News for hackers. Stuff that matters.
Re:Compiler Optimizations are Awesome by mpweiher on Thursday June 01, 2017 @05:32AM (#14458966)
> replacing your divisions by a different algorithm

No they're not. I don't actually specify the algorithm to be used for division, I just write "/" (usually), and so the compiler (or runtime or whoever) is certainly free to implement that "high level" specification (I want these two quantities divided) in any way it sees fit.

Re:Compiler Optimizations are Awesome by DannyBee on Thursday June 01, 2017 @04:37PM (#14464388)
I'm really unsure why you want to be this pedantic, but okay. You are making a lot of assumptions here. My point is that bunch will idiom recognize division algorithms and change them. So yes, yes they do!
Re:Compiler Optimizations are Awesome by mpweiher on Friday June 02, 2017 @03:16AM (#14468089)
> pedantic

I am being precise where you are incredibly sloppy.

> You are making a lot of assumptions here.

Really? Which assumptions am I making? Try to be precise.

Considering the fact that you are calling everyone else idiots who don't have any data, the fact that what you write has (a) no data to back it up and (b) is usually trivially, obviously and comically wrong doesn't exactly help your argument.

> ...will idiom recognize division algorithms and change them

Really? Which ones? How is this useful? How many real-world programs actually code a division "algorithm"? I don't think I've seen that once in the last 20+ years or so, but of course YMMV.

UPDATE: Last I remember, "idiom recognition" was invented for the sole purpose of cheating on industry standard benchmarks, for example replacing the Dhrystone with a computed result. Has this changed?