⇤ ← Revision 1 as of 2010-12-05 17:20:25
838
Comment: Added a new page describing the jython code experiments that I'm conducting
|
← Revision 2 as of 2010-12-05 17:23:39 ⇥
853
changed the link
|
Deletions are marked like this. | Additions are marked like this. |
Line 14: | Line 14: |
Before venturing into these experiments, I tried to speedup the {{{PyByteCode}}}. You can view the page on that at PyByteCodeSpeedups | Before venturing into these experiments, I tried to speedup the {{{PyByteCode}}}. You can view the page on that at CodeSpeedupExperiments/PyByteCode |
Code Speedup Experiments
This is a log of the speed up experiments that we are conducting.
Here are some experiments we could conduct. For each, we would compare against a pure Python implementation: Fib hand optimized to Java, no frames or other call setup support as seen in PyBaseCode.
- Fib pure Python, but with indy.
- Nested loops. Array multiplication?
Guard failure of a loop, call into interpreter (PyBytecode) with well-known offset and synthesized frame. (Limited CPS.)
- Etc
Related Speedups
Before venturing into these experiments, I tried to speedup the PyByteCode. You can view the page on that at CodeSpeedupExperiments/PyByteCode
Repository
All the work is being done at a mercurial repo at http://bitbucket.org/shashank/jython-experiments/