1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
|
/*========================== begin_copyright_notice ============================
Copyright (C) 2017-2021 Intel Corporation
SPDX-License-Identifier: MIT
============================= end_copyright_notice ===========================*/
// SVML code
static float __ocl_svml_sinpif_noLUT( float a )
{
float result;
float sAbsX = SPIRV_OCL_BUILTIN(fabs, _f32, )( a );
if( sAbsX > as_float( 0x4A800000 ) ) // 2^22
{
float sShifter =
( sAbsX < as_float( 0x4F000000 ) ) ? // 2^31
as_float( 0x4FC00000 ) :
0.0f;
a = a - (sShifter + a - sShifter);
}
float sN = a + as_float( 0x4B400000 );
uint usN = as_uint( sN ) << 31;
sN = sN - as_float( 0x4B400000 );
float sR = M_PI_F * (a - sN);
float sR2 = sR * sR;
sR = as_float( as_uint( sR ) ^ usN );
float sP = as_float( 0x362EDEF8 );
sP = SPIRV_OCL_BUILTIN(fma, _f32_f32_f32, )( sP, sR2, as_float( 0xB94FB7FF ) );
sP = SPIRV_OCL_BUILTIN(fma, _f32_f32_f32, )( sP, sR2, as_float( 0x3C08876A ) );
sP = SPIRV_OCL_BUILTIN(fma, _f32_f32_f32, )( sP, sR2, as_float( 0xBE2AAAA6 ) );
sP = sP * sR2;
result = SPIRV_OCL_BUILTIN(fma, _f32_f32_f32, )( sP, sR, sR );
float n = SPIRV_OCL_BUILTIN(nan, _i32, )(0);
result = __intel_relaxed_isinf( sAbsX ) ? n : result;
return result;
}
|
