This post originated from an RSS feed registered with Python Buzz by Ryan Tomayko. Original Post: Performance Retrospective in PEP 3146 -- Merging Unladen Swallow into CPython Feed Title: Ryan Tomayko (weblog/python) Feed URL: http://tomayko.com/feed/ Feed Description: Entries classified under Python. Latest Python Buzz Posts Latest Python Buzz Posts by Ryan Tomayko Latest Posts From Ryan Tomayko (weblog/python)

Shame:

Our initial goal for Unladen Swallow was a 5x performance improvement over CPython 2.6. We did not hit that, nor to put it bluntly, even come close. Why did the project not hit that goal, and can an LLVM-based JIT ever hit that goal?

Here’s the performance comparison itself. Most gains were under 1.5x and memory usage grew significantly in every benchmark. Startup time also suffered.

As mentioned in the Performance Retrospective, they had to divert a lot of energy from performance work to fixing LLVM bugs and axe grinding on debugging/profiling tools. I have my fingers crossed that they can pull off the 5x gains in the next round. Let’s hope so, I'd demand at least that much to justify this:

In order to use LLVM, Unladen Swallow has introduced C++ into the core CPython tree and build process.

Slipperyest of slopes.

# | Discuss