arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1304.4292math.NAcs.NA
收藏

浮点运算中单比特翻转引起的相对误差

Relative error due to a single bit-flip in floating-point arithmetic

Bradley R. Lowery

AI总结:

研究单比特翻转在浮点数运算中的相对误差影响,采用IEEE 754双精度标准,分析不同误差阈值下的概率分布。

摘要:

我们考虑浮点数中单比特翻转引起的误差。假设使用IEEE 754双精度运算,将二进制浮点数编码在64位字中。假设比特翻转是随机发生的,等概率地影响任意一位。为了减少对初始浮点数的假设,我们假设其在所有归一化数中均匀分布。在此框架下,我们的发现可总结为:在归一化浮点数中,单比特翻转导致相对误差小于10^-11的概率超过25%;导致相对误差小于10^-6的概率超过50%;等等。

英文摘要:

We consider the error due to a single bit-flip in a floating point number. We assume IEEE 754 double precision arithmetic, which encodes binary floating point numbers in a 64-bit word. We assume that the bit-flip happens randomly so it has equi-probability (1/64) to hit any of the 64 bits. Since we want to mitigate the assumption on our initial floating-point number, we assume that it is uniformly picked among all normalized number. With this framework, we can summarize our findings as follows. The probability for a single bit flip to cause a relative error less than 10^-11 in a normalized floating-point number is above 25%; The probability for a single bit flip to cause a relative error less than 10^-6 in a normalized floating-point number is above 50%; Etc.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑