vitalif
/
gf-complete
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merged in jimplank/gf-complete-unit-bytwo (pull request #1) 

Unit & w=64 bytwo.  Made a few more changes.
master
Ethan Miller 2013-03-04 10:20:44 -08:00
parent cf6a5dfa29 82a365dae1
commit 5c719db2b9
2 changed files with 102 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
gf_unit.c
View File

@ -190,6 +190,8 @@ int main(int argc, char **argv)
gf_general_set_zero(a, w);
} else if (i < 40) {
gf_general_set_one(a, w);
} else if (i < 60) {
gf_general_set_two(a, w);
} else {
gf_general_set_random(a, w, 1);
}

110
gf_w64.c
View File

@ -57,6 +57,8 @@ gf_val_64_t gf_w64_inverse_from_divide (gf_t *gf, gf_val_64_t a)
return gf->divide.w64(gf, 1, a);
}
#define MM_PRINT8(s, r) { uint8_t blah[16], ii; printf("%-12s", s); _mm_storeu_si128((__m128i *)blah, r); for (ii = 0; ii < 16; ii += 1) printf("%s%02x", (ii%4==0) ? " " : " ", blah[15-ii]); printf("\n"); }
static
inline
gf_val_64_t gf_w64_divide_from_inverse (gf_t *gf, gf_val_64_t a, gf_val_64_t b)
@ -163,7 +165,7 @@ gf_w64_shift_multiply (gf_t *gf, gf_val_64_t a64, gf_val_64_t b64)
pl ^= bl;
pr ^= br;
}
/* printf("P: %016llx %016llx ", pl, pr); printf("B: %016llx %016llx\n", bl, br); */
/* printf("P: %016llx %016llx ", pl, pr); printf("B: %016llx %016llx\n", bl, br); */
bl <<= 1;
if (br & lbit) bl ^= 1;
br <<= 1;
@ -185,7 +187,6 @@ gf_w64_shift_multiply (gf_t *gf, gf_val_64_t a64, gf_val_64_t b64)
return pr;
}
#ifdef INTEL_PCLMUL
/*
* ELM: Use the Intel carryless multiply instruction to do very fast 64x64 multiply.
*/
@ -194,6 +195,7 @@ inline
gf_val_64_t
gf_w64_clm_multiply (gf_t *gf, gf_val_64_t a64, gf_val_64_t b64)
{
#ifdef INTEL_PCLMUL
__m128i a, b;
__m128i result;
__m128i prim_poly;
@ -201,7 +203,7 @@ gf_w64_clm_multiply (gf_t *gf, gf_val_64_t a64, gf_val_64_t b64)
gf_internal_t * h = gf->scratch;
a = _mm_insert_epi64 (_mm_setzero_si128(), a64, 0);
b = _mm_insert_epi64 (_mm_setzero_si128(), b64, 0);
b = _mm_insert_epi64 (a, b64, 0);
prim_poly = _mm_set_epi32(0, 0, 0, (uint32_t)(h->prim_poly & 0xffffffffULL));
/* Do the initial multiply */
result = _mm_clmulepi64_si128 (a, b, 0);
@ -216,8 +218,8 @@ gf_w64_clm_multiply (gf_t *gf, gf_val_64_t a64, gf_val_64_t b64)
result = _mm_xor_si128 (result, w);
return ((gf_val_64_t)_mm_extract_epi64(result, 0));
}
#endif
}
void
gf_w64_split_4_64_lazy_multiply_region(gf_t *gf, void *src, void *dest, uint64_t val, int bytes, int xor)
@ -409,8 +411,15 @@ gf_w64_split_16_64_lazy_multiply_region(gf_t *gf, void *src, void *dest, uint64_
static
int gf_w64_shift_init(gf_t *gf)
{
/* gf->multiply.w64 = gf_w64_shift_multiply; */
gf->multiply.w64 = gf_w64_clm_multiply;
gf_internal_t *h;
h = (gf_internal_t *) gf->scratch;
gf->multiply.w64 = gf_w64_shift_multiply;
#ifdef INTEL_PCLMUL
if (h->region_type != GF_REGION_NOSSE) gf->multiply.w64 = gf_w64_clm_multiply;
#endif
gf->inverse.w64 = gf_w64_euclid;
gf->multiply_region.w64 = gf_w64_multiply_region_from_single;
return 1;
@ -950,7 +959,6 @@ gf_w64_bytwo_b_nosse_multiply_region(gf_t *gf, void *src, void *dest, gf_val_64_
}
gf_do_final_region_alignment(&rd);
}
#define MM_PRINT8(s, r) { uint8_t blah[16], ii; printf("%-12s", s); _mm_storeu_si128((__m128i *)blah, r); for (ii = 0; ii < 16; ii += 1) printf("%s%02x", (ii%4==0) ? " " : " ", blah[15-ii]); printf("\n"); }
#define SSE_AB2(pp, m1 ,m2, va, t1, t2) {\
t1 = _mm_and_si128(_mm_slli_epi64(va, 1), m1); \
@ -1031,6 +1039,72 @@ void gf_w64_bytwo_p_sse_multiply_region(gf_t *gf, void *src, void *dest, gf_val_
#endif
}
static
void
gf_w64_bytwo_b_sse_region_2_xor(gf_region_data *rd)
{
#ifdef INTEL_SSE4
int i;
uint64_t one64, amask;
uint8_t *d8, *s8, tb;
__m128i pp, m1, m2, t1, t2, va, vb;
gf_internal_t *h;
s8 = (uint8_t *) rd->s_start;
d8 = (uint8_t *) rd->d_start;
h = (gf_internal_t *) rd->gf->scratch;
one64 = 1;
amask = -1;
amask ^= 1;
pp = _mm_set1_epi64x(h->prim_poly);
m1 = _mm_set1_epi64x(amask);
m2 = _mm_set1_epi64x(one64 << 63);
while (d8 < (uint8_t *) rd->d_top) {
va = _mm_load_si128 ((__m128i *)(s8));
SSE_AB2(pp, m1, m2, va, t1, t2);
vb = _mm_load_si128 ((__m128i *)(d8));
vb = _mm_xor_si128(vb, va);
_mm_store_si128((__m128i *)d8, vb);
d8 += 16;
s8 += 16;
}
#endif
}
static
void
gf_w64_bytwo_b_sse_region_2_noxor(gf_region_data *rd)
{
#ifdef INTEL_SSE4
int i;
uint64_t one64, amask;
uint8_t *d8, *s8, tb;
__m128i pp, m1, m2, t1, t2, va;
gf_internal_t *h;
s8 = (uint8_t *) rd->s_start;
d8 = (uint8_t *) rd->d_start;
h = (gf_internal_t *) rd->gf->scratch;
one64 = 1;
amask = -1;
amask ^= 1;
pp = _mm_set1_epi64x(h->prim_poly);
m1 = _mm_set1_epi64x(amask);
m2 = _mm_set1_epi64x(one64 << 63);
while (d8 < (uint8_t *) rd->d_top) {
va = _mm_load_si128 ((__m128i *)(s8));
SSE_AB2(pp, m1, m2, va, t1, t2);
_mm_store_si128((__m128i *)d8, va);
d8 += 16;
s8 += 16;
}
#endif
}
static
void
gf_w64_bytwo_b_sse_multiply_region(gf_t *gf, void *src, void *dest, gf_val_64_t val, int bytes, int xor)
@ -1049,6 +1123,16 @@ gf_w64_bytwo_b_sse_multiply_region(gf_t *gf, void *src, void *dest, gf_val_64_t
gf_set_region_data(&rd, gf, src, dest, bytes, val, xor, 16);
gf_do_initial_region_alignment(&rd);
if (val == 2) {
if (xor) {
gf_w64_bytwo_b_sse_region_2_xor(&rd);
} else {
gf_w64_bytwo_b_sse_region_2_noxor(&rd);
}
gf_do_final_region_alignment(&rd);
return;
}
s8 = (uint8_t *) rd.s_start;
d8 = (uint8_t *) rd.d_start;
h = (gf_internal_t *) gf->scratch;
@ -1417,8 +1501,13 @@ int gf_w64_split_init(gf_t *gf)
/* Defaults */
gf->multiply_region.w64 = gf_w64_multiply_region_from_single;
gf->multiply.w64 = gf_w64_clm_multiply;
/* gf->multiply.w64 = gf_w64_shift_multiply; */
gf->multiply.w64 = gf_w64_shift_multiply;
#ifdef INTEL_PCLMUL
if (h->region_type != GF_REGION_NOSSE) gf->multiply.w64 = gf_w64_clm_multiply;
#endif
gf->inverse.w64 = gf_w64_euclid;
if ((h->arg1 == 4 && h->arg2 == 64) || (h->arg1 == 64 && h->arg2 == 4)) {
@ -1489,7 +1578,8 @@ int gf_w64_scratch_size(int mult_type, int region_type, int divide_type, int arg
switch(mult_type)
{
case GF_MULT_SHIFT:
if (arg1 != 0 || arg2 != 0 || region_type != 0) return -1;
if (arg1 != 0 || arg2 != 0) return -1;
if (region_type != GF_REGION_NOSSE && region_type != GF_REGION_SSE && region_type != GF_REGION_DEFAULT) return -1;
return sizeof(gf_internal_t);
break;
case GF_MULT_BYTWO_p: