frederik
/
oscar64
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Improve precision of float log function

This commit is contained in:
drmortalwombat 2022-11-16 20:51:07 +01:00
parent 42b4f46356
commit 74f93ca4fb
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
include/math.c
View File

@ -143,12 +143,12 @@ float exp(float f)
return s * x.f; return s * x.f;
} }
#define F_LOG_0 -2.79423993 #define F_LOG_0 -3.78712618
#define F_LOG_1 5.05984692 #define F_LOG_1 10.0957081
#define F_LOG_2 -3.49247429 #define F_LOG_2 -13.9747486
#define F_LOG_3 1.58181193 #define F_LOG_3 12.7568806
#define F_LOG_4 -0.396821126 #define F_LOG_4 -6.48114552
#define F_LOG_5 0.041845518 #define F_LOG_5 1.39045416
float log(float f) float log(float f)
{ {
@ -162,8 +162,8 @@ float log(float f)
x.f = f; x.f = f;
int ei = x.i[1]; int ei = x.i[1];
int ex = (ei >> 7) - 0x7f; int ex = (ei >> 7) - 0x7e;
x.i[1] = (ei & 0x007f) | 0x3f80; x.i[1] = (ei & 0x007f) | 0x3f00;
float g = x.f; float g = x.f;