I watched the new Pyjion video. Their current excuse for not using PyPy is that it uses too much memory on the n-body benchmark; it uses around 6x what CPython uses. Is this a reasonable complaint?
gef has joined #pypy
aleale has quit [Quit: Connection closed for inactivity]
Corbin: I think n-body benchmark itself consumes virtually no memory. PyPy generally consumes quite a bit to start (way more than cpython) and consumes some memory for the JIT. It's a legit complaint for certain kind of applications I think, but I would expect in this particular case that they were looking at post-factum rationalization
at least suspect, not expect
the "pypy always uses 6x more memory because of n body benchmark" is just not true, I don't think you get much info from n-body benchmark other than interpreter startup memory
That's a fair suspicion.
The entire video collection is a little depressing. It feels like folks really are shackling themselves to C extensions forever.
gef has quit [Ping timeout: 264 seconds]
glyph has quit [Read error: Connection reset by peer]