The upper bound of mean square error is given for the two algorithms of fixed and block floating point.
给出了定点与块浮点两种算法时的均方误差上限。
Present a improved block floating point algorithm which solves the overflow in long point FFT effectively and improves precision.
采用了一种改进的块浮点算法，有效避免了溢出问题并且提高了精度。
Block floating-point arithmetic is used to enhance the dynamic range and computation accuracy.
采用块浮点算法以提高动态范围和运算精度。
应用推荐