Cog/Frameworks/lazyusf/lazyusf/rsp/vu/vabs.h

102 lines
3.3 KiB
C

/******************************************************************************\
* Authors: Iconoclast *
* Release: 2013.11.26 *
* License: CC0 Public Domain Dedication *
* *
* To the extent possible under law, the author(s) have dedicated all copyright *
* and related and neighboring rights to this software to the public domain *
* worldwide. This software is distributed without any warranty. *
* *
* You should have received a copy of the CC0 Public Domain Dedication along *
* with this software. *
* If not, see <http://creativecommons.org/publicdomain/zero/1.0/>. *
\******************************************************************************/
#include "vu.h"
/*
* -1: VT *= -1, because VS < 0 // VT ^= -2 if even, or ^= -1, += 1
* 0: VT *= 0, because VS = 0 // VT ^= VT
* +1: VT *= +1, because VS > 0 // VT ^= 0
* VT ^= -1, "negate" -32768 as ~+32767 (corner case hack for N64 SP)
*/
INLINE static void do_abs(usf_state_t * state, short* VD, short* VS, short* VT)
{
ALIGNED short neg[N], pos[N];
ALIGNED short nez[N], cch[N]; /* corner case hack -- abs(-32768) == +32767 */
ALIGNED short res[N];
register int i;
vector_copy(res, VT);
#ifdef ARCH_MIN_ARM_NEON
int16x8_t vs = vld1q_s16((const int16_t*)VS);
int16x8_t resi = vld1q_s16((const int16_t*)VT);
int16x8_t zero = vdupq_n_s16(0);
int16x8_t eightk = vdupq_n_s16(0x8000);
int16x8_t one = vdupq_n_s16(1);
uint16x8_t negi = vcltq_s16(vs,zero);
int16x8_t posi = vaddq_s16((int16x8_t)negi,one);
posi = vorrq_s16(posi,(int16x8_t)negi);
resi = veorq_s16(resi,posi);
uint16x8_t ccch = vcgeq_s16(resi,eightk);
int16x8_t ch = vnegq_s16((int16x8_t)ccch);
resi = vaddq_s16(resi, (int16x8_t)ch);
vst1q_s16(VACC_L, resi);
vector_copy(VD, VACC_L);
return;
#else
#ifndef ARCH_MIN_SSE2
#define MASK_XOR
#endif
for (i = 0; i < N; i++)
neg[i] = (VS[i] < 0x0000);
for (i = 0; i < N; i++)
pos[i] = (VS[i] > 0x0000);
for (i = 0; i < N; i++)
nez[i] = 0;
#ifdef MASK_XOR
for (i = 0; i < N; i++)
neg[i] = -neg[i];
for (i = 0; i < N; i++)
nez[i] += neg[i];
#else
for (i = 0; i < N; i++)
nez[i] -= neg[i];
#endif
for (i = 0; i < N; i++)
nez[i] += pos[i];
#ifdef MASK_XOR
for (i = 0; i < N; i++)
res[i] ^= nez[i];
for (i = 0; i < N; i++)
cch[i] = (res[i] != -32768);
for (i = 0; i < N; i++)
res[i] += cch[i]; /* -(x) === (x ^ -1) + 1 */
#else
for (i = 0; i < N; i++)
res[i] *= nez[i];
for (i = 0; i < N; i++)
cch[i] = (res[i] == -32768);
for (i = 0; i < N; i++)
res[i] -= cch[i];
#endif
vector_copy(VACC_L, res);
vector_copy(VD, VACC_L);
return;
#endif
}
static void VABS(usf_state_t * state, int vd, int vs, int vt, int e)
{
ALIGNED short ST[N];
SHUFFLE_VECTOR(ST, state->VR[vt], e);
do_abs(state, state->VR[vd], state->VR[vs], ST);
return;
}