3 ms·

It should have a trophy case then :) About nice found optimizations

• celrod 6y ago
  Yeah, I'd like to know more about what it can do. Can it optimize sequences of shuffles? If so, it'd be worth a try, even if it takes hours (probably doesn't take that long?), I could let it run until it finds faster sequences than what I have now, and update my code. But for it to be worth the time for me to figure out how to install and use it, I need some sort of expectation for what I'd get out of trying this instead of something else.
  • kzrdude 6y ago
    it's not an optimizer and so it's runtime doesn't matter really. It's intended to find new rules (patterns) that can in turn be implemented in llvm as new optimizations. So this thing is just a research step.
• millstone 6y ago
  The trophy case is here: https://www.cs.utah.edu/~regehr/souper/ https://www.cs.utah.edu/~regehr/souper/ Example program 6: https://www.cs.utah.edu/~regehr/souper/output_6.html https://www.cs.utah.edu/~regehr/souper/output_6.html int a, b; void fn1() { b = a * ~1 == 1; } Here we multiply an int by -2. LLVM emitted the multiply and compare, but Souper proved that no int multiplied by -2 can equal 1, so this could be optimized to 0. By its nature Souper becomes less effective over time as its findings are incorporated into LLVM, so most of the trophies are old!