You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+1Lines changed: 1 addition & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -30,6 +30,7 @@ Common uses include computing dot products, matrix multiplication, evaluating po
30
30
For a simple example of how one can use fused multiply-add, here's how we can compute the exact product of Numbers $a$ and $b$ with no rounding:
31
31
32
32
`let high = a * b` (using standard IEEE 754 multiplication)
33
+
33
34
`let err = Math.fma(a, b, -high)`
34
35
35
36
At this point, barring overflow or underflow of Number's exponent range, $MV(a) × MV(b) = MV(high) + MV(err)$, with $high$ containing the most significant bits of the result and $err$ containing the least significant bits of the result.
0 commit comments