeval_mla(x, a5, a4); // Group degree 6–7 let e3 = T::polyeval_mla(x, a7, a6); // Group degree 8–9 let e4 = T::polyeval_mla(x, a9, a8); // Group degree 10–11 let e5 = T::polyeval_mla(x, a11, a10); // Group degree 12–13 let e6 = T::polyeval_mla(x, a13, a12); // Group degree 14–15 let e7 = T::polyeval_mla(x, a15, a14); // Group degree 16–17 let e8 = T::polyeval_mla(x, a17, a16); // Group degree 18–19 let e9 = T::polyeval_mla(x, a19, a18); // Group degree 20–21 let e10 = T::polyeval_mla(x, a21, a20); // Group degree 22–23 let e11 = T::polyeval_mla(x, a23, a22); // Now group into x2 terms let f0 = T::polyeval_mla(x2, e1, e0); let f1 = T::polyeval_mla(x2, e3, e2); let f2 = T::polyeval_mla(x2, e5, e4); let f3 = T::polyeval_mla(x2, e7, e6); let f4 = T::polyeval_mla(x2, e9, e8); let f5 = T::polyeval_mla(x2, e11, e10); // Now group into x4 terms let g0 = T::polyeval_mla(x4, f1, f0); let g1 = T::polyeval_mla(x4, f3, f2); let g2 = T::polyeval_mla(x4, f5, f4); // Now group into x8 terms let h0 = T::polyeval_mla(x8, g1, g0); let h1 = g2; // Final step (x16 term) T::polyeval_mla(x16, h1, h0) }