100 lines
2.2 KiB
C
100 lines
2.2 KiB
C
|
|
#include "math.h"
|
|
|
|
#include <stdint.h>
|
|
|
|
void ccr_vec_xor (int bits, ccr_mtx a, ccr_mtx b, ccr_mtx r)
|
|
{
|
|
/* possible speedup for wideword architectures
|
|
while(bits>=32) {
|
|
*(uint32_t*)r = *(uint32_t*)a ^ *(uint32_t*)b;
|
|
a+=4;b+=4;r+=4;bits-=32;
|
|
} */
|
|
while (bits > 0) {
|
|
* (uint8_t*) r = * (uint8_t*) a ^ * (uint8_t*) b;
|
|
a += 1;
|
|
b += 1;
|
|
r += 1;
|
|
bits -= 8;
|
|
}
|
|
/* we can safely ignore padding bytes at the end of the vector */
|
|
}
|
|
|
|
void ccr_vec_and (int bits, ccr_mtx a, ccr_mtx b, ccr_mtx r)
|
|
{
|
|
while (bits > 0) {
|
|
* (uint8_t*) r = * (uint8_t*) a & * (uint8_t*) b;
|
|
a += 1;
|
|
b += 1;
|
|
r += 1;
|
|
bits -= 8;
|
|
}
|
|
}
|
|
|
|
int ccr_vec_parity (int bits, ccr_mtx a)
|
|
{
|
|
/* first, xor everything to one byte */
|
|
uint8_t b = 0;
|
|
while (bits >= 8) {
|
|
b ^= * (uint8_t*) a;
|
|
a += 1;
|
|
bits -= 8;
|
|
}
|
|
if (bits > 0) /* overflow padding bits away */
|
|
b ^= * (uint8_t*) a << (8 - bits);
|
|
|
|
/* squash the result in a single bit */
|
|
b ^= b >> 4;
|
|
b ^= b >> 2;
|
|
b ^= b >> 1;
|
|
return b & 1;
|
|
}
|
|
|
|
void ccr_vec_bit_set (ccr_mtx a, int offset, int bit)
|
|
{
|
|
if (bit)
|
|
( (uint8_t*) a) [offset/8] |= (uint8_t) (1 << (offset % 8) );
|
|
else
|
|
( (uint8_t*) a) [offset/8] &= ~ (uint8_t) (1 << (offset % 8) );
|
|
}
|
|
|
|
uint8_t ccr_vec_bit_get (ccr_mtx a, int offset)
|
|
{
|
|
return 1 & ( ( (uint8_t*) a) [offset/8] >> (offset % 8) );
|
|
}
|
|
|
|
void ccr_mtx_add (int cols, int rows,
|
|
ccr_mtx a, ccr_mtx b, ccr_mtx r)
|
|
{
|
|
int i, t;
|
|
for (i = 0; i < cols; ++i) {
|
|
t = ccr_mtx_vec_offset (rows, i);
|
|
ccr_vec_xor (rows, a + t, b + t, r + t);
|
|
}
|
|
}
|
|
|
|
int ccr_mtx_dotproduct (ccr_mtx a, ccr_mtx b,
|
|
int aoff, int aheight, int boff, int len)
|
|
{
|
|
uint8_t r = 0;
|
|
int i;
|
|
for (i = 0; i < len; ++i)
|
|
r ^= ccr_vec_bit_get (a + ccr_mtx_vec_offset (aheight, i), aoff)
|
|
& ccr_vec_bit_get (b + ccr_mtx_vec_offset (len, boff), i);
|
|
return r;
|
|
}
|
|
|
|
void ccr_mtx_multiply (int rows, int veclen, int cols,
|
|
ccr_mtx a, ccr_mtx b, ccr_mtx r)
|
|
{
|
|
/* TODO use faster algorithm */
|
|
|
|
int i, j;
|
|
for (i = 0; i < cols; ++i)
|
|
for (j = 0; j < rows; ++j)
|
|
ccr_vec_bit_set (r + ccr_mtx_vec_offset (rows, i), j,
|
|
ccr_mtx_dotproduct (a, b,
|
|
j, rows, i,
|
|
veclen) );
|
|
}
|