Commits

Author Commit Message Labels Comments Date
Carl Friedrich Bolz
don't actually write anything to disc, to make the result repeatable
Carl Friedrich Bolz
gah
Carl Friedrich Bolz
run psyco more robustly
Carl Friedrich Bolz
(l.diekmann): check track_memory capability with own process, so we don't have to run it with root anymore
Carl Friedrich Bolz
this runner script now lives in another repo
Carl Friedrich Bolz
do parsing manually. Also stop counting virtual_ref as news
Carl Friedrich Bolz
fix warning when trying to track memory
Carl Friedrich Bolz
also track memory
Carl Friedrich Bolz
different approach to doing psyco
Carl Friedrich Bolz
diff many files at once
Carl Friedrich Bolz
a better set of optimizations: string does virtualization as well and heap dilutes the results
Carl Friedrich Bolz
make a main method
Carl Friedrich Bolz
put logs into their own directory
Carl Friedrich Bolz
a script to summarize and diff operation logs
Carl Friedrich Bolz
remove warmups
Carl Friedrich Bolz
a benchmark script to run all the executables needed for my thesis
Carl Friedrich Bolz
hack to allow the setting of PYPYLOG
Carl Friedrich Bolz
nicer json files
Carl Friedrich Bolz
psyco runs only with 2.6, not 2.7
Carl Friedrich Bolz
run sympy benchmarks the normal number of times, they are fast
Maciej Fijalkowski
yet another necessary fix (this is getting obscure)
Amaury Forgeot d'Arc
Simplify BUILD_LIST_FROM_ARG again, and hope that benchmarks pass
Maciej Fijalkowski
hopefully the minimal change to make benchmark run again
Maciej Fijalkowski
I think I disabled translate by chance
Maciej Fijalkowski
disable pypy translation benchmark until we decide what to do
Maciej Fijalkowski
reenable web bench
Maciej Fijalkowski
hopefully this is the last special case for errors
Maciej Fijalkowski
hopefully fix uploading
Maciej Fijalkowski
Merge csenger benchmark runner changes, IMO makes sense
Carsten Senger
change --pypy-c to --changed and leave b/w compat option
  1. Prev
  2. Next