https://github.com/mupq/pqm4
Tip revision: 604ca2da84ebf8b068b5d54629fdc42649163c9d authored by Amin Abdulrahman on 13 October 2022, 15:57:26 UTC
Add reductions to iNTT in stack optimized code
Add reductions to iNTT in stack optimized code
Tip revision: 604ca2d
bc_2.c
// Implemented by Ming-Shing Chen, Tung Chou and Markus Krausz.
// public domain
#include "bc_2.h"
void bc_1_256( uint32_t *poly , unsigned n_256bit )
{
for(int i=n_256bit;i>0;i--) {
uint32_t p0 = poly[0];
uint32_t p1 = poly[1];
uint32_t p2 = poly[2];
uint32_t p3 = poly[3];
uint32_t p4 = poly[4];
uint32_t p5 = poly[5];
uint32_t p6 = poly[6];
uint32_t p7 = poly[7];
//layer 13: [256] s4^8:(128,8) suggest unit:1
p4 ^= p7>>24;
p3 ^= (p7<<8)^(p6>>24);
p2 ^= (p6<<8)^(p5>>24);
p1 ^= (p5<<8)^(p4>>24);
p0 ^= (p4<<8);
//layer 14: [128] s4^4:(64,4) suggest unit:1
p2 ^= p3>>28;
p1 ^= (p3<<4)^(p2>>28);
p0 ^= (p2<<4);
p6 ^= p7>>28;
p5 ^= (p7<<4)^(p6>>28);
p4 ^= (p6<<4);
//layer 15: [64] s4^2:(32,2) suggest unit:1
p1 ^= p1>>30;
p3 ^= p3>>30;
p5 ^= p5>>30;
p7 ^= p7>>30;
p0 ^= p1<<2;
p2 ^= p3<<2;
p4 ^= p5<<2;
p6 ^= p7<<2;
//layer 16: [32] s4^1:(16,1) suggest unit:1
p0 ^= ((p0&0x80000000)>>15);
p1 ^= ((p1&0x80000000)>>15);
p2 ^= ((p2&0x80000000)>>15);
p3 ^= ((p3&0x80000000)>>15);
p4 ^= ((p4&0x80000000)>>15);
p5 ^= ((p5&0x80000000)>>15);
p6 ^= ((p6&0x80000000)>>15);
p7 ^= ((p7&0x80000000)>>15);
p0 ^= ((p0&0x7fff0000)>>15);
p1 ^= ((p1&0x7fff0000)>>15);
p2 ^= ((p2&0x7fff0000)>>15);
p3 ^= ((p3&0x7fff0000)>>15);
p4 ^= ((p4&0x7fff0000)>>15);
p5 ^= ((p5&0x7fff0000)>>15);
p6 ^= ((p6&0x7fff0000)>>15);
p7 ^= ((p7&0x7fff0000)>>15);
//layer 17: [256] s2^2:(128,32) suggest unit:16
p4 ^= p7;
p3 ^= p6;
p2 ^= p5;
p1 ^= p4;
//layer 18: [128] s2^1:(64,16) suggest unit:16
p2 ^= p3>>16;
p1 ^= (p3<<16)^(p2>>16);
p0 ^= (p2<<16);
p6 ^= p7>>16;
p5 ^= (p7<<16)^(p6>>16);
p4 ^= (p6<<16);
//layer 19: [256] s1^1:(128,64) suggest unit:64
p5 ^= p7;
p4 ^= p6;
p3 ^= p5;
p2 ^= p4;
//layer 20: [64] s1^1:(32,16) suggest unit:16
p1 ^= p1>>16;
p3 ^= p3>>16;
p5 ^= p5>>16;
p7 ^= p7>>16;
p0 ^= p1<<16;
p2 ^= p3<<16;
p4 ^= p5<<16;
p6 ^= p7<<16;
//layer 21: [16] s2^2:(8,2) suggest unit:1
p0 ^= (p0&0xc000c000)>>6;
p1 ^= (p1&0xc000c000)>>6;
p2 ^= (p2&0xc000c000)>>6;
p3 ^= (p3&0xc000c000)>>6;
p4 ^= (p4&0xc000c000)>>6;
p5 ^= (p5&0xc000c000)>>6;
p6 ^= (p6&0xc000c000)>>6;
p7 ^= (p7&0xc000c000)>>6;
p0 ^= (p0&0x3f003f00)>>6;
p1 ^= (p1&0x3f003f00)>>6;
p2 ^= (p2&0x3f003f00)>>6;
p3 ^= (p3&0x3f003f00)>>6;
p4 ^= (p4&0x3f003f00)>>6;
p5 ^= (p5&0x3f003f00)>>6;
p6 ^= (p6&0x3f003f00)>>6;
p7 ^= (p7&0x3f003f00)>>6;
//layer 22: [8] s2^1:(4,1) suggest unit:1
p0 ^= ((p0&0x80808080)>>3);
p1 ^= ((p1&0x80808080)>>3);
p2 ^= ((p2&0x80808080)>>3);
p3 ^= ((p3&0x80808080)>>3);
p4 ^= ((p4&0x80808080)>>3);
p5 ^= ((p5&0x80808080)>>3);
p6 ^= ((p6&0x80808080)>>3);
p7 ^= ((p7&0x80808080)>>3);
p0 ^= ((p0&0x70707070)>>3);
p1 ^= ((p1&0x70707070)>>3);
p2 ^= ((p2&0x70707070)>>3);
p3 ^= ((p3&0x70707070)>>3);
p4 ^= ((p4&0x70707070)>>3);
p5 ^= ((p5&0x70707070)>>3);
p6 ^= ((p6&0x70707070)>>3);
p7 ^= ((p7&0x70707070)>>3);
//layer 23: [16] s1^1:(8,4) suggest unit:4
p0 ^= ((p0&0xf000f000)>>4);
p1 ^= ((p1&0xf000f000)>>4);
p2 ^= ((p2&0xf000f000)>>4);
p3 ^= ((p3&0xf000f000)>>4);
p4 ^= ((p4&0xf000f000)>>4);
p5 ^= ((p5&0xf000f000)>>4);
p6 ^= ((p6&0xf000f000)>>4);
p7 ^= ((p7&0xf000f000)>>4);
p0 ^= ((p0&0x0f000f00)>>4);
p1 ^= ((p1&0x0f000f00)>>4);
p2 ^= ((p2&0x0f000f00)>>4);
p3 ^= ((p3&0x0f000f00)>>4);
p4 ^= ((p4&0x0f000f00)>>4);
p5 ^= ((p5&0x0f000f00)>>4);
p6 ^= ((p6&0x0f000f00)>>4);
p7 ^= ((p7&0x0f000f00)>>4);
//layer 24: [4] s1^1:(2,1) suggest unit:1
p0 ^= ((p0&0x88888888)>>1);
p1 ^= ((p1&0x88888888)>>1);
p2 ^= ((p2&0x88888888)>>1);
p3 ^= ((p3&0x88888888)>>1);
p4 ^= ((p4&0x88888888)>>1);
p5 ^= ((p5&0x88888888)>>1);
p6 ^= ((p6&0x88888888)>>1);
p7 ^= ((p7&0x88888888)>>1);
p0 ^= ((p0&0x44444444)>>1);
p1 ^= ((p1&0x44444444)>>1);
p2 ^= ((p2&0x44444444)>>1);
p3 ^= ((p3&0x44444444)>>1);
p4 ^= ((p4&0x44444444)>>1);
p5 ^= ((p5&0x44444444)>>1);
p6 ^= ((p6&0x44444444)>>1);
p7 ^= ((p7&0x44444444)>>1);
poly[0] = p0;
poly[1] = p1;
poly[2] = p2;
poly[3] = p3;
poly[4] = p4;
poly[5] = p5;
poly[6] = p6;
poly[7] = p7;
poly += 8;
}
}
void ibc_1_256( uint32_t *poly , unsigned n_256bit )
{
for(int i=n_256bit;i>0;i--) {
uint32_t p0 = poly[0];
uint32_t p1 = poly[1];
uint32_t p2 = poly[2];
uint32_t p3 = poly[3];
uint32_t p4 = poly[4];
uint32_t p5 = poly[5];
uint32_t p6 = poly[6];
uint32_t p7 = poly[7];
//layer 7: [4] s1^1:(2,1) suggest unit:1
p0 ^= ((p0&0xcccccccc)>>1);
p1 ^= ((p1&0xcccccccc)>>1);
p2 ^= ((p2&0xcccccccc)>>1);
p3 ^= ((p3&0xcccccccc)>>1);
p4 ^= ((p4&0xcccccccc)>>1);
p5 ^= ((p5&0xcccccccc)>>1);
p6 ^= ((p6&0xcccccccc)>>1);
p7 ^= ((p7&0xcccccccc)>>1);
//layer 6: [16] s1^1:(8,4) suggest unit:4
p0 ^= ((p0&0xff00ff00)>>4);
p1 ^= ((p1&0xff00ff00)>>4);
p2 ^= ((p2&0xff00ff00)>>4);
p3 ^= ((p3&0xff00ff00)>>4);
p4 ^= ((p4&0xff00ff00)>>4);
p5 ^= ((p5&0xff00ff00)>>4);
p6 ^= ((p6&0xff00ff00)>>4);
p7 ^= ((p7&0xff00ff00)>>4);
//layer 5: [8] s2^1:(4,1) suggest unit:1
p0 ^= ((p0&0xf0f0f0f0)>>3);
p1 ^= ((p1&0xf0f0f0f0)>>3);
p2 ^= ((p2&0xf0f0f0f0)>>3);
p3 ^= ((p3&0xf0f0f0f0)>>3);
p4 ^= ((p4&0xf0f0f0f0)>>3);
p5 ^= ((p5&0xf0f0f0f0)>>3);
p6 ^= ((p6&0xf0f0f0f0)>>3);
p7 ^= ((p7&0xf0f0f0f0)>>3);
//layer 4: [16] s2^2:(8,2) suggest unit:1
p0 ^= ((p0&0xff00ff00)>>6);
p1 ^= ((p1&0xff00ff00)>>6);
p2 ^= ((p2&0xff00ff00)>>6);
p3 ^= ((p3&0xff00ff00)>>6);
p4 ^= ((p4&0xff00ff00)>>6);
p5 ^= ((p5&0xff00ff00)>>6);
p6 ^= ((p6&0xff00ff00)>>6);
p7 ^= ((p7&0xff00ff00)>>6);
//layer 3: [64] s1^1:(32,16) suggest unit:16
p0 ^= p1<<16;
p2 ^= p3<<16;
p4 ^= p5<<16;
p6 ^= p7<<16;
p1 ^= p1>>16;
p3 ^= p3>>16;
p5 ^= p5>>16;
p7 ^= p7>>16;
//layer 2: [256] s1^1:(128,64) suggest unit:64
p3 ^= p5;
p2 ^= p4;
p5 ^= p7;
p4 ^= p6;
//layer 1: [128] s2^1:(64,16) suggest unit:16
p0 ^= (p2<<16);
p1 ^= (p3<<16)^(p2>>16);
p2 ^= p3>>16;
p4 ^= (p6<<16);
p5 ^= (p7<<16)^(p6>>16);
p6 ^= p7>>16;
//layer 0: [256] s2^2:(128,32) suggest unit:16
p1 ^= p4;
p2 ^= p5;
p3 ^= p6;
p4 ^= p7;
//layer 16: [32] s4^1:(16,1) suggest unit:1
p0 ^= ((p0&0xffff0000)>>15);
p1 ^= ((p1&0xffff0000)>>15);
p2 ^= ((p2&0xffff0000)>>15);
p3 ^= ((p3&0xffff0000)>>15);
p4 ^= ((p4&0xffff0000)>>15);
p5 ^= ((p5&0xffff0000)>>15);
p6 ^= ((p6&0xffff0000)>>15);
p7 ^= ((p7&0xffff0000)>>15);
//layer 15: [64] s4^2:(32,2) suggest unit:1
p0 ^= p1<<2;
p2 ^= p3<<2;
p4 ^= p5<<2;
p6 ^= p7<<2;
p1 ^= p1>>30;
p3 ^= p3>>30;
p5 ^= p5>>30;
p7 ^= p7>>30;
//layer 14: [128] s4^4:(64,4) suggest unit:1
p0 ^= (p2<<4);
p1 ^= (p3<<4)^(p2>>28);
p2 ^= p3>>28;
p4 ^= (p6<<4);
p5 ^= (p7<<4)^(p6>>28);
p6 ^= p7>>28;
//layer 13: [256] s4^8:(128,8) suggest unit:1
p0 ^= (p4<<8);
p1 ^= (p5<<8)^(p4>>24);
p2 ^= (p6<<8)^(p5>>24);
p3 ^= (p7<<8)^(p6>>24);
p4 ^= p7>>24;
poly[0] = p0;
poly[1] = p1;
poly[2] = p2;
poly[3] = p3;
poly[4] = p4;
poly[5] = p5;
poly[6] = p6;
poly[7] = p7;
poly += 8;
}
}
/////////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////////
static
void cvt_to_fft_256_16384( uint32_t *poly )
{
const unsigned w = 16384;
uint32_t *ptr;
//layer 6: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 7: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 8: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 12: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
static
void icvt_to_poly_256_16384( uint32_t *poly )
{
const unsigned w = 16384;
uint32_t *ptr;
//layer 12: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 8: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 7: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 6: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
static
void cvt_to_fft_256_32768( uint32_t *poly )
{
const unsigned w = 32768;
uint32_t *ptr;
//layer 7: [32768] s4^4:(16384,1024) suggest unit:256
#define DEG_H 16384
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 8: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [32768] s2^1:(16384,4096) suggest unit:4096
#define DEG_H 16384
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 12: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 13: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 14: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 15: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
static
void icvt_to_poly_256_32768( uint32_t *poly )
{
const unsigned w = 32768;
uint32_t *ptr;
//layer 15: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 14: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 13: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 12: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [32768] s2^1:(16384,4096) suggest unit:4096
#define DEG_H 16384
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 8: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 7: [32768] s4^4:(16384,1024) suggest unit:256
#define DEG_H 16384
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
static
void cvt_to_fft_256_65536( uint32_t *poly )
{
const unsigned w = 65536;
uint32_t *ptr;
//layer 8: [65536] s4^8:(32768,2048) suggest unit:256
#define DEG_H 32768
#define DEG_L 2048
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [32768] s4^4:(16384,1024) suggest unit:256
#define DEG_H 16384
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 12: [65536] s2^2:(32768,8192) suggest unit:4096
#define DEG_H 32768
#define DEG_L 8192
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 13: [32768] s2^1:(16384,4096) suggest unit:4096
#define DEG_H 16384
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 14: [65536] s1^1:(32768,16384) suggest unit:16384
#define DEG_H 32768
#define DEG_L 16384
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 15: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 16: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 17: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 18: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 19: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
static
void icvt_to_poly_256_65536( uint32_t *poly )
{
const unsigned w = 65536;
uint32_t *ptr;
//layer 19: [1024] s1^1:(512,256) suggest unit:256
#define DEG_H 512
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 18: [4096] s1^1:(2048,1024) suggest unit:1024
#define DEG_H 2048
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 17: [2048] s2^1:(1024,256) suggest unit:256
#define DEG_H 1024
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 16: [4096] s2^2:(2048,512) suggest unit:256
#define DEG_H 2048
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 15: [16384] s1^1:(8192,4096) suggest unit:4096
#define DEG_H 8192
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 14: [65536] s1^1:(32768,16384) suggest unit:16384
#define DEG_H 32768
#define DEG_L 16384
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 13: [32768] s2^1:(16384,4096) suggest unit:4096
#define DEG_H 16384
#define DEG_L 4096
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 12: [65536] s2^2:(32768,8192) suggest unit:4096
#define DEG_H 32768
#define DEG_L 8192
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 11: [8192] s4^1:(4096,256) suggest unit:256
#define DEG_H 4096
#define DEG_L 256
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 10: [16384] s4^2:(8192,512) suggest unit:256
#define DEG_H 8192
#define DEG_L 512
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 9: [32768] s4^4:(16384,1024) suggest unit:256
#define DEG_H 16384
#define DEG_L 1024
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 8: [65536] s4^8:(32768,2048) suggest unit:256
#define DEG_H 32768
#define DEG_L 2048
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
/////////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////////
static
void repr_s8_1_16384( uint32_t *poly )
{
uint32_t *ptr;
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
for(int i=(8192/32)-4;i>=0;i-=4) {
uint32_t p0 = poly[i+(8192/32)];
uint32_t p1 = poly[i+(8192/32)+1];
uint32_t p2 = poly[i+(8192/32)+2];
uint32_t p3 = poly[i+(8192/32)+3];
uint32_t q0 = poly[i+1];
uint32_t q1 = poly[i+1+1];
uint32_t q2 = poly[i+1+2];
uint32_t q3 = poly[i+1+3];
poly[i+1] = q0^p0;
poly[i+1+1] = q1^p1;
poly[i+1+2] = q2^p2;
poly[i+1+3] = q3^p3;
}
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(16384/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=(4096/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>16)^(leftover<<16);
ptr[i+1+2] = q2^(p2>>16)^(p3<<16);
ptr[i+1+1] = q1^(p1>>16)^(p2<<16);
ptr[i+1] = q0^(p0>>16)^(p1<<16);
leftover= p0;
}
ptr[0] ^= (leftover<<16);
ptr += (8192/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(16384/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=(2048/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>24)^(leftover<<8);
ptr[i+1+2] = q2^(p2>>24)^(p3<<8);
ptr[i+1+1] = q1^(p1>>24)^(p2<<8);
ptr[i+1] = q0^(p0>>24)^(p1<<8);
leftover= p0;
}
ptr[0] ^= (leftover<<8);
ptr += (4096/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(16384/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=(1024/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>28)^(leftover<<4);
ptr[i+1+2] = q2^(p2>>28)^(p3<<4);
ptr[i+1+1] = q1^(p1>>28)^(p2<<4);
ptr[i+1] = q0^(p0>>28)^(p1<<4);
leftover= p0;
}
ptr[0] ^= (leftover<<4);
ptr += (2048/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(16384/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=(512/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>30)^(leftover<<2);
ptr[i+1+2] = q2^(p2>>30)^(p3<<2);
ptr[i+1+1] = q1^(p1>>30)^(p2<<2);
ptr[i+1] = q0^(p0>>30)^(p1<<2);
leftover= p0;
}
ptr[0] ^= (leftover<<2);
ptr += (1024/32);
}
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(16384/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=(256/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>31)^(leftover<<1);
ptr[i+1+2] = q2^(p2>>31)^(p3<<1);
ptr[i+1+1] = q1^(p1>>31)^(p2<<1);
ptr[i+1] = q0^(p0>>31)^(p1<<1);
leftover= p0;
}
ptr[0] ^= (leftover<<1);
ptr += (512/32);
}
}
static
void irepr_s8_1_16384( uint32_t *poly )
{
uint32_t *ptr;
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(16384/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(256/32);i+=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>31)^(p3<<1);
ptr[i+2] = q2^(p1>>31)^(p2<<1);
ptr[i+1] = q1^(p0>>31)^(p1<<1);
ptr[i] = q0^(leftover>>31)^(p0<<1);
leftover= p3;
}
ptr[(256/32)] ^= (leftover>>31);
ptr += (512/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(16384/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(512/32);i+=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>30)^(p3<<2);
ptr[i+2] = q2^(p1>>30)^(p2<<2);
ptr[i+1] = q1^(p0>>30)^(p1<<2);
ptr[i] = q0^(leftover>>30)^(p0<<2);
leftover= p3;
}
ptr[(512/32)] ^= (leftover>>30);
ptr += (1024/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(16384/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(1024/32);i+=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>28)^(p3<<4);
ptr[i+2] = q2^(p1>>28)^(p2<<4);
ptr[i+1] = q1^(p0>>28)^(p1<<4);
ptr[i] = q0^(leftover>>28)^(p0<<4);
leftover= p3;
}
ptr[(1024/32)] ^= (leftover>>28);
ptr += (2048/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(16384/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(2048/32);i+=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>24)^(p3<<8);
ptr[i+2] = q2^(p1>>24)^(p2<<8);
ptr[i+1] = q1^(p0>>24)^(p1<<8);
ptr[i] = q0^(leftover>>24)^(p0<<8);
leftover= p3;
}
ptr[(2048/32)] ^= (leftover>>24);
ptr += (4096/32);
}
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(16384/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(4096/32);i+=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>16)^(p3<<16);
ptr[i+2] = q2^(p1>>16)^(p2<<16);
ptr[i+1] = q1^(p0>>16)^(p1<<16);
ptr[i] = q0^(leftover>>16)^(p0<<16);
leftover= p3;
}
ptr[(4096/32)] ^= (leftover>>16);
ptr += (8192/32);
}
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
for(int i=0;i<(8192/32);i+=4) {
uint32_t p0 = poly[i+(8192/32)];
uint32_t p1 = poly[i+(8192/32)+1];
uint32_t p2 = poly[i+(8192/32)+2];
uint32_t p3 = poly[i+(8192/32)+3];
uint32_t q0 = poly[i+1];
uint32_t q1 = poly[i+1+1];
uint32_t q2 = poly[i+1+2];
uint32_t q3 = poly[i+1+3];
poly[i+1] = q0^p0;
poly[i+1+1] = q1^p1;
poly[i+1+2] = q2^p2;
poly[i+1+3] = q3^p3;
}
}
//////////////////
static
void repr_s8_1_32768( uint32_t *poly )
{
const unsigned w = 32768;
uint32_t *ptr;
//layer 0: [32768] s8^64:(16384,64) suggest unit:1
#define DEG_H 16384
#define DEG_L 64
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
#define DEG_H 8192
#define DEG_L 32
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(w/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=(4096/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>16)^(leftover<<16);
ptr[i+1+2] = q2^(p2>>16)^(p3<<16);
ptr[i+1+1] = q1^(p1>>16)^(p2<<16);
ptr[i+1] = q0^(p0>>16)^(p1<<16);
leftover= p0;
}
ptr[0] ^= (leftover<<16);
ptr += (8192/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(w/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=(2048/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>24)^(leftover<<8);
ptr[i+1+2] = q2^(p2>>24)^(p3<<8);
ptr[i+1+1] = q1^(p1>>24)^(p2<<8);
ptr[i+1] = q0^(p0>>24)^(p1<<8);
leftover= p0;
}
ptr[0] ^= (leftover<<8);
ptr += (4096/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(w/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=(1024/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>28)^(leftover<<4);
ptr[i+1+2] = q2^(p2>>28)^(p3<<4);
ptr[i+1+1] = q1^(p1>>28)^(p2<<4);
ptr[i+1] = q0^(p0>>28)^(p1<<4);
leftover= p0;
}
ptr[0] ^= (leftover<<4);
ptr += (2048/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(w/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=(512/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>30)^(leftover<<2);
ptr[i+1+2] = q2^(p2>>30)^(p3<<2);
ptr[i+1+1] = q1^(p1>>30)^(p2<<2);
ptr[i+1] = q0^(p0>>30)^(p1<<2);
leftover= p0;
}
ptr[0] ^= (leftover<<2);
ptr += (1024/32);
}
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(w/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=(256/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>31)^(leftover<<1);
ptr[i+1+2] = q2^(p2>>31)^(p3<<1);
ptr[i+1+1] = q1^(p1>>31)^(p2<<1);
ptr[i+1] = q0^(p0>>31)^(p1<<1);
leftover= p0;
}
ptr[0] ^= (leftover<<1);
ptr += (512/32);
}
}
static
void irepr_s8_1_32768( uint32_t *poly )
{
const unsigned w = 32768;
uint32_t *ptr;
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(w/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(256/32);i+=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>31)^(p3<<1);
ptr[i+2] = q2^(p1>>31)^(p2<<1);
ptr[i+1] = q1^(p0>>31)^(p1<<1);
ptr[i] = q0^(leftover>>31)^(p0<<1);
leftover= p3;
}
ptr[(256/32)] ^= (leftover>>31);
ptr += (512/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(w/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(512/32);i+=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>30)^(p3<<2);
ptr[i+2] = q2^(p1>>30)^(p2<<2);
ptr[i+1] = q1^(p0>>30)^(p1<<2);
ptr[i] = q0^(leftover>>30)^(p0<<2);
leftover= p3;
}
ptr[(512/32)] ^= (leftover>>30);
ptr += (1024/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(w/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(1024/32);i+=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>28)^(p3<<4);
ptr[i+2] = q2^(p1>>28)^(p2<<4);
ptr[i+1] = q1^(p0>>28)^(p1<<4);
ptr[i] = q0^(leftover>>28)^(p0<<4);
leftover= p3;
}
ptr[(1024/32)] ^= (leftover>>28);
ptr += (2048/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(w/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(2048/32);i+=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>24)^(p3<<8);
ptr[i+2] = q2^(p1>>24)^(p2<<8);
ptr[i+1] = q1^(p0>>24)^(p1<<8);
ptr[i] = q0^(leftover>>24)^(p0<<8);
leftover= p3;
}
ptr[(2048/32)] ^= (leftover>>24);
ptr += (4096/32);
}
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(w/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(4096/32);i+=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>16)^(p3<<16);
ptr[i+2] = q2^(p1>>16)^(p2<<16);
ptr[i+1] = q1^(p0>>16)^(p1<<16);
ptr[i] = q0^(leftover>>16)^(p0<<16);
leftover= p3;
}
ptr[(4096/32)] ^= (leftover>>16);
ptr += (8192/32);
}
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
#define DEG_H 8192
#define DEG_L 32
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [32768] s8^64:(16384,64) suggest unit:1
#define DEG_H 16384
#define DEG_L 64
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
//////////////
static
void repr_s8_1_65536( uint32_t *poly )
{
const unsigned w = 65536;
uint32_t *ptr;
//layer 0: [65536] s8^128:(32768,128) suggest unit:1
#define DEG_H 32768
#define DEG_L 128
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [32768] s8^64:(16384,64) suggest unit:1
#define DEG_H 16384
#define DEG_L 64
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
#define DEG_H 8192
#define DEG_L 32
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=(DEG_H/32)-4;i>=0;i-=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(w/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=(4096/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>16)^(leftover<<16);
ptr[i+1+2] = q2^(p2>>16)^(p3<<16);
ptr[i+1+1] = q1^(p1>>16)^(p2<<16);
ptr[i+1] = q0^(p0>>16)^(p1<<16);
leftover= p0;
}
ptr[0] ^= (leftover<<16);
ptr += (8192/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(w/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=(2048/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>24)^(leftover<<8);
ptr[i+1+2] = q2^(p2>>24)^(p3<<8);
ptr[i+1+1] = q1^(p1>>24)^(p2<<8);
ptr[i+1] = q0^(p0>>24)^(p1<<8);
leftover= p0;
}
ptr[0] ^= (leftover<<8);
ptr += (4096/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(w/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=(1024/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>28)^(leftover<<4);
ptr[i+1+2] = q2^(p2>>28)^(p3<<4);
ptr[i+1+1] = q1^(p1>>28)^(p2<<4);
ptr[i+1] = q0^(p0>>28)^(p1<<4);
leftover= p0;
}
ptr[0] ^= (leftover<<4);
ptr += (2048/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(w/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=(512/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>30)^(leftover<<2);
ptr[i+1+2] = q2^(p2>>30)^(p3<<2);
ptr[i+1+1] = q1^(p1>>30)^(p2<<2);
ptr[i+1] = q0^(p0>>30)^(p1<<2);
leftover= p0;
}
ptr[0] ^= (leftover<<2);
ptr += (1024/32);
}
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(w/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=(256/32)-4;i>=0;i-=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+1+3];
uint32_t q2 = ptr[i+1+2];
uint32_t q1 = ptr[i+1+1];
uint32_t q0 = ptr[i+1];
ptr[i+1+3] = q3^(p3>>31)^(leftover<<1);
ptr[i+1+2] = q2^(p2>>31)^(p3<<1);
ptr[i+1+1] = q1^(p1>>31)^(p2<<1);
ptr[i+1] = q0^(p0>>31)^(p1<<1);
leftover= p0;
}
ptr[0] ^= (leftover<<1);
ptr += (512/32);
}
}
static
void irepr_s8_1_65536( uint32_t *poly )
{
const unsigned w = 65536;
uint32_t *ptr;
//layer 5: [512] s8^1:(256,1) suggest unit:1
ptr = poly;
for(int j=(w/512);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(256/32);i+=4) {
uint32_t p3 = ptr[i+(256/32)+3];
uint32_t p2 = ptr[i+(256/32)+2];
uint32_t p1 = ptr[i+(256/32)+1];
uint32_t p0 = ptr[i+(256/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>31)^(p3<<1);
ptr[i+2] = q2^(p1>>31)^(p2<<1);
ptr[i+1] = q1^(p0>>31)^(p1<<1);
ptr[i] = q0^(leftover>>31)^(p0<<1);
leftover= p3;
}
ptr[(256/32)] ^= (leftover>>31);
ptr += (512/32);
}
//layer 4: [1024] s8^2:(512,2) suggest unit:1
ptr = poly;
for(int j=(w/1024);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(512/32);i+=4) {
uint32_t p3 = ptr[i+(512/32)+3];
uint32_t p2 = ptr[i+(512/32)+2];
uint32_t p1 = ptr[i+(512/32)+1];
uint32_t p0 = ptr[i+(512/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>30)^(p3<<2);
ptr[i+2] = q2^(p1>>30)^(p2<<2);
ptr[i+1] = q1^(p0>>30)^(p1<<2);
ptr[i] = q0^(leftover>>30)^(p0<<2);
leftover= p3;
}
ptr[(512/32)] ^= (leftover>>30);
ptr += (1024/32);
}
//layer 3: [2048] s8^4:(1024,4) suggest unit:1
ptr = poly;
for(int j=(w/2048);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(1024/32);i+=4) {
uint32_t p3 = ptr[i+(1024/32)+3];
uint32_t p2 = ptr[i+(1024/32)+2];
uint32_t p1 = ptr[i+(1024/32)+1];
uint32_t p0 = ptr[i+(1024/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>28)^(p3<<4);
ptr[i+2] = q2^(p1>>28)^(p2<<4);
ptr[i+1] = q1^(p0>>28)^(p1<<4);
ptr[i] = q0^(leftover>>28)^(p0<<4);
leftover= p3;
}
ptr[(1024/32)] ^= (leftover>>28);
ptr += (2048/32);
}
//layer 2: [4096] s8^8:(2048,8) suggest unit:1
ptr = poly;
for(int j=(w/4096);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(2048/32);i+=4) {
uint32_t p3 = ptr[i+(2048/32)+3];
uint32_t p2 = ptr[i+(2048/32)+2];
uint32_t p1 = ptr[i+(2048/32)+1];
uint32_t p0 = ptr[i+(2048/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>24)^(p3<<8);
ptr[i+2] = q2^(p1>>24)^(p2<<8);
ptr[i+1] = q1^(p0>>24)^(p1<<8);
ptr[i] = q0^(leftover>>24)^(p0<<8);
leftover= p3;
}
ptr[(2048/32)] ^= (leftover>>24);
ptr += (4096/32);
}
//layer 1: [8192] s8^16:(4096,16) suggest unit:1
ptr = poly;
for(int j=(w/8192);j>0;j--) {
uint32_t leftover = 0;
for(int i=0;i<(4096/32);i+=4) {
uint32_t p3 = ptr[i+(4096/32)+3];
uint32_t p2 = ptr[i+(4096/32)+2];
uint32_t p1 = ptr[i+(4096/32)+1];
uint32_t p0 = ptr[i+(4096/32)];
uint32_t q3 = ptr[i+3];
uint32_t q2 = ptr[i+2];
uint32_t q1 = ptr[i+1];
uint32_t q0 = ptr[i];
ptr[i+3] = q3^(p2>>16)^(p3<<16);
ptr[i+2] = q2^(p1>>16)^(p2<<16);
ptr[i+1] = q1^(p0>>16)^(p1<<16);
ptr[i] = q0^(leftover>>16)^(p0<<16);
leftover= p3;
}
ptr[(4096/32)] ^= (leftover>>16);
ptr += (8192/32);
}
//layer 0: [16384] s8^32:(8192,32) suggest unit:1
#define DEG_H 8192
#define DEG_L 32
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [32768] s8^64:(16384,64) suggest unit:1
#define DEG_H 16384
#define DEG_L 64
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
//layer 0: [65536] s8^128:(32768,128) suggest unit:1
#define DEG_H 32768
#define DEG_L 128
ptr = poly;
for(int j=(w/(DEG_H*2));j>0;j--) {
for(int i=0;i<(DEG_H/32);i+=4) {
uint32_t p0 = ptr[i+(DEG_H/32)];
uint32_t p1 = ptr[i+(DEG_H/32)+1];
uint32_t p2 = ptr[i+(DEG_H/32)+2];
uint32_t p3 = ptr[i+(DEG_H/32)+3];
uint32_t q0 = ptr[i+(DEG_L/32)];
uint32_t q1 = ptr[i+(DEG_L/32)+1];
uint32_t q2 = ptr[i+(DEG_L/32)+2];
uint32_t q3 = ptr[i+(DEG_L/32)+3];
ptr[i+(DEG_L/32)] = q0^p0;
ptr[i+(DEG_L/32)+1] = q1^p1;
ptr[i+(DEG_L/32)+2] = q2^p2;
ptr[i+(DEG_L/32)+3] = q3^p3;
}
ptr += (DEG_H*2/32);
}
#undef DEG_H
#undef DEG_L
}
////////////////////////////////////////////////////////////////////////
void bc_1_16384( uint32_t *poly )
{
repr_s8_1_16384( poly );
cvt_to_fft_256_16384( poly );
bc_1_256( poly , 16384/256 );
}
void ibc_1_16384( uint32_t *poly )
{
ibc_1_256( poly , 16384/256 );
icvt_to_poly_256_16384( poly );
irepr_s8_1_16384( poly );
}
void bc_1_32768( uint32_t *poly )
{
repr_s8_1_32768( poly );
cvt_to_fft_256_32768( poly );
bc_1_256( poly , 32768/256 );
}
void ibc_1_32768( uint32_t *poly )
{
ibc_1_256( poly , 32768/256 );
icvt_to_poly_256_32768( poly );
irepr_s8_1_32768( poly );
}
void bc_1_65536( uint32_t *poly )
{
repr_s8_1_65536( poly );
cvt_to_fft_256_65536( poly );
bc_1_256( poly , 65536/256 );
}
void ibc_1_65536( uint32_t *poly )
{
ibc_1_256( poly , 65536/256 );
icvt_to_poly_256_65536( poly );
irepr_s8_1_65536( poly );
}
