[pypy-svn] r25603 - pypy/dist/pypy/doc/discussion

pedronis at codespeak.net pedronis at codespeak.net
Sat Apr 8 23:54:49 CEST 2006


Author: pedronis
Date: Sat Apr  8 23:54:48 2006
New Revision: 25603

Modified:
   pypy/dist/pypy/doc/discussion/translation-swamp.txt
Log:
(cfbolz, pedronis)

update status of work/issues.



Modified: pypy/dist/pypy/doc/discussion/translation-swamp.txt
==============================================================================
--- pypy/dist/pypy/doc/discussion/translation-swamp.txt	(original)
+++ pypy/dist/pypy/doc/discussion/translation-swamp.txt	Sat Apr  8 23:54:48 2006
@@ -5,27 +5,23 @@
 
  * understand nondeterminism after rtyping
  
- * nondeterminism of insert stackcheck:
-
-    * (DONE) move it after inlining
-    * (DONE) avoid inserting it multiple times in the same block
-
  * experiment with different heuristics:
  
-    * weigh backedges more
+    * weigh backedges more (TESTING)
     * consider size of outer function
-    * consider number of arguments
+    * consider number of arguments (TESTING)
 
- * find a more deterministic inlining order (number of calls?)
+ * find a more deterministic inlining order (TESTING using number of callers)
 
  * experiment with using a base inlining threshold and then drive inlining by
    malloc removal possibilities (using escape analysis)
 
  * move the inlining of gc helpers just before emitting the code.
-   throw the graph away (TESTING)
+   throw the graph away (TESTING, need to do a new framework translation)
 
  * for gcc: use just one implement file (TRIED: turns out to be a bad idea,
-   because gcc uses too much ram)
+   because gcc uses too much ram). Need to experiment more now that
+   inlining should at least be more deterministic!
 
 things to improve the framework gc
 ==================================



More information about the Pypy-commit mailing list