00:00
Atque has joined #pypy
00:19
slav0nic has quit [Ping timeout: 252 seconds]
05:14
Atque has quit [Remote host closed the connection]
05:15
Atque has joined #pypy
05:26
luckydonald has quit [*.net *.split]
05:26
lastmikoi has quit [*.net *.split]
05:26
Dejan has quit [*.net *.split]
05:26
dmalcolm has quit [*.net *.split]
05:26
samth has quit [*.net *.split]
05:26
Corbin has quit [*.net *.split]
05:26
lesshaste has quit [*.net *.split]
05:26
[Arfrever] has quit [*.net *.split]
05:26
xcm_ has quit [*.net *.split]
05:26
atomizer_ has quit [*.net *.split]
05:26
iwkse has quit [*.net *.split]
05:26
mgorny has quit [*.net *.split]
05:26
arigo has quit [*.net *.split]
05:26
energizer has quit [*.net *.split]
05:26
antocuni has quit [*.net *.split]
05:26
benjamin has quit [*.net *.split]
05:27
luckydonald has joined #pypy
05:27
lesshaste has joined #pypy
05:27
antocuni has joined #pypy
05:27
Dejan has joined #pypy
05:27
lastmikoi has joined #pypy
05:27
[Arfrever] has joined #pypy
05:27
Corbin has joined #pypy
05:27
xcm_ has joined #pypy
05:27
benjamin has joined #pypy
05:27
atomizer_ has joined #pypy
05:27
samth has joined #pypy
05:27
dmalcolm has joined #pypy
05:27
mgorny has joined #pypy
05:27
iwkse has joined #pypy
05:27
arigo has joined #pypy
05:27
energizer has joined #pypy
05:30
tazle has quit [Ping timeout: 256 seconds]
05:31
tazle has joined #pypy
05:37
jstoker has quit [*.net *.split]
05:37
LarstiQ has quit [*.net *.split]
05:37
LarstiQ has joined #pypy
05:37
LarstiQ has quit [Changing host]
05:37
LarstiQ has joined #pypy
05:38
jstoker has joined #pypy
05:51
_whitelogger has joined #pypy
05:55
tumbleweed has joined #pypy
05:56
graingert[m] has joined #pypy
07:53
otisolsen70 has joined #pypy
09:30
otisolsen70 has quit [Quit: Leaving]
09:45
jacob22 has quit [Ping timeout: 256 seconds]
09:50
glyph has quit [Quit: End of line.]
09:50
glyph has joined #pypy
11:15
otisolsen70 has joined #pypy
11:22
jacob22 has joined #pypy
12:28
Atque has quit [Quit: ...]
15:12
otisolsen70 has quit [Quit: Leaving]
16:08
x88x88x has joined #pypy
16:19
x88x88x- has joined #pypy
17:01
x88x88x- has quit [Ping timeout: 265 seconds]
17:38
Julian has joined #pypy
19:11
slav0nic has joined #pypy
19:21
Guest96_ has joined #pypy
19:35
marky1991 has joined #pypy
20:15
<
marky1991 >
i'm doing some silly programming problems this year and I was testing one of my solutions with larger and larger Ns to see how it scaled
20:16
<
marky1991 >
one thing I've found is that my latest solution has a performance cliff it falls off if I make my iteration count too large
20:18
<
marky1991 >
but if i just run this as is, it takes more than 90 seconds, probably more. I killed it when I got impatient. I'll give it more time now
20:18
<
marky1991 >
but I'm wondering why it scales soo poorly at that value, the implementation seems like it should be roughly constant time
20:19
<
marky1991 >
i also tested it with cpython and get the same huge slowdown at the same value, so I'm guessing that there's something wrong with my algorithm
20:19
<
marky1991 >
but I was wondering if anyone else saw what I'm doing wrong
20:20
<
marky1991 >
feel free to ignore if it's not interesting, it's obviously not real production code anywhere, so not that important
20:20
<
cfbolz >
marky1991: isn't that because at some point you switch to long integers
20:20
<
cfbolz >
the result grows exponentially
20:21
<
cfbolz >
and at some point the answer doesn't fit into 64 bits any more
20:21
<
marky1991 >
hm, it had grown to large integers before that
20:21
<
marky1991 >
but that makes some sense
20:21
<
cfbolz >
ok, but it gets slower and slower to add them
20:21
<
marky1991 >
maybe it finally went past some specific size that just kills it
20:23
<
cfbolz >
marky1991: maybe plot the time?
20:32
<
marky1991 >
it looks like pypy is much faster than cpython for this value, so I guess that's still good
20:32
<
marky1991 >
still waiting on cpython to finish
20:32
<
marky1991 >
pypy took 75s wall time
20:32
<
marky1991 >
but that's surprising, I would expect them to take roughly te same time if the problem were 8 really-big-bignums
20:33
<
cfbolz >
marky1991: we have different bignum implementations
20:33
<
marky1991 >
oof, 5m for cpython
20:33
<
marky1991 >
thanks for the explanation
20:33
<
marky1991 >
i wonder how decimal would perform, will test
20:35
<
cfbolz >
decimal would only help if printing was the problem (unlikely)
20:35
<
cfbolz >
you could try gmpy
20:35
<
marky1991 >
my gut feeling was that it would be slower
20:36
<
cfbolz >
nope, factor 4 faster for me
20:36
<
cfbolz >
marky1991: you should switch to using a list instead of a dict though, that helps cpython
20:44
<
marky1991 >
rurprising, my implementatoin took longer than 5m if i replaced it with decimal objects
20:46
<
marky1991 >
ha, floats just don't work, returning a nan
20:46
<
marky1991 >
must exceed the limit
20:46
<
marky1991 >
interesting, very different
20:47
Julian has quit [Ping timeout: 265 seconds]
20:47
<
marky1991 >
the index trick is very clever
20:47
<
marky1991 >
much cheaper than my lazy dict-shuffling approach
20:51
Julian has joined #pypy
21:00
marky1991 has quit [Ping timeout: 252 seconds]
21:02
Julian has quit [Quit: leaving]
21:05
marky1991 has joined #pypy
21:30
marky1991 has quit [Ping timeout: 265 seconds]
22:18
Guest96_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
22:21
Atque has joined #pypy