cfbolz changed the topic of #pypy to: #pypy PyPy, the flexible snake | IRC logs: and | the pypy angle is to shrug and copy the implementation of CPython as closely as possible, and staying out of design decisions
MiguelX413 has quit [*.net *.split]
xcm_ has quit [*.net *.split]
antocuni_ has quit [*.net *.split]
antocuni has joined #pypy
xcm_ has joined #pypy
xcm_ has quit [Signing in (xcm_)]
xcm_ has joined #pypy
sthalik has joined #pypy
<sthalik> hey. remember that problem you had with making too many jit specializations?
<sthalik> you've had some counter for each function and it was growing into infinity and with it, the amount of jitted functions
<sthalik> you eventually arrived at some heuristic. it was something like -- every 100k iterations, divide every counter by 2
<sthalik> I'm having a very similar problem in my (completely unrelated) code. I wonder what a good heuristic for decaying counters is in the first place
<nimaje> as sat-solvers have great needs for heuristics, maybe look what they do, what you have now sounds similar to Variable State-Independent Decaying Sum (VSIDS) a heuristic many sat-solvers use, because it is fast to calculate and pretty good for their problem
mattip has quit [Ping timeout: 244 seconds]
mattip has joined #pypy
[Arfrever] has quit [Read error: Connection reset by peer]
[Arfrever] has joined #pypy