Tuesday, May 14



It's never that simple.

I downloaded and installed Julia Studio and started tinkering, and things were going pretty smoothly, and I ran some simple benchmarks, and the best case was nearly as fast as Python.

Wait, what?


Python has a remarkably efficient core of functions, but its code execution is rather sluggish.  That is, Python itself is fast, but code written in Python is slow.  So, for example, splitting sentences into words in Python runs about five times faster than in Julia, according to my little benchmark.  But the more complicated your code gets, the more that should tilt in Julia's favour.

Indeed, Julia's standard library is written mostly in Julia, so the performance you get for built-in functions is the same as for your own code, where with Python there can be orders of magnitude between the two.

Still, /5 performance on early tests when you're seeking *20 is not encouraging.

Next week: Pixy investigates Numba.

Posted by: Pixy Misa at 02:46 AM | No Comments | Add Comment | Trackbacks (Suck)
Post contains 159 words, total size 1 kb.

Comments are disabled. Post is locked.
44kb generated in CPU 0.0203, elapsed 0.1289 seconds.
54 queries taking 0.1171 seconds, 339 records returned.
Powered by Minx 1.1.6c-pink.