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
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
|
/*========================== begin_copyright_notice ============================
Copyright (C) 2017-2021 Intel Corporation
SPDX-License-Identifier: MIT
============================= end_copyright_notice ===========================*/
#include "../include/BiF_Definitions.cl"
#include "../../Headers/spirv.h"
#include "../IMF/FP32/ln_s_la.cl"
#if defined(cl_khr_fp64)
#include "../IMF/FP64/ln_d_la.cl"
#include "../IMF/FP64/ln_d_la_noLUT.cl"
#endif // defined(cl_khr_fp64)
float __attribute__((overloadable)) __spirv_ocl_log( float x )
{
#if 0
// This version is ever so slightly faster (<1%) than the version below,
// however it is almost a full ULP less precise in some cases, so we'll
// stick with the full expansion for now.
return __spirv_ocl_log2(x) * M_LN2_F;
#else
float result;
if(BIF_FLAG_CTRL_GET(FastRelaxedMath))
{
result = __spirv_ocl_native_log(x);
}
// Denorm checking is to work-around a llvm issue that demote
// "(float) x > 0.0f" to " (half)x > (half)0.0f" (log(half).
// This causes the inaccurate result with -cl-denorms-are-zero.
else if( __intel_relaxed_isfinite(x) &
((!BIF_FLAG_CTRL_GET(FlushDenormals) & (x > 0.0f)) |
( BIF_FLAG_CTRL_GET(FlushDenormals) & (as_int(x) > 0x7FFFFF))) )
//else if( __intel_relaxed_isfinite(x) & ( x > 0.0f ) )
{
if(BIF_FLAG_CTRL_GET(UseMathWithLUT))
{
result = __ocl_svml_logf(x);
}
else
{
// We already know that we're positive and finite, so
// we can use this very cheap check for normal vs.
// subnormal inputs:
float s = x * ( 1 << FLOAT_MANTISSA_BITS );
float e = ( x < FLT_MIN ) ? -FLOAT_MANTISSA_BITS : 0.0f;
x = ( x < FLT_MIN ) ? s : x;
const int magic = 0x3f2aaaab;
int iX = as_int(x) - magic;
int iR = ( iX & FLOAT_MANTISSA_MASK ) + magic;
e += iX >> FLOAT_MANTISSA_BITS;
float sR = as_float(iR) - 1.0f;
float sP = as_float(0xbe0402c8);
sP = __spirv_ocl_fma( sP, sR, as_float(0x3e0f335d));
sP = __spirv_ocl_fma( sP, sR, as_float(0xbdf9889e));
sP = __spirv_ocl_fma( sP, sR, as_float(0x3e0f6b8c));
sP = __spirv_ocl_fma( sP, sR, as_float(0xbe2acee6));
sP = __spirv_ocl_fma( sP, sR, as_float(0x3e4ce814));
sP = __spirv_ocl_fma( sP, sR, as_float(0xbe7fff78));
sP = __spirv_ocl_fma( sP, sR, as_float(0x3eaaaa83));
sP = __spirv_ocl_fma( sP, sR, as_float(0xbf000000));
sP = sP * sR;
sP = __spirv_ocl_fma( sP, sR, sR);
sP = __spirv_ocl_fma( e, as_float(0x35bfbe8e), sP);
sP = __spirv_ocl_fma( e, as_float(0x3f317200), sP);
result = sP;
}
}
else
{
// If we get here, we're either infinity, NaN, or negative.
// The native log2 handles all of these cases. Note, we don't
// have to multiply by M_LN2_F, since the result in
// these cases is NaN or +/- infinity, therefore the multiply
// is irrelevant and unnecessary.
result = __spirv_ocl_native_log2(x);
}
return result;
#endif
}
GENERATE_SPIRV_OCL_VECTOR_FUNCTIONS_1ARG_LOOP( log, float, float, f32 )
#if defined(cl_khr_fp64)
INLINE double __attribute__((overloadable)) __spirv_ocl_log( double x )
{
double result;
if (BIF_FLAG_CTRL_GET(UseHighAccuracyMath)) {
result = __ocl_svml_log_noLUT(x);
} else {
result = __ocl_svml_log(x);
}
return result;
}
GENERATE_SPIRV_OCL_VECTOR_FUNCTIONS_1ARG_LOOP( log, double, double, f64 )
#endif // defined(cl_khr_fp64)
#if defined(cl_khr_fp16)
INLINE half __attribute__((overloadable)) __spirv_ocl_log( half x )
{
return (half)__spirv_ocl_log((float)x);
}
GENERATE_SPIRV_OCL_VECTOR_FUNCTIONS_1ARG_LOOP( log, half, half, f16 )
#endif // defined(cl_khr_fp16)
#if defined(IGC_SPV_INTEL_bfloat16_arithmetic)
INLINE bfloat __attribute__((overloadable)) __spirv_ocl_log( bfloat x )
{
return __spirv_ocl_native_log(x);
}
GENERATE_SPIRV_OCL_VECTOR_FUNCTIONS_1ARG_LOOP( log, bfloat, bfloat, )
#endif // defined(IGC_SPV_INTEL_bfloat16_arithmetic)
|
