在AArch64状态下，哪种FPCR配置下的ARM FP指令符合IEEE 754-2008？-解网

问：

一个简单的问题：在哪种 FPCR 配置 ARM FP 指令下符合 IEEE 754-2008？

我假设符合 IEEE 754-2008 的 FPCR 配置是：

- AHP (Alternative half-precision control bit): 0
- DN (Default NaN use for NaN propagation): 0
- FZ (Flushing denormalized numbers to zero control bit): 0
- RMode (Rounding Mode control field): 0 (Round to Nearest (RN) mode)
- Stride (This field has no function in AArch64 state): x (don't care)
- FZ16 (Flushing denormalized numbers to zero control bit on half-precision): 0
- Len (This field has no function in AArch64 state): x (don't care)
- IDE (Input Denormal floating-point exception trap enable): 0
- EBF (Controls the numeric behaviors of BFloat16 dot product calculations performed by the BFDOT, BFMMLA, BFMOPA, and BFMOPS instructions): x (don't care)
- IXE (Inexact floating-point exception trap enable): x (don't care)
- UFE (Underflow floating-point exception trap enable): x (don't care)
- OFE (Overflow floating-point exception trap enable): x (don't care)
- DZE (Divide by Zero floating-point exception trap enable): x (don't care)
- IOE (Invalid Operation floating-point exception trap enable): x (don't care)
- NEP (Controls how the output elements other than the lowest element of the vector are determined for Advanced SIMD scalar instructions): 0
- AH (Alternate Handling. Controls alternate handling of floating-point numbers): 0
- FIZ (Flush Inputs to Zero. Controls whether single-precision, double-precision and BFloat16 input operands that are denormalized numbers are flushed to zero): 0

这是对的吗？

IEEE-754 标准的浮点 ARM FPU

@old_timer 正如我所看到的，关于IEEE 754-2008的一致性，没有一个放之四海而皆准的声明。相反，有每个功能的语句。例如：“将非正规化刷新为零与 IEEE 754 标准不兼容，当需要 IEEE 754 兼容性时，不得使用”。因此，、和需要为 0。但是，并非所有 FP 功能都具有这些（或类似）语句。困惑（例如）。FZ16FZFIZ

答： 暂无答案

上一个：“四舍五入前”（而不是“四舍五入后”）的微小的基本原理是什么？

下一个：从 C 中的双精度中提取“真实”科学指数#

在AArch64状态下，哪种FPCR配置下的ARM FP指令符合IEEE 754-2008？

Under which FPCR configuration ARM FP instructions conform to IEEE 754-2008 in AArch64 state?

评论