Flop Count Calculations + x86 Measurements + c13 Fix

This commit is contained in:
[Quy Anh] «Elliot» Nguyen
2026-06-17 07:28:11 +02:00
parent 42b3c7097e
commit a7d45665da
225 changed files with 19699 additions and 172 deletions
@@ -180,6 +180,7 @@ static inline __m256 exp256_ps(__m256 x) {
const __m256 ln2 = _mm256_set1_ps(0.6931471805599453f);
__m256 r = _mm256_fnmadd_ps(nf, ln2, x); // # 8 MULTs, 8 ADDs
const __m256 c13 = _mm256_set1_ps(c13_scalar);
const __m256 c12 = _mm256_set1_ps(c12_scalar);
const __m256 c11 = _mm256_set1_ps(c11_scalar);
const __m256 c10 = _mm256_set1_ps(c10_scalar);
@@ -194,7 +195,7 @@ static inline __m256 exp256_ps(__m256 x) {
const __m256 c1 = _mm256_set1_ps(c1_scalar);
const __m256 c0 = _mm256_set1_ps(c0_scalar);
__m256 p = c12;
__m256 p = c13;
// # 13 * (8 MULTs, 8 ADDs)
// # = 104 MULTs, 104 ADDs