+Speeding up existing code using PyPy
+I spent quite some time profiling existing python programs under PyPy.
+This talk will walk through an existing Python library (undecided which yet)
+and showcase how to write benchmarks, how to find bottlenecks, how to analyze
+them and how to improve them when running on the PyPy interpreter and what
+are the theoretical and pracitcal limits.
+In this talk I would like to share my experience when optimizing existing
+Python codebases. I spend copious amounts of time staring at profiling data,
+improving profilers to see anything and improving PyPy to work better
+on real-life workloads. I would like to give the audience insight what
+sort of constructs are optimized by PyPy, what sort of constructs can
+possibly be optimized and which ones are out of question. This talk is
+an intermediate one and assumes good enough knowledge of Python to understand
+code of a given library (Twisted, Django, Flask, Gunicorn and some stdlib
+module are potential candidates), however no prior knowledge of PyPy or
+the processor performance characteristics is necessary.