Hi
Exploring Go, 1st project is to port over my favourite playground (prime
number factorisation) from Python to Go.
Aside: I was a big Oberon-2 fan back in the day, so I'm pretty excited
about Go making possible what never was in Oberon.
Back to business: I was disappointed that Go is only a factor of 3 or so
faster than Python. The numbers I'm dealing with are big enough that they
don't fit into a uint64 so I'm using math/big ... turns out that about 70%
of the runtime is spent in the Mod function of math/big, and mostly this is
inside the underlying function divLarge. See profile info below.
The naive explanation would be that Mod in Go math/big is about just as
fast as the % primitive in python (Python has big integers automatically
where needed) but the rest of the program is a factor of 10 or so faster in
Go than Python. I wonder though, especially if there is a better way to
take a Mod. The numbers are not THAT large (less than 100 decimal digits
in most cases) ...
(pprof) top10 -cum
13.01s of 41.11s total (31.65%)
Dropped 86 nodes (cum <= 0.21s)
Showing top 10 nodes out of 74 (cum >= 6.66s)
flat flat% sum% cum cum%
0 0% 0% 40.45s 98.39% runtime.goexit
0 0% 0% 40.20s 97.79% main.main
0 0% 0% 40.20s 97.79% runtime.main
0.28s 0.68% 0.68% 40.19s 97.76%
_/Users/templon/gofact/factoring.Mybrent
0 0% 0.68% 40.19s 97.76%
_/Users/templon/gofact/factoring.Trial_n_X
0.44s 1.07% 1.75% 29.42s 71.56% math/big.(*Int).Mod
0.77s 1.87% 3.62% 28.98s 70.49% math/big.(*Int).QuoRem
0.51s 1.24% 4.86% 28.18s 68.55% math/big.nat.div
10.39s 25.27% 30.14% 27.47s 66.82% math/big.nat.divLarge
0.62s 1.51% 31.65% 6.66s 16.20% runtime.makeslice
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
