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

100 lines
3.6 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"
INLINE static void do_madl(usf_state_t * state, short* VD, short* VS, short* VT)
{
#ifdef ARCH_MIN_ARM_NEON
uint16x8_t vs = vld1q_u16((const uint16_t*)VS);
uint16x8_t vt = vld1q_u16((const uint16_t*)VT);
uint16x8_t v_vaccl = vld1q_u16((const uint16_t*)VACC_L);
uint16x8_t v_vaccm = vld1q_u16((const uint16_t*)VACC_M);
uint16x8_t v_vacch = vld1q_u16((const uint16_t*)VACC_H);
uint32x4_t zero = vdupq_n_u32(0);
uint16x4_t vs_low = vget_low_u16(vs);
uint16x4_t vs_high = vget_high_u16(vs);
uint16x4_t vt_low = vget_low_u16(vt);
uint16x4_t vt_high = vget_high_u16(vt);
uint32x4_t product_L = vmulq_u32( vmovl_u16(vs_low), vmovl_u16(vt_low) );
uint32x4_t product_H = vmulq_u32( vmovl_u16(vs_high), vmovl_u16(vt_high) );
uint16x4_t addend_L = vaddhn_u32(product_L, zero);
uint16x4_t addend_H = vaddhn_u32(product_H, zero);
uint32x4_t exceed1 = vaddl_u16(addend_L, vget_low_u16(v_vaccl));
uint32x4_t exceed2 = vaddl_u16(addend_H, vget_high_u16(v_vaccl));
v_vaccl = vcombine_u16(vmovn_u32(exceed1), vmovn_u32(exceed2));
addend_L = vaddhn_u32(exceed1, zero);
addend_H = vaddhn_u32(exceed2, zero);
exceed1 = vaddl_u16(addend_L, vget_low_u16(v_vaccm));
exceed2 = vaddl_u16(addend_H, vget_high_u16(v_vaccm));
v_vaccm = vcombine_u16(vmovn_u32(exceed1), vmovn_u32(exceed2));
addend_L = vaddhn_u32(exceed1, zero);
addend_H = vaddhn_u32(exceed2, zero);
uint16x8_t v_vacch2 = vcombine_u16(addend_L, addend_H);
v_vacch = vaddq_u16(v_vacch, v_vacch2);
vst1q_s16(VACC_L, (int16x8_t)v_vaccl);
vst1q_s16(VACC_M, (int16x8_t)v_vaccm);
vst1q_s16(VACC_H, (int16x8_t)v_vacch);
SIGNED_CLAMP_AL(state, VD);
return;
#else
ALIGNED int32_t product[N];
ALIGNED uint32_t addend[N];
register int i;
for (i = 0; i < N; i++)
product[i] = (unsigned short)(VS[i]) * (unsigned short)(VT[i]);
for (i = 0; i < N; i++)
addend[i] = (product[i] & 0x0000FFFF0000) >> 16;
for (i = 0; i < N; i++)
addend[i] = (unsigned short)(VACC_L[i]) + addend[i];
for (i = 0; i < N; i++)
VACC_L[i] = (short)(addend[i]);
for (i = 0; i < N; i++)
addend[i] = (unsigned short)(addend[i] >> 16);
for (i = 0; i < N; i++)
addend[i] = (unsigned short)(VACC_M[i]) + addend[i];
for (i = 0; i < N; i++)
VACC_M[i] = (short)(addend[i]);
for (i = 0; i < N; i++)
VACC_H[i] += addend[i] >> 16;
SIGNED_CLAMP_AL(state, VD);
return;
#endif
}
static void VMADL(usf_state_t * state, int vd, int vs, int vt, int e)
{
ALIGNED short ST[N];
SHUFFLE_VECTOR(ST, state->VR[vt], e);
do_madl(state, state->VR[vd], state->VR[vs], ST);
return;
}