Yes, I know PyPy, while I use C or Cython heavily when I want performance. I've setuped PyPy on http://www.techempower.com/benchmarks/ As you can see on there, Go is faster than PyPy. Additionally, Go is memory efficient than PyPy. Since Go can use multicore without forking, there are no thundering herd problem.
I hope asyncio communicates with Go nicely. gRPC [1] may be one answer about building high performance web service with Python and Go. _[1]: https://github.com/grpc/grpc On Fri, Mar 13, 2015 at 2:32 AM, Glyph Lefkowitz <[email protected]> wrote: > On Mar 12, 2015, at 6:19 AM, INADA Naoki <[email protected]> wrote: > > > IMO, When I really needs high performance (1000~ req/(sec*cores)), I > use Go in these days. > > > While I can see why you would use Go, comparing it to CPython is really not > fair. If you want performance in Python, try PyPy, which is on the order of > 7x faster than CPython, in aggregate: see http://speed.pypy.org for more > detailed analysis. > > You won't be able to use Tulip with it just yet (although they are of course > working towards a version of pypy3 which will support it), but as you can > see from the right side of the graph, PyPy has been tracking its performance > on Twisted for quite some time. > > It bothers me that so many people try "Python" (meaning CPython) and then > try Go and decide "go is faster" when in fact PyPy can often get competitive > performance; if you care about performance, PyPy should always be your > default. > > -glyph > -- INADA Naoki <[email protected]>
