GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-08-12 11:49:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

Also replaced uint32_t with unsigned int to make the code more portable

Browse Source
This commit is contained in:
Benoit Steiner 2016-03-11 19:34:21 -08:00
parent 1ca8c1ec97
commit eecd914864
1 changed files with 11 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
Eigen/src/Core/arch/CUDA/Half.h
View File

@ -210,7 +210,7 @@ static inline EIGEN_DEVICE_FUNC __half raw_uint16_to_half(unsigned short x) {
} }
union FP32 { union FP32 {
uint32_t u; unsigned int u;
float f; float f;
}; };
@ -223,10 +223,10 @@ static inline EIGEN_DEVICE_FUNC __half float_to_half_rtne(float ff) {
const FP32 f32infty = { 255 << 23 }; const FP32 f32infty = { 255 << 23 };
const FP32 f16max = { (127 + 16) << 23 }; const FP32 f16max = { (127 + 16) << 23 };
const FP32 denorm_magic = { ((127 - 15) + (23 - 10) + 1) << 23 }; const FP32 denorm_magic = { ((127 - 15) + (23 - 10) + 1) << 23 };
uint32_t sign_mask = 0x80000000u; unsigned int sign_mask = 0x80000000u;
__half o = { 0 }; __half o = { 0 };
uint32_t sign = f.u & sign_mask; unsigned int sign = f.u & sign_mask;
f.u ^= sign; f.u ^= sign;
// NOTE all the integer compares in this function can be safely // NOTE all the integer compares in this function can be safely
@ -246,10 +246,10 @@ static inline EIGEN_DEVICE_FUNC __half float_to_half_rtne(float ff) {
// and one integer subtract of the bias later, we have our final float! // and one integer subtract of the bias later, we have our final float!
o.x = f.u - denorm_magic.u; o.x = f.u - denorm_magic.u;
} else { } else {
uint32_t mant_odd = (f.u >> 13) & 1; // resulting mantissa is odd unsigned int mant_odd = (f.u >> 13) & 1; // resulting mantissa is odd
// update exponent, rounding bias part 1 // update exponent, rounding bias part 1
f.u += ((uint32_t)(15 - 127) << 23) + 0xfff; f.u += ((unsigned int)(15 - 127) << 23) + 0xfff;
// rounding bias part 2 // rounding bias part 2
f.u += mant_odd; f.u += mant_odd;
// take the bits! // take the bits!
@ -267,11 +267,11 @@ static inline EIGEN_DEVICE_FUNC float half_to_float(__half h) {
return __half2float(h); return __half2float(h);
#else #else
const FP32 magic = { 113 << 23 }; const FP32 magic = { 113 << 23 };
const uint32_t shifted_exp = 0x7c00 << 13; // exponent mask after shift const unsigned int shifted_exp = 0x7c00 << 13; // exponent mask after shift
FP32 o; FP32 o;
o.u = (h.x & 0x7fff) << 13; // exponent/mantissa bits o.u = (h.x & 0x7fff) << 13; // exponent/mantissa bits
uint32_t exp = shifted_exp & o.u; // just the exponent unsigned int exp = shifted_exp & o.u; // just the exponent
o.u += (127 - 15) << 23; // exponent adjust o.u += (127 - 15) << 23; // exponent adjust
// handle exponent special cases // handle exponent special cases