shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[libclc] erfc: fix fp32 implementation in FTZ mode (#132390)

Browse Source 
On some implementations, the current implementation leads to slight
accuracy issues.

While the maths behind this implementation is correct, it does not take
into account the accumulation of errors coming from other operators that
do not provide correct rounding (like the exp function).

To avoid it, compute statically exp(-0.5625).

Fixes #124939
This commit is contained in:
Romaric Jodin 2025-03-24 17:08:54 +01:00 committed by GitHub
parent f10dc76f03
commit a6a56a326a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 12 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
libclc/generic/lib/math/erf.cl
View File

@ -23,7 +23,7 @@
#define erx 8.4506291151e-01f /* 0x3f58560b */
// Coefficients for approximation to erf on [00.84375]
// Coefficients for approximation to erf on [0, 0.84375]
#define efx 1.2837916613e-01f /* 0x3e0375d4 */
#define efx8 1.0270333290e+00f /* 0x3f8375d4 */
@ -39,7 +39,7 @@
#define qq4 1.3249473704e-04f /* 0x390aee49 */
#define qq5 -3.9602282413e-06f /* 0xb684e21a */
// Coefficients for approximation to erf in [0.843751.25]
// Coefficients for approximation to erf in [0.84375, 1.25]
#define pa0 -2.3621185683e-03f /* 0xbb1acdc6 */
#define pa1 4.1485610604e-01f /* 0x3ed46805 */
@ -55,7 +55,7 @@
#define qa5 1.3637083583e-02f /* 0x3c5f6e13 */
#define qa6 1.1984500103e-02f /* 0x3c445aa3 */
// Coefficients for approximation to erfc in [1.251/0.35]
// Coefficients for approximation to erfc in [1.25, 1/0.35]
#define ra0 -9.8649440333e-03f /* 0xbc21a093 */
#define ra1 -6.9385856390e-01f /* 0xbf31a0b7 */
@ -74,7 +74,7 @@
#define sa7 6.5702495575e+00f /* 0x40d23f7c */
#define sa8 -6.0424413532e-02f /* 0xbd777f97 */
// Coefficients for approximation to erfc in [1/.3528]
// Coefficients for approximation to erfc in [1/0.35, 28]
#define rb0 -9.8649431020e-03f /* 0xbc21a092 */
#define rb1 -7.9928326607e-01f /* 0xbf4c9dd4 */
@ -130,7 +130,8 @@ _CLC_OVERLOAD _CLC_DEF float erf(float x) {
// |x| < 6
float z = as_float(ix & 0xfffff000);
float r = exp(mad(-z, z, -0.5625f)) * exp(mad(z-absx, z+absx, q));
float r = exp(-z * z) * exp(mad(z - absx, z + absx, q));
r *= 0x1.23ba94p-1f; // exp(-0.5625)
r = 1.0f - MATH_DIVIDE(r, absx);
ret = absx < 6.0f ? r : ret;

11
libclc/generic/lib/math/erfc.cl
View File

@ -23,7 +23,7 @@
#define erx_f 8.4506291151e-01f /* 0x3f58560b */
// Coefficients for approximation to erf on [00.84375]
// Coefficients for approximation to erf on [0, 0.84375]
#define efx 1.2837916613e-01f /* 0x3e0375d4 */
#define efx8 1.0270333290e+00f /* 0x3f8375d4 */
@ -39,7 +39,7 @@
#define qq4 1.3249473704e-04f /* 0x390aee49 */
#define qq5 -3.9602282413e-06f /* 0xb684e21a */
// Coefficients for approximation to erf in [0.843751.25]
// Coefficients for approximation to erf in [0.84375, 1.25]
#define pa0 -2.3621185683e-03f /* 0xbb1acdc6 */
#define pa1 4.1485610604e-01f /* 0x3ed46805 */
@ -55,7 +55,7 @@
#define qa5 1.3637083583e-02f /* 0x3c5f6e13 */
#define qa6 1.1984500103e-02f /* 0x3c445aa3 */
// Coefficients for approximation to erfc in [1.251/0.35]
// Coefficients for approximation to erfc in [1.25, 1/0.35]
#define ra0 -9.8649440333e-03f /* 0xbc21a093 */
#define ra1 -6.9385856390e-01f /* 0xbf31a0b7 */
@ -74,7 +74,7 @@
#define sa7 6.5702495575e+00f /* 0x40d23f7c */
#define sa8 -6.0424413532e-02f /* 0xbd777f97 */
// Coefficients for approximation to erfc in [1/.3528]
// Coefficients for approximation to erfc in [1/0.35, 28]
#define rb0 -9.8649431020e-03f /* 0xbc21a092 */
#define rb1 -7.9928326607e-01f /* 0xbf4c9dd4 */
@ -131,7 +131,8 @@ _CLC_OVERLOAD _CLC_DEF float erfc(float x) {
float ret = 0.0f;
float z = as_float(ix & 0xfffff000);
float r = exp(mad(-z, z, -0.5625f)) * exp(mad(z - absx, z + absx, q));
float r = exp(-z * z) * exp(mad(z - absx, z + absx, q));
r *= 0x1.23ba94p-1f; // exp(-0.5625)
r = MATH_DIVIDE(r, absx);
t = 2.0f - r;
r = x < 0.0f ? t : r;