"On the DLX with this implementation, there is no way to avoid a one- cycle stall on branch taken through programming or compiler optimiza- tion. However, some RISC architectures provide a “branch-and-execute” instruction, which always executes the next instruction in sequence before performing the b"
"DEFINITION 2.5 Two computations are equivalent if, on the same inputs, they produce identical values for output variables at the time output statements are executed and the output statements are executed in the same order. ... It is clear that a compiler transformation should never introduce an erro"
Except for the nasty translation, this book is awesome for those who want to understand more about compiler based on architecture.
I would recommend those to read the original English version.