121 lines
5.4 KiB
C
121 lines
5.4 KiB
C
/* llroundf4 - rounds four floats in to four nearest 64bit integer.
|
|
0.5 will be rounded to far from 0
|
|
Copyright (C) 2006, 2007 Sony Computer Entertainment Inc.
|
|
All rights reserved.
|
|
|
|
Redistribution and use in source and binary forms,
|
|
with or without modification, are permitted provided that the
|
|
following conditions are met:
|
|
* Redistributions of source code must retain the above copyright
|
|
notice, this list of conditions and the following disclaimer.
|
|
* Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
* Neither the name of the Sony Computer Entertainment Inc nor the names
|
|
of its contributors may be used to endorse or promote products derived
|
|
from this software without specific prior written permission.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
|
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
|
|
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
|
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
POSSIBILITY OF SUCH DAMAGE.
|
|
*/
|
|
|
|
#ifndef ___SIMD_MATH_LLROUNDF4_H___
|
|
#define ___SIMD_MATH_LLROUNDF4_H___
|
|
|
|
#include <simdmath.h>
|
|
#include <spu_intrinsics.h>
|
|
|
|
//
|
|
// Handles no exception
|
|
// over flow will return unspecified data
|
|
|
|
static inline llroundf4_t
|
|
_llroundf4 (vector float in)
|
|
{
|
|
llroundf4_t res;
|
|
vec_int4 exp;
|
|
vec_uint4 mant0, mant1, mant2, mant3;
|
|
vec_uint4 mask, mask0, mask1;
|
|
vec_uint4 sign, sign0, sign1;
|
|
vec_uint4 addend0, addend1;
|
|
vec_uint4 borrow0, borrow1;
|
|
vec_uint4 res0, res1;
|
|
int shift0, shift1, shift2, shift3;
|
|
|
|
/* Place mantissa bits (including implied most signficant
|
|
* bit) into the most significant bits of element 3. Elements
|
|
* 0, 1, and 2 are zeroed.
|
|
*/
|
|
mant0 = spu_sel(spu_rlmaskqwbyte((vec_uint4)in,-11), ((vec_uint4){0, 0, 0, 0x80000000}), ((vec_uint4){-1, -1, -1, 0x800000FF}));
|
|
mant1 = spu_sel(spu_rlmaskqwbyte((vec_uint4)in, -7), ((vec_uint4){0, 0, 0, 0x80000000}), ((vec_uint4){-1, -1, -1, 0x800000FF}));
|
|
mant2 = spu_sel(spu_rlmaskqwbyte((vec_uint4)in, -3), ((vec_uint4){0, 0, 0, 0x80000000}), ((vec_uint4){-1, -1, -1, 0x800000FF}));
|
|
mant3 = spu_sel( spu_rlqwbyte((vec_uint4)in, 1), ((vec_uint4){0, 0, 0, 0x80000000}), ((vec_uint4){-1, -1, -1, 0x800000FF}));
|
|
|
|
/* Determine how many bits to shift the mantissa to correctly
|
|
* align it into long long element 0.
|
|
*/
|
|
exp = spu_and(spu_rlmask((vec_int4)in, -23), 0xFF);
|
|
exp = spu_add(exp, -94);
|
|
shift0 = spu_extract(exp, 0);
|
|
shift1 = spu_extract(exp, 1);
|
|
shift2 = spu_extract(exp, 2);
|
|
shift3 = spu_extract(exp, 3);
|
|
|
|
/* Algn mantissa bits
|
|
*/
|
|
mant0 = spu_slqwbytebc(spu_slqw(mant0, shift0), shift0);
|
|
mant1 = spu_slqwbytebc(spu_slqw(mant1, shift1), shift1);
|
|
mant2 = spu_slqwbytebc(spu_slqw(mant2, shift2), shift2);
|
|
mant3 = spu_slqwbytebc(spu_slqw(mant3, shift3), shift3);
|
|
|
|
mask = spu_cmpgt(exp, 0);
|
|
mask0 = spu_shuffle(mask, mask, ((vec_uchar16){0,0,0,0,0,0,0,0, 4, 4, 4, 4, 4, 4, 4, 4}));
|
|
mask1 = spu_shuffle(mask, mask, ((vec_uchar16){8,8,8,8,8,8,8,8, 12,12,12,12,12,12,12,12}));
|
|
|
|
res0 = spu_shuffle(mant0, mant1,((vec_uchar16){0,1,2,3,4,5,6,7, 16,17,18,19,20,21,22,23}));
|
|
res1 = spu_shuffle(mant2, mant3,((vec_uchar16){0,1,2,3,4,5,6,7, 16,17,18,19,20,21,22,23}));
|
|
res0 = spu_and(res0, mask0);
|
|
res1 = spu_and(res1, mask1);
|
|
|
|
/* Perform round by adding 1 if the fraction bits are
|
|
* greater than or equal to .5
|
|
*/
|
|
addend0 = spu_shuffle(mant0, mant1, ((vec_uchar16){0x80,0x80,0x80,0x80,0x80,0x80,0x80,8, 0x80,0x80,0x80,0x80,0x80,0x80,0x80,24}));
|
|
addend1 = spu_shuffle(mant2, mant3, ((vec_uchar16){0x80,0x80,0x80,0x80,0x80,0x80,0x80,8, 0x80,0x80,0x80,0x80,0x80,0x80,0x80,24}));
|
|
addend0 = spu_rlmask(addend0, -7);
|
|
addend1 = spu_rlmask(addend1, -7);
|
|
// addend0 = spu_and(spu_rlqw(res0, 1), ((vec_uint4){ 0,1,0,1}));
|
|
// addend1 = spu_and(spu_rlqw(res1, 1), ((vec_uint4){ 0,1,0,1}));
|
|
res0 = spu_addx(res0, addend0, spu_rlqwbyte(spu_genc(res0, addend0), 4));
|
|
res1 = spu_addx(res1, addend1, spu_rlqwbyte(spu_genc(res1, addend1), 4));
|
|
|
|
/* Compute the two's complement of the mantissa if the
|
|
* input is negative.
|
|
*/
|
|
sign = (vec_uint4)spu_rlmaska((vec_int4)in, -31);
|
|
sign0 = spu_shuffle(sign, sign, ((vec_uchar16){0,0,0,0,0,0,0,0, 4, 4, 4, 4, 4, 4, 4, 4}));
|
|
sign1 = spu_shuffle(sign, sign, ((vec_uchar16){8,8,8,8,8,8,8,8, 12,12,12,12,12,12,12,12}));
|
|
|
|
res0 = spu_xor(res0, sign0);
|
|
res1 = spu_xor(res1, sign1);
|
|
borrow0 = spu_genb(res0, sign0);
|
|
borrow1 = spu_genb(res1, sign1);
|
|
borrow0 = spu_shuffle(borrow0, borrow0, ((vec_uchar16){4,5,6,7,0xc0,0xc0,0xc0,0xc0, 12,13,14,15,0xc0,0xc0,0xc0,0xc0}));
|
|
borrow1 = spu_shuffle(borrow1, borrow1, ((vec_uchar16){4,5,6,7,0xc0,0xc0,0xc0,0xc0, 12,13,14,15,0xc0,0xc0,0xc0,0xc0}));
|
|
res.vll[0] = (vec_llong2)spu_subx(res0, sign0, borrow0);
|
|
res.vll[1] = (vec_llong2)spu_subx(res1, sign1, borrow1);
|
|
|
|
return res;
|
|
}
|
|
|
|
#endif
|