Also, to make sure that it's not some MathConst-related thing I changed `2pi` to `2.0f0` in the loop and saw no difference.
peace, s On Sat, Jun 28, 2014 at 12:50 PM, Spencer Russell <[email protected]> wrote: > Dear big Julia meta-brain, > > I'm seeing some strange allocation behavior that I don't understand, so I > built a minimal example that seems to be doing the same things. Here's the > sample: > > https://gist.github.com/ssfrr/b4885e3fc10ea3f2ab9d > > 1. I'm getting different allocation numbers when I use @allocated and > @time > 2. My loop functions seem to allocate differently depending on whether > they're called from within a function or at global scope, though they're > only referencing their arguments > 3. When called from global scope the while loop appears to allocate > much less than the for loop. (by over 150x) > > Maybe this is just a bug where @time is over-reporting allocation? > > peace, > s >
