You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
72 lines
2.8 KiB
72 lines
2.8 KiB
/*
|
|
* Copyright (c) 2014 Advanced Micro Devices, Inc.
|
|
* Copyright (c) 2016 Aaron Watry
|
|
*
|
|
* Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
* of this software and associated documentation files (the "Software"), to deal
|
|
* in the Software without restriction, including without limitation the rights
|
|
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
* copies of the Software, and to permit persons to whom the Software is
|
|
* furnished to do so, subject to the following conditions:
|
|
*
|
|
* The above copyright notice and this permission notice shall be included in
|
|
* all copies or substantial portions of the Software.
|
|
*
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
|
|
* THE SOFTWARE.
|
|
*/
|
|
#if __CLC_FPSIZE == 32
|
|
#ifdef __CLC_SCALAR
|
|
_CLC_OVERLOAD _CLC_DEF __CLC_GENTYPE fdim(__CLC_GENTYPE x, __CLC_GENTYPE y) {
|
|
if (__builtin_isnan(x) || __builtin_isnan(y))
|
|
return as_float(QNANBITPATT_SP32);
|
|
return fmax(x - y, 0.0f);
|
|
}
|
|
#define __CLC_FDIM_VEC(width) \
|
|
_CLC_OVERLOAD _CLC_DEF float##width fdim(float##width x, float##width y) { \
|
|
/* Determine if x or y is NaN. */ \
|
|
/* Vector true is -1, i.e. all-bits-set, and NaN==NaN is false. */ \
|
|
/* If either is NaN, then ~((x==x) & (y==y)) will be 0 (e.g. ~(-1)), as will n. */ \
|
|
int##width n = ~((x == x) & (y == y)) & QNANBITPATT_SP32; \
|
|
/* Calculate x-y if x>y, otherwise positive 0, again taking */ \
|
|
/* advantage of vector true being all-bits-set. */ \
|
|
int##width r = (x > y) & as_int##width(x - y); \
|
|
return as_float##width(n | r); \
|
|
}
|
|
__CLC_FDIM_VEC(2)
|
|
__CLC_FDIM_VEC(3)
|
|
__CLC_FDIM_VEC(4)
|
|
__CLC_FDIM_VEC(8)
|
|
__CLC_FDIM_VEC(16)
|
|
#undef __CLC_FDIM_VEC
|
|
#endif
|
|
#endif
|
|
|
|
#if __CLC_FPSIZE == 64
|
|
#ifdef __CLC_SCALAR
|
|
_CLC_OVERLOAD _CLC_DEF __CLC_GENTYPE fdim(__CLC_GENTYPE x, private __CLC_GENTYPE y) {
|
|
long n = -(isnan(x) | isnan(y)) & QNANBITPATT_DP64;
|
|
long r = -(x > y) & as_long(x - y);
|
|
return as_double(n | r);
|
|
}
|
|
#define __CLC_FDIM_VEC(width) \
|
|
_CLC_OVERLOAD _CLC_DEF double##width fdim(double##width x, double##width y) { \
|
|
/* See comment in float implementation for explanation. */ \
|
|
long##width n = ~((x == x) & (y == y)) & QNANBITPATT_DP64; \
|
|
long##width r = (x > y) & as_long##width(x - y); \
|
|
return as_double##width(n | r); \
|
|
}
|
|
__CLC_FDIM_VEC(2)
|
|
__CLC_FDIM_VEC(3)
|
|
__CLC_FDIM_VEC(4)
|
|
__CLC_FDIM_VEC(8)
|
|
__CLC_FDIM_VEC(16)
|
|
#undef __CLC_FDIM_VEC
|
|
#endif
|
|
#endif
|