23#ifndef __LMMP_INLINES_H__
24#define __LMMP_INLINES_H__
44#if defined(__GNUC__) || defined(__clang__)
46#if defined(LMMP_ASM_X64)
47 __asm__
volatile(
"popcnt %1, %0" :
"=r"(count) :
"r"(x) :
"cc");
49 count = __builtin_popcountll(x);
63 if (x == 0)
return 64;
65 return __builtin_clzll(x);
68 if (x <= 0x00000000FFFFFFFF) {
n += 32; x <<= 32; }
69 if (x <= 0x0000FFFFFFFFFFFF) {
n += 16; x <<= 16; }
70 if (x <= 0x00FFFFFFFFFFFFFF) {
n += 8; x <<= 8; }
71 if (x <= 0x0FFFFFFFFFFFFFFF) {
n += 4; x <<= 4; }
72 if (x <= 0x3FFFFFFFFFFFFFFF) {
n += 2; x <<= 2; }
73 if (x <= 0x7FFFFFFFFFFFFFFF) {
n += 1; x <<= 1; }
79 if (x == 0)
return 64;
81 return __builtin_ctzll(x);
84 if ((x & 0x00000000FFFFFFFF) == 0) {
n += 32; x >>= 32; }
85 if ((x & 0x000000000000FFFF) == 0) {
n += 16; x >>= 16; }
86 if ((x & 0x00000000000000FF) == 0) {
n += 8; x >>= 8; }
87 if ((x & 0x000000000000000F) == 0) {
n += 4; x >>= 4; }
88 if ((x & 0x0000000000000003) == 0) {
n += 2; x >>= 2; }
89 if ((x & 0x0000000000000001) == 0) {
n += 1; x >>= 1; }
95#if (defined(__GNUC__) || defined(__clang__)) && defined(__SIZEOF_INT128__)
96 __uint128_t
prod = (__uint128_t)a * b;
100 uint64_t ah = a >> 32, bh = b >> 32;
101 a = (uint32_t)a, b = (uint32_t)b;
102 uint64_t
r0 = a * b,
r1 = a * bh,
r2 = ah * b,
r3 = ah * bh;
103 r3 += (
r1 >> 32) + (
r2 >> 32);
104 r1 = (uint32_t)
r1,
r2 = (uint32_t)
r2;
107 dst[1] =
r3 + (
r1 >> 32);
108 dst[0] = (
r1 << 32) | (uint32_t)
r0;
113#if (defined(__GNUC__) || defined(__clang__)) && defined(__SIZEOF_INT128__)
114 __uint128_t
t = (__uint128_t)a * (__uint128_t)b;
117 uint64_t ah = a >> 32, bh = b >> 32;
118 a = (uint32_t)a, b = (uint32_t)b;
119 uint64_t
r0 = a * b,
r1 = a * bh,
r2 = ah * b,
r3 = ah * bh;
120 r3 += (
r1 >> 32) + (
r2 >> 32);
121 r1 = (uint32_t)
r1,
r2 = (uint32_t)
r2;
124 return r3 + (
r1 >> 32);
162#define lmmp_leading_zeros_ __lmmp_leading_zeros_
163#define lmmp_tailing_zeros_ __lmmp_tailing_zeros_
164#define lmmp_limb_bits_ __lmmp_limb_bits_
165#define lmmp_limb_popcnt_ __lmmp_limb_popcnt_
166#define lmmp_mullh_ __lmmp_mullh_
167#define lmmp_mulh_ __lmmp_mulh_
168#define lmmp_sqr_ __lmmp_sqr_
169#define lmmp_mul_n_ __lmmp_mul_n_
static int __lmmp_limb_bits_(mp_limb_t x)
Copyright (C) 2026 HJimmyK (Jericho Knox)
static void __lmmp_mul_n_(mp_ptr restrict dst, mp_srcptr restrict numa, mp_srcptr restrict numb, mp_size_t n)
static mp_limb_t __lmmp_mulh_(mp_limb_t a, mp_limb_t b)
static int __lmmp_tailing_zeros_(mp_limb_t x)
static int __lmmp_limb_popcnt_(mp_limb_t x)
static void __lmmp_mullh_(mp_limb_t a, mp_limb_t b, mp_ptr restrict dst)
static void __lmmp_sqr_(mp_ptr restrict dst, mp_srcptr restrict numa, mp_size_t na)
static int __lmmp_leading_zeros_(mp_limb_t x)
const mp_limb_t * mp_srcptr
#define MUL_FFT_THRESHOLD
#define SQR_TOOM6_THRESHOLD
#define MUL_TOOM22_THRESHOLD
#define MUL_TOOM8H_THRESHOLD
#define SQR_FFT_THRESHOLD
#define SQR_TOOM44_THRESHOLD
#define MUL_TOOM33_THRESHOLD
#define SQR_TOOM8_THRESHOLD
#define MUL_TOOM6H_THRESHOLD
#define SQR_TOOM22_THRESHOLD
#define SQR_TOOM33_THRESHOLD
#define MUL_TOOM44_THRESHOLD
void lmmp_mul_toom22_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
Toom-22乘法运算 [dst,na+nb] = [numa,na] * [numb,nb]
void lmmp_mul_toom44_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
Toom-44乘法运算 [dst,na+nb] = [numa,na] * [numb,nb]
void lmmp_mul_toom6h_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
Toom-6.5乘法运算 [dst,na+nb] = [numa,na] * [numb,nb] (非对称切分+条件inf点,原生支持不平衡;评估点 inf,±4,...
void lmmp_sqr_toom8_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
Toom-8平方运算 [dst,2*na] = [numa,na]^2 (评估点 ±8,±4,±2,±1,±1/2,±1/4,±1/8,0,无inf点)
void lmmp_sqr_fft_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
FFT平方运算 [dst,2*na] = [numa,na]^2
void lmmp_sqr_toom2_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
Toom-2平方运算 [dst,2*na] = [numa,na]^2
void lmmp_sqr_toom3_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
Toom-3平方运算 [dst,2*na] = [numa,na]^2
void lmmp_mul_toom8h_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
Toom-8.5乘法运算 [dst,na+nb] = [numa,na] * [numb,nb] (非对称切分+条件inf点,原生支持不平衡;评估点 inf,±8,...
void lmmp_mul_fft_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
FFT乘法运算 [dst,na+nb] = [numa,na] * [numb,nb]
void lmmp_sqr_toom6_(mp_ptr dst, mp_srcptr numa, mp_size_t na)
Toom-6平方运算 [dst,2*na] = [numa,na]^2 (评估点 ±4,±2,±1,±1/2,±1/4,0,无inf点)
void lmmp_mul_toom33_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_srcptr numb, mp_size_t nb)
Toom-33乘法运算 [dst,na+nb] = [numa,na] * [numb,nb]
void lmmp_sqr_toom4_(mp_ptr dst, mp_srcptr numa, mp_size_t an)
Toom-4平方运算 [dst,2*na] = [numa,na]^2
#define restrict
Copyright (C) 2026 HJimmyK (Jericho Knox)