Will It Optimize? See how well you can anticipate gcc's optimizer (2010)

Re:Will It Optimize? See how well you can anticipate gcc's o... by ISL on Sunday April 07, 2013 @11:01AM (#5507410)

Question 4 points at a huge surprise. With a scientific computing project, I got (if I recall correctly), correct and expected output at no optimization and -O1, and patently incorrect output at -O2 and -O3.

Cost: one sleepless night in college."What do you mean, optimization doesn't just make it faster?!?"

[ Parent | on HN ]

Re:Will It Optimize? See how well you can anticipate gcc's o... by haberman on Sunday April 07, 2013 @07:32PM (#5509206)

It's possible that both were "correct" but that -O2 and -O3 gave different results due to x86 extended precision. For some interesting reading, see:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=323 (a centi-bug of back and forth between "it's broken!" and "no it's not, and we're not going to change the compiler.")

http://gcc.gnu.org/ml/gcc/2003-08/msg01183.html (a really informative thread about the issue)

[ Parent | on HN ]

Re:Will It Optimize? See how well you can anticipate gcc's o... by widdershins on Sunday April 07, 2013 @07:07PM (#5509126)

Really? Wow, I thought optimisations had to provably avoid affecting the correctness of your program. Hence:

"This need to avoid imprecise intermediate results often prevents compilers from exploiting mathematical properties like associativity when optimizing FP arithmetic."

from the explanation under Q4.

[ Parent | on HN ]